Democracy and Economic Openness in an Interconnected System: Complex transformations

DEMOCRACY AND ECONOMIC OPENNESS IN AN INTERCONNECTED SYSTEM

In this book, Quan Li and Rafael Reuveny combine the social scientific approach with a broad, interdisciplinary scope to address some of the most intriguing and important political, economic, and environmental issues of our times. Their book employs formal and statistical methods to study the interactions of economic globalization, democratic governance, income inequality, economic development, military violence, and environmental degradation. In doing so, Li and Reuveny cross multiple disciplinary boundaries, engage various academic debates, bring the insights from compartmentalized bodies of literature into direct dialogue, and uncover policy trade-offs in a growingly interconnected system of polity, economy, and environment. They show that growing interconnectedness in the global system increases the demands on national leaders and their advisors; academicians and policymakers will need to cross disciplinary boundaries if they seek to better understand and address the policy trade-offs of even more complex processes than the ones investigated here. Quan Li is Professor of Political Science and Director of the Program on International Conflict and Cooperation (PICC) at Texas A&M University, which he joined in 2008. Previously, he was a faculty member in political science at the Pennsylvania State University, where he codirected the Multidisciplinary Seminar Series on Globalization in the College of Liberal Arts and served on the inaugural Faculty Governing Council of the School of International Affairs. Professor Li served on the editorial board of the Journal of Politics and is serving on the editorial boards of International Studies Quarterly and International Interactions. Professor Li holds a Ph.D. in political science and international relations. His research interests focus on the causes and consequences of economic globalization (international trade, foreign direct investment, financial openness, and capital account liberalization), democratic governance, political violence (interstate military conflict, civil conflict, transnational terrorism), and macroeconomic policymaking and cooperation. His research has appeared in numerous journals, including the British Journal of Political Science, Comparative Political Studies, International Organization, International Studies Quarterly, the Journal of Conflict Resolution, the Journal of Peace Research, the Journal of Politics, and Political Research Quarterly. Professor Li is the corecipient of the 2003 Best Article on Democratization Award from the American Political Science Association. Rafael Reuveny is Professor of International Political Economy at the School of Public and Environmental Affairs, Indiana University, Bloomington. His research focuses on the causes and effects of economic globalization, democracy, international military conflict, and sustainable development. He is the author and coauthor of numerous articles and book chapters. Professor Reuveny’s work has appeared in journals such as the American Journal of Political Science, the Journal of Politics, International Studies Quarterly, International Organization, the Journal of Conflict Resolution, Ecological Economics, and Environmental and Resource Economics. He is the coauthor or coeditor of five books, the most recent of which is North and South in the World Political Economy (2008). He was also a guest coeditor of a special issue of International Studies Review (2007). Professor Reuveny was program chair of the 2006 meeting of the International Studies Association and the North America program chair of the 2008 meetings of the Global International Studies Conference. Reuveny has won two teaching awards at Indiana University and is the 2007 corecipient of the Award of Excellence in World Society Research, First Place, given by the World Society Foundation, Zurich, Switzerland. Professor Reuveny is also the corecipient of the 2003 Best Article on Democratization Award from the American Political Science Association. He holds a double-major Ph.D. in business economics and political science.

x

Acknowledgments

Chapter 3 is a thoroughly revised and extended version of the following article: Reuveny, Rafael, and Quan Li. 2003. “Economic Openness, Democracy and Income Inequality: An Empirical Analysis,” Comparative Political c 2003 Sage Publishing. Studies 36(5):575–601. Copyright
Chapter 5 is a thoroughly revised and extended version of the following article: Reuveny, Rafael, and Quan Li. 2003. “The Joint Democracy–Dyadic Conflict Nexus: A Simultaneous Equations Model,” International Studies c 2003 Blackwell Publishing. Quarterly 47(3):325–347. Copyright
Chapter 7 is a thoroughly revised and extended version of the following article: Li, Quan, and Rafael Reuveny. 2006. “Democracy and Environmental c Degradation,” International Studies Quarterly 50(4):935–956. Copyright
2006 Blackwell Publishing. Chapter 8 is a thoroughly revised and extended version of the following article: Li, Quan, and Rafael Reuveny. 2007. “The Effects of Liberalism on the Terrestrial Environment,” Conflict Management and Peace Science c 2007 Taylor & Francis. 24(3):219–238. Copyright
This book has grown out of our decade-long joint research and collaboration, friendship, and many debates since we first met in 1999. The quality of our joint scholarship has benefited from both our agreements and our disagreements, all of which we were ultimately able to resolve with a smile. Without our mutual willingness to listen and compromise, and our continuous mutual support and trust, this book would not have come into being. Finally, our families have been important to the completion of this book. We owe many thanks to our respective parents in China and Israel, Li Maoji and Kuang Juying, and Mordechai and Ora Reuveny, for their unending support. Our spouses and children, Liu, Ellen, and Andrew, and Ronit, Adi, and Noam, have given so much support, encouragement, and meaning to our work that we feel we must dedicate this book to them.

Democracy and Economic Openness in an Interconnected System Complex Transformations QUAN LI Texas A&M University

RAFAEL REUVENY Indiana University

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Dubai, Tokyo Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521491433 © Quan Li and Rafael Reuveny 2009 This publication is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published in print format 2009 ISBN-13

978-0-511-65157-1

eBook (NetLibrary)

ISBN-13

978-0-521-49143-3

Hardback

ISBN-13

978-0-521-72890-4

Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of urls for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

List of Figures and Tables

page vii

Acknowledgments 1

ix

Introduction

1

PART I: THE DEMOCRACY–ECONOMY NEXUS

2

Democracy and Economic Openness

23

3

Democracy, Economic Openness, and Income Inequality

62

4

Democracy and Development

89

PART II: BRINGING IN CONFLICT

5

Democracy and Conflict

125

6

Economic Openness and Conflict

158

PART III: BRINGING IN THE ENVIRONMENT

7

Democracy and the Environment

205

8

Economic Openness and the Environment

239

9

Conflict and the Environment

266

Conclusion

292

10

References

309

Author Index

337

Subject Index

344

v

List of Figures and Tables

FIGURES 1.1. Graphical layout of the book 6.1. Disaggregated bilateral trade equilibrium 10.1. Graphical layout of key findings

page 5 167 294

TABLES 2.1. 2.2. 2.3. 2.4. 2.A1. 2.A2. 2.A3. 3.1. 3.A1. 3.A2. 3.A3. 4.1. 4.2. 4.A1. 4.A2.

Globalization promotes democracy Globalization obstructs democracy Globalization does not necessarily affect democracy Effects of economic globalization on democracy Pooled time-series cross-sectional models of democracy by decade OLS estimates with additional control variables or Freedom House data Parameter estimates from alternative estimators Income inequality, democracy, and economic openness Interactive effect of democracy and FDI on income inequality Income inequality, democracy, and economic openness (all countries) Effects of democracy and economic openness in DCs and LDCs Variables and expected effects in the simultaneous equations Democracy and development, 2SLS Democracy and development, 3SLS Democracy and development, 2SLS-Kiviet

vii

28 31 34 39 54 55 57 74 84 86 88 100 103 120 121

viii

List of Figures and Tables

5.1. Expectation of direction of effects in the simultaneous equations 5.2. Interactions among joint democracy, regime dissimilarity, and dyadic conflict 5.3. Probabilities and relative risks of MID involvement 5.A1. Interactions among joint democracy, regime dissimilarity, and dyadic conflict, controlling for affinity 6.1. Trade categories and conflict expectations 6.2. Effects of bilateral import and export in five sectors on MID initiation 6.A1. Equality tests on significant positive and negative effects of sectoral trade flows 6.A2. In-sample prediction of MID initiation 6.A3. Effects of bilateral sectoral flows on display of force and use of force 7.1. Effect of level of democracy on environmental degradation 7.2. Effect of democracy on environmental degradation 7.3. Effect of autocracy on environmental degradation 7.4. Effect of political regime type on environmental composites 8.1A–C. Summary of causal mechanisms 8.2. Effects of democracy and trade openness on deforestation and land degradation 8.A1. Effects of democracy and trade openness on deforestation 8.A2. Effects of democracy and trade openness on land degradation 9.1. War and CO2 emissions per capita 9.2. War and deforestation 9.A1. War, cubic GDP per capita, CO2 emissions per capita, and deforestation 9.A2. War and CO2 emissions per capita, omitting the lagged dependent variable

139 142 143 156 171 176 200 200 201 217 220 221 223 249 254 262 264 275 277 289 290

Acknowledgments

This book would not have been possible without the support of many friends, colleagues, and students at the Pennsylvania State University, Indiana University, and Texas A&M University. Our many friends and colleagues in political science and economics also offered numerous useful comments, criticisms, and suggestions regarding the various parts of our project, which were presented in earlier forms at many professional conferences, workshops, and seminars. We thank them all even though it is not possible to enumerate all those who played a role in some stage of our research project. We give special thanks to Jim Eisenstein, Frank Baumgartner, Evan Ringquist, and the reviewers for Cambridge University Press for their comments and suggestions directed at various parts of our book manuscript. Daehee Bak, Andreea Mihalache, Sam Sniderman, Ashley Allen Peterson, Leslie McDonald, Melanie Arnold, and Matt Warhol provided valuable research and editorial assistance, for which we are grateful. We also warmly thank Scott Parris at Cambridge University Press, who is the best editor an author could hope for. We deeply value his guidance throughout the different phases of this project. Scott’s assistant, Adam Levine, provided excellent editorial assistance. Himanshu Abrol effectively oversaw the production of our book, and Heather Phillips did a thorough job copyediting the manuscript. Five chapters of our book draw upon our previously published journal articles. We want to thank Blackwell Publishing, Cambridge University Press, Sage Publishing, and Taylor & Francis for their permission to use materials from those journal articles for this project. Chapter 2 is a thoroughly revised and extended version of the following article: Li, Quan, and Rafael Reuveny. 2003. “Economic Globalization and Democracy: An Empirical Analysis,” British Journal of Political Science c 2003 Cambridge University Press. 33(1):29–54. Copyright
ix

ONE

Introduction

Two important and complex transformations have characterized the global political–economic system since the end of World War II in 1945. One transformation is the rise and spread of democracy over time and across countries. By the term democracy, we mean a national system of political governance based on free elections and broad political representation. We conceptualize this concept as a continuum of political regime types, ranging from full autocracy at one end to full democracy at the other end. The second transformation is the expansion and deepening of globalization. In recent decades, globalization has been a popular term but is more complex to define than democracy. Scholars typically employ this term to describe recent global transformations toward growing cross-national interconnectedness. Although many scholars tend to focus on the interconnectedness features of particular interest to their own disciplines, some take a broader perspective. For example, Held et al. (2009) cast a wide net, covering essentially all the transformations that have increased international interconnectedness, including political-legal (e.g., growth of international treaties and institutions), military (e.g., disputes, growth of armies, and weapon proliferation), communication and informational (e.g., Internet, telephone, and media), economic flows (e.g., trade and investments), knowledge flows (technology transfers and education), taste convergence (e.g., consumption preferences), social contacts (e.g., migration and tourism), and environmental developments (e.g., pollution and changes in the biosphere), as well as the national and international income inequalities resulting from these processes. Whereas the scope and meaning of globalization are controversial, most scholars agree that, at a minimum, economic globalization implies that countries are becoming more integrated into the world economy, with increasing information flows among them. Greater economic integration, in turn, implies more trade, investment, and financial 1

2

Democracy and Economic Openness in an Interconnected System

flows, or rising economic openness. We also adopt this conceptualization in our book. Questions that pertain to democracy and globalization have attracted the attention of scholars, policymakers, and the public at large. These questions stand at the center of our book. We believe it is best to describe our analytical focus by situating it within a big picture of world affairs. THE BIG PICTURE AND OUR ANALYTICAL FOCUS

Let us assume we are charged by an intergovernmental organization, comprising all the countries in the world, with the mission to address two questions. First, assuming other things do not change (e.g., public policies, state of technology, and human preferences), what will be the national trajectories of key forces such as income inequality, economic development, democratic governance, military conflict, and environmental degradation in the next decade? Second, assuming other things do not change, how will these economic, political, and environmental trajectories change if we alter certain policies to promote democracy and encourage economic openness? Fundamentally these questions point to the implications of democracy and economic openness for the world in which we live. Answering these questions helps to predict the evolution of the global system. Although no one can absolutely predict the future, we might be able to say something informative about it if we can explain current and past developments. Indeed, almost all the predictions in the social sciences use the present and recent past as a baseline. The logic driving this approach is straightforward. Because we try to say something about the near future, not the very distant future, we and our immediate offspring will probably still be around. It follows logically that the behavior of the socioeconomic– political–environmental system in the near future is connected to the state of affairs in the present and recent past. This book is not about forecasting the future but rather provides a sound analytical basis for such an exercise. The ultimate goal is to say something new and informative about the direction and magnitude of global transformations, defined in the spirit of Held et al. (2009), that are expected for the near future. To that effect, we must understand theoretically and empirically the interconnections among the relevant forces in a big picture of the world, which is what this book is about. Going deeper into our purpose in this book, we note that both economic openness and democracy have expanded dramatically over the past six decades. The number of democracies has increased considerably, spreading

Introduction

3

from Western Europe and North America to almost all the regions in the world. Meanwhile, the world economy has witnessed an unprecedented expansion in the volume and patterns of international commerce, crossborder investments, and international production.1 It is reasonable to assume, which is the basic argument we make and evaluate in this book, that democracy and economic openness, as two important and rapidly changing forces, must have left their marks on many other forces in the international system. We are particularly interested in the consequences of democracy and economic openness for three important aspects of life on Earth: the economy, the polity, and the environment. One can study the economic, political, and environmental aspects of life on Earth at different levels of analysis, including individual, subnational, national, regional, continental, and planetary. We focus on two levels of analysis: the monadic state and the dyadic interstate. Our choice is not arbitrary, for we submit that the nation-state remains the most important and influential actor in the international system despite the rising challenges from various nonstate actors. What should we study in terms of the economy, the polity, and the environment? The national economy covers many dimensions, such as the level of development, income distribution, money, investment, saving, taxation, innovation, and so on, whereas the national polity may concern forces, such as the government, the party system, civil society, social cleavages, the law and property rights institutions, and so forth. We could also examine the international aspects of the national polity that involve how states interact with each other. In the environmental domain, there are many aspects that concern nation-states, beginning with how they reach agreements over the environment, the way they treat their own environments, their abilities to provide resources, their levels of food sufficiency (as all types of food ultimately come from environmental sources), and so on. We recognize that, naturally, we cannot study everything at once. Such a study is not only overly complex but also largely intractable. We also face considerable difficulty communicating our findings to the scholarly and policymaking communities, toward which we aim our book. Therefore, we concentrate on the consequences of democracy and economic openness for some key aspects of the national economy, polity, and environment. 1

Information and communication technology (ICT) and immigration are also relevant. But in this book we emphasize on economic globalization (i.e., flows of goods and capital). One should note that the current globalization wave, which stands at the center of our book, involves much smaller immigration flows than globalization of the late nineteenth and early twentieth centuries, whereas ICT is highly correlated with GDP per capita, which is part of our analysis.

4

Democracy and Economic Openness in an Interconnected System

For the national economy, we focus on the national level of economic development and the national distribution of income. These two forces are important for any country. They inform us about the average standard of living in a country and the distribution of this standard across the population. The importance of economic development for human welfare goes without saying, and the effects exceed economics. The lack of economic development tends to beget grievances, increasing the likelihood of political instability. Spillovers into the environment also occur as poor countries, preoccupied with the immediate needs of subsistence, spend less on environmental regulation and cleanups. A skewed income distribution will most likely make things even worse. As the poor compare their situation to that of the rich, their grievances grow stronger, causing the political regime to break down in some cases. In sum, we are interested in furthering our understanding of how democracy and economic openness influence income distribution and economic development. For the polity, we are interested in the determinants of national democratic governance and interstate military conflict. It is no exaggeration to claim that these two phenomena are among the most salient issues in international relations and in political science in general. From a normative perspective, democracy and world peace are desirable, for they appeal to the longings of most human beings. In reality, however, democracy and world peace are often difficult to achieve and sustain. The scarcity of resources and the human desire to dominate others have also been with us for millennia, leading to conflicts of interest among individuals and between nation-states. Democratic governance is an important human-engineered institutional solution for resolving conflicts among people and groups living within national territorial boundaries. In the current international system, which operates on principles of anarchy and state sovereignty, interstate military conflict is often employed as an instrument for resolving disagreements among nation states. We seek to find out how the forces of globalization affect the prospect of democratic governance and how democracy and economic openness influence interstate military conflict. For many years, the environment was on the margin of scientific inquiry and policy analysis. As long as the economy was small in size relative to the environment in which it resided, and as long as the activities of the polity did not exert a large effect on the environment, one could arguably accept this neglect both intellectually and in public policy. Yet conditions have changed in recent decades. Human activities have increasingly caused environmental degradations of various types, impacting human conditions around the world and the global climate system. Environmental concerns

Introduction

5

Figure 1.1. Graphical layout of the book.

are no longer marginal issues. In light of these changing conditions, we believe it is important to expand research on the links between our forces of interest and the environment and to address such critical questions as whether spreading democracy, rising trade, and political violence are good or bad for the environment. CONCEPTUAL FRAMEWORK, ANALYTICAL APPROACH, AND CONTRIBUTIONS

The previous section described the primary elements of the big picture we study in this book. This section presents our analytical approach, lays out the conceptual framework within which we operate, and highlights the new things we bring to the table. Figure 1.1 displays a graphical blueprint of our inquiry. Although the interactions we study are in fact more complex and intricate than the figure illustrates, we believe this blueprint is a useful way to guide and visualize the overall structure of our inquiry. The forces of democracy and economic openness stand at the center of the figure, indicating the analytical foci of our conceptual framework. One set of arrows flows from the center to the national economy in the upper part of Figure 1.1, corresponding to the focus on the economy. The national

6

Democracy and Economic Openness in an Interconnected System

economy is represented by the national level of economic development and the national level of income inequality. The two-sided arrow flowing between democracy and the national economy represents their effects on each other. We will study the reciprocal effects between democracy and the level of economic development as well as the effects between democracy and income inequality. The arrow from economic openness to the national economy indicates that we will investigate how openness influences income inequality and development. The arrow from economic openness to democracy and the set of arrows from the center to military conflict in the bottom right of Figure 1.1 indicate the part of our analysis focusing on the polity. With respect to the polity, we conduct two analyses. First, we examine how economic openness affects democracy. Second, we study the effects of economic openness and democracy on military conflict. The set of arrows that flows from the center to the environment in the bottom left of Figure 1.1 represents our focus on the environment. The arrows leading from economic openness and democracy to the environment indicate that we study how democracy and economic openness affect the environment. Finally, we consider the effect of military conflict on the environment. We seek to shed new light on the causal interactions indicated in Figure 1.1, but the channels depicted in Figure 1.1 do not exhaust all possibilities. The arrows shown in the figure represent the channels we study in this book. They involve some of the most important and intriguing questions in the fields of international political economy, international relations, comparative politics, environmental economics, and global environmental politics. Thus, the reader should construe the conceptual layout in Figure 1.1 as a road map for the structure of our book, not as a model per se. Each of the primary causes and effects in Figure 1.1 is addressed in a separate chapter in the book. We adopt a quantitative and sometimes formal approach to study the complex interconnections in Figure 1.1 for several reasons. The various interconnections are modeled separately in the respective chapters, assuming other things do not change, but the findings are integrated in the concluding chapter. In the context of Figure 1.1, each primary cause and effect translates into a statistical model in the respective chapter. The effect (i.e., the phenomenon we seek to explain) is called the dependent variable or the left-hand-side variable in a statistical model. The causal determinants of the effect are referred to as independent or right-hand-side variables in the statistical model. We typically refer to the primary causal factor in a

Introduction

7

chapter as the key independent variable, and we call the other relevant, but secondary, independent variables in the model control variables. Of course, a key independent variable in one chapter often becomes a control variable in another chapter, and vice versa. In the quantitative approach, we conduct statistical tests of the hypotheses to ascertain whether the effect of each independent variable on the dependent variable is statistically different from zero (where zero means no effect), and we compute the size of the effect for each relationship that is significantly different from zero. We employ large samples of real-world data and, where appropriate, we also evaluate the robustness of our findings in subsamples and with different measurements and statistical techniques. We seek robust, valid statistical inferences that answer the important questions we and others raise. The advantages of the quantitative approach are fully exploited. First, systematic quantitative analysis allows us to study some forces in detail while controlling for many other relevant factors to prevent spurious findings. Second, the quantitative analysis identifies the average cause-and-effect statistical association and allows the analyst to evaluate whether a theoretically hypothesized relationship is statistically different from zero in the empirical data. This approach is important if we seek to build theories that are empirically valid and generalizable and if we intend to adopt policies that may actually work. Third, the use of large samples of real-world data often prevents subjective case-selection bias and uncovers general patterns, which is critical if we want to understand how the world generally works. Fourth, systematic statistical analysis also enables us to evaluate the robustness of our findings when we vary measurements, samples, and statistical techniques. Finally, the quantitative approach offers a valuable opportunity for us to gauge the substantive size of an effect that is statistically different from zero, which provides useful information for policymakers who have to contemplate the effectiveness of public policies and the various trade-offs associated with competing policy objectives. Therefore, we believe that the statistical approach contributes with rigor and clarity new insights to our analysis of significant issues. Of course, the statistical modeling approach is not without limitations common to all studies of this type. First, we are aware that our approach may be more technical and numerical than that in many, though not all, of the books on these topics, possibly posing challenges to some readers. To make the findings more accessible to the general readership, we write out explicitly the estimated statistical models accompanied by detailed explanations, present one main set of statistical results, and describe the formal

8

Democracy and Economic Openness in an Interconnected System

and statistical models and their results in a way that does not require prior training in statistics or mathematics for comprehension. We relegate all the technical details to the chapter appendices. Readers who are interested in the substance of our findings only need to read the main text; those who are interested in the technical details can obtain the relevant information from the appendices. We believe that such organization of the presentation makes the book highly readable for a wide audience. Second, statistical findings imply probabilistic statements regarding relationships of interest, and they are not intended to depict or forecast each particular case. Where appropriate, we highlight some real-world cases for illustration. But in essence, our analysis is statistical and reflects average patterns in the real world. Finally, statistical models typically include a limited number of variables, relegating other possible influences to the error term. In other words, we should and do make further assumptions concerning which variables to include in each model. At the same time, these assumptions are not arbitrary; rather they are based on theoretical considerations and, to a large extent, previous studies. Our book brings at least three new things to the table, which other studies of our topics have typically not attempted to achieve thus far. First, we model a relatively large number of forces and topics pertaining to globalization and democracy. Previous scholarly works on these topics tend to be compartmentalized, focusing on one or two aspects of the global system. We acknowledge and model explicitly the economic, political, and environmental dimensions of the global system and their interconnections. Taken together, the findings from the different chapters form a relatively more holistic and integrative view of the inner workings of our world than previous research. Second, because we study seemingly diverse topics in a conceptually coherent framework and treat these topics as different components of a big picture, we are able to uncover interconnections that have thus far received little attention. Finally, because all the analyses employ statistical modeling, we have one coherent methodological approach throughout the book, which controls for various confounding forces, rigorously tests the relationships of interest, estimates the magnitude of each key relationship, and enables us to uncover general patterns. STRUCTURE OF THE BOOK

Our book consists of three parts, and democracy and economic openness run through all of them. The first part examines the relationships among

Introduction

9

democracy, openness, and the national economy. The second part investigates the relationships between democracy and openness, on the one hand, and interstate military conflict, on the other hand. The third part of the book studies the relationships among openness, democracy, and conflict, on the one hand, and various indicators of environmental degradation, on the other. Given the complexity of our topics, each chapter typically employs, as an anchor, one of the two core concepts – democracy or openness – relegating the other to the secondary role. As a preview, we discuss the gist of each chapter, focusing on the organization and flow of the argument and summarizing the key findings. The book includes eight main analyses or chapters, each pertaining to a part of the big picture. The organizational structure of each substantive chapter (2–9) is similar. We begin each chapter by stating the research question(s) and the theoretical argument(s) concerning the relationship(s) studied. On the basis of the theoretical argument(s), we turn to the empirical research that specifies the statistical model, clarifies important research design issues, and describes the primary findings in light of the arguments posited at the beginning of the chapter. The technical details pertaining to each chapter are in the chapter appendix, which follows the general flow of the discussion in the body of the chapter. The first and last chapters of the book integrate the eight analyses into a larger picture, but do so with different levels of detail. The current chapter introduces the book, focusing on the conceptual framework, analytical focus and approach, and structure. The concluding chapter of the book summarizes the key findings by incorporating them into a revised and much more elaborate version of Figure 1.1, suggests avenues for future research, emphasizes key policy implications, and evaluates the trade-offs and tensions that result from different policy objectives such as economic development versus environmental quality, economic development versus equitable distribution of income, democracy versus environmental quality, economic liberalization versus political liberalization, and democracy versus national security.

Part I: The Democracy–Economy Nexus Part I of the book focuses on the democracy–economy nexus. We begin our analyses by looking at the two key variables in our book, democracy and economic openness. In Chapter 2, we study the effects of economic openness on democracy, controlling for the important influences of economic development and income distribution. We ask whether more integration

10

Democracy and Economic Openness in an Interconnected System

into the world economy and the wider spread of democratic ideas and norms across countries lead to a rise or decline of democratic governance. This question has captured the attention of policymakers and academic scholars alike, who have provided various answers and policy recommendations. Anecdotal evidence is typically invoked in debates, but systematic evidence is scarce. To answer our question, we examine the effects of trade flows, foreign direct investment flows, financial investment flows, and the spread of democratic ideas on the level of democracy in a country. Our statistical analysis suggests that trade openness and portfolio investment inflows reduce democracy. The effect of trade openness is constant over time, whereas the negative effect of financial investment grows stronger. Foreign direct investment inflows promote democracy, but the effect weakens over time. The spread of democratic ideas is persistently conducive to democracy over time. In Chapter 3, we examine the effects of democracy and economic openness on income inequality within countries. Although the issue of income inequality was central in classical economics – the body of thought that emanated from the writings of the liberal philosophers of the nineteenth century – it has received relatively little attention in neoclassical economics, the body of thought and knowledge in modern mainstream economics. We believe that the issue of income inequality deserves more attention, because commercial liberalism, or free market–oriented capitalism, and republican liberalism, or democracy as a form of political governance, are not easy companions. Whereas democracy is based on the principles of “one person, one vote” and representative government, capitalism is based on the principles of laissez-faire and private enterprise. Furthermore, democracy is often associated with redistributive policies (e.g., progressive taxation), but capitalism typically rewards heterogeneous individuals with different levels of income. Hence, democracy may suppress income inequality, but capitalism promotes income inequality. A skewed income distribution under capitalism could lead to an asymmetric distribution of political power, which could undermine democracy and, therefore, its effect on income inequality within a country. Turning to the empirical analysis of Chapter 3, we argue that the effects of democracy and economic openness on income inequality should be analyzed together rather than in separate models that include one force and exclude the other. Economic openness is measured based on national trade flows, foreign direct investment inflows, and financial capital inflows; income inequality is measured by the Gini coefficient of each country in the sample. We find that democracy and trade openness reduce income

Introduction

11

inequality, foreign direct investments increase income inequality, and financial capital inflows do not affect income inequality. We also find, however, that democratic governance mitigates the inequality-increasing effect of foreign direct investments in the advanced industrial democracies. In Chapter 4, we bring development into the story, focusing on the relationship between democracy and economic development. We open the democracy–development box, this time relegating economic openness to a control variable, and study the interactions between democracy and economic development. The democracy–development relationship not only is central to much academic research but also has significant policy implications. A recent essay in Foreign Affairs argues that “democracies consistently outperform autocracies” in economic development (Siegle et al., 2004: 57). U.S. President George W. Bush, the authors write, supports having democratic governance “as qualifying criteria for countries to receive assistance . . . The U.S. Agency for International Development (USAID), still the major development actor in the U.S. government, should also offer preferential treatment to democracies and target its assistance to help countries undertaking democratic reforms” (2004: 67). Normatively speaking, democratic governance, which favors equal participation, equal political rights, and equal civil liberties, is of course worthy of promotion. The argument in the Foreign Affairs essay, however, justifies the promotion of democracy as a means to increase economic development. Does democracy really encourage economic development? The Foreign Affairs essay presents anecdotal evidence to support its argument, but it does not take into account the possibility that the causality between democracy and development could flow from development to democracy. We argue that sorting out the causal direction between democracy and development is very important. If the goal is to promote democracy but the causation goes from development to democracy, we should not condition the distribution of aid to poor countries on instituting democratic reforms. In this case, we should condition aid on forces such as the quality of projects, market reforms, or investment in education. The issue of causation between democracy and development stands at the center of Chapter 4. We specify and estimate a statistical simultaneousequations model, which allows for reciprocal effects between the level of development and the level of democracy. This modeling approach rigorously evaluates the nature of the relationship, that is, whether democracy affects development only, development affects democracy only, they affect each other, or they are not related statistically. The empirical results indicate that a rise in the level of democracy reduces the level of economic development,

12

Democracy and Economic Openness in an Interconnected System

whereas a rise in the level of economic development promotes democracy. It is not difficult to imagine the important policy implications of these findings.

Part II: Bringing in Conflict In Part II, we introduce into our book a new force – interstate military conflict. Interstate political relations behave like a continuum of types of interstate interactions, spanning from full cooperation to full-scale war. In measuring the concept of conflict, some scholars consider it a continuum, whereas others focus on the discrete events that involve violence only. In this book, we employ the latter approach, focusing on militarized interstate disputes, for these types of interstate interactions have the most damaging consequences to human society. In Chapter 5, we focus on how democracy affects conflict, controlling for economic openness. This concerns the famous democratic peace proposition in the field of international relations. Democracies are highly unlikely to go to war with each other, which suggests that the spread of democracy in the international system implies the spread of peace. This idea has been around for many years and can be traced to the writings of German philosopher Immanuel Kant. The modern, empirical-statistical incarnation of this argument takes the form of a simple and powerful claim: democracies rarely, if at all, go to war with one another. This is not to say democracies may not exhibit conflict with one another at times, or that they may never fight with one another. It only means that they are less likely to fight each other than to fight autocracies, and they are less likely to do so than two autocracies. The empirical literature concerning the democratic peace proposition has been arguably the largest in the field of international relations over the past two decades. The bulk of this literature shares a similar design, employing the dyadic level of analysis and estimating single-equation models. The dependent variable is typically dichotomous, measuring the presence or absence of a militarized interstate dispute (MID). The models typically include a measure of joint (dyadic) democracy as the key independent variable and control for various confounding forces. The vast majority of studies report that the probability of a MID between two countries declines as joint democracy increases. A few studies, however, reject this argument and do not find supporting evidence. At the same time, another separate, growing body of literature in international relations argues that conflict affects democracy. Although scholars debate the direction of the effect of conflict on democracy, they seem to

Introduction

13

agree that some type of effect exists. Conceptually, these studies form a subset of the large monadic country-level literature within comparative politics that studies the determinants of national democracy. The argument that conflict affects democracy is intriguing because it implies the possibility that results of the effect of democracy on conflict may in fact have merely captured the effect of conflict on democracy. At any rate, the democratic peace literature and studies that address the effect of conflict on democracy have basically ignored the insights offered by one another. In particular, the effect of joint democracy on conflict has been treated as unidirectional by international relations scholars even though it has been demonstrated elsewhere that conflict affects democracy. By taking the two bodies of literature into account, one can reasonably argue that conflict and democracy affect each other simultaneously. Chapter 5 evaluates this possibility empirically. We develop an innovative simultaneous-equations model of democracy and dyadic conflict for a large number of countries over many years. We find that dyadic military dispute reduces the level of joint democracy of two countries, and that the level of joint democracy reduces the probability of MIDs. Our findings vindicate the main arguments in two separate bodies of literature. But we offer important caveats concerning the size of the effect of democracy on conflict, which is somewhat smaller than in previous research. In Chapter 6, we study how trade, as an important aspect of economic openness, affects conflict. This involves another salient controversy – the effect of trade on peace – among academics and practitioners, which has important theoretical and policy implications. We contribute to this debate with new theoretical and empirical insights. The intellectual history of competing positions on the relationship between trade and conflict is long. The idea that international trade promotes peace traces back to the great works of the eighteenth-century philosopher Immanuel Kant, French social commentator and political thinker Baron de Montesquieu, English philosopher and economist Adam Smith, and early-nineteenth-century English journalist and member of the Parliament Norman Angell. The antithesis that trade begets political disagreement and even military conflict also has a long history, appearing in the writings of such important policymakers as Vladimir Ilich Lenin, political scientist Kenneth Waltz, and political economist Albert Hirschman. Among contemporary social scientific studies, the pacifying effect of trade has two explanations: the liberal argument and the bargaining argument, whereas the conflict-generating effect of trade is often explained from a neo-Marxist view or a neomercantilist derivative of realism. Scholars have

14

Democracy and Economic Openness in an Interconnected System

proved formally the logical consistency of the liberal and bargaining arguments. However, both formal models consider aggregated or total trade, ignoring variations of trade across economic sectors and flow directions (export and import). Almost all recent empirical studies have focused on the dyad as the unit of analysis and have employed data on total bilateral trade flows, computed as the sum of export and import flows in a dyad. Only a few studies examine the effect of bilateral trade in various sectors on international political conflict, but they fail to provide a microfoundation formally, they do not focus on military conflict, they have small samples, and they fail to distinguish between imports and exports. In this chapter, we challenge the prevalent approach and offer a theoretical formal model to explain how export and import flows in specific economic sectors influence the decision to initiate military conflict, a question almost all previous theories have ignored. Our model indicates that one country’s sectoral imports and exports influence the conflict it initiates toward another country through the expected effects the conflict will have on the prices of these trade flows. These effects reflect the sensitivities of sectoral import demand and sectoral export supply to military violence and, fundamentally, whether a country expects to benefit from initiating conflict in terms of less cost on imports or more revenue from exports. We employ these principles to generate hypotheses regarding the effects of export and import in specific goods on the likelihood of military conflict. We test the hypotheses using a data set covering many countries for almost three decades. The findings support our expectations on the effects of specific imports and exports broken down along sectors or goods. In particular, we are able to identify traded goods whose export and import flows increase the likelihood of conflict, goods whose trade flows reduce the likelihood of conflict, and goods that have no effect on conflict. Our research in this chapter leads to new insights on a long-standing debate in international relations, compelling scholars to rethink the logic of how trade affects conflict. The theory encompasses the liberal argument as a special case and offers an alternative explanation to the bargaining, neo-Marxist, neomercantilist realist, and classical realist arguments. It also has various important policy implications.

Part III: Bringing in the Environment Part III focuses on how the polity and the economy influence the physical and biological environments of countries or their biospheres. Specifically, we study the effects of democracy, economic openness, and military conflict on

Introduction

15

human activities that degrade the environment. The impact of these forces on the state of the environment has not received wide attention from political scientists, economists, or policymakers. We believe that this relative neglect will likely change as the global environment continues to degrade and environmental concerns and issues acquire salience in the public discourse. Chapter 7 focuses on the effect of democratic governance in a country on various aspects of its environment such as land, air, water, and vegetation. In recent years, the democracy–environment relationship has received some attention from theorists and is controversial. We begin the chapter by discussing various theoretical arguments on the effects of democracy on the environment. Some theorists claim that democracy reduces environmental degradation, whereas others argue that democracy has no effect on environmental degradation at best and may harm the environment at worst. Despite the growing importance of the issue, existing empirical evidence is relatively scant and mixed. Extant results also do not pertain to the same type of environmental variables. Some scholars examine government commitment to environmental quality in terms of signing international agreements to protect the environment, others investigate resource scarcity and access to environmental amenities such as safe water or sanitation, and a third group explores human activities harmful to the environment. Although the focal points of these analyses are undoubtedly important, we believe it is essential to study human actions that directly damage the environment because the best way to protect the environment is to minimize the damage in the first place. The empirical analysis examines five salient, specific types of humaninduced degradation: carbon dioxide (CO2) emissions, nitrogen oxide (NOx) emissions, land degradation, rate of deforestation, and organic pollution in water. We also look at an aggregate indicator of environmental degradation that may be interpreted as a measure of sustainable development. The results demonstrate that a rise in democracy reduces CO2 emissions, NOx emissions, land degradation, water pollution, and the aggregate measure of environmental degradation, but it raises the rate of deforestation. The size of the effect varies from small to substantial, depending on the particular environmental attribute at issue. The findings suggest that one should not treat democracy as a magic solution to environmental problems. Chapter 8 examines how international trade and its interaction with democratic governance influence the terrestrial environment. As international trade flows have become more and more important in the global economy, their effects on the environment have attracted more and more

16

Democracy and Economic Openness in an Interconnected System

attention. In principle, international trade may directly increase or decrease environmental degradation by changing the patterns of consumption, production, and investment; affecting the production technology; and influencing government regulatory policies. However, scholars disagree on the nature of the overall effect of trade on the environment. In one view, trade reduces environmental degradation and promotes environmental quality. In another view, trade can harm the environment, increasing environmental degradation. Most of these studies do not consider the effects of both trade and democracy, and certainly not their interactive effect on the environment. The empirical analysis in Chapter 8 studies the effects of trade on land degradation and the rate of deforestation within countries. These attributes are important because land quality is crucial for food production and deforestation is important to the water cycle, land erosion, and climate change. We include both democracy and trade in the model, examine the effect of trade conditional on democracy by including their interactive term, account for possible differences across continents, and distinguish between developed countries (DCs) and less developed countries (LDCs). We find that a rise in trade openness reduces deforestation in autocracy but raises it in democracy. These effects are similar for LDCs and DCs. A rise in trade openness reduces land degradation, but the effect is not robust and does not depend on regime type. A rise in democracy increases deforestation and reduces land degradation, but these effects are weaker in LDCs than in DCs. The effect of democracy on deforestation is stronger when trade openness increases, and the effect of democracy on land degradation does not depend on trade openness. Chapter 9 studies the effect of warfare on the environment. Many scholars in international relations have examined the effect of environmental factors on warfare. The expectation is that as climate change progresses and environmental degradation intensifies in the world, countries vie for depleted or degraded resources or quarrel about who should carry the costs of environmental damages, resulting in more warfare. This possibility has received attention, but the effect of warfare on the environment has largely been ignored. In empirical studies, the effect of war on the environment has virtually not been studied statistically. To understand how wars affect the environment, we separate them into those fought at home and those fought abroad, because they are likely to have different effects on the environment. We find that the effect of war depends on the location of the conflict and the environmental indicator examined. Wars fought at home or abroad reduce carbon dioxide emissions at home; wars fought at home speed up the deforestation in a country, but

Introduction

17

wars fought abroad slow it down. War could be a boon for the environment sometimes, though we obviously do not think that it is a good idea to use this knowledge to promote war. Chapter 10 concludes our book by performing three tasks. First, we summarize the analyses performed in this book and take stock of our key findings. We flesh out the conceptual framework of Figure 1.1 with our findings, explicitly identifying the sign of each cause–effect arrow we have analyzed and found to be statistically significant. Second, we offer some caveats regarding our analyses and propose some general directions in which one could usefully extend our research. As we noted earlier, one book-length manuscript cannot address all issues. Despite the richness of our models, we believe room always exists for additional research on the complex topics we study in this book. Finally, we highlight the key policy implications and trade-offs that have been revealed by our analyses. They pertain to economic, political, and environmental policies at both national and international levels. We believe that the issues we study and the results we obtain in this book should be of great interest to both academics and practitioners. SUMMARY AND OUTLOOK

In this chapter, we have presented our book in a nutshell by highlighting the key questions, issues, motivations, arguments, and findings. What we offer is a fresh look – either theoretically or empirically, or theoretically and empirically – at some of the most intriguing and important questions in contemporary social sciences. Our analysis in this book is relatively complex. We seek theoretical and empirical insights by reaching into a considerable number of issue areas, crossing multiple social science disciplines, linking different bodies of literature, engaging various academic debates, and applying formal and statistical social scientific methods. Our coherent conceptual, methodological, and organizational approach enables us to demonstrate the interconnectedness of different issues, debates, and disciplines among domestic and international forces in the economy, polity, and environment. We believe that our book is unique in that it combines the social scientific approach with a broad, interdisciplinary scope. In terms of the range of topics covered, the one book that comes closest to ours is Global Transformations by David Held and his coauthors (1999). This popular text, however, is more of a survey of the state of knowledge, whereas our book is a statement of an ongoing research program that aims at expanding the extant

18

Democracy and Economic Openness in an Interconnected System

body of knowledge.2 In terms of the methodological approach, many books use the same approach as ours, but their respective substantive coverages are much narrower, focusing on only one or some of the interconnected processes we investigate here. For example, Democracy, Governance, and Economic Performance by Yi Feng (2003) studies the interrelationship between democracy, regime type, inflation, and economic growth at the nation-state level, ignoring much of the globalization, environmental, and interstate conflict processes. Triangulating Peace: Democracy, Interdependence, and International Organization by Bruce Russett and John Oneal (2001) investigates the interrelationships among military conflict, democracy, trade, and international organizations at the dyadic level, ignoring income inequality, development, environment, and aspects of economic globalization other than trade. Growth, Trade, and Systemic Leadership by Rafael Reuveny and William Thompson (2004) explores the long-term dynamics among growth, trade, and a few great powers at the system level, ignoring aspects of globalization other than trade, issues of democracy, income inequality, and the environment. These books are important in their own right, but they fail to address the interconnectedness of several key processes in the global system. We believe that to understand the impact of broad transformations like the spread of democracy and rising economic openness, one has to cross and transcend disciplinary boundaries. To that effect, our book confronts debates in various fields and subfields in the social sciences such as international relations, comparative politics, political economy, environmental politics, environmental economics, and globalization studies. It brings the insights from compartmentalized bodies of literature into direct dialogues and exchanges to produce a better understanding of the inner workings of the international system. Our book is academic, scholarly, and policy-oriented in nature. We orient this book toward a rather broad readership that encompasses scholars, graduate students, undergraduate students in upper-division honors programs, practitioners, and policymakers in the areas of international political economy, international relations, comparative politics, international economics, sociology, political geography, human geography, environmental economics, ecological economics, and environmental studies, as well as those who have broad interdisciplinary interests in the topics we analyze here. Our book ought to help graduate students, scholars, and policymakers get familiar with different debates, literatures, and new ideas in different 2

For another example by Held and his coauthors, see Held and McGrew (2007).

Introduction

19

fields that are interconnected in one framework. It also provides a springboard from which one can continue to investigate both the linkages we study and those linkages we do not study. Because we relegate all the technical materials to the chapter appendices, the general readership interested in the topics and relevant debates can be fully informed of the issues, arguments, and findings simply by reading the main text. Most important, we expect our readers to benefit from the opportunity to confront the various new connections, solutions, and policy trade-offs one could only identify within a larger picture of the world we live in. In an age of complex transformations, it is important to realize that policies and processes often interact in unexpected manners and produce unintended consequences. We must note that we do not cover all the possible cause–effect links, nor is it desirable to do so in one book. Yet it is important to discuss the missing links at the outset so that the analytical boundaries of the book are transparent. The reader may recall the big picture with which we started and gradually narrowed down to Figure 1.1 that represents the essence of our book. Naturally, the causal arrows shown in Figure 1.1 do not exhaust all possibilities. Figure 1.1 demonstrates the causal arrows we study in this book, not all the possible cause–effect links among those forces. One may group the causal arrows “missing” from Figure 1.1 into two types. One type concerns some channels that we include but treat as secondary in the book. For example, the channels from economic development to income distribution (the so-called Kuznets curve) and from development to the environment (the so-called environmental Kuznets curve) are discussed theoretically and analyzed statistically, albeit not in the spotlight of our analysis. A second type of missing arrows involves possible linkages between forces in Figure 1.1, but they are neither shown in the figure nor studied here. Most notably, income distribution, democracy, and interstate military conflict may influence economic openness. In fact, we intentionally treat economic openness as exogenous in this book for two reasons. We do so to focus on the consequences of globalization, for our interest lies in the effects of economic openness on democracy, inequality, development, the environment, and conflict, and the interactions among these forces facing globalization. In addition, this keeps our analysis tractable because the sources of economic openness and globalization are controversial and involve multiple bodies of literature on trade, foreign direct investment, and financial liberalization. Although we do not study these and perhaps other potentially missing channels in this book, we discuss them again in the last chapter as possibilities for future research.

20

Democracy and Economic Openness in an Interconnected System

In closing, this book has grown out of our joint research program in recent years. The fact that several of our chapters have been published in different forms in high-quality journals and scrutinized by peer scholars multiple times speaks to one important strength of our project. But our book does not simply replicate our previous work word-for-word. Rather, we build on, extend, and rewrite our various projects from recent years to integrate them together into one coherent picture and to cater to the broad readership we have in mind. As such, this book amounts to a product whose total size is bigger than the sum of its individual parts. As we explained, the global system has experienced rising interdisciplinary interconnectedness among its different processes and mechanisms. Our previous works, like those of our peers, have emphasized single parts of the ensuing global transformations. We believe that only by putting together separate analyses and identifying their interrelationships can one understand the complex interconnections of these global transformations. With the introduction of the book behind us, we proceed to the next chapter. Chapter 2 begins by observing the growth of globalization forces in recent decades. The question we seek to answer is whether and how these processes affect democracy within national boundaries. To that effect, we conceptualize globalization as the interconnectivity of a country to the world system in the realms of economic and information flows. As we shall see, this interconnectivity may have significant and controversial ramifications for national democratic governance.

PART I

THE DEMOCRACY–ECONOMY NEXUS

TWO

Democracy and Economic Openness

INTRODUCTION

We begin our analytical journey with an analysis of the causal flow from economic openness to democracy, which stands at the center of the conceptual layout of the book in Figure 1.1. The issue we seek to investigate in this chapter is in fact much broader than the components of openness itself; it falls under the umbrella of economic globalization. A popular interpretation of this phenomenon is the gradual turning of separate national economies, each operating in its own domain, into a set of national economies heavily entangled with one another, affecting and being affected by the economic forces operating in other countries in the world much in the same way that the economy of the state of Texas is affected by and affects, for example, the economies of the states of California or Florida in the United States. As noted, the causal relationship of interest falls under the umbrella of the globalization discourse; therefore, we use the terms “economic globalization” and “economic openness” interchangeably. Two questions stand at the core of our investigation in this chapter. Does economic globalization affect the level of democracy? Is deepening integration into the world economy associated with a decline or rise of democratic governance? These questions have captured the imagination of policymakers and academic scholars alike. Various answers have been provided, and policy recommendations have been made. Anecdotal evidence is typically invoked in debates, but systematic evidence is scarce. This chapter seeks to fill this empirical lacuna. The notions of globalization and democracy are widely discussed in the literature. Most scholars agree that, at the minimum, globalization implies that countries are becoming more integrated into the world economy, with increasing information flows among them (e.g., Held et al., 1999, 2009). 23

24

Democracy and Economic Openness in an Interconnected System

Greater economic integration, in turn, implies more trade and financial openness. A rise in information flow implies, arguably, the spread of ideas and cultural convergence across countries. As we discussed in the previous chapter, most scholars also agree that democracy implies a national political regime based on free elections and broad political representation, and this is the conceptualization of democracy that we continue to use in this chapter and, for that matter, throughout the book. Before we delve further into this chapter, some clarifications are in order. Although the historical developments of both globalization and democracy have been long and cyclical, we will not take a long-term historical slant in this chapter, primarily due to the lack of economic performance and openness data over a long period of time.1 We also note that throughout this book, democracy is conceptualized as a level variable, a continuum of regime types, from fully autocratic at one end to fully democratic at the other. Hence, we use the terms “democracy” and “level of democracy” interchangeably. Returning to our questions for this chapter, the theoretical literature presents conflicting positions on the effect of globalization on the level of democracy: one position claims a positive effect, a second position asserts a negative effect, and a third position argues that globalization does not necessarily affect democracy. The position that globalization promotes democracy is entrenched in the views of recent American presidents, and it has been used as a rationale for promoting international economic liberalization. For example, U.S. President William Clinton (1996) argues that “commerce helps make the world safe for democracy” and U.S. President George W. Bush (2001) claims that “societies that open to commerce across their borders will open to democracy within their borders.” This view garners some support in stylized observations. Lin and Nugent (1995) argue that Korea, Taiwan, and Chile exemplify the positive influence exerted by economic openness on the democratic transitions of their authoritarian regimes. As another example, the opening up of Indonesia to the world economy, despite causing some financial difficulties in the late 1990s, seems to have facilitated the evolution of a more competitive political system in that country. In contrast, history also provides some opposing cases in which economic openness had little or even a negative effect on democracy. For instance, as noted by Flandreau (2007), the liberalization of trade between France 1

On the historical development of globalization, see, e.g., Wallerstein (1974), Cameron (1997), Held et al. (1999), and Held and McGrew (2007). On the historical development of democracy, see, e.g., Huntington (1991), Potter et al. (1997), and Diamond (1999).

Democracy and Economic Openness

25

and Britain in the 1860s was not associated with a change in democracy in France, which was then led by an autocratic regime. The contraction of trade liberalization in the late nineteenth century did not affect the levels of British or French democracy. The opening up of Russia to the world economy in the late nineteenth century, which involved deepening economic relations with democracies, was associated with an increase in its level of autocracy. Democratic governance in the United States evolved largely as an internal matter, not affected much by its increasing contacts with the world in the nineteenth century. Finally, recent increases in the economic openness of Russia and China seem to have led to a decline in the level of democracy in Russia since the late 1990s and have had little impact on the level of democracy in China. Of course, the anecdotal evidence does not demonstrate generalizable patterns across countries. Moreover, historical cases may not be applicable to the contemporary era. After all, one may argue that the globalization of our times is much more all-embracing than the one that intensified during the latter half of the nineteenth century. This argument in and of itself can make a difference in terms of the direction, not only the size, of the effect of increasing connectivity to the world economy on the level of democracy within a country. Progress in explaining the effect of globalization on democracy requires an empirical evaluation of the three competing theoretical positions. We seek to detect generalizable empirical patterns in the effect of globalization on democracy. As noted, democracy is conceptualized as a continuum. Four national aspects of globalization are examined: trade openness, foreign direct investment (FDI) inflows, portfolio (financial) investment inflows, and the spread of democratic ideas. Our analysis in this chapter covers 127 countries from 1970 to 1996 in a pooled time-series cross-sectional statistical model. Our primary findings can be summarized as follows. Trade openness and portfolio investment inflows negatively affect democracy. The effect of trade openness is constant over time, whereas the negative effect of portfolio investment inflows strengthens democracy. FDI inflows positively affect democracy, but the effect weakens over time. The spread of democratic ideas promotes democracy persistently over time. The chapter is organized as follows. The next section briefly reviews the studies on the determinants of democracy. The section that follows discusses the effects of globalization on democracy, which, as noted and as we shall further demonstrate, are highly controversial. This discussion is followed by a section that describes both research design and findings from the statistical

26

Democracy and Economic Openness in an Interconnected System

analysis. Understanding this section does not require statistical expertise, for the materials of more technical nature are delegated to the chapter appendix. The last section of this chapter summarizes our main findings and discusses their implications. THE LITERATURE ON DETERMINANTS OF DEMOCRACY

This section reviews the literature on the determinants of democracy. Our review is not meant to be exhaustive but rather seeks to illustrate how our analysis fits into the larger picture. We categorize this voluminous literature into three groups. One group consists of detailed case studies. A second group includes statistical analyses. These two groups, by and large, focus on domestic political and economic variables and pay relatively little attention to international factors. A third group (discussed in the next section) includes largely theoretical studies of the effects of globalization on democracy. In their synthesis of a number of case studies, O’Donnell, Schmitter, and Whitehead (1986) conclude that the effect of international factors on democracy is indirect and marginal. Challenging this conclusion, Pridham (1994) labels international factors as the “forgotten dimensions in the study of domestic transition,” and Schmitter (1996: 27) argues “perhaps, it is time to reconsider the impact of international context upon regime change.” Attempting to bridge this gap among case studies, Whitehead (1996) and Drake (1998) argue that international factors such as the diffusion of democratic ideas and global markets are important determinants of democracy. Among the statistical studies, the dependent variable is democracy, but the independent variables vary. As discussed in the previous chapter, one group of studies argues that economic development positively affects democracy.2 A second group debates the effects of economic crisis, such as recessions and high inflation (Helliwell, 1994; O’Donnell, 1973; Haggard and Kaufman, 1995; Gasiorowski, 1995). A third group expects positive influences by Christianity and negative effects of social cleavages (Huntington, 1984; Muller, 1988; Gasiorowski, 1995). A fourth group examines the effects of institutions such as constitutional arrangements, nonfragmented party systems, and parliamentary versus presidential systems (e.g., Lijphart, 1977; Mainwaring, 1993; Linz, 1994). Finally, some statistical studies 2

See, e.g., Lipset (1959), Dahl (1989), Huntington (1991), Burkhart and Lewis-Beck (1994), Muller (1995), Londregan and Poole (1996), and Feng and Zak (1999).

Democracy and Economic Openness

27

consider external factors but not globalization-related variables, such as core-periphery status and diffusion.3 The strengths and weaknesses of case studies and statistical analyses of democracy are debated (e.g., see Przeworski and Limongi, 1997). Briefly, statistical studies focus on the macro conditions that facilitate or hinder democracy. Although they can test general theoretical claims and control for competing forces, they are less able to explain the micro processes that affect democracy. Case studies identify detailed microlevel influences, but they are less able to test general theories or to assess the relative strength of causal factors. Our study in this chapter takes a macro approach. However, we believe that the macro and micro approaches to our question provide complementary insights into the determinants of democracy, much as macroeconomic and microeconomic analyses provide complementary insights into the operation of the economy. Our inquiry adopts the spirit of the important qualitative comparative study by Rueschemeyer et al. (1992). Like their study, we also study many countries and emphasize the importance of transnational power relations, particularly economic and information flows. In the next section, we discuss how these external forces affect democracy. THE GLOBALIZATION–DEMOCRACY CONTROVERSY

The literature on the effects of globalization on democracy is quite large and mostly theoretical. It posits three competing theoretical positions: globalization promotes democracy, globalization obstructs democracy, and globalization has no systematic effect on democracy. To streamline the presentation of these positions, Tables 2.1, 2.2, and 2.3 summarize the arguments from studies supporting each of the three theoretical positions, respectively.

Globalization Promotes Democracy The first proposition listed in Table 2.1 is that globalization promotes democracy by encouraging economic development. The notion that free markets facilitate democracy can be traced back to the late eighteenth century. 3

For example, see Bollen (1983) and Burkhart and Lewis-Beck (1994), who study position in the world system. Starr (1991) and Przeworksi et al. (1996) study diffusion of democratic ideas. Gasiorowski (1995) focuses on the effect of economic crisis on democratization but also includes trade openness as a control variable.

28

Democracy and Economic Openness in an Interconnected System Table 2.1. Globalization promotes democracy

Number

Argument

Discussed in

1

Globalization promotes economic development.

2

Globalization increases the demand of international business for democracy. Globalization reduces the incentives of the authoritarian leaders to cling to power. Globalization reduces information costs, increasing contact with other democracies and making the prodemocracy international nongovernmental organizations more effective. Globalization pushes the authoritarian states to decentralize power. Globalization promotes domestic institutions that support democracy.

Schumpeter (1950), Held (1992), Platner (1993), Weitzman (1993), Bhagwati (1994), Lipset (1994), Muller (1995), Im (1996) Kant (1795), Bhagwati (1994), Schmitter (1996), Oneal and Russett (1997, 1999a) Rueschemeyer and Evans (1985), Diamond (1995), Drake (1998)

3

4

5

6

7

Globalization intensifies the diffusion of democratic ideas.

Van Hanen (1990), Brunn and Leinback (1991), Diamond (1992a), Schmitter (1996), Keck and Sikkink (1998), Kummell (1998), Boli and Thomas (1999), Risse and Sikkink (1999) Self (1993), Sheth (1995), Roberts (1996) Roberts (1996), Fruhling (1998), Keck and Sikkink (1998), Stark (1998), Boli and Thomas (1999), Risse and Sikkink (1999) Kant (1795), Whitehead (1986, 1996), Huntington (1991), Starr (1991), Przeworski et al. (1996)

In this view, globalization promotes economic growth, increases the size of the middle class, promotes education, and reduces income inequality, all of which foster democracy. Trade, FDI, and financial capital flows are said to allocate resources to their most efficient use; democracy is said to allocate political power to its most efficient use. The outcome in both cases represents the free will of individuals.4 According to a second view, globalization increases the demand of international business for democracy. Business prosperity requires peace and political stability. Because democracies rarely, if ever, fight each other, commercial interests pursue democracy to secure peace and stability. As the 4

See Schumpeter (1942), Held (1992), Platner (1993), Weitzman (1993), Bhagwati (1994), Lipset (1994), Muller (1995), and Im (1996).

Democracy and Economic Openness

29

economic links between states develop, commercial interests strengthen, and the demand for democracy rises. Authoritarian countries that open their economies face greater pressures from international business for political liberalization. Furthermore, in economies open to the transnational flows of commodities and capital, the government and the central bank have to be more transparent in their procedures and policymaking to attract and maintain international business. The increased transparency, in turn, implies less power for the autocratic regime, which can facilitate democratization.5 A third argument is that globalization reduces the incentives of the authoritarian leaders to cling to power. Because the state can extract rents from society, losing office implies the forfeit of these rents. Hence, autocratic rulers cling to power, resisting democracy. However, globalization reduces the capacity of the state to extract rents from society by increasing competition and weakening the effectiveness of economic policies. It follows that leaders of autocracies whose economies are more open are less likely to resist democratization. To the extent that the autocrats believe that providing social welfare for the losers of globalization reduces the latter’s incentive to contest their grip on power, the autocratic leaders should be willing to give their citizens more political rights and civil liberties.6 A fourth view argues that globalization reduces information costs, increasing contacts with other democracies and making the prodemocracy international nongovernmental organizations (INGOs) more effective. A prosperous democracy requires well-informed actors. With increasing globalization, citizens have access to more information, supplied not just by their own governments. Economic openness enables the established democracies, aided by their developed media, to export their values to autocracies. Authoritarian regimes now have less control over information. More exposure to the media also strengthens the effectiveness of transnational advocacy networks and the INGOs, helping them protect prodemocracy forces in authoritarian regimes and promote democracy.7 According to a fifth view, globalization pushes the authoritarian state to decentralize power. As globalization deepens, states relinquish control over economic and social progress to the market, which is “inherently democratic” – as if millions of economic agents cast their “votes” voluntarily. 5 6 7

See Kant (1795), Bhagwati (1994), Schmitter (1996), Oneal and Russett (1997, 1999a), and Dailami (2000). See Rueschemeyer and Evans (1985), Diamond (1995), and Drake (1998). For the effect of social welfare spending and globalization, see Rudra (2005). See Van Hanen (1990), Brunn and Leinback (1991), Diamond (1992a), Schmitter (1996), Keck and Sikkink (1998), Kummell (1998), Boli and Thomas (1999), and Risse and Sikkink (1999).

30

Democracy and Economic Openness in an Interconnected System

The weakened state also implies the entrance of grassroots groups into the political arena (e.g., business and professional associations, labor unions). Citizens become more involved in the day-to-day governance of the country, facilitating democracy.8 A sixth view argues that globalization strengthens domestic institutions that support democracy. Because the efficient operation of the market requires an enforceable system of property rights and impartial courts, economic openness compels the popularization of norms respecting the rule of law and civil and human rights. The increased involvement of international business and INGOs in the domestic economy further promotes the transparency and accountability of domestic institutions and reduces state intervention, all of which are said to facilitate democracy.9 According to a seventh view, economic globalization intensifies the diffusion of democratic ideas across borders. Transnational movements of commodities, services, and factors of production facilitate and encourage the exchange of norms and ideas, increasing the diversity of political views in a country and indirectly promoting political competition. Scholars argue that the more democracies that surround a certain nondemocratic country, the more likely it is that this country will become democratic. Because greater economic openness is associated with more information flow and transnational contacts, the diffusion of democratic ideas across borders is expected to intensify with growing economic integration. The diffusion of democratic ideas and norms will affect the people in the nondemocratic countries, motivating them to demand democratic reforms.10

Globalization Obstructs Democracy The first argument in Table 2.2 is that globalization reduces state policy autonomy and brings about public policies that please foreign investors instead of the common people. Globalization increases financial capital mobility across countries and facilitates relocation of the means of production, which, in turn, reduces the ability of states to implement domestically oriented economic policies. Another consequence is that governments now try to compete for foreign capital and design their policies to please global

8 9 10

See Self (1993), Sheth (1995), and Roberts (1996). See Roberts (1996), Fruhling (1998), Keck and Sikkink (1998), Stark (1998), Boli and Thomas (1999), and Risse and Sikkink (1999). Kant (1795), Whitehead (1986, 1996), Huntington (1991), Starr (1991), Przeworski et al. (1996), and Dailami (2000).

Democracy and Economic Openness

31

Table 2.2. Globalization obstructs democracy Number 1

2

3

4

5

6

7

Argument

Discussed in

Globalization reduces state policy autonomy and brings about public policies that please foreign investors instead of the common people. Globalization produces more domestic losers than winners, at least in the short run, and it also diminishes the ability of the state to compensate the losers financially. Globalization enables the fast movement of money between countries, resulting in frequent balance of payment crises and unstable domestic economic performance. Globalization deepens ethnic and class cleavages and diminishes the national-cultural basis of democracy. Globalization enables the state and multinational corporations to control and manipulate information supplied to the public. Globalization degrades the concept of citizenship, an important prerequisite for a functioning and stable democracy. Globalization widens the economic gap between the North and the South.

Lindblom (1977), Held (1991), Diamond (1995), Gill (1995), Jones (1995), Gray (1996), Schmitter (1996), Cox (1997), Cammack (1998) Drucker (1994), Muller (1995), Beck (1996), Bryan and Farrel (1996), Cox (1996), Moran (1996), Marquand (1997), Martin and Schumann (1997), Rodrik (1997), Longworth (1998) Im (1987), MacDonald (1991), Diamond (1992a, 1999), O’Donnell (1994), Trent (1994), Haggard and Kaufman (1995), Cammack (1998)

Robertson (1992), Dahl (1994), Im (1996)

Gill (1995), Im (1996), Martin and Schumann (1997)

O’Donnell (1993), Whitehead (1993), Im (1996), Sassen (1996), Cox (1997), Boron (1998) Wallerstein (1974), Bollen (1983), Tarkowski (1989), Przeworski (1991), Gill (1995), Amin (1996), Cox (1996), Im (1996), Kummell (1998)

investors and firms, who may not act in the best interest of, or be held accountable to, the voters. It follows that the level of democracy declines.11 According to a second argument, globalization produces more domestic losers than winners, at least in the short run, and it also diminishes the 11

See Lindblom (1977); Held (1991), Diamond (1995), Gill (1995), Jones (1995), Gray (1996), Cox (1996), Schmitter (1996), and Cammack (1998).

32

Democracy and Economic Openness in an Interconnected System

ability of the state to compensate the losers financially. Domestic producers that cannot compete internationally lose from more economic openness. Trade and financial liberalization also limit the ability of governments to deploy regulation and risk-sharing policies that can redistribute income from winners to losers. Governments that want to compensate the victims of economic openness confront the phenomenon of footloose capital, which in turn shrinks the tax base and penalizes deficit spending. Consequently, governments reduce the scope of welfare programs, and the poor increasingly feel this pinch. The result is rising income inequality and class polarization, which serve to weaken democracy.12 A third view argues that globalization enables the fast movement of money between countries, resulting in frequent balance-of-payment crises and unstable domestic economic performance. In such situations, the less developed countries (LDCs) are compelled to accept economic reforms imposed by the developed countries (DCs) and international organizations, and those reforms typically involve austerity measures. Economic crises hurt the poor more than the rich, raising domestic income inequality. Social unrest then rises, and support for radical opposition groups grows. In an attempt to reassert power, weak democracies resort to authoritarian measures. The electoral technicalities are seemingly retained, but civil rights and the inputs from the elected legislators are increasingly ignored.13 According to a fourth argument, globalization deepens ethnic and class cleavages and diminishes the national-cultural basis of democracy. The losers from economic openness tend to seek a united identity based on ethnicity or religion. The winners may promote discriminatory measures to maintain their edge over the losers. Globalization also induces labor migration across countries. The old comers typically attempt to restrict or eliminate the participation of the immigrants in the political system to reduce their competitiveness. All these actions intensify social cleavages and undermine the consolidation of democracy.14 A fifth argument is that globalization enables the state and the multinational corporations (MNCs) to control and manipulate information supplied to the public. With the help of new information technologies, the state and the MNCs feed to the public processed information that represents 12 13 14

See Drucker (1994), Muller (1995), Beck (1996), Cox (1996), Moran (1996), Marquand (1997), Martin and Schumann (1997), Rodrik (1997), Longworth (1998), and Dailami (2000). See Im (1987), Diamond (1992a, 1999), MacDonald (1991), O’Donnell (1994), Trent (1994), Haggard and Kaufman (1995), and Cammack (1998). See Robertson (1992), Dahl (1994), and Im (1996).

Democracy and Economic Openness

33

only certain views. They are also better able to tightly monitor the people’s information sources. The result is a disconnection between the way government decisions are made and the way the public thinks they are made. The government becomes less transparent and less accountable to the people, and the level of democracy declines.15 According to a sixth argument, globalization degrades the concept of citizenship, which is an important prerequisite for a functioning and stable democracy. The global market transforms the individual into a common “Homo economicus” who cares more about profits than public and civic commitments. Individuals pursue their own economic interests, disregarding whether governments practice democratic decision making. Since the public has less interest in the conduct and content of public policy, democracy gradually weakens.16 A seventh argument is that globalization widens the economic gap between the North and the South. Globalization involves mostly DCs, draining capital, technology, and skilled labor from LDCs. With the gap in wealth rising, social unrest increases in LDCs, their elites cling to power, and their governments become less democratic. The dependency story evolves along similar lines. In a world composed of a rich core and a poor periphery, the core dominates the periphery. The elite in the periphery unite with the elite in the core to exploit the masses in the periphery. MNCs relocate to LDCs to enjoy both lower wages and more lax labor and environmental standards and then repatriate profits to the core. The penetration by MNCs distorts the economies in LDCs and sways domestic politics in their own favor, all of which obstruct democracy.17

Globalization Does Not Necessarily Affect Democracy Table 2.3 shows three arguments that question whether globalization has any general effect on democracy. The first argument holds that the extent of globalization is greatly exaggerated. The world economy is not as integrated as is commonly believed. Most international trade takes place within geographical regions, MNCs typically have a home-country bias, and most FDI concentrate in a few advanced countries. Because LDCs generally do not participate in the global economy, the effect of economic openness on 15 16 17

See Gill (1995), Im (1996), and Martin and Schumann (1997). O’Donnell (1993), Im (1996), Sassen (1996), Whitehead (1996), Cox (1997), and Boron (1998). See Wallerstein (1974), Tarkowski (1989), Przeworski (1991), Gill (1995), Amin (1996), Cox (1995), Im (1996), and Kummell (1998).

34

Democracy and Economic Openness in an Interconnected System Table 2.3. Globalization does not necessarily affect democracy

Number

Arguments

1

The extent of globalization is exaggerated.

2

Globalization does not render the welfare state powerless.

3

Effects of globalization on countries vary.

Discussed in Scharpf (1991), Jones (1995), Hirst and Thompson (1996), Wade (1996), Hirst (1997) Vernon (1971), Kurzer (1993), Frieden and Rogowski (1996), Garrett (1999) Haggard and Kaufman (1995), Frieden and Rogowski (1996), Milner and Keohane (1996), Armijo (1998), Longworth (1998)

their democracy should not be large to begin with. Because DCs are already stable democracies, globalization does not affect their levels of democracy.18 A second argument asserts that globalization does not necessarily render the welfare state powerless. Increased national economic openness originates from the deliberate choices of states. Governments still exert considerable control over their own economies. Moreover, the modern welfare state is still effective because it provides important collective goods undersupplied by markets (e.g., social stability, property rights, infrastructure) and compensates the losers of economic openness. By implication, one can argue that the level of democracy does not necessarily have to decline with economic openness.19 According to a third view, the effects of globalization vary across countries and depend on government policies, a country’s location in the global pecking order, the domestic political institutions, the identity of the domestic winners and losers, whether economic sectors are privatized or not, and the current level of democracy. For example, though globalization-induced economic crises may force the authoritarian regime to exit and be replaced by democracy, these crises, if managed effectively, may instead increase the public support for the authoritarian leader. Hence, the effects of globalization on democracy may not be uniform.20 The arguments summarized in Tables 2.1, 2.2, and 2.3 all appear to exhibit face validity in terms of logic consistency, and it is not possible to evaluate 18 19 20

See Scharpf (1991), Jones (1995), Hirst and Thompson (1996), Wade (1996), and Hirst (1997). See Frieden and Rogowski (1996) and Garrett (1999). See Haggard and Kaufman (1995), Frieden and Rogowski (1996), Milner and Keohane (1996), Armijo (1998), and Longworth (1998).

Democracy and Economic Openness

35

them based on pure theoretical considerations. In other words, because these arguments reach conflicting conclusions, they need to be evaluated empirically, to which we turn next. EMPIRICAL MODEL AND ANALYSIS

This section first presents our statistical model for the empirical analysis and then discusses several research design issues. Next, the section presents the key results of the empirical analysis. The discussion in this section is self-contained and does not require any specific statistical expertise. Interested readers seeking further details should consult the chapter’s appendix, which discusses various technical issues related to the statistical design and implementation of the regression-based model presented here. The chapter appendix follows the same structure of presentation as in this section.

Empirical Model To assess the competing claims about the effects of globalization on democracy, we specify and estimate the following statistical model of democracy. We denote variables with uppercase letters and their coefficients with Greek notations. Each coefficient indicates the effect of the independent variable on the dependent variable – the phenomenon we seek to explain. The notation εt denotes the random error that is not explained by the statistical model. The variable subscripts t and t −1 indicate the time period of the variable, where t represents the current period and t −1 the previous time period (a lagged variable). To simplify the presentation, we refer to the variables without their time subscripts t or t −1. The model specification is based on the preceding theoretical discussions and the literature on democratization. It provides a structure for us to guide the statistically uninitiated readers through the empirical exercise. democracyt = ␤0 + ␤1 tradet−1 + ␤2 fdit−1 + ␤3 portfoliot−1 + ␤4 diffusiont−1 + ␤5 inflationt−1 + ␤6 growtht−1 + ␤7 developmentt−1 + ␤8 prior democracyt−1 + ␤9 year + ␤10 year∗ fdit−1 + ␤11 year∗ portfoliot−1 + ␤12 year∗ inflationt−1 +␤13 year∗ developmentt−1 +εt . (2.1) In the model, democracy denotes the left-hand-side variable or the dependent variable (the phenomenon we explain), whose determinants are

36

Democracy and Economic Openness in an Interconnected System

specified on the right-hand side of the equation and dubbed the independent variables. The democracy variable represents the level of democracy in a country in a given year. The level of democracy can be measured quantitatively over a continuum of national attributes ranging from complete autocracy or dictatorship at one end of the continuum to full democracy at the other end of the continuum. The measure construction, described in the appendix, includes both democratic and autocratic characteristics, an approach widely used in the literature. The independent variables divide into two types. One type involves the globalization-related variables, which stand at the center of our study. A second type of independent variable involves other determinants of democracy that have been identified in various other studies. We call the second type of factor control variables. Here we discuss briefly the measurement of these variables. We start with how to measure the notion of globalization. As a multidimensional concept, the integration of states into the world economy needs to be measured from multiple indicators. In our empirical model, globalization is measured by four indicators: trade, foreign direct investment, portfolio investment, and diffusion of democratic ideas across countries. The trade variable is a measure of trade openness – the extent to which a country depends on trade with other countries. The fdi variable is a measure of the openness of a country to foreign production capital flowing into a country, which refers to investments that enter a country for the purpose of building plants or acquiring management positions in plants. The portfolio variable is a measure of the openness of a country to financial capital flowing into a country, meaning portfolio investments that enter for the purpose of buying financial securities. The diffusion variable represents the diffusion of democratic ideas into a country, which is measured here by the extent to which a country is affected by democratic ideologies emanating from neighboring countries. Our study is one of the first attempts to assess the effects of globalization on democracy by using multiple indicators in one empirical statistical analysis. Because globalization produces competing effects on democracy through many mechanisms, these globalization indicators help estimate the short-run net effects. Obviously, democracy does not only depend on a country’s economic openness. Control variables must be included to avoid spurious statistical inferences. Although existing studies of the causes of democracy differ when identifying these additional determinants, many studies find that domestic economic factors such as economic development, economic growth,

Democracy and Economic Openness

37

and inflation influence democracy.21 The inflation variable measures the yearly rise in price level in a country. The development variable measures the level of economic development in a country and is based on the gross domestic product (GDP) per capita, which is the value of final goods and services produced in a country divided by the country’s population. The growth variable is the yearly growth rate of GDP, which captures the vitality and improvement of economic performance over time in a country. The prior democracy variable denotes the level of democracy in a country in the previous year. This variable controls for the fact that political regimes tend to have inertia and change slowing over time. Theoretically, domestic structural variables (e.g., Protestant population, institutional qualities of the regime, and party fragmentation) and the attributes of the international system may also affect democracy. These additional factors are relatively stable over time, implying that democracy exhibits inertia over time or that the past level of democracy affects the present. The year variable is a yearly counter that serves two functions. By itself, year tests whether the level of democracy has a linear trend, and it also interacts with other variables that have time-varying effects over democracy. It is possible that the effects of economic globalization on the level of democracy at home may change over time. For example, the scope and types of FDI, financial capital investments, and information flows took on a relatively more prominent role in the 1990s than in the 1970s. This change is not unique to the globalization variables. Gasiorowski, for example, finds that the effect of inflation on the likelihood of democratization changes over time. Ignoring this issue altogether may lead to model misspecification. In our exploratory by-decade analysis of which variables have time-varying effects on democracy (presented in Table 2.A1 of the appendix), we find that the effects of development, fdi, portfolio, and inflation on the level of democracy in a country change their signs over time. Hence, our statistical model includes interaction terms between year and these four variables, respectively.

Research Design Issues To implement our statistical model, several design issues must be raised that require some clarification. Although we address the technical details 21

For the three economic variables, see Bollen (1979), Muller (1988), Lipset et al. (1993), Burkhart and Lewis-Beck (1994), Helliwell (1994), Muller and Seligson (1994),Gasiorowski (1995), Muller (1995), and Feng (1997).

38

Democracy and Economic Openness in an Interconnected System

of these issues in the appendix, here we provide a brief overview for the general reader. First, the variable εt is the so-called error term in the statistical model, that is, the part of the variations in democracy the model does not explain. For statistical inferences to be valid, the error term needs to satisfy some assumptions. If the error term does not have several desirable properties, the accuracy of the results and our ability to make correct inferences is compromised. We address the related assumption violations using appropriate econometric techniques. Second, the relationship between democracy and other variables on the right-hand side may be reciprocal or simultaneous. For example, whereas economic integration variables may affect democracy, they themselves may be affected by democracy. A higher level of democracy may imply lower political risks and more secure property rights, attracting trade and investments (Olson, 1993). Ignoring this possibility can produce incorrect results. We deal with this risk by using the first lag of the right-hand-side variables, as denoted by the subscript t −1 in the model. The third issue concerns sample selection. We take the approach of largeN studies, employing a sample for many countries, where each country is represented by yearly data. More specifically, our sample includes 127 countries from 1970 to 1996. The unit of analysis is the country year. Such a sample design allows us to assess the effect of globalization on democracy over time and across countries and to generalize the inferred results across cases. A fourth issue concerns a complication in sample selection, which has to do with a possible distinction between DCs and LDCs. Most DCs have already achieved high levels of democracy at the beginning of the sample period. On the other hand, many LDCs experienced large variations in their levels of democracy from 1970 to 1996. Although the effects of globalization on the level of democracy may be independent of the development level of a country, it is equally possible that LDCs exhibit patterns in the effects of globalization on democracy distinct from those of DCs. In fact, several authors argue that the adverse effects of globalization on democracy may be stronger in DCs than in LDCs. We investigate this possibility by analyzing the effects of globalization for all countries and then for a sample that excludes DCs as measured by membership in the Organization for Economic Cooperation and Development (OECD).

Empirical Findings We present the main results of our analysis in Table 2.4 for two samples. One sample includes all the countries for which we have data, whereas the

Democracy and Economic Openness

39

Table 2.4. Effects of economic globalization on democracy

TRADE FDI PORTFOLIO DIFFUSION INFLATION GROWTH DEVELOPMENT PRIOR DEMOCRACY YEAR YEAR∗ FDI YEAR∗ PORTFOLIO YEAR∗ INFLATION YEAR∗ DEVELOPMENT Constant Observations Adjusted R2

All countries

Non-OECD countries

−0.0029∗∗∗ (0.0010) 5.1906∗ (3.3016) 4.7240∗∗∗ (1.8888) 0.2437∗∗∗ (0.1058) 0.0599∗∗ (0.0350) 0.0068 (0.0084) 34.2597∗∗∗ (11.5024) 0.9269∗∗∗ (0.0109) 0.1514∗∗∗ (0.0505) −0.0026∗ (0.0017) −0.0024∗∗∗ (0.0009) −0.00003∗∗ (0.000017) −0.0172∗∗∗ (0.0058) −301.8202∗∗∗ (100.2552) 2021 0.93

−0.0032∗∗∗ (0.0011) 5.8359∗ (3.7712) 5.3892∗∗ (2.4406) 0.2922∗∗∗ (0.1219) 0.0537∗ (0.0344) 0.0069 (0.0087) 36.7637∗∗∗ (16.8662) 0.9242∗∗∗ (0.0115) 0.1593∗∗∗ (0.0670) −0.0029∗ (0.0019) −0.0027∗∗ (0.0012) −0.000027∗ (0.000017) −0.0184∗∗ (0.0085) −317.8469∗∗∗ (132.8935) 1640 0.90

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

second sample includes only LDCs. The findings for the so-called control variables are consistent with those in previous studies in the democratization literature. This general pattern provides support for our statistical modeling. To sharpen our presentation, we delegate the discussion of control variables to the chapter appendix and focus on the findings of the globalization variables instead. Beginning with the all-countries sample, as shown in the second column of the table, the effect of a rise in trade (the level of trade openness of

40

Democracy and Economic Openness in an Interconnected System

a country) on democracy (the level of democracy in a country) is found to be negative. The finding for trade is consistent with the well-known tendency of trade to generate economic winners and losers in the short run. Although the overall gain from trade outweighs the cost of trade, the process by which the winners compensate the losers is political and slow, and the endogenous domestic economic adjustment process (e.g., movement of labor from losing to winning sectors) is also relatively slow. As such, trade can generate changes in the distribution of income in the short run, giving rise to the type of processes described in the section on the obstructing effects of globalization on democracy. The effect of a rise in fdi (the openness of a country to FDI inflow) on the level of democracy in that country is positive and statistically significant. However, this positive effect declines over time, as indicated by the negative and statistically significant coefficient of the interaction term year∗ fdi. More specifically, the total effect of a rise in fdi on the level of democracy, which is given by (5.1906 − 0.0026∗ year), is 0.055 for 1975, 0.029 for 1985, and 0.001 for 1996. Hence, the total effect of a rise in FDI on the level of democracy in a country is positive, but its size is declining over time. These results suggest that a rise in FDI exerts both positive and negative effects on the level of democracy in a country, as discussed earlier, but the negative effect grows over time. On the positive side, FDI can promote economic growth and technological progress. On the negative side, FDI can raise domestic income inequality. The tendency of FDI to quickly relocate to cheaper production places in other nations and the tendency of MNCs to interfere in host countries’ domestic politics (see “The Globalization– Democracy Controversy,” this chapter) may also account for the negative effect of FDI on democracy. The direct effect of a rise in portfolio (the openness of a country to financial flows) on the level of democracy of a country is also positive. However, once the dependence of this particular effect on time is incorporated, as indicated by the interaction term of year∗ portfolio, the total effect comes out negative. Specifically, the total effect of portfolio investment inflows in the all-countries sample, which is given by (4.7240 − 0.0024∗ year), is −0.016 for 1975, −0.04 for 1985, and −0.066 for 1996. Hence, the total effect of portfolio on democracy is negative and strengthens over time. The results for portfolio suggest that a rise in portfolio investments also has both positive and negative effects on democracy, as discussed earlier in the chapter, but the negative effect grows to become dominant in the all-countries sample. The positive main effect of portfolio investments on democracy may reflect the economic discipline financial markets impose on

Democracy and Economic Openness

41

governments, which has contributed to greater transparency, better management practices, and stronger property rights institutions. The negative effect may capture the dramatic increase in the ability of portfolio investments to move quickly among countries, which at times leads to financial collapse and devastating sociopolitical outcomes. An increase in diffusion (the number of democracies in the region around a country) raises the level of democracy in that country. Interpreting this variable as a proxy for information flows induced by integration of a country into the world economy, one may argue that democratic ideas and values are more likely to transmit and diffuse to geographically proximate countries. Indirectly, then, this result does not support the claim discussed earlier in the chapter that the increased information flows into a country would reduce the level of democracy in that country. The spread of democratic ideas from one’s neighboring countries improves, rather than hinders, the country’s own level of democracy. The third column in Table 2.4 shows the estimation results obtained for a sample that includes only LDCs. These results match well with those we reported earlier for the sample of all countries. However, two primary differences in results exist between the two samples. One difference involves the sizes of the direct effects of trade, fdi, portfolio, and diffusion (our measures of globalization) on democracy. The coefficients of these variables appear to be larger in the LDC sample than those obtained in the full sample of all countries. We interpret this outcome as a demonstration that a rise in globalization has a larger effect on democracy in LDCs than in DCs. In a way, this result is intuitive since, in the period covered by our sample, the democracy levels of DCs are stable for the most part and hover at, or close to, the highest possible positive level of the democracy indicator. The democracy levels in LDCs, on the other hand, exhibit much larger variations and are more sensitive to the external influences of globalization. A second difference between the two samples concerns the total effect of portfolio on democracy. As in the all-countries sample, once one incorporates the dependence of the democracy effect of portfolio on time, indicated by the interaction term year∗ portfolio, the total effect (5.3892 − 0.0027∗ year) is 0.0567 for 1975, 0.0297 for 1985, and 0 for 1996. Hence, the total effect of a rise in portfolio on democracy for the LDC sample declines over time, as in the all-countries sample, but it has not yet turned negative in our sample period. We think this result is also intuitive. The direct effect of a rise in portfolio on an average country in the LDC sample is larger than the direct effect on an average country in the all-countries sample (5.8359 vs. 4.724,

42

Democracy and Economic Openness in an Interconnected System

respectively). A typical LDC gains more from a rise in portfolio investment than a typical country in the all-countries sample in terms of a rise in its level of democracy. Over time, the positive effect of a rise in portfolio is counteracted by a growing negative effect that reflects the overall rise in global capital mobility and increasing volatility of financial markets. The size of this negative component is larger in the LDC sample (−0.0027∗ year) than in the all-countries sample (−0.0024∗ year), but because the direct positive effect is larger in the LDC sample than in the all-countries sample, the overall effect in the LDC sample remains positive. This outcome is consistent with the observation that capital flows into a typical LDC are relatively smaller than those into a typical country in the all-countries sample. As such, the share of the growing negative effect of capital flows over time out of the total effect is smaller in the LDC sample than in the all-countries sample. While we have discussed the signs of the effects of globalization on the level of democracy, the statistical findings also can inform us about the substantive significance of these effects. In evaluating the substantive effect, we assume that trade, fdi, portfolio, or diffusion exhibit certain increases, and then we compute the changes in the level of democracy induced by these variables ceteris paribus. Studies often assume a one-standard-deviation increase for a certain variable, as identified by the estimation sample, and we also employ this approach. Using the coefficient of each variable for the all-countries sample in Table 2.4, we compute the percent change in democracy relative to its sample mean as the baseline. Turning to the size of effect, a one-standard-deviation rise in trade (46.83%) leads to a decline of about 8% in the average level of democracy in the sample (1.72). A one-standard-deviation rise in diffusion (0.79) raises the level of democracy by 6.5% relative to the baseline case. A onestandard-deviation rise in fdi (2.04%) in 1975 raises the level of democracy by 6.5% relative to our baseline, ceteris paribus, whereas a similar rise in fdi in 1996 raises the level of democracy only by 0.12%. A one-standarddeviation rise in portfolio (2.5%) in 1975 reduces the level of democracy by 2.3% relative to the baseline, whereas a similar change in 1996 reduces the level of democracy by about 10%. Therefore, the immediate responses of democracy to changes in globalization do not appear to be large in absolute size; this indicates that the globalization-induced changes in democracy tend to happen slowly over time. What about the long-run dynamic changes in democracy resulting from globalization forces? Indeed, the changes noted earlier in globalization also affect democracy in the next period through their effects on the lagged value of democracy. As detailed in the appendix, we can compute the effect of a

Democracy and Economic Openness

43

change in, say, trade on democracy in the long run. On employing those one-standard-deviation changes, the long-term percentage changes in the average democracy level in the sample register a decline by 108% due to trade and a rise by 153% due to diffusion. The long-term effect of fdi change is a rise in the average level of democracy by 89% in 1975 and only 1.6% in 1996. With respect to portfolio, its long-term effect is a decline in the average level of democracy by 32% in 1975 and 131% in 1996. Hence, globalization forces could generate large changes in democratic governance over the long run.

Sensitivity Analyses The robustness of statistical findings is important for establishing empirical regularity, proving the validity of theoretical arguments, and facilitating public policy positions. Do the empirical results reported earlier hold when other possible confounding control variables enter the statistical model, or when other statistical estimation procedures are used? To answer this particular question, we conduct a number of sensitivity analyses to verify the robustness of the reported results. In one type of sensitivity analysis, we employ different statistical estimators, which are methods to obtain the quantitative effects of variables from historical data. In a second type of sensitivity analysis, we use various measures of democracy based on different types of data. In the third and final type of sensitivity analysis, we add various additional control variables to the model, including income inequality in a country, the level of education in a country, the number of INGOs to which a country belongs, and variables denoting whether a country belongs to periphery or semiperiphery of the world system (yet another indicator of its level of development). The details of these sensitivity analyses and their results are reported in the chapter appendix. We can summarize these additional statistical results by noting that the statistical findings we have reported, both in terms of the effects on democracy of the key globalization variables and the effects on democracy of the original control variables specified in this section, are found to hold under the large majority of the sensitivity analyses. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

In general, the effects of changes in globalization on the level of democracy in a country tend to be small in the short run but could accumulate to large magnitudes over the long run. The effects of FDI, however, tend to dwindle

44

Democracy and Economic Openness in an Interconnected System

over time. We believe that these results have important implications not only for future research on the determinants of democracy but also for the larger public policy debate about globalization. If maximizing economic efficiency is the only public policy objective, then the negative effects of globalization on democracy do not, and should not, matter. However, if the democratic form of governance is deemed a desirable public policy objective, our empirical findings presented in this chapter suggest a policy dilemma between economic efficiency gains and democratic decline. Our analysis implies that globalization erodes the prospects for democracy. The post-1945 model of embedded liberalism served to sustain democratic governance in DCs with open economies. However, this model cannot be easily implemented in LDCs in the current globalization. The emerging democracies among LDCs lack the financial and managerial resources needed to build social safety nets. As trade liberalization continues, the negative effect of trade on democracy may increase. Similarly, the growing capital mobility accompanying globalization produces a political dilemma for governments who want both economic competitiveness and democratic political accountability. Footloose capital is generally not accountable to the public. Capital mobility reduces the democratic government’s ability to respond to popular demands for social welfare and effective economic management. Our findings imply that under economic openness, the room for policy maneuvering is obviously reduced. Hence, the threats to democracy from financial inflows and FDI are substantial. If the current trend in the effect of globalization on democracy continues, our finding that the increasing number of democracies in a region is associated with greater democracy may reveal a “diabolic” flip side. That is, as trade, FDI, and financial flows become increasingly associated with democratic decline, the number of democracies in a region may also decline, which in turn may contribute to a decrease in the number of democracies. In light of our findings, it is only natural that one may ask the following question: If democracy falls due to trade openness, is openness a mistake, say, for Eastern Europe? Openness may generate economic gains. Countries open up because they seek the gains, or actors poised to gain are able to get their way, but some arguments suggest openness can also reduce democracy, which is what we find for trade. Thus, if both democracy and openness are desirable, countries face a trade-off. We further discuss this trade-off in the context of our larger conceptual picture in Chapter 10. In addition, one should note that our findings apply to the large-N sample, detecting average statistical patterns rather than explaining any particular country or region; Eastern Europe is just part of a larger sample.

Democracy and Economic Openness

45

How can one alleviate the negative effects of globalization on democracy? We believe there are basically three possibilities. The first possibility is for there to be one world government, managing one economic system, without any barriers to the flows of goods, financial capital, and other factors of production. The trade-off between efficiency gains and democratic decline can then be effectively managed. But this option is currently impractical. A second possibility is to enhance the socioeconomic policy coordination among governments within the existing international system. For example, concerted tax policies could minimize excessive financial capital volatility in an integrated global financial market. With policy coordination, countries could continue to benefit from the globalizing economy while minimizing the harmful effects on democracy. However, one also notices the weakness of such an option. Historically, international socioeconomic policy coordination was generally unstable. Within the existing international political system, each national government is accountable to its own people. The policy coordination option is difficult to achieve and sustain over time. A third possibility is to slow down the rate of globalization. For example, governments could design tax and subsidy policies to compensate the losers from economic openness, slow down capital movements and reduce excessive volatility, and curtail by law excessive rent-seeking by MNCs. To be successful, this approach would require coordinated market regulation that involves the government, the private sector, and citizen advocacy groups, implying wide consultation with and under close monitoring by the private sector and citizen advocacy groups. The purpose of such coordination is to minimize the reliance on the government alone to design and implement regulation. Governments, particularly in LDCs, often suffer from inefficiency and inflexibility. Relying solely on governments may suffocate the market. Where democratic rules are not well established, as in many LDCs, the notion that only the government can correct market failures may cause excessive government intervention and lower social welfare, and it eventually precipitates a decline in democratic governance. Although this option may yield lower social welfare relative to the first two possibilities, it may be the only practical approach available at present for reducing the negative effects of globalization on democracy. SUMMARY AND OUTLOOK

The theoretical literature presents conflicting expectations of the effect of globalization on democracy. These conflicting expectations are also reflected in the public debate on globalization. In this chapter, we make a systematic empirical effort to assess the controversial effects of globalization on

46

Democracy and Economic Openness in an Interconnected System

democracy. We find that trade openness and portfolio investment inflows exert negative effects on democracy. The negative effect of trade openness on the level of democracy is stable over time, whereas the negative effect of portfolio investment inflow into a country strengthens over time. FDI inflows into a country are conducive to its democratic governance, but the strength of this particular effect weakens over time. The spread of democratic ideas flowing into a country from other countries in its vicinity contributes to increasing the level of democracy in that country persistently over time. These effects are all substantively important, particularly in light of their long-term implications for democracy. These empirical regularities are found to hold for both the all-country sample and LDCs alone. They are also robust across various model specifications, different measures of democracy, and alternative statistical estimators. In sum, the economic aspects of integration into the world economy are beginning to cause a decline in national democratic governance, which ought to cause concern for both policymakers and academicians. In this chapter, we investigated the effect of aspects of economic openness, or economic globalization, on the level of democracy within countries. The reader may recall that, in our sensitivity analyses, the distribution of income within countries, as measured by the Gini coefficient, was one of the control variables. In the next chapter, we change the focus of our inquiry, making income inequality the main variable of interest. What is the effect of a rise in the level of democracy on the national distribution of income? What is the effect of our measures of economic openness (i.e., trade, FDI, and portfolio flows) on the national distribution of income? Do these forces make the rich richer and the poor poorer, further skewing the income distribution, or do they make the rich poorer and the poor richer, reducing the income gap? Historical observation suggests that these are important questions in that they can have political ramifications, particularly for the stability of the government and civic life. Extremely skewed distributions of income are known to have contributed to the violent revolutions in France in the eighteenth century, Russia in the early twentieth century, and China in the mid-twentieth century. Now it is time for us to step into Chapter 3.

APPENDIX EMPIRICAL MODEL AND ANALYSIS

This appendix accompanies the section “Empirical Model and Analysis” in the main text. We provide various details on the measurement of our

Democracy and Economic Openness

47

variables, their data sources, issues pertaining to the technical design of the statistical model, and the expected effects of various control variables. We also report and discuss the results for the control variables as well as the various additional analyses.

Empirical Model The dependent variable democracy denotes the level of democracy for a country in any given year, based on the POLITY III data (Jaggers and Gurr, 1995; Gurr and Jaggers, 1999). The widely used POLITY III data register various attributes of regime type on an annual basis for many countries from about 1800 to 1998, using two indices. The 10-point democracy index (DEMOC) measures the democratic characteristics of the regime. The 10point autocracy index (AUTOC) measures the autocratic characteristics of the regime. As pointed out by Oneal and Russett, Londregan and Poole, Mansfield and Snyder, among others, because many governments have both democratic and autocratic characteristics, DEMOC and AUTOC do not provide redundant information about regime type, and both should be used to measure the level of democracy.22 We follow these studies and measure democracy as the difference between DEMOC and AUTOC, generating an index ranging between −10 (for the most autocratic regime) and 10 (for the most democratic regime). Globalization is measured by four indicators: trade openness, FDI inflows, portfolio investment inflows, and the spread of democratic ideas. The three economic variables are collected from the World Bank’s World Development Indicators CD-ROM. The trade variable denotes the yearly level of trade openness for a country. As conventional in the literature, trade is defined as the sum of the value of imports and the value of exports of goods and services of a country with the rest of the world, measured as a percentage of the country’s GDP.23 The fdi variable denotes the yearly value of net inflows of FDI as a percentage of a country’s GDP for each country. FDI involves either the acquisition of a lasting management interest (10% or more of voting stock) in an enterprise operating in an economy other than that of the investor or the creation of a new subsidiary of a firm in a foreign country. It is given here as the sum of equity capital, reinvestment of earnings, and other long- and short-term capital as shown in the balance of payments. The portfolio variable denotes the yearly value of net 22 23

See, e.g., Mansfield and Snyder (1995), Londregan and Poole (1996), and Oneal and Russett (1997, 1999a). We also estimated the models using the log of trade/GDP. All the results remain robust.

48

Democracy and Economic Openness in an Interconnected System

inflows of portfolio investments as a percentage of GDP for each country. Portfolio investments (excluding liabilities constituting foreign authorities’ reserves) cover transactions in equity securities (the sum of country funds, depository receipts, and direct purchases of shares by foreign investors) and debt securities (publicly guaranteed and nonguaranteed debt from privately placed bonds). The measure of the spread of democratic ideas is coded using POLITY III data; diffusion denotes the yearly number of democracies within a region around each country. The regions include Europe, Middle East, Africa, Asia, and North and South America. We define a country as democratic if the difference between its DEMOC and AUTOC scores is greater than or equal to six (as in, e.g., Mansfield and Snyder, 1995; Oneal and Russett, 1997, 1999a, 1999b, 1999c). Like Starr (1991) and Gasiorowski (1995), we assume that the type of political regime in a country is more likely to be affected by the type of political regime of its neighboring countries than by more distant countries. In general, the diffusion of democratic norms is said to work through contact-based mechanisms, which involve information flows, contact in the marketplace, communication networks, tourism, and so forth. Geographically proximate countries typically have more contact. Hence, in our context, diffusion may be thought of as a proxy for information and communication flows of democratic ideas among countries. It should be noted that we compute the correlation between a yearly measure of the number of main phone lines and television sets for each country and diffusion. The correlation (r = 0.45) is statistically significant at the 1% level. We do not include this measure as an independent variable because it is highly collinear (r = 0.8) with GDP per capita in the model. Hence, it is possible to argue that GDP per capita also serves as a proxy of the communication aspect of national integration into the world economy. Moving to the economic control variables, inflation denotes the yearly growth rate of the GDP deflator, which is the ratio of nominal GDP (measured in current prices) to real GDP (measured in constant prices). Several studies use inflation as a proxy of economic crisis, the effect of which on democracy is debated. Some studies argue the effect is positive, whereas others argue the effect is negative (Huntington, 1991; Gasiorowski, 1995; Haggard and Kaufman, 1995; Drake, 1998). The development variable denotes the logged yearly GDP per capita in purchasing power parity– adjusted real international prices. As discussed at length in this chapter, many statistical studies of the determinants of democracy use this variable as an indicator of the level of economic development, either as a central

Democracy and Economic Openness

49

variable or as a control variable.24 The growth variable denotes the annual percentage growth rate of real GDP.

Research Design Issues Several empirical design issues influence our model specification and estimation in this chapter. First, in the main text, we note that, to control for the effect of inertia, we include the lagged dependent variable (the level of democracy) as an independent variable, a strategy also adopted in other studies.25 In addition to the theoretical reason, this lagged variable helps capture the effects of variables not present in the model. “With such a pervasive control in place, it is more difficult for spurious effects to be reported” (Burkhart and Lewis-Beck, 1994: 905). It is worth noting that the inclusion of the lagged dependent variable might soak up the variations in the dependent variable that could be explained by other independent variables, making it harder for us to find statistically significant results. Hence, our approach can be described as conservative. Second, to address the possibility that the effects of some of the determinants of democracy may change over time, we use the systematically varying parameter approach (Judge et al., 1988: 435; Griffiths et al., 1993: 421). To determine which variables have a time-varying effect, we first estimate a pooled time-series cross-sectional model for each decade (1970s, 1980s, and 1990s). The decade selection has no particular theoretical justification, but the results provide a general sense about the temporal stability of the effects on democracy. Variables whose parameters exhibit reversals of signs across those pooled decades are suspected to have time-varying effects on democracy. Interaction terms between these variables and a calendar year variable are created and included in the model.26 Third, the error term in the model faces the risks of heteroskedasticity and serial correlation. The estimated coefficients are still consistent with heteroskedastic variance and serial correlation in the error term, but their standard errors are not efficient and are most likely biased. To deal with this problem, we estimate our statistical model using a variant of the White estimator of robust standard errors, which adjusts for clustering over country 24 25 26

In addition to sources cited in Chapter 2, see, e.g., Rueschemeyer (1991), Diamond (1992a), Gasiorowski (1995), and Przeworski and Limongi (1997). See, e.g., Bollen (1979), Muller (1988), Burkhart and Lewis-Beck (1994), and Muller and Seligson (1994). Gasiorowski (1995) employs such an interactive model to examine the time-varying effects of inflation on democratization.

50

Democracy and Economic Openness in an Interconnected System

(White, 1980). This estimator yields consistent estimation of the covariance matrix under very general conditions of heteroskedasticity and serial correlation (Wiggins, 1999). In addition, as Beck and Katz suggest, the lagged dependent variable models the temporal dynamics and also helps to address serial correlation (Beck and Katz, 1995a, 1995b). Fourth, consider the design issue for resolving the simultaneity bias. Ignoring the possible effect of democracy (the dependent variable in this chapter) on the right-hand-side variables in the model (globalizationrelated and others) implies the risk of simultaneity bias, which generates misleading estimated coefficients. The ideal solution to this problem is to estimate a fully specified simultaneous equations system, endogenizing democracy, trade, FDI, portfolio investment, and information flows. At the current stage of our knowledge of how these forces interact, it is not feasible to specify such a system of equations. In contrast, many studies deal with this problem by lagging the independent variables one period, which in our case amounts to one year (e.g., Muller and Seligson, 1994; Oneal and Russett, 1997, 1999a, 1999b, 1999c). Though this is not an optimal solution to the issue of simultaneity, we follow the same approach here. Fifth, we need to consider which statistical significance level to use when we interpret the results. In evaluating the effect of globalization on democracy, we have discussed three types of theories. One type expects that globalization will promote democracy, the other type expects the opposite, and the third type expects no effect at all. As we noted, the three types of theory appear logically consistent, given their own assumptions, regarding the effect of globalization forces on democracy. The empirical model can only identify the net effect of globalization on democracy. We therefore only test the sign of this net effect on democracy against the null hypothesis of no effect – or rather the two competing effects of globalization on democracy are equal in size – employing the one-tailed test in reporting the results. Many other studies have employed this approach (see, e.g., Morrow et al., 1998; Oneal and Russett, 1999a, 1999b, 1999c; Li and Reuveny, 2003; Reuveny and Li, 2003). Sixth, there is the issue of what thresholds to use for statistical significance. In interpreting the results, we employ 1, 5, and 10% significance levels. We take this approach for a couple of reasons. Some of our samples are relatively small because of limitations on data availability. In addition, the lagged dependent variable in the model soaks up the variations in the dependent variable to be explained by other explanatory variables, making it difficult to find significant effects. Seventh, we need to measure the size of globalization’s effect on democracy. In the text, we discuss not only the direct effect of the globalization

Democracy and Economic Openness

51

variable but also the long-run effect through the lagged dependent variable. For the long-run effect, the impact of a globalization variable on the level of democracy from previous periods is absorbed into the effect of the lagged democracy variable, which is also on the right-hand side. The globalization variable affects the current level of democracy via the direct effect and continues to affect the level of democracy in the next period via the lagged democracy variable (Londregan and Poole, 1996). The long-run impact of a change in a globalization variable produces the following change in the level of democracy in a country: [coefficient of globalization variable/ (1–coefficient of lagged democracy)] ∗ (change in globalization variable). We compute the long-run effects for all our globalization variables. Finally, one may question in particular why we measure the impact of FDI by its flow rather than by stock. As noted, FDI flows may affect democracy, and FDI flows through many mechanisms and generates competing effects on democracy. The goal of the empirical analysis is to measure the immediate net effect of FDI flows on democracy. This measurement requires the use of FDI inflows (incurring during a year) because the stock variable, cumulated over time since the beginning of FDI, conflates the short- and long-run effects of FDI. As explained earlier, we estimate the long-run impact of FDI through the inclusion of the lagged dependent variable in the model. In addition, the FDI stock data are statistically problematic. Whereas flows are reported by national governments to the IMF, stock data, which are only available from the United Nations Conference on Trade and Development (UNCTAD), are often estimated and involve large measurement errors. Stock data sum FDI flows and reinvestment earnings since the start of FDI. Reinvestment earning data are often missing, generating measurement errors that continue in the stock data in future years. In contrast, the effect of such measurement error in the flow data is limited to the year of the missing data. Moreover, the initial baseline stock of FDI in a country is often unknown for sure, and the year UNCTAD uses as the start of stock accumulation is the year in which a country began to report flows, not necessarily the year when FDI first entered a country. This generates another type of measurement error. Due to these errors, the stock data are not appropriate for cross-national comparisons.

Empirical Findings Table 2.4 presents the results for all countries from 1970 to 1996 in the first column and for the non-OECD countries in the second column. The adjusted R2 for both models are 0.93 and 0.90, respectively, indicating that the models explain most of the observed variations in democracy.

52

Democracy and Economic Openness in an Interconnected System

All the interaction terms are statistically significant, as are all the main effects of the variables for which an interaction was included. Another general observation is that all the globalization-related variables are also statistically significant. In fact, except for economic growth, all the effects are statistically significant, which supports our model specification. Because we have discussed the results for the globalization variables in detail, here we focus on the control variables. In terms of the specific results of the control variables in column 1, prior democracy has a positive and statistically significant effect on current democracy. Other studies that include this variable report similar results, illustrating that democracy as a social phenomenon tends to have inertia. As Przeworski et al. (1996) argue, it almost seems tautological to argue that the current level of democracy rises when the past level of democracy rose. However, this result has an important substantive implication. One simply cannot expect to find a higher level of democracy by strengthening dictatorship or authoritarianism ceteris paribus. If the level of democracy fell in the past, one should expect to find a lower level of democracy in the future. Therefore, international interactions that strengthen an authoritarian regime are likely to hinder the movement toward democracy in that country in the near future. Similarly, reforms toward a higher level of democracy tend to accumulate over time, leading to democratic consolidation. The effect of inflation is statistically significant and positive, but its interaction term year∗ inflation is significant and negative. The total effect of inflation on the level of democracy is positive and was the largest in the 1980s; economic growth has a positive sign, but it is not statistically significant. As in many studies, economic development has a positive and statistically significant effect on the level of democracy. The interaction term year∗ economic development is statistically significant but negative, suggesting that its effect weakens over time. The calendar year variable year is statistically significant and positive, indicating in this sample a partial tendency for democracy to grow over time. The results regarding the domestic economic effects on democracy replicate the spirit of previous studies, which supports our model. To further investigate the robustness of the results, we estimated the model for a sample of only non-OECD countries. Compared with the results for all countries in column 1, the adjusted R2 for the non-OECD countries in column 2 is a bit lower because of the smaller sample size and reduced variation in the independent variables. However, the signs and statistical significance levels for all the variables in column 2 match well with those in column 1,

Democracy and Economic Openness

53

indicating that the results in column 1 are not an artifact of the inclusion of the OECD countries in the sample. We also estimate the models in Table 2.4 by excluding the lagged dependent variable. Although the inclusion of the lagged dependent variable usually makes other independent variables insignificant, its exclusion from our model actually renders the results statistically less significant in both samples. This suggests that we did not get null results in Table 2.A2 because of the lagged dependent variable. These results are available from the authors upon request. Now, as noted, to identify which variables have varying effects on democracy over time, we first estimate pooled time-series cross-sectional models for each decade for all countries and for LDCs. We report those results in Table 2.A1. The effect of trade on democracy is always negative and statistically significant, and the effect of diffusion is always positive and statistically significant in four out of six cases. The effect of prior democracy is always positive and statistically significant. The effect of economic growth is positive but statistically insignificant in all cases (except for the case of all countries from 1970 to 1979). Because the signs of these four variables are consistent across decades, we include them in the model as main effects only. The signs of economic development, fdi, portfolio, and inflation change over decades. Hence, they enter the model both individually and as interactive terms with year. For the overlapping variables (inflation, economic growth, economic development, trade openness), the results generally agree with those reported by Gasiorowski (1995), although his dependent variable measures the event of democratization (a 1, 0 variable), whereas our dependent variable measures the level of democracy. Overall, these results provide justification for why we specify the interactive model as such in the main text.

Additional Analyses Model specification, the measures of democracy, and the estimation techniques vary across statistical studies of democracy. To investigate the effects of these variations in our case, we conduct various sensitivity analyses. Specifically, we add new control variables, use Freedom House data to measure democracy, and apply different estimators. The results are presented in Tables 2.A2 and 2.A3. Overall, the effects of globalization on democracy reported in Table 2.4 are replicated across the 25 experiments reported in Tables 2.A2 and 2.A3. Therefore, we judge our results to be robust. Later we discuss Tables 2.A2 and 2.A3 in detail. The results for the original control

54

−0.0042∗∗ (0.0022) 0.0141 (0.0178) 0.0031 (0.0183) 0.6034∗∗ (0.2662) −0.0031 (0.0028) −0.0019 (0.0153) 0.1098 (0.1143) 0.9253∗∗∗ (0.0237) −1.4142 (1.0956) 353 0.93

−0.0021∗ (0.0014) −0.0357 (0.0340) −0.0211 (0.0196) 0.1501 (0.1559) 0.0002∗∗∗ (0.0000) 0.0067 (0.0122) 0.4081∗∗∗ (0.0999) 0.9253∗∗∗ (0.0161) −3.0927∗∗∗ (0.8233) 948 0.94

−0.0041∗∗ (0.0021) −0.0188 (0.0391) 0.0021 (0.0069) 0.2251∗ (0.1445) −0.0001 (0.0001) 0.0115 (0.0156) −0.0820 (0.0797) 0.9202∗∗∗ (0.0226) 1.0082 (0.7240) 720 0.90

−0.0049∗∗ (0.0025) 0.0110 (0.0192) −0.0118 (0.0216) 0.9548∗∗∗ (0.3172) −0.0045∗ (0.0029) 0.0012 (0.0165) 0.1523 (0.1509) 0.9141∗∗∗ (0.0244) −2.2999 (1.3985) 273 0.89

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Observations Adjusted R2

Constant

PRIOR DEMOCRACY

DEVELOPMENT

GROWTH

INFLATION

DIFFUSION

PORTFOLIO

FDI

TRADE

−0.0029∗∗ (0.0016) −0.0426 (0.0385) −0.0247 (0.0210) 0.1981 (0.1777) 0.0001∗∗∗ (0.0000) 0.0071 (0.0125) 0.5352∗∗∗ (0.1199) 0.9290∗∗∗ (0.0170) −4.0277∗∗∗ (0.9187) 772 0.91

−0.0032∗ (0.0024) −0.0200 (0.0424) 0.0092 (0.0093) 0.2378∗ (0.1617) −0.0001 (0.0001) 0.0127 (0.0160) −0.1938∗∗ (0.1120) 0.9170∗∗∗ (0.0238) 1.7565∗ (0.9460) 595 0.88

ALL (1970–1979) ALL (1980–1989) ALL (1990–1996) LDC (1970–1979) LDC (1980–1989) LDC (1990–1996)

Table 2.A1. Pooled time-series cross-sectional models of democracy by decade

55 −0.0012 8.0087∗∗ 5.7176∗∗ 0.3073∗∗∗ −0.0040∗∗ −0.0029∗∗ 0.0007∗∗∗

−0.0018∗∗ 6.2954∗∗ 4.5246∗∗∗ 0.2576∗∗∗ −0.0032∗∗ −0.0023∗∗∗ 0.0002∗∗

1568 0.90

Model 10 LDC

Model 9 ALL

1949 0.93

1640 0.90

2021 0.93

−0.0033∗∗∗ 6.3762∗∗ 4.9084∗∗ 0.2971∗∗∗ −0.0032∗∗ −0.0025∗∗ −0.9099∗

−0.0028∗∗∗ 5.4310∗ 4.1119∗∗ 0.2369∗∗ −0.0027∗ −0.0021∗∗ −0.6336∗

0.0148 −0.7589∗∗ −0.5427∗∗ −0.1591 1879 0.93

−0.0016∗∗ 6.8347∗∗ 3.9381∗∗∗ 0.1345 −0.0035∗∗ −0.0020∗∗∗

Model 11 ALL

0.1417 1662 0.93

−0.0025∗∗∗ 5.5593∗ 2.7475 0.3141∗∗∗ −0.0028∗ −0.0014

Model 3 ALL

Model 5 ALL

Model 13 ALL

1826 0.93

0.0170∗∗ −0.0046

1705 0.94

−0.0035 −0.0117

Freedom House Data −0.0026∗∗∗ −0.0006 4.8632∗ 5.8462∗∗ 5.1121∗∗∗ 2.3640∗ 0.2902∗∗∗ 0.1151∗∗∗ −0.0025∗ −0.0030∗∗ −0.0026∗∗∗ −0.0012∗

Model 12 ALL

POLITY III data −0.0027∗∗∗ −0.0025∗∗∗ 5.7162 5.6112∗ 3.6058 2.6123 0.3915∗∗∗ 0.3124∗∗∗ −0.0029 −0.0028∗ −0.0018 −0.0013 −0.1091 0.1137 0.1343 1281 1662 0.91 0.93

Model 4 LDC

1784 0.94

−0.0007∗ 4.3957∗ 2.4035∗∗ 0.0995∗∗∗ −0.0022∗ −0.0012∗∗

Model 14 ALL

−0.0027∗∗ 6.1508∗ 3.1767 0.3928∗∗∗ −0.0031∗ −0.0016 −0.3664 0.0879 1281 0.91

Model 6 LDC

1421 0.90

−0.0007∗ 5.3203∗ 1.6529 0.1209∗∗∗ −0.0027∗ −0.0008

Model 15 LDC

−0.0023∗∗∗ 8.4656∗∗ 5.6914 0.2844∗∗∗ −0.0043∗∗ −0.0029 0.0950 0.2889∗∗∗ 1783 0.93

Model 7 ALL −0.0025∗∗ 10.5950∗∗ 3.4633 0.3558∗∗∗ −0.0053∗∗ −0.0017 −0.2263 0.2457∗∗ 1353 0.91

Model 8 LDC

Note: ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%. Other independent variables not reported to save space. Models 5 and 6 include while Models 7 and 8 exclude economic development and its interaction term.

TRADE FDI PORTFOLIO DIFFUSION YEAR∗ FDI YEAR∗ PORTFOLIO INGOS SEMIPERIPHERY∗ DEVELOPMENT PERIPHERY∗ DEVELOPMENT EUROPE MIDDLE EAST AFRICA ASIA N Adjusted R2

TRADE FDI PORTFOLIO DIFFUSION YEAR∗ FDI YEAR∗ PORTFOLIO GINI EDUCATION N Adjusted R 2

Model 2 LDC

Model 1 ALL

Table 2.A2. OLS estimates with additional control variables or Freedom House data

56

Democracy and Economic Openness in an Interconnected System

variables from the models in Tables 2.A2 and 2.A3 are broadly consistent with Table 2.4. Therefore, we do not discuss these results again in this subsection. Table 2.A2 reports the results from 15 experiments, denoted Model 1 through Model 15, which investigated the effects of adding new control variables and/or using a different measure of democracy. In Models 1–12, democracy is measured based on POLITY III data as in Table 2.4 whereas in Models 13–15, democracy is based on Freedom House data. As for the new control variables, gini denotes the level of national income inequality, measured by the Gini coefficient. The variable education denotes the level of national education measured by the average number of years of education in the population. For gini and education, we use the same data and variable transformation as in Feng and Zak (1999). The variable gini comes from the high-quality data from Deininger and Squire (1996). We log and lag the raw data. The missing values of gini are filled with predictions from estimating gini as a function of GDP per capita, GDP per capita squared, and regional dummies. The variable education comes from Barro and Lee (2000) and is also logged and lagged. The variable ingos denotes the number of INGOs to which individuals or organizations from each country belonged in a given year. The data on INGOs come from Boli and Thomas (1999). The two dummy variables semiperiphery and periphery denote whether a country belongs to the world-system semiperiphery or periphery, respectively. In the model, they interact with economic development as in Burkhart and Lewis-Beck (1994). Similar to this study, we have also estimated the models using only semiperiphery and periphery dummies by themselves and found similar results; europe, middle east, africa, and asia are dummy variables that denote the geographical region of a country. Across Models 1–12 in Table 2.A2, the effect of trade is significant in 11 out of 12 cases and is always negative. The main effect of fdi is significant in 11 cases and is always positive. The interaction effect of fdi is significant in 11 cases and is always negative. The effect of diffusion is significant in 11 cases and is always positive. The main effect of portfolio is significant in 6 cases and is always positive. The interaction effect of this variable is significant in 6 cases and is always negative. These results strongly support those in Table 2.A2. Regarding the dependent variable, although we measure democracy based on the POLITY III data, some scholars employ the Freedom House data (e.g., Burkhart and Lewis-Beck, 1994; Diamond, 1999). It is worth noting that some scholars point out that the Freedom House data are not suitable for a pooled time-series analysis because the data are not fully comparable

57

∗

significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

−0.0030∗∗∗ 5.183∗ 4.652∗ 0.358∗∗∗ 0.024 0.004 41.139∗∗ 0.922∗∗∗ 0.171∗∗∗ −0.00001 −0.003∗ −0.002∗ −0.021∗∗ −340.864∗∗ 1640 0.91

−0.000005 11.075∗∗ −1.311 0.185 0.006 0.003 24.572 0.792∗∗∗ 0.186∗ −0.000003 −0.006∗∗ 0.001 −0.013 −368.001∗ 1640 0.91

TRADE FDI PORTFOLIO DIFFUSION INFLATION GROWTH DEVELOPMENT PRIOR DEMOCRACY YEAR YEAR∗ INFLATION YEAR∗ FDI YEAR∗ PORTFOLIO YEAR∗ DEVELOPMENT Constant N Adjusted R2

Note:

−0.0027∗∗∗ 4.786∗ 3.545∗ 0.285∗∗∗ 0.039 0.005 36.566∗∗∗ 0.925∗∗∗ 0.157∗∗∗ −0.00002 −0.002∗ −0.002∗ −0.018∗∗∗ −312.65∗∗∗ 2021 0.93

0.0004 8.796∗∗ 0.515 0.197 0.006 0.004 45.851∗∗∗ 0.797∗∗∗ 0.262∗∗∗ −0.000003 −0.004∗∗ −0.0003 −0.023∗∗∗ −517.172∗∗∗ 2021 0.93

Fixed effects (year)

TRADE FDI PORTFOLIO DIFFUSION INFLATION GROWTH DEVELOPMENT PRIOR DEMOCRACY YEAR YEAR∗ INFLATION YEAR∗ FDI YEAR∗ PORTFOLIO YEAR∗ DEVELOPMENT Constant N Adjusted R2

Fixed effects (country) −0.0029∗∗∗ 5.191∗ 4.724 0.244∗∗∗ 0.060 0.007 34.260∗∗ 0.927∗∗∗ 0.151∗∗ −0.00003 −0.003∗ −0.002 −0.017∗∗ −301.820∗∗ 2021 0.93 −0.0032∗∗∗ 5.836∗ 5.389 0.292∗∗∗ 0.054 0.007 36.764∗ 0.924∗∗∗ 0.159∗ −0.00003 −0.003∗ −0.003 −0.018∗ −317.847∗ 1640 0.90

LDCs −0.0032∗∗∗ 5.836 5.389 0.292∗∗∗ 0.054 0.007 36.764∗∗ 0.924∗∗∗ 0.159∗∗ −0.00003 −0.003 −0.003 −0.018∗∗ −317.847∗∗ 1640 0.90

OLS and PCSE

All countries −0.0030∗∗∗ 5.196 4.724 0.244∗∗∗ 0.060 0.007 34.290∗∗∗ 0.927∗∗∗ 0.151∗∗∗ −0.00003 −0.003 −0.002 −0.017∗∗∗ −302.064∗∗∗ 2021 0.93

Random effects

Table 2.A3. Parameter estimates from alternative estimators

−0.0029∗∗∗ 5.536∗ 5.882∗∗ 0.290∗∗∗ 0.061∗∗ 0.007 35.577∗∗ 0.934∗∗∗ 0.154∗∗∗ −0.00003∗∗ −0.003∗ −0.003∗∗ −0.018∗∗ −306.119∗∗∗ 1640

−0.0027∗∗∗ 4.903∗ 4.704∗∗ 0.237∗∗∗ 0.067∗∗ 0.007 32.634∗∗∗ 0.934∗∗∗ 0.145∗∗∗ −0.00003∗∗ −0.002∗ −0.002∗∗ −0.016∗∗∗ −288.835∗∗∗ 2021

GEE

58

Democracy and Economic Openness in an Interconnected System

across years due to changes in methodology and scale (Gates, 2002).27 We use these data here because they offer a new take on qualities of democracies, despite their limitations. POLITY III data emphasize constraints on executive political recruitment, contestation, and functioning. Based on the Freedom House data, Diamond (1999) argues that in recent years some countries have become pseudodemocracies, where the rights of contestation are protected but civil rights are precarious. In addition to institutional aspects directly measured in POLITY III, Freedom House data capture aspects that are only indirectly implied in POLITY III, such as the de facto power of the opposition, freedom from foreign domination, minority rights, freedom of expression and belief, association rights, rule of law, human rights, and personal economic rights. Unlike POLITY III, Freedom House data are available only from 1973 onward. In their overlapping period, the cross-correlation between the two data sets is relatively high: 0.88 from a sample of all countries and 0.83 from a sample of LDCs. Indeed, the inferences from the Freedom House– based models in Table 2.A2 do not differ much from the inferences from the POLITY III–based models in Table 2.4. Like Burkhart and Lewis-Beck (1994), we sum the two 7-point scales of “political rights” and “civil liberties” in Freedom House. This procedure gives an index ranging from 2 (lowest democracy) to 14 (highest democracy). We also recompute diffusion and prior democracy based on this score. Across Models 13–15, the effect of trade is always negative and statistically significant in two cases. The effect of diffusion is always positive and significant. The positive effect of fdi and the negative effect of its interaction term are always significant. The positive effect of portfolio and the negative effect of its interaction term are significant in two cases. Compared with Table 2.4, the results for Models 14 and 15, based on the Freedom House data, are weaker in terms of statistical significance for portfolio investments. We believe this is because Freedom House also measures civil rights and democratic norms that take a relatively long time to develop. The Freedom House index typically lags behind the POLITY III index and has a smaller variance.28 Because the Freedom House index varies less, the results are less significant. Furthermore, because portfolio 27 28

For additional issues on the measurement of democracy, see Bollen (1991). For example, for Bulgaria, the POLITY score rose from −7 in 1989 to +8 in 1991, a 75% increase (within a range of 20 from −10 to +10), whereas the Freedom House score rose from 2 in 1989 to 9 in 1991, a 53% increase (within a range of 13 from 2 to 14). For another example, for Russia over the same time frame, the POLITY score rose from −4 to +6, a 50% increase, while its Freedom House score rose from 5 to 7, a 15% increase.

Democracy and Economic Openness

59

investments in LDCs are small and concentrated in the elite, their effect can be large in terms of constraints on the executive, which is the focus of POLITY III, but is likely to be smaller in terms of the civil rights, which is captured by the Freedom House. Table 2.A3 presents results for five estimators: country fixed effects, year fixed effects, random effects, OLS with panel corrected standard errors (PCSE), and general estimating equation (GEE) model. The country fixedeffects estimator assumes that the intercept term varies across panels, the year fixed-effects estimator assumes that the intercept term varies across years, and the random effects estimator assumes that the error term varies across panels. The PCSE method assumes that the variance of the error term is heteroskedastic across panels and homoskedastic within panels (Beck and Katz, 1995b). GEE is a population average–based estimator that is frequently used for panel data (Liang and Zeger, 1986). All models, including those in Table 2.4, have roughly the same goodness of fit. The estimates in Table 2.A3 are generally similar to those in Table 2.4, except for the country fixed-effects estimator. Scholars debate the merits of panel data estimation methods. Although this chapter is not about methods, we believe the estimator used in Table 2.4 (OLS with White robust standard errors adjusted for clustering over countries) is the most appropriate for our case. Beginning with the fixed-effects estimators, King (2001) argues that using more elegant statistical techniques is preferable to using fixed effects estimation, which is the approach taken in Table 2.4. In simulations, fixed-effects estimators are known to soak up excessive between-country variations attributable to substantive variables and to reduce degrees of freedom due to the inclusion of many dummy variables. Unless the number of time periods approaches infinity, the estimated effects are biased for the country fixed-effects estimator (Green, 2007). Because we have 27 periods, our analysis may suffer from this problem, which could explain why the results from the country fixed-effect estimator differ noticeably from those based on the other estimators. The applicability of the random effects model for a certain data set can be tested using an appropriate Lagrange multiplier test. The null hypothesis is that the random effect model is not supported by the data (Green, 2007). From conducting this test we are not able to reject the null hypothesis, because we find no evidence in favor of the random effects model. The OLS with PCSE estimator assumes that the variance of the error term is heteroskedastic and contemporaneously correlated across panels, and homoskedastic within panels (Beck and Katz, 1995b). The PCSE estimator may not be appropriate in our case because in Table 2.4 several variables have

60

Democracy and Economic Openness in an Interconnected System

time-varying effects, suggesting the possibility of heteroskedastic errors over time. The estimator we used in Table 2.4 does not assume a certain structure of heteroskedasticity or serial correlation, making it more appropriate for our purpose. One advantage of the PCSE method is its ability to correct for correlated errors across panels. However, as Beck and Katz (1995b) note, a better strategy is to model this correlation theoretically, which is achieved by our diffusion variable. The results from the GEE method in Table 2.A3 are identical to the results in Table 2.4 in terms of hypothesis testing. This outcome further supports the claim that our results are generally robust when different estimators are used (except for the country-fixed effects variant). Turning to the results for the new controls in Table 2.A2, Feng and Zak (1999) argue that a fall in income inequality and a rise in education raise the likelihood of democratic transitions. Whereas our analysis differs from Feng and Zak (we include economic openness, study the level of democracy, and our sample is larger), our results generally agree with theirs. In Models 1 and 2, the effect of gini is negative and significant (a rise in inequality reduces democracy). In Models 7 and 8, the effect of education is positive and significant when economic development is excluded. As in the study by Feng and Zak, when economic development is included (Models 3–6), the effect of education is positive but insignificant. As they explain, this is because education and economic development are correlated. When included with education (Models 5–8), the effect of gini is negative, as in Feng and Zak, but it is not significant, which we attribute to the aforementioned differences between our model and theirs. The effect of INGOs on democracy is positive and significant in Models 9 and 10. These results demonstrate that membership in INGOs increases the level of democracy in a member country, supporting the arguments of Keck and Sikkink (1998) and Risse and Sikkink (1999). The coefficients of the regional dummies in Model 11 measure the effect on democracy of being in some region, relative to the American continent (i.e., we exclude one dummy to prevent falling into the dummy variable trap). The results indicate that relative to the American continent, the level of democracy in the Middle East, Africa, and Asia is lower but roughly the same as in Europe. As could be expected, when the regional dummies are included, the effect of diffusion (which correlates with regional effects), although still positive as in Table 2.4, is insignificant. The effects of the world position variables on democracy are examined in Models 12 and 13. In Model 12, the effect of semiperiphery is positive and

Democracy and Economic Openness

61

significant, and the effect of periphery is negative and insignificant. However, in Model 13, which uses the Freedom House data as in Burkhart and Lewis-Beck (1994), both variables are negative and neither is significant. Our results differ from the results of Burkhart and Lewis-Beck because they find negative and significant effects for both periphery and semiperiphery interaction terms. We believe one reason for the difference has to do with the sample. Their data end in 1989 whereas ours end in 1996, capturing post–Cold War democratic transitions in many of the semiperiphery countries (e.g., Bulgaria, Czechoslovakia). A second reason has to do with the difference between the POLITY III and Freedom House data. As noted, the measure based on POLITY III has a larger variance than that of Freedom House. POLITY III focuses on the changes in institutional constraints on the leaders of the executive branch, whereas Freedom House also covers the gradual changes in civil rights. These are also reflected in the world position variables, particularly semiperiphery.

THREE

Democracy, Economic Openness, and Income Inequality

INTRODUCTION

In the previous chapter, we saw that the different aspects of national economic openness affect the level of democracy in a country. In that analysis, the level of national income inequality acted as a control variable in the statistical model of the sensitivity analysis. We found that a rise in the level of income inequality could reduce the level of democracy in a country. This chapter changes the causal arrow between income inequality and democracy, and we now focus on identifying the causes of national income inequality. We seek to study two interrelated important questions pertaining to this phenomenon: How does the level of economic openness of a country affect its distribution of income? How does the level of democracy influence a country’s distribution of income? These two questions are very important for social scientists because a country with a highly skewed income distribution tends to be politically unstable. Political instability often may lead to intrastate conflict and ultimately civil war. So far, these two questions of interest have been addressed in two separate bodies of literature. The effect of economic openness on the distribution of income has been debated in the literature on economic globalization (e.g., Rodrik, 1997; Held et al., 1999). The issue has also been debated in policy and popular circles (e.g., Soros, 1997; Wolf, 2000; World Bank Poverty Net, 2000). As discussed in the previous chapter, although the exact definition and scope of globalization are debated, scholars agree that current globalization implies growing economic openness of countries to trade, foreign direct investments (FDIs), financial capital flows, and – to a lesser extent – labor mobility (e.g., Held et al., 1999; Hughes, 2000). Existing studies offer conflicting theoretical expectations of the effects of economic

62

Democracy, Economic Openness, and Income Inequality

63

openness on the distribution of income, but these studies do not provide rigorous empirical analyses of their claims. Meanwhile, the effect of democracy on income distribution is an important topic in the literature on democracy (Chan, 1997b). Scholars agree that democracy implies a national political regime based on free elections and broad political representation (e.g., Diamond, 1999). Existing studies argue that democracy reduces income inequality, but empirical evidence is inconclusive. Some studies find a negative effect of democracy on income inequality, other studies find a positive effect, and a third group finds no effect. These two bodies of literature have remained largely on separate courses. Taking a broader view, one may position economic openness and democracy at the center of liberalism. The literature on liberalism, which emphasizes the importance of free choice, comprises two main streams (Zacher and Matthew, 1995). Republican liberalism implies that democracy, the preferred political regime, reduces income inequality. Commercial liberalism (classical and neoclassical economics) argues that free market capitalism is the preferred economic regime, and income inequality is an inevitable by-product.1 Although the issue of income inequality was central for classical economists, it has received relatively little attention in neoclassical economics (Ferreira, 1999). We believe the relationship between income distribution, on the one hand, and democracy and open market, on the other, deserves more attention. We argue that the effects of these forces on income distribution need to be studied together. Capitalism and democracy are not easy companions. Whereas democracy is based on the principles of “one person, one vote” and representative government, capitalism is based on the principles of laissez-faire and private enterprise. Democracy is inherently associated with redistributive policies (e.g., progressive taxation), but capitalism typically rewards individuals with different levels of income and wealth. Hence, democracy may promote income equality, whereas capitalism may promote income inequality. Democracy and economic openness may cancel out or mediate each other in affecting the income distribution within a society. Since many studies2 also argue that economic openness affects democracy, empirical studies of income inequality that exclude either economic openness or democracy as a causal determinant can incorrectly attribute the effect of one force to the other. 1 2

For a review article focusing on the variants of liberalism, see Zacher and Matthew (1995). See, e.g., Im (1996), Whitehead (1996), Drake (1998), Held et al. (1999), and Li and Reuveny (2003).

64

Democracy and Economic Openness in an Interconnected System

Our empirical work in this chapter differs from previous studies in several ways. We study the effects of both democracy and economic openness on income inequality. We examine how democracy, trade, FDI, and financial capital inflows influence income inequality measured by the Gini coefficient. Furthermore, we pay particular attention to the possible interactive effect between democracy and openness. Using the World Bank data on national Gini coefficients, we employ a sample of 69 countries over the period 1960– 1996, which is larger than the samples in previous studies. Finally, we perform separate analyses for developed countries (DCs) and less developed countries (LDCs). Our primary findings can be summarized as follows: democracy reduces the level of income inequality within countries; trade openness induces more equitable income distribution within countries; foreign financial capital inflows do not affect income inequality; and FDI leads to greater income inequality, but this effect is mediated and weakened in democracy. The chapter is organized as follows. The next section discusses the effects of democracy and economic openness on income inequality. The section that follows describes the empirical research design and findings from the statistical analysis. The last section of the chapter summarizes our findings and discusses their implications.

EFFECTS OF DEMOCRACY AND ECONOMIC OPENNESS ON INCOME INEQUALITY

Effects of Democracy The claim that democracy promotes an egalitarian distribution of income can be traced back to the late eighteenth century. Many scholars argue that democracy increases the opportunities for political participation, which allows the poor to demand more equitable income distribution (e.g., Chan, 1997b; Boix, 1998).3 As suffrage expands, reelection-oriented leaders are held accountable to the voters and become increasingly attuned to their needs. Democratic governments are inclined to help the lower and middle classes by adopting redistributive policies such as welfare spending, progressive taxation, minimum wage laws, price subsidies, and public works. Authoritarian leaders, in contrast, are mainly accountable to a powerful and

3

See also Bollen and Jackman (1985), Sirowy and Inkeles (1991), Lappe et al. (1998), and Human Development Report (2000).

Democracy, Economic Openness, and Income Inequality

65

rich minority. They tend to adopt public policies that benefit this minority – policies that tend to maintain or increase income inequality.4 Democracy affects the distribution of income through the process of competing pressures: the government is subject to pressure from interest groups. By promoting a more equal distribution of political power, democracy gives rise to labor unions and political parties that represent the lower and middle classes, as well as public policies that redistribute income to their constituents. The more organized and vital these groups are, the more successful they are at influencing policymaking. As described by Lenski (1966), democracy redistributes political power in favor of the majority and, therefore, leads to policies that reduce inequality.5 Previous empirical studies of the effect of democracy on income inequality exhibited several limitations. Some authors describe a few specific historical episodes, arguing, for example, that the expansion of suffrage in Western Europe in the early 1900s reduced income inequality (Lindert, 1994; Justman and Gradstein, 1999). When income inequality was measured directly in statistical studies, the data were limited. Muller (1988), for example, measured income inequality using the Gini coefficient (as we do), but only for 1970. Other scholars used larger samples but employed indirect measures of income inequality.6 The statistical evidence on the effect of democracy on income inequality is mixed. Muller (1988), Moon (1991), and Rodrik (1998) report that democracy reduces inequality. However, Bollen and Jackman (1985), Deininger and Squire (1996), and Gasiorowski (1997) report that the effect of democracy on income inequality is statistically insignificant. Chan (1997b) reports mixed findings, whereas Simpson (1990) argues that income inequality first rises with democracy and then declines after a certain threshold level of democracy.7 4

5

6

7

Some scholars argue that autocracies suppress wages to industrialize their economies (e.g., Schamis, 1991). Others note that industrialization in autocracies can also reduce inequality (e.g., Birdsall, 1998). For details, see, e.g., Lipset (1959), Muller (1988), Lindert (1989), Moon (1991), and Gasiorowski (1997). Alesina and Rodrik (1994) formalize these ideas by presenting an economic model in which the expansion of suffrage makes a poor individual the swing vote, leading to redistributive taxes that provide the poor with more income. For example, Chan (1997b) used government expenditures in areas expected to transfer income from the rich to the poor (e.g., health, housing, and education). Gasiorowski (1997) used the growth rates of industrial wages to gauge changes in income inequality. Rodrik (1998) used labor wages as a percentage of GDP. In their literature survey, Sirowy and Ingeles (1991) report that six studies find the effect of democracy on income inequality is negative and six studies find the effect is positive or statistically insignificant.

66

Democracy and Economic Openness in an Interconnected System

Effects of Economic Openness In broad terms, economic openness creates winners and losers in a society. The effects of openness on income distribution depend on the identity of the winners and losers. We discuss these effects in four categories: international trade, FDI, foreign financial capital, and international labor mobility. Trade According to the Heckscher–Ohlin model of international trade, countries export goods that use their abundant factors of production intensively and import goods that use their scarce factors intensively. Building on this model, Stolper and Samuelson (1941) predict that trade would raise the incomes of the owners of abundant factors and reduce the incomes of the owners of scarce factors.8 Because DCs are relatively well endowed with skilled labor and capital, their imports are expected to hurt their unskilled labor while their exports should benefit the capital owners and skilled labor. In contrast, LDCs are relatively well endowed with unskilled labor. Therefore, their imports should hurt their capital owners and skilled labor while their exports should benefit the unskilled labor. Hence, trade should raise income inequality in DCs and reduce it in LDCs. Wood (1994) finds evidence in favor of the Stolper–Samuelson model. Robbins (1996), however, argues that wage inequality rose in many LDCs, contrary to the prediction of Stolper and Samuelson. It is also reported that, in many LDCs, trade shifts income to the resource-intensive sectors (Inter-American Development Bank, 1998).9 The Stolper–Samuelson theorem does not exhaust the channels through which trade can affect income inequality. Rodrik (1997) argues that trade makes it easier for firms in DCs to substitute the unskilled labor with cheap imports, reducing their bargaining power and wage rates. But to the extent that trade reduces the wages of the unskilled labor, it provides incentives for workers to acquire education and for firms to employ more unskilled labor, which may reduce inequality (Blanchard, 2000). Furthermore, according to Birdsall (1998), trade intensifies economic competition, which reduces the 8 9

The Stolper–Samuelson theorem deals only with the distribution of income. It is well established in the trade literature that economic gains from trade outweigh economic losses. Slaughter and Swagel (1997) attribute wage inequality in DCs to the technological-change bias toward skilled labor, whereas Minford et al. (1997) argue that trade and technological innovation contribute equally to wage inequality. However, Rodrik (1997) argues that the evidence in favor of a technological-change driver of wage inequality in DCs is weak.

Democracy, Economic Openness, and Income Inequality

67

prices of basic consumption goods, benefiting the poor more than the rich because the poor spend relatively larger shares of their incomes on basic consumption goods. The competition also diminishes the monopoly position enjoyed by the upper class, which may reduce income inequality (Birdsall, 1998). Yet another argument is that trade increases labor productivity, which increases wages and reduces inequality (Held et al., 1999). Finally, the winners from trade could compensate the losers, reducing inequality, although such compensation typically is not offered voluntarily (Rodrik, 1997; Salvatore, 1998). Foreign Direct Investment FDI typically involves multinational corporations (MNCs), whose effects on income inequality are debated.10 Some analysts argue that FDI raises inequality. First, MNCs can pressure host governments to cut welfare expenditures and repress labor unions to reduce wages, both of which hurt the lower and middle classes. The threat of MNCs to leave the country also reduces workers’ wages by weakening their bargaining power (Nafziger, 1997; Salvatore, 1998). Second, MNCs are said to repatriate profits from LDCs, leaving LDCs underdeveloped and hurting their poor (Baran, 1973; Lall, 1974; Jenkins, 1996; Nafziger, 1997). Third, the capital-intensive techniques used by MNCs are said to create a dual economy with a small advanced sector and a large backward sector, which promotes unemployment among the unskilled laborers and distorts income distribution (Muller, 1979; Lall, 1985; Jenkins, 1996; Robbins, 1996; Nafziger, 1997). Fourth, MNCs are said to pay low wages in labor-intensive industries such as footwear and clothing, and to push domestic suppliers to follow suit in order to reduce the MNCs’ purchasing costs (Barnet and Cavanagh, 1994; Held et al., 1999). Fifth, domestic tax systems are not well suited to tax MNCs. The smaller tax base reduces government revenue and, therefore, welfare expenditures, which hurts the poor more than the wealthy (Hatzius, 1997; Human Development Report, 1999). In contrast, several studies argue that MNCs provide LDCs with capital and technology, improve their corporate governance, and propagate better management practices. These forces, in turn, raise labor productivity and 10

FDI entails the acquisition of a lasting management interest in an enterprise operating in an economy other than that of the investor, or the creation of a subsidiary of a domestic firm in a foreign country. Portfolio investments cover private transactions in securities. For details, see World Development Indicators (World Bank, 2002).

68

Democracy and Economic Openness in an Interconnected System

promote economic growth.11 It follows that FDI could eventually reduce income inequality via the Kuznets (1955) effect, which is discussed in detail in the next section. Dollar and Kraay (2000) also support this view; they argue that economic growth raises the income of the poor proportionally more than that of the rich, making FDI a useful tool in reducing poverty (Stiglitz, 1998). In this spirit, Borenszstein et al. (1994) find that MNCs promote economic growth in LDCs, and Blomstrom et al. (1992) find that FDI promotes technology transfer from DCs to LDCs. Other scholars argue that MNCs cannot easily relocate from one country to another to reduce their labor costs because lower wages also are associated with lower labor productivity. Hence, labor’s bargaining power is not necessarily diminished by FDI (Lawrence, 1994). Furthermore, the host governments can regulate the operation of MNCs, thus controlling their effects on host economies (Vernon, 1971; Kurzer, 1993). Relatively few studies investigate empirically the effects of FDI on wages in LDCs (Klein et al., 2001). They find that MNCs tend to pay higher wages for skilled labor, indirectly supporting the claim that FDI promotes income inequality by enhancing the premium on skilled labor.12 Several studies in the field of sociology argue that FDI promotes income inequality by reducing the power of labor unions to pull low-wage workers up the income ladder, thus distorting the occupational structure of LDCs and marginalizing workers displaced as a result of the capital-intensive nature of FDI.13 A couple of other studies (Bussman et al., 2005; Sylwester, 2005) find FDI and income inequality are uncorrelated, but these studies appear to suffer from important data problems. Bussman et al. (2005) use FDI stock over GDP, but as we explained at length in the appendix of Chapter 2, FDI stock data have serious measurement errors, making them inappropriate for cross-national comparisons. Also, using FDI stock, one conflates the shortand long-term effects of foreign investment. Sylwester (2005) uses net FDI inflows over GDP, but because he focuses on the change of income inequality between 1970 and 1990 and his sample is so small (29 observations), it is not surprising to find insignificant results. 11 12 13

See, e.g., Haddad and Harrison (1993), Coe et al. (1994), OECD (1994a), Blomstrom and Kokko (1996), Batra and Tan (1997), and Markusen and Venables (1999). See case studies by Aitken et al. (1996) for Mexico and Venezuela, Feenstra and Gordon (1995) and Graham and Wada (2000) for Mexico, and Mazumdar and Mazaheri (2002) for Africa. See, e.g., Sullivan (1983) and Bornschier and Chase-Dunn (1985). After empirical analysis, Tsai (1995) rejects this argument, whereas Dixon and Boswell (1996) and Alderson and Nielson (1999) support it.

Democracy, Economic Openness, and Income Inequality

69

Foreign Financial Capital Several authors argue that to attract foreign financial capital, governments engage in liberal reforms (e.g., they reduce taxation and government expenditures, privatize state-owned enterprises, and deregulate markets). These reforms may hurt labor and increase income inequality (Strange, 1996; Germain, 1997; Held et al., 1999). A related issue is that financial openness tends to bring about volatile financial flows across countries, increasing the frequency and magnitude of financial crises. These crises typically hurt the poor more than the wealthy (Human Development Report, 1999). As the financial crisis ensues, the economy enters a recession and the rate of unemployment rises while the tax base erodes, public budgets are slashed, and welfare expenditures decline. Since the wealthy can better weather recessions than the poor, the poor suffer more and income inequality increases. Some statistical evidence (Quinn, 1997) indicates that financial openness raises income inequality. A competing view argues that financial openness does not necessarily imply the end of the welfare state. Financial markets allocate funds to places that need them the most, as they pay higher rates of return (Wooddall, 1995). Foreign capital also allows countries to consume more than they produce and to invest more than they save, promoting economic development and reducing income inequality (Nafziger, 1997). Financial openness forces governments to become more prudent, efficient, and accountable to the public, because financial markets penalize corrupt, deficit-oriented, inflationary, and inefficient governments with higher interest rates and currency crises (Wooddall, 1995). Better governmental policymaking can reduce income inequality by improving tax systems, property rights, and public welfare programs (Held et al., 1999). Foreign money also can be used to fund social security. Consequently, governments can spend more of their budgets on poverty reduction (Normand, 1996). International Labor Mobility Labor mobility across countries (migration) also affects the distribution of income within countries. We consider this empirically complex topic to be outside the scope of our chapter and defer it to future research. In the past several decades, international labor flows have generally been from LDCs to DCs. Accordingly, let us consider a two-country, North–South theoretical framework. The North is endowed with more skilled labor, and less unskilled labor, than the South. The wages for skilled labor are higher than they are for unskilled labor. It follows that unskilled labor migration from South to North reduces income inequality in the South and raises it

70

Democracy and Economic Openness in an Interconnected System

in the North (O’Rourke, 2001). Contemporary migration to DCs typically involves unskilled labor from LDCs. Some immigrants from LDCs have joined the middle class in DCs. However, immigrants from LDCs typically take on the lower-paying jobs in the DCs. As a result, the unskilled workers in DCs can move to better-paying jobs (Held et al., 1999). It follows that the effect of labor migration on income inequality is theoretically ambiguous. Empirically, contemporary international labor migration generally has been restricted and small in size, and its effects on wages have been marginal and mixed.14

Hypotheses The literature argues that democracy reduces income inequality, as in Hypothesis H1, which follows. However, because some empirical studies report that the effect of democracy on inequality is not statistically significant, Hypothesis H1 may not be supported in our empirical test. In contrast to that of democracy, the overall (or net) effect of economic openness on income inequality is theoretically ambiguous. To focus our analysis, we posit directional Hypotheses H2–H4 on the effect of openness. Since the competing theories on the effects of openness all seem plausible, Hypotheses H2–H4 may be rejected in our empirical test. H1: H2: H3: H4:

Democracy reduces income inequality. Trade increases income inequality in DCs and reduces it in LDCs. FDIs reduce income inequality. Financial capital inflows reduce income inequality. EMPIRICAL MODEL AND ANALYSIS

Similar to the previous chapter, this section first presents our statistical model for the empirical analysis and then discusses several research design issues. The section then presents key results from the empirical analysis. The discussion in this section is self-contained and does not require any specific statistical expertise. Interested readers seeking further technical details should consult the chapter’s appendix. 14

For a survey of empirical results, see Ichino (1993). For results from the United States, Canada, and Australia, see Borjas (1987, 1990, 1993). Labor migration flows were much higher in the late 1800s and early 1900s than today. Data from 1870 to 1913 show that inequality fell in the originating poor European countries and rose in the destination New World countries (Williamson, 1997).

Democracy, Economic Openness, and Income Inequality

71

Empirical Model To test the effects of democracy and economic openness, we specify and estimate the following statistical model of income inequality. As noted, it provides a structure for us to guide the statistically uninitiated readers through the empirical exercise. income inequalityt = ␤0 + ␤1 tradet−l + ␤2 fdit−1 + ␤3 portfoliot−1 + ␤4 democracyt−1 + ␤5 gdppct−1 + ␤6 gdppc2t−1 + ␤7 past inequalityt−1 + εt .

(3.1)

The model specification is based on the theoretical discussion and the literature on income inequality. As in the previous chapter, we denote variables with small capital letters and their coefficients with Greek notation. Each coefficient indicates the effect of the independent variable on the dependent variable, or the phenomenon we seek to explain. The notation ε t denotes the random error that is not explained by the statistical model. The variable subscripts t and t − 1 indicate the time period of the variable, where t represents the current period and t − 1 the previous time period (a lagged variable). To simplify the presentation, we refer to the variables without the time subscripts t or t − 1. In the model, income inequality within countries is the dependent variable (the phenomenon we explain). The right-hand side of the model includes independent variables that explain income inequality. The variable income inequality represents the level of income inequality in a country in a time period. An ideal measure of income inequality is the Gini coefficient, which is computed on the basis of national income surveys. As noted, previous studies have used alternative measures of income inequality due to the lack of comprehensive Gini data. We take advantage of the income inequality data collected by Deininger and Squire (1996). However, even though their data set has the most comprehensive coverage, it contains many missing values because national income surveys are never conducted year after year. To address this problem, we follow the practice of Easterly (1999) and Higgins and Williamson (1999), and we use the decade averages of the Gini data for each country. Although the technical details of the Gini data are presented in the appendix, it is worth noting that the Gini coefficient ranges between 0 and 100, where 100 denotes perfect income inequality and zero denotes perfect equality. To apply the appropriate statistical technique, we transform the Gini coefficient into an unbounded measure.

72

Democracy and Economic Openness in an Interconnected System

To test the effects of both economic openness and democracy on income inequality in the same model, we include four key independent variables: trade, fdi, portfolio, and democracy. Corresponding to the income inequality measure, these variables take on average yearly values during each decade. As in the previous chapter, we measure economic openness by using flows of trade, fdi, and portfolio investments. The importance of these flows to a country depends on their magnitude relative to the size of the domestic economy. Accordingly, we divide each of these flows by GDP. The trade variable measures the share of trade within a country’s national economy. The variable fdi reflects the share of foreign production capital in the country’s national economy, and portfolio is the share of foreign financial capital in the country’s national economy. Also, as in the previous chapter, democracy measures the level of democracy in a country. It ranges from –10 (the most autocratic regime) to +10 (the most democratic regime) and combines information on both the democratic and the autocratic characteristics of the political regime. The model also controls for the effects of economic development and past inequality. Kuznets (1955) hypothesizes that below some level of GDP per capita, income inequality rises with GDP per capita; above this level, income inequality declines with GDP per capita. This pattern is known as the Kuznets curve. Previous empirical results on the Kuznets curve are mixed. Ahluwalia et al. (1979) and Higgins and Williamson (1999), for example, found evidence supporting the Kuznets hypothesis; Deininger and Squire (1998) found no supporting evidence. We use gdppc and gdppc2 to capture the Kuznets effect. gdppc represents GDP per capita, expressed in purchasing power parity – adjusted international dollars – and gdppc2 is gdppc squared. If the Kuznets curve is correct, the coefficients of gdppc and gdppc2 should be positive and negative, respectively. The level of education and the share of agriculture in GDP, which can also affect income inequality, are indirectly included in the model. Both variables tend to be highly correlated with GDP per capita. The level of past inequality is measured by the one-decade lagged Gini coefficient. The inclusion of this variable is consistent with the observed tendency of inequality to persist over time. Several theoretical reasons account for this tendency. First, wealth concentration typically correlates positively with political influence, generating arrangements that favor wealth owners.15 Second, people tend to marry those from the same socioeconomic group. Consequently, the children of the rich (or poor) group remain in 15

For example, in 1950, 65% of all the agricultural land in Latin America was held by 1.5% of the farm owners. This inequality has increased since (Birdsall, 1998).

Democracy, Economic Openness, and Income Inequality

73

the original group, perpetuating income differences across groups. Third, in cases where the poor and the rich belong to different ethnic groups, racial discrimination can institutionalize the current income distribution (Lewis, 1994). Fourth, education can promote upward social mobility, but acquiring education is costly. Poor people tend to have more children than rich people (Heerink, 1994). Thus, education spending per child tends to be smaller for the poor, ensuring a vicious circle. The poor remain less educated and earn less, and income inequality persists (Dasgupta, 1993).

Research Design Issues As in the previous chapter, to justify the validity of our analysis we highlight several key design issues that require some clarification. As before, the related technical details of these issues are relegated to the appendix. First, ε t in the model is the so-called error term, that is, the part of the variations in income inequality that are not explained by the model. For statistical inferences to be valid, the error term needs to satisfy some assumptions. We address the related assumption violations using appropriate econometric techniques. Second, the sample includes 69 countries over the period 1960–1996. Since economic openness and the level of democracy are generally higher in DCs than in LDCs, we examine the robustness of the results from the full sample in separate samples for DCs and LDCs. The DCs in our analysis are members of the Organization for Economic Cooperation and Development (OECD), whereas the LDCs are non-OECD countries. Third, we design our research to measure the net effects of economic openness and the level of democracy on income inequality. We do not intend to measure the strength of each of the competing forces discussed in the previous section. To assess the effect of each of the causal mechanisms would require a very large data-collection effort that would include variables such as wages, international technology transfers, government welfare expenditures, breakdown of fdi and trade by sectors, taxes, government transfer payments, and various attributes of political parties. The collection of these data is better deferred to future research. EMPIRICAL FINDINGS

Table 3.1 presents the statistical results for the full sample, the DCs, and the LDCs.16 The models exhibit reasonably good explanatory power, and the 16

The number of countries in our full sample is smaller than the number of countries in the Deininger and Squire (1996) data, due to missing data points for the other variables in our model.

74

Democracy and Economic Openness in an Interconnected System Table 3.1. Income inequality, democracy, and economic openness

DEMOCRACY TRADE PORTFOLIO FDI GDPPC GDPPC2 PAST INEQUALITY Constant Observations Adjusted R2

All countries

LDCs

DCs

−0.0125∗∗∗ (0.0033) −0.0013∗∗ (0.0006) 0.0074 (0.0166) 0.0632∗∗∗ (0.0229) −1.91e-06 (1.08e-05) 1.90e-10 (4.40e-10) 0.7181∗∗∗ (0.0590) −0.1307∗∗∗ (0.0402) 142 0.69

−0.0112∗∗∗ (0.0037) −0.0013∗∗ (0.0008) 0.0340 (0.0367) 0.0518∗∗ (0.0290) 3.92e-05∗∗ (2.20e-05) −2.98e-09∗∗ (1.28e-09) 0.7163∗∗∗ (0.0693) −0.1898∗∗∗ (0.0510) 99 0.62

−0.0125∗∗ (0.0056) −0.0026∗∗ (0.0013) 0.0059 (0.0157) 0.0590∗∗∗ (0.0218) 8.51e-07 (1.54e-05) 2.07e-10 (4.85e-10) 0.4901∗∗∗ (0.0755) −0.2718∗∗∗ (0.0682) 43 0.52

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

effects of the control variables are largely consistent with our expectations. The results provide us with confidence in the models. As before, we leave the discussion of the results of the control variables to the appendix. We begin with the result for democracy. As shown in the table, democracy reduces income inequality, and the effect is consistent in all three samples (all countries, DCs, and LDCs). When using better data of income inequality than previous studies of the effect of democracy on inequality – including economic openness in the model – and controlling for the Kuznets curve and past inequality, our analysis provides empirical support for the expected negative effect of democracy on income inequality. As Table 3.1 indicates, trade openness also reduces income inequality in all three samples. The results support Hypothesis H2, with regard to LDCs but not DCs. As discussed earlier in the chapter, trade generates both inequality-increasing and inequality-decreasing effects. Hence, our findings can be interpreted as representing the net effect of trade on income inequality within countries, which, in turn, reduces income inequality. In contrast to the effects of democracy and trade openness, fdi inflows raise income inequality in all three samples (all countries, DCs, and LDCs).

Democracy, Economic Openness, and Income Inequality

75

These results reject Hypothesis H3. Again, fdi can generate both inequalityincreasing and inequality-decreasing effects. According to our results, the net effect of fdi is to raise income inequality. In all three samples, the effect of portfolio inflow on income inequality is positive. But the effect is never different from zero statistically, meaning that the effect is too weak to be different from what one would expect due to chance. These results are consistent with the observation that the rise in portfolio investment inflows is a relatively recent phenomenon. If financial market integration continues to deepen, portfolio investments may significantly affect income inequality in the future. In any case, our results do not support Hypothesis H4. How do the statistical findings in Table 3.1 inform us about the substantive significance of the effects of democracy, trade openness, and fdi inflows? Similar to what we did in the previous chapter, we first compute a baseline Gini coefficient by setting all variables in Model 1 of Table 3.1 at their respective means. We then raise democracy, trade, and fdi by one standard deviation at a time and compute Gini again. We illustrate the sizes of these effects in two ways. First, we compute the percentage change in Gini across the two scenarios, showing the absolute influence of these three variables on Gini. Second, we compute the ratio of each of the changes in Gini over the range between the average of the Gini maxima (42.93) and the average of the Gini minima (38.98) across the 69 countries in the sample (see Appendix 3.1 for the country statistics). This computation indicates the influence of changes in trade, democracy, or fdi on Gini relative to the variability of Gini in our sample. In absolute terms, a rise in democracy by one standard deviation above its mean reduces Gini by 1.5% (from 40.56 to 39.06). A rise in trade openness by one standard deviation reduces Gini by 1.39% (from 40.56 to 39.17). A rise in fdi by one standard deviation raises Gini by 2.17% (from 40.56 to 42.73). In relative terms, these changes are 38%, 35%, and 55% over the range of Gini values between the average of the Gini maxima and the average of the Gini minima in the sample for democracy, trade, and fdi, respectively. These estimated effects of democracy, trade, and fdi on income inequality are substantial.17 17

Alternatively, one could express the sizes of these effects in terms of their shares out of one standard deviation of Gini (9.77). This would give 15.46%, 14.22%, and 22.21% for democracy, trade, and FDI, respectively, which are also substantial. A third way is to examine the size of our effects as shares of the range between the Gini sample minimum (22.46) and sample maximum (65.38). This range understates the size of our effects, because the majority of the observations in the sample are far from the tails of the Gini distribution.

76

Democracy and Economic Openness in an Interconnected System

Our results indicate that trade reduces income inequality, whereas fdi raises it. This result may reflect the fact that trade is often associated with welfare programs that compensate the losers from trade, whereas fdi typically is not associated with such programs. Additionally, the benefits of trade tend to diffuse throughout the domestic economy, benefiting all consumers. In contrast, the benefits of fdi tend to be concentrated in the industries or the individual firms (in the case of joint ventures) in which the investment takes place. Such concentration of benefits tends to increase income inequality. In summary, our results support Hypothesis H1 that democracy reduces income inequality. We find that trade openness reduces income inequality, supporting Hypothesis H2 with regard to LDCs, but not DCs. fdi increases income inequality, rejecting Hypothesis H3. Foreign financial capital inflows reduce income inequality, which does not support Hypothesis H4.

Additional Analyses We conduct several additional analyses to complement the main results reported earlier. As in the previous chapter, we report the details of these analyses in the chapter appendix. The first additional analysis is substantive and policy-relevant. In Table 3.1, we find that fdi increases income inequality and that democracy reduces it. As fdi flows into many democracies, an important policy question is whether democratic governance helps to mitigate the positive impact of fdi on inequality. If democratic institutions help reduce the negative impact of foreign capital on the host country, we uncover another benefit of spreading democracy, acquire more confidence in our ability to harness the powerful forces of economic globalization, and bring good news to the many democracies of the world. On the other hand, if democratic governance does not reduce the impact of foreign capital on income inequality, we should worry about how we could take advantage of the benefits of globalization and minimize its drawbacks. The interaction between fdi and democracy is thus an important empirical question. To evaluate this effect, we create an interaction term between fdi and democracy and reestimate the models in Table 3.1 with this additional interaction term. Table 3.A1 reports the results for these interaction models. The fdi∗ democracy interaction term does not have any significant effect in the all-countries sample or the LDC sample, but it has a significant negative coefficient in the OECD sample. Specifically, as democracy rises from −1.7 to +10 within the OECD sample, the significant total effect of fdi declines

Democracy, Economic Openness, and Income Inequality

77

from about 0.30 to about 0.057. fdi increases income inequality across the range of all possible democracy values, but its effect shrinks dramatically as we move from Greece and Spain in the early years of the OECD into the mature democracies in the OECD world. This effect, however, does not exist in many other democracies in the developing world. Second, we evaluate the robustness of the main results in Table 3.1 by using alternative measures of income inequality, excluding the lagged income inequality from the model, and employing alternative estimators. The results of these analyses are reported in Table 3.A2. Overall, the effects of democracy and economic openness remain robust. democracy reduces income inequality. portfolio investment inflows do not produce any significant effect on inequality in any of the six models. fdi inflows always raise income inequality. trade openness is correlated with less inequality in all cases, and the effect is weak in all but the case of the country fixedeffects estimator. We believe that this insignificant result reflects the noted limitations of the fixed-effects estimator. Finally, one may wonder if the effects of economic openness and democracy on income inequality are different in size between the DC and LDC samples. We estimate a statistical model to compare the effects of the variables on the two samples; the results are in Table 3.A3. The effect of democracy remains the same between the two samples, as do the effects of trade, portfolio, and fdi. Interestingly, the effects of the Kuznets curve and past inequality differ between the two samples. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Our analysis demonstrates the need to study the effects on income inequality of economic openness and democracy together. Better theoretical explanations of inequality should consider economic openness, democracy, and their interaction. Our statistical findings indicate that trade openness tends to be associated with more equitable income distribution within countries, but FDI leads to greater income inequality, and foreign financial capital inflows do not have any significant effect on income inequality. A rise in democracy reduces the level of income inequality within countries. In addition, democratic governance mediates and weakens the undesirable impact of FDI on income inequality in the OECD world but not in DCs. One may wonder why trade and FDI, often highly correlated with each other, exert different effects on inequality. It is important to note that trade and FDI do not coincide perfectly, because they are generally operated by different actors and are often driven by different costs and benefits.

78

Democracy and Economic Openness in an Interconnected System

Had they perfectly correlated, we could not have estimated their individual effects statistically. Theoretically, as discussed earlier, trade and FDI often influence the distribution of income through different causal mechanisms. Also, government policies regulating and responding to trade and FDI tend to be different. Trade adjustment assistance has a much longer history and is much more common and widespread. What are the public policy implications of our findings? To the extent that their goal is to bring about a more equitable distribution of income, international organizations such as the International Monetary Fund, the World Bank, and the World Trade Organization may condition new memberships, continued memberships, or various assistance programs on improvements of democracy. Industrialized democracies that are interested in reducing income inequality in other countries may also make liberal political reforms a condition of foreign aid. In DCs, democracy enables people to appeal to all levels of their governments, which then regulate and make sure FDI is not damaging to income distribution. In LDCs, whereas democratic governance by itself reduces income inequality, it does not restrain the impact of FDI. People in LDCs are not as effectively organized and empowered politically. As their governments invite MNCs and give them preferential treatment, it comes at the expense of the poor, causing an even more skewed income distribution. Importantly, because most FDI flows among DCs and a much smaller part goes to LDCs, the democratic DCs are in a win-win situation: they are able to enjoy the benefits of FDI and neutralize one of its bad side effects (skewing income distribution). In contrast, LDCs are likely unable to grapple politically with this side effect of FDI because their democratic institutions may be too weak to make a difference. Both proponents and critics of economic openness should reassess their claims if their arguments hinge solely on income inequality. Economic openness may improve or worsen income equality, depending on its type. Policymakers seeking a more equitable income distribution could promote international trade while compensating the losers from trade openness. On average, it appears to be the case that governments tend to compensate those who lose from trade openness, preventing income inequality from getting worse. Policymakers may also take measures to reduce the negative effects of FDI. For example, governments may offer subsidies or tax breaks to sectors or regions without FDI and curtail the occurrence of FDI-induced dual economy. At the same time, governments need to be careful not to impose restrictions that entirely drive away FDI. Although the inequality-increasing effect of portfolio investments in our results is statistically insignificant, we

Democracy, Economic Openness, and Income Inequality

79

believe that the consistently positive coefficient across samples may indicate a need for caution on the part of governments considering liberalization of this type. Finally, our results indicate that income inequality declines with economic development, but it also tends to persist over time. Governments interested in reducing income inequality need to design welfare programs that help the impoverished. Financial aid from DCs and international organizations to LDCs should be designed to reduce poverty. Regardless of the exact policies taken, we believe that reducing income inequality is important. If history is any guide, failure to reduce income inequality may well result in domestic and international political instability. SUMMARY AND OUTLOOK

Scholars have investigated the effects of economic openness and democracy on income inequality in two separate bodies of literature. In studies of democracy, scholars agree that democracy reduces income inequality, but the empirical evidence is mixed. In the literature on globalization, the effects of economic openness on income inequality are debated, but they have not been examined rigorously. To the best of our knowledge, our analysis is the first systematic statistical study of the effects of both economic openness and democracy on income inequality. We argue that the two effects should be studied together. In other words, the exclusion of one variable or the other can lead to incorrect inferences of the determinants of income inequality. Our empirical analysis comprised 69 countries over the period 1960– 1996. We have focused on the effects of trade, FDI, foreign financial capital inflows, and democracy on income inequality. We find that a higher level of democracy reduces the level of income inequality within countries. Trade openness is associated with more equitable income distribution within countries, FDI is associated with greater income inequality, and foreign financial capital inflows have no statistically significant effect on income inequality. All the empirical findings in this chapter are found to be robust across different measures of income inequality, alternative statistical estimators, and model specification. In this chapter, we investigated the relationship between economic openness and democracy, on the one hand, and income inequality, on the other. In this analysis, economic development acted as a control variable. In the following chapter, we shift our attention to the relationship between economic development and the level of democracy. Is economic development a prerequisite for democratization? Is democratization good for the

80

Democracy and Economic Openness in an Interconnected System

economy? These two questions have generated large bodies of literature focusing on either economic development or democracy as the dependent variable. Our interest in the next chapter is to analyze the potentially reciprocal relationship between democracy and development that has been largely ignored in previous research.

APPENDIX EMPIRICAL MODEL AND ANALYSIS

As before, this appendix accompanies the section “Empirical Model and Analysis” in the main text of the chapter. We provide various details on the measurement of our variables, their data sources, issues pertaining to the technical design of the statistical model, and the expected effects of various control variables. We also report and discuss all the results for the control variables as well as various additional analyses.

Empirical Model The dependent variable – income inequality within countries – is measured via the Gini coefficient, which is computed on the basis of the Lorenz curve. In this curve, the cumulative percentage of income held by shares of society appears on the y-axis and the percentage of population that holds the particular income share appears on the x-axis. The line that is 45 degrees from the origin denotes perfect income equality (e.g., 10% of the people hold 10% of national income). However, perfect income equality is never observed empirically, and so the Lorenz curve is below this line. The Gini coefficient measures the area between the 45-degree line and the Lorenz curve and is expressed as the percentage of the area between the x-axis and the line. Thus, a Gini coefficient of 100 denotes perfect income inequality, whereas a Gini coefficient of 0 denotes perfect equality.18 The Gini coefficient is an ideal measure of income inequality. As noted, previous studies have used alternative measures of income inequality due to the lack of comprehensive Gini data. We take advantage of the income inequality data collected by Deininger and Squire (1996). However, whereas their data set has the most comprehensive coverage, it contains missing values. To that effect, as in Easterly (1999) and Higgins and Williamson 18

For details, see, e.g., Nafziger (1997) and World Development Indicators (World Bank, 2002).

Democracy, Economic Openness, and Income Inequality

81

(1999), all the variables in the model are computed as decade averages. Hence, the unit of analysis is country decade.19 Deininger and Squire (1996) note that although their Gini data generally are of good quality, differences in data sources and data analysis methods across countries may cause measurement errors in a pooled sample.20 Using these data, Easterly (1999) applies all the available information from Deininger and Squire (1996) in computing the Gini decade averages, under the assumption that the measurement errors are randomly distributed and are captured in the statistical model’s error term. We use the data computed by Easterly. To examine the robustness of our Gini-based results, we also employ the share of income held by the top 20% of the national population as an alternative measure of income inequality.21 Because the Gini coefficient is bounded between 0 and 100, using ordinary least squares (OLS) regression could be problematic (OLS assumes that the dependent variable is unbounded). The usual practice is to transform the bounded variable into an unbounded indicator. We transform the Gini coefficient into an unbounded measure using the formula log[Gini/(100 − Gini)].22 We also assess the robustness of results using the untransformed Gini coefficient. As noted, we measure economic openness by using flows of trade, fdi, and portfolio investments. We divide each of these flows by GDP to capture their importance in a country’s domestic economy. The economic openness data are taken from the World Development Indicators (World Bank, 2002). As in the previous chapter, democracy is the level of democracy in a country and is computed on the basis of the POLITY III data set (Jaggers and Gurr, 1995; Gurr and Jaggers, 1999). This widely used data set provides two indices of political regime characteristics. The 10-point democracy index

19

20

21

22

The Deininger and Squire data include 111 countries, covering the period from 1890 to 1996, but the data coverage is very limited before 1960. The Gini decade averages we use were computed by Easterly (1999) and cover four decades (1960s to 1990s). These data are available at www.worldbank.org/research/growth/ddlife.htm. For example, the Gini coefficient may be computed from the level of income or the level of expenditure. In the case of income, individual welfare may be measured before or after payment of taxes. The aggregation of individual levels of welfare may vary across countries in the mixing of individual and household units. Appendix 3.1 provides the mean, minimum, and maximum values for the Gini data for each country in the sample. The country with the highest mean level of inequality is South Africa (Gini coefficient = 62.3), and the country with the lowest mean level of inequality is Hungary (Gini coefficient = 25.17). Log denotes natural logarithm. The transformed variable equals −∞ for Gini = 0 and +∞ for Gini = 100. See Pindyck and Rubenfeld (1991) for technical details.

82

Democracy and Economic Openness in an Interconnected System

(DEMOC) measures the democratic characteristics of the regime. The 10point autocracy index (AUTOC) measures the autocratic characteristics of the regime. Mansfield and Snyder (1995), Londregan and Poole (1996), and others observed that many governments may have both democratic and autocratic characteristics. Consequently, these studies measure the level of democracy as the difference between DEMOC and AUTOC, a practice we also adopt here. The democracy measure ranges between −10 (the most autocratic regime) and 10 (the most democratic regime). In the model, gdppc is GDP per capita, expressed in purchasing power parity–adjusted international dollars.

Research Design Issues Several design issues require further clarification. First, we employ a pooled time-series, cross-sectional research design. The sample includes 69 countries over the period 1960–1996. Statistical models for pooled time-series cross-sectional data may exhibit heteroskedasticity and serial correlation. Although these problems do not bias the estimated coefficients, they could result in inefficient and biased standard errors for the coefficients. To deal with these potential problems, we estimate the model using OLS regression with Huber–White robust standard errors clustered over countries. These estimated standard errors are robust both to heteroskedasticity and to a general type of serial correlation (Wiggins, 1999). The inclusion of the lagged dependent variable in the model helps to absorb temporal dependence in the data, as shown by Beck and Katz (1995a, 1995b). Second, in addition to the substantive reasons noted in the main text, the inclusion of past inequality in the model helps to control for the effect of potentially relevant but omitted structural variables, such as the ethnic and demographic structures of society. Many studies adopt this modeling strategy to prevent spurious findings (e.g., Bollen, 1979; Burkhart and Lewis-Beck, 1994; Muller and Seligson, 1994; Muller, 1995). Third, as noted, we design our research to measure the net effects of economic openness and the level of democracy on income inequality. As is typically done, the statistical significance levels of the estimated coefficients are investigated with a one-tailed t-test since our hypotheses are signed. Fourth, we need to clarify how the size of the effect is computed for the statistically significant key variables. The coefficient sizes cannot be interpreted linearly from Table 3.1, because the dependent variable is a nonlinear transformation of Gini. As noted, we first compute a baseline Gini value by setting all variables in Model 1 of Table 3.1 at their respective means.

Democracy, Economic Openness, and Income Inequality

83

We then raise democracy, trade, and fdi by one standard deviation at a time and compute Gini on the basis of the following equation: log[Gini/(100 − Gini)] = a + bx, where a is a constant, b is a vector of coefficients, and x is a vector of independent variables. We solve for Gini in terms of a, b, and x. Next we illustrate the sizes of these effects in two ways, as noted in the main text. EMPIRICAL FINDINGS

Table 3.1 presents the estimated coefficients and their standard errors from the full sample, the DC sample, and the LDC sample.23 All the models in Table 3.1 have a relatively good fit to the data, with the adjusted R2 ranging from 0.52 (in the DC sample) to 0.69 (in the all-country sample). The joint significance level of the model’s independent variables from the F-test is statistically significant at a level better than 1% in all the samples, which supports our model specification.24 The effect of democracy on income inequality is statistically significant at the 5% level for the DC sample and at the 1% level for the LDC and all-country samples. The effect of trade openness on income inequality is negative and statistically significant at the 5% level for all the samples, indicating that trade openness reduces income inequality. The effect of FDI inflows on income inequality is positive and statistically significant at the 5% level for the DC and LDC samples and at the 1% level for all countries. The effect of portfolio inflow on income inequality is positive in all samples, but it is never statistically significant. The effects of gdppc and gdppc2 on income inequality are statistically significant in the LDC sample at the levels of 5% and 1%, respectively. gdppc has a positive effect on income inequality, whereas gdppc2 has a negative effect. These results support the existence of a Kuznets curve for LDCs. In the DC and all-country samples, the Kuznets curve is not statistically significant. The insignificance of the Kuznets curve for DCs is to be expected. The Western European countries experienced a Kuznets curve transformation 23

24

The number of countries in our full sample is smaller than the number of countries in the Deininger and Squire (1996) data due to missing data points for the other variables in our model. We also check for high collinearity using the variance inflation factor. We find no high collinearity in the sample of LDCs, but high collinearity in the DC and all-country samples. Using the matrix of variance decomposition, we identify the sources of high collinearity to be gdppc and gdppc2 . These variables are correlated by construction, a quality shared by all models that include the Kuznets curve. Nevertheless, this result suggests caution in interpreting the Kuznets curve coefficients for the DC and all-country samples.

84

Democracy and Economic Openness in an Interconnected System Table 3.A1. Interactive effect of democracy and FDI on income inequality

DEMOCRACY FDI FDI∗ DEMOCRACY TRADE PORTFOLIO GDPPC GDPPC2 PAST INEQUALITY Constant Observations

All countries

LDCs

DCs

−0.0123∗∗∗ (0.0040) 0.0637∗∗∗ (0.0266) −0.0001 (0.0022) −0.0014∗∗ (0.0007) 0.0074 (0.0167) −1.92e-06 (0.00001) 1.91e-10 (4.42e-10) 0.7183∗∗∗ (0.0583) −0.1305∗∗∗ (0.0407) 142

−0.0097∗∗ (0.0053) 0.0545∗∗ (0.0299) −0.0011 (0.0027) −0.0013∗ (0.0008) 0.0352 (0.0378) 0.00004∗∗ (0.00002) −3.07e-09∗∗ (1.38e-09) 0.7181∗∗∗ (0.0684) −0.1889∗∗∗ (0.0516) 99

0.0041 (0.0130) 0.2680∗∗ (0.1063) −0.0211∗∗ (0.0105) −0.0026∗∗ (0.0013) 0.0063 (0.0159) 2.85e-06 (0.000015) 1.41e-10 (4.71e-10) 0.4870∗∗∗ (0.0718) −0.4523∗∗∗ (0.1176) 43

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

in the late nineteenth and early twentieth centuries (Kuznets, 1955). Since this period is not included in our sample, we do not detect a Kuznets curve effect for DCs. The effect of past inequality on current inequality is statistically significant at the level of 1% in all samples. The positive sign of this effect corroborates our expectation that income inequality exhibits inertia. The realization of equitable income distribution is a lengthy process; hence, one should take a long-run view of the dynamics of income inequality. The relatively large size of the inertia effect in our results suggests that models of income inequality that do not control for this effect may omit an important variable.

Additional Analyses The first additional analysis concerns the interactive effect of democracy and fdi on income inequality. Table 3.A1 presents the results. The total effect of fdi is (0.0637 − 0.0001∗ democracy) for all countries, (0.0545 − 0.0011∗ democracy) for LDCs, and (0.2680 − 0.0211∗ democracy) for

Democracy, Economic Openness, and Income Inequality

85

DCs. We test the significance of the total effect of fdi across the different values of democracy for each sample. The total effect of fdi is significant only in the DC sample throughout the values of democracy. The second additional analysis focuses on the robustness of our results in Table 3.1. We use alternative measures of income inequality, exclude the lagged income inequality from the model, and employ alternative estimators. The results for the all-country sample are presented in Table 3.A2. The results for the LDC and DC samples are consistent with those in Table 3.A2 and are not reported here.25 Overall, the effects of democracy and economic openness in Table 3.1 are replicated and robust across the six experiments reported in Table 3.A2. The models presented in Table 3.1 employ a transformed Gini coefficient as the dependent variable. To assess the robustness of these results, we use alternative measures of inequality. In Column 1 of Table 3.A2, income inequality is measured as the share of national income held by the top 20% of the population. The data for this particular measure come from the studies of Deininger and Squire (1996) and Easterly (1999). In Column 2 of Table 3.A2, we use a nontransformed Gini coefficient, as in Easterly (1999) and Muller (1988).26 The inclusion of the lagged Gini coefficient in the models presented in Table 3.1 may reduce the variance to be explained by other variables. In column 3 of Table 3.A2, we exclude the lagged Gini coefficient for a robustness check. Recall, however, that the lagged Gini is part of our theoretical model specification. The results reported in Table 3.1 are generated from OLS regression with robust standard errors. Columns 4–6 of Table 3.A2 present the results from two other estimators: fixed effects with robust standard errors, and random effects. We employ two versions of the fixed-effects estimator, introducing country or decade dummies into the model. The fixed-effects estimator controls for the possibility that some countries or decades are marked by large income inequality. As noted in the literature, the fixed-effects estimator has serious limitations. First, the country and decade dummy variables are atheoretical and absorb many of the variations in the dependent variable that are attributable to the model’s independent variables.27 Second, the country dummy variables and other country-dependent variables (e.g., GDP per 25 26 27

Statistical results for the LDC and DC samples, as well as the by-country statistics for Gini and the sample descriptive statistics, are available from the authors on request. As noted, the transformation of the Gini coefficient is needed, since it is bounded. However, this transformation may introduce nonlinear effects and so is not used in some studies. As summarized by King (2001), most methodologists agree that using better models, better measures, and robust estimation is preferable to fixed-effects estimation.

86

−0.3004∗∗∗ (0.0797) −0.0316∗∗ (0.0153) 0.2196 (0.3847) 1.4552∗∗∗ (0.5495) −1.50e-05 (0.0003) 3.37e-09 (1.05e-08) 0.7141∗∗∗ (0.0629) 11.0658∗∗∗ (3.2733) 142 0.68

−0.0032∗∗∗ (0.0010) −0.0002∗∗ (0.0001) 0.0041 (0.0035) 0.0103∗∗∗ (0.0029) 2.15e-06 (2.18e-06) −7.47e-11 (9.12e-11) 0.7352∗∗∗ (0.0521) 0.1166∗∗∗ (0.0252) 104 0.75 −0.2021∗∗ (0.0950) 142 0.25

−0.0125∗∗ (0.0073) −0.0014∗ (0.0010) −0.0014 (0.0233) 0.0884∗∗∗ (0.0318) −4.19e-05∗ (2.54e-05) 8.61e-10 (1.06e-09)

Without past inequality −0.0111∗ (0.0076) −0.0025 (0.0026) 0.0096 (0.0157) 0.0586∗∗ (0.0335) −3.96e-05∗ (2.62e-05) 1.27e-09∗ (8.11e-10) 0.0290 (0.1839) −0.1399 (0.1323) 142 0.92

Country fixed effects

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Observations Adjusted R2

Constant

PAST INEQUALITY

GDPPC2

GDPPC

FDI

PORTFOLIO

TRADE

DEMOCRACY

Untransformed Gini

Top 20% income −0.0130∗∗∗ (0.0036) −0.0013∗∗ (0.0006) 0.0040 (0.0151) 0.0566∗∗∗ (0.0223) 2.78e-06 (1.42e-05) −4.89e-11 (5.81e-10) 0.7266∗∗∗ (0.0588) −0.1337∗∗∗ (0.0505) 142 0.71

Decade fixed effects

Table 3.A2. Income inequality, democracy, and economic openness (all countries)

−0.0147∗∗∗ (0.0043) −0.0012∗∗ (0.0007) 0.0057 (0.0188) 0.0686∗∗∗ (0.0187) −5.02e-06 (1.56e-05) 3.17e-10 (6.28e-10) 0.6033∗∗∗ (0.0616) −0.1635∗∗∗ (0.0560) 142 0.70

Random effect

Democracy, Economic Openness, and Income Inequality

87

capita) tend to be collinear. The random-effects estimator parameterizes the error term associated with different cross sections and is useful for cross-nationally dominated panels.28 Table 3.A2 suggests that the results from Table 3.1 are robust against the exclusion of the lagged dependent variable, alternative estimators, and alternative measures of income inequality. In all cases, the effect of democracy is statistically significant and negative, reducing income inequality. The effect of portfolio investment inflows is insignificant in all six models, as in Table 3.1. The effect of fdi inflows is always statistically significant and positive, again as in Table 3.1. The effect of trade openness is negative in all cases, reducing inequality, and is statistically significant in all but the case of the country fixed-effects estimator. We believe that this insignificant result reflects the noted limitations of the fixed-effects estimator. The control variables in Table 3.A2 replicate the spirit of Table 3.1. The effect of past inequality is positive in the five cases reported and is significant in all cases except for the country fixed-effects estimator. The effects of gdppc and gdppc2 are consistent with Table 3.1 in four of the six cases. Columns 3 (without past inequality) and 4 (country fixed effects) are the exceptions. The weaker results obtained without past inequality support our model specification. An argument against using the lagged dependent variable is that it causes insignificant results. Our results from the model including the lagged dependent variable, however, are more significant than those from a model without this variable. Since we also have theoretical reasons for including past inequality in the model, we believe Column 3 suffers from model misspecification bias. In Column 4, the effects of gdppc and gdppc2 are significant, but their signs are not consistent with the Kuznets curve. We believe this weaker result reflects the aforementioned limitations of the fixed-effects estimator.29 Finally, one may question whether the effects of economic openness and democracy on income inequality are statistically different between the DC and LDC samples. To answer this question, we constructed two variables for each independent variable in the original model. For example, instead of one democracy variable, we now have two separate variables: oecd democracy and ldc democracy. We estimated a full model of all variables thus created 28

29

An alternative estimator is the panel corrected standard error (PCSE; Beck and Katz, 1995b). Because our sample is cross-nationally dominated with only four decades, the PCSE estimator is not suitable for our application. We thank Neal Beck for this comment. We have also estimated the model in Table 3.1 with a dummy variable for communist regime status, because these countries tend to emphasize income equality. The results for the hypothesis testing are similar to those reported in Table 3.1.

88

Democracy and Economic Openness in an Interconnected System Table 3.A3. Effects of democracy and economic openness in DCs and LDCs

DCs DEMOCRACY TRADE PORTFOLIO FDI GDPPC GDPPC2 PAST INEQUALITY Constant Observations Adjusted R2 Note:

∗

−0.0142∗∗∗ (0.0047) −0.0027∗∗ (0.0012) 0.0052 (0.0151) 0.0591∗∗∗ (0.0205) −8.36e-06 (0.00001) 5.20e-10 (4.71e-10) 0.4868∗∗∗ (0.0756) −0.1945∗∗∗ (0.0490) 142 0.69

LDCs −0.0113∗∗∗ (0.0038) −0.0012∗ (0.0008) 0.0346 (0.0370) 0.0516∗∗ (0.0293) 0.00004∗∗ (0.00002) −3.06e−09∗∗∗ (1.27e-09) 0.7155∗∗∗ (0.0703)

Coefficient equality test 0.26 1.02 0.54 0.04 5.10∗∗ 8.74∗∗∗ 4.89∗∗

significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

and report the results in Table 3.A3. We then tested whether the coefficient of, say, oecd democracy equals that of ldc democracy, and we did so for each variable. These equality test results are in the last column of Table 3.A3. The test results indicate that the sizes of the effects of economic openness and democracy on income inequality, although statistically significant in both DC and LDC samples, are not significantly different across the two samples.

FOUR

Democracy and Development

INTRODUCTION

The effect of democracy on economic development has captivated many thinkers. De Tocqueville (1835) and Schumpeter (1942), for example, believe democracy provides the social securities required for development. Hayek (1944) argues that democracies exhibit relatively less unrest and political instability, promoting development. Friedman (1962) asserts that democracy promotes development by keeping state power in check. North (1990) expects autocrats to prey on their subjects, and Olson (1993) argues that autocrats are corrupt and promote their cronies, which are bad for development. Going a step further, Fukuyama (1992) argues that capitalism and democracy work in unison to promote welfare, a view many observers share (e.g., The Economist, 1994; Cohen, 2007). In contrast, Marx (1867, 1871) and Lenin (1911) argued that democracy came out of capitalist development as a tool for the elite to deprive and appease the masses. John S. Mill (1860), a supporter of democracy, is skeptical about its ability to promote development. Huntington (1968) argues that democracies exhibit high state expenditures in response to their citizens’ demands, reducing the excess that is expendable for development. Olson (1982) argues that special interests in democracy shape public policy to promote their own interests, undermining the economy. Wade (1996) argues that strong autocrats can resist popular pressure for policies to save doomed industries, which is good for development. Barro (2000) and Bueno de Mesquita and Downs (2005) argue that autocrats can effectively pursue prodevelopment policies. Despite the controversy, the view that democracy promotes development has recently become prevalent in the policy circles of the developed world. For example, a number of senior practitioners recently argued that poor 89

90

Democracy and Economic Openness in an Interconnected System

countries undertaking democratic reforms outperform poor autocracies in development. The United States should, therefore, target its aid to help countries that attempt democratic reforms. President Bush, the authors note, supports having democratic governance “as qualifying criteria for countries to receive assistance” (Siegle et al., 2004). Similarly, Bush’s Under Secretary of State Dobriansky argues that solid democratic institutions are the basis for economic development (U.S. Department of State, 2007), and Bush himself apparently believes that political and economic liberties, both of which promote social welfare, are the natural by-products of each other (Cohen, 2007). Democracy, as a regime favoring equal political rights and civil liberties, is definitely worthy of promotion. We have seen that many writers and policymakers justify the promotion of democracy as a means to encourage economic development, but does a rise in democracy really lead to a higher level of development? The argument that democracy promotes development considers neither the possibility that democracy could undermine development nor the possibility that the causality may flow from development to democracy. In fact, as we show in the next section, the effect of development on democracy has received ample scholarly attention. The issue of causation stands at the center of this chapter. We argue that sorting out the causal direction between democracy and economic development is important. For example, if the goal is to promote democracy but the causation goes from development to democracy, we should not condition the distribution of aid to poor countries on instituting democratic reforms; rather, we should condition aid on forces such as economic reforms that encourage development or on the quality of the projects themselves. The relationship between democracy and development raises two salient questions. Does development lead to democracy? Does democracy encourage or undermine development? Together, these questions imply that the two forces may affect each other in a reciprocal manner. But statistical studies typically employ single-equation models, which specify either democracy as the dependent variable and development as the key independent variable, or development as the dependent variable and democracy as the key independent variable. As several scholars noted recently, these single-equation models cannot address the possible reciprocal relationship between democracy and development, and their inferences may be misleading (Chan, 2002; Midlarsky, 2002; Przeworski and Limongi, 2003).1 1

Chan (2002: 118) warns that even if some results indicate that democracy boosts development, “the relationship may well be the other way around [development boosts democracy].” Midlarsky (2002: 672) calls “for the greater use of structural equations modeling” in the

Democracy and Development

91

The development–democracy reciprocity issue has so far received very limited attention. Burkhart and Lewis-Beck (1994) tested the causality between democracy and energy consumption per capita by using Granger’s method, but they did not specify a simultaneous equations model (SEM). Przeworski et al. (2000) used dynamic probit to study the effect of development on regime transition and a selection-corrected estimator to assess the effect of regime type on economic growth, but they did not specify a full SEM. Helliwell (1994) and Feng (2003) employed SEMs, but their analyses raise a number of concerns, to which we will return. Given the significance of and the inadequate attention to the issue, further analysis of the simultaneity between democracy and development is warranted. Some readers may view this issue as a methodological nuisance, not a theoretical problem. We consider it to have both theoretical and methodological implications. If democracy and development affect each other, unidirectional models are theoretically incomplete and, as Alvarez and Glasgow (2000) noted, will produce biased estimates. We construct a SEM of the levels of development and democracy. By employing this approach, we are able to decide which of the following four theoretically distinct possibilities holds: (1) development affects democracy, but not the other way around; (2) democracy affects development, but not the other way around; (3) the two forces affect each other; and (4) the two forces are unrelated. We also identify the signs and sizes of the statistically significant effects. We estimate this model using a large-N sample of countries across nearly four decades. The empirical results suggest that a rise in development in a country increases its democracy, whereas a rise in democracy reduces development. These findings have important policy implications. The rest of this chapter is organized as follows. The next section discusses the theoretical arguments concerning the relationship between democracy and development, and the following section provides an overview of previous empirical efforts and raises several important but overlooked design issues. The fourth section discusses our statistical model and presents the main empirical findings. The last section summarizes the findings and explores their implications. As in previous chapters, technical details and additional results are delegated to the chapter appendix. development–democracy literature. Przeworski and Limongi (2003: 348) argue that “inferences based on standard [single-equation] regression models [of democracy and development] are invalid.”

92

Democracy and Economic Openness in an Interconnected System

THEORETICAL ARGUMENTS

This section provides the theoretical context for our analysis rather than an exhaustive review of two very large bodies of literature. We demonstrate that a large number of arguments exist for both claims: economic development influences democracy, and democracy affects development. Acknowledging these plausible arguments, one should entertain the theoretical possibility that development and democracy affect each other simultaneously.

Causality from Development to Democracy The hypothesis that economic development promotes democracy has received wide attention in the literature. Lipset traces this so-called modernization thesis back to the Greek philosopher Aristotle. “From Aristotle down to the present,” he writes, “men have argued that only in a wealthy society in which relatively few citizens lived in real poverty could a situation exist in which the mass of the population could intelligently participate in politics and could develop the self-restraint necessary to avoid succumbing to the appeals of irresponsible demagogues” (1959: 75). Development arguably promotes democracy through several channels. First, as economy develops, the need for an educated labor force rises and people can afford to spend more on education. Educated people, in turn, often demand greater government transparency, political freedom, and participation in policymaking (Diamond et al., 1987; Lipset et al., 1993), which are all elements of democracy. Educated people also tend to be more receptive to the idea of resolving political disagreement through nonviolent, majority-based decision making. They also demand freer flows of information, more mass media, and communication outlets, which assist the diffusion of democratic ideas (Lipset, 1959; Dahl, 1989; Diamond and Plattner, 1994). In a second channel, economic development promotes civic society and socioeconomic progress, which increase the power of the middle class at the expense of the ruling elite and constrain authoritarianism (de Tocqueville, 1835; Lipset, 1959; Huber et al., 1993; Putnam et al., 1993). Also, as the number of urban dwellers rises, the masses can more easily get organized to demand political rights (Lipset, 1959; Lipset et al., 1993). Third, underdevelopment and poverty cause discontent, accentuate grievances, and polarize societies. These tensions increase the attractiveness of violence as a means to secure survival and acquire social mobility, which can increase political instability and even lead to civil war. In

Democracy and Development

93

trying to assert power, the state may become oppressive, reducing democracy. Economic development promotes democracy by alleviating these tensions (Im, 1987; Huntington, 1991; Haggard and Kaufman, 1995). A fourth channel considers the appeal of autocracy to the elite. Political leaders constantly contemplate whether to accept the outcome of a democratic competition that mandates the distribution of the national pie or to fight to establish autocratic control so as to acquire a larger share of the pie. Assuming the marginal utility of consumption and the marginal return of capital decrease with affluence, the attractiveness of dictatorship is expected to wane when affluence rises. Also, as the economy and civil society develop, defeated politicians have more opportunities for respectable employment, enabling their acceptance of electoral defeats. Because economic development increases affluence and supplies more acceptable alternatives, dictatorship becomes less attractive as a political solution to distributional conflicts (Lipset, 1959, 1981; Przeworski et al., 2000). Finally, poverty and underdevelopment erode public confidence in the ability of nascent democracies to resolve problems, reducing the legitimacy of the regime and increasing the appeal of authoritarianism to the public. When developmental problems become salient and pressing, the democratic government may lose public support altogether, raising the likelihood of a forceful transition to autocracy. Development, in contrast, alleviates these pressures and enables democracy to take hold and consolidate (Linz, 1978; Seligson and Muller, 1987; Huntington, 1991; Haggard and Kaufman, 1995). The positive effect of development on democracy, however, is not the only possible outcome. Development may delay democratization in two ways. First, economic development may strengthen an authoritarian government (Bueno de Mesquita and Downs, 2005); development provides more financial resources for the autocrat, not only the middle class, particularly when the government owns some means of production. Because a financially stronger government can sustain a larger security force and develop better monitoring and surveillance methods, it becomes more capable of suppressing the grassroots efforts toward democracy. Second, as long as the economy develops strongly, the public in an autocracy may not always be interested in changing the status quo. If the public values robust economic development more than democratic norms such as freedom of speech, freedom of association, or freedom of political participation, economic development may serve to maintain the autocratic regime. Some stylized observations support the argument that development may not bring about democratization. China, for example, has been vigorously

94

Democracy and Economic Openness in an Interconnected System

developing its economy since the late 1970s, but it remains a one-party state. Since the late 1990s, Russia has achieved considerable economic development, but various reports suggest its level of democracy is on the decline. Strong developments in Japan from the late nineteenth century to 1945 and in Germany from 1933 to 1945 coincided with highly undemocratic yet popular regimes. Finally, despite Singapore’s stellar economic development record, the country is still autocratic.

Causality from Democracy to Development Theoretical considerations suggest the effect of democracy on development can be negative or positive. Starting with the negative impact, one channel argues democracies are bogged down by special interests and rentseeking, which hinders development. Special interests use the democratic political process to redistribute wealth from the public to themselves, and rent-seeking allocates resources to unproductive activities. A developmentoriented autocrat may restrain the overbearing rent-seeking interests (Olson, 1982; Lal, 1983; Jackman, 1993). In this vein, regimes based on alliances between the military and technocrats may be efficient in promoting development, because the technocrats devise required policies and the determined military regime implements them (O’Donnell, 1979; Cohen, 1985). In a second channel, free people may be reluctant to curtail their consumption in order to save. Low savings, in turn, leads to low investments, which slows down development. Democratic governments, depending on the electoral support to stay in office, are often afraid to impose unpopular measures to increase savings and investments. Depending relatively less on public support, autocrats can implement unpopular policies to boost savings and investments (Hewlett, 1979; Rao, 1985). Third, democratic governments tend to spend much on social welfare programs because the public values them. Massive government expenditures, in turn, crowd out resources for development projects. Democracies may also have to confront strong populist pressures to redistribute wealth from the rich to the poor in various ways, reducing the willingness of the wealthy to save and invest. Authoritarian leaders may be better able to ward off these pressures, encouraging economic development (Haggard, 1990; Barro, 2000). In a fourth channel, democracies are less able to suppress social unrest than autocracies because they have to protect civil liberties. Social unrest, in turn, slows down production, undermines effective business planning, and

Democracy and Development

95

scares away investors. Because it is more capable of enforcing social order, an autocratic regime can promote economic development better than a democratic government (Pye, 1966; Hewlett, 1980). Finally, less developed countries (LDCs) feature agrarian societies, high income inequality, youth-oriented demographics, low levels of technology, underdeveloped financial markets, and dependence on international capital and technology. In such situations, the state necessarily must play an active and interventionist role in promoting development, suggesting that authoritarian decision making may outperform the diffused democratic process (Chirot, 1977; Cohen, 1985). In contrast, several arguments indicate that democracy can facilitate development. First, political freedom leads to economic freedom. Free markets, in turn, allocate financial resources and factors of production to their most efficient use, bringing about development (Goodin, 1979; Goodell and Powelson, 1982). Second, democracy respects individual rights, including private property rights. Secure property rights, in turn, are critical for trade, savings, and investments, which promote development (Goodell and Powelson, 1982; North, 1990; Olson, 1993; Clauge et al., 1996; Li and Resnick, 2003). Autocrats may also secure property rights, but democratic leaders are more likely to do so since they are subject to checks and balances (North and Weingast, 1989). Autocrats may abruptly change the rules of the game, increasing uncertainty and, therefore, the willingness to invest. Autocrats may also be more predatory than democratic leaders, because they have fewer incentives to consider public welfare, as opposed to their own interests and those of their cronies (Goodell, 1985; Findlay, 1990; Olson, 1991). Third, because democratic regimes are more responsive to public concerns and more likely to solve problems through political compromise, they may experience less domestic unrest than autocracies. Because democratic governments represent the public, they may seek to achieve relatively more equitable income distributions than autocracies. All these forces work to defuse public grievances and reduce the likelihood of social strife and revolution. Political stability, in turn, leads to a longer time horizon for economic activities, increasing the allocation efficiency and investments required for development (Przeworski, 1985; Barro, 2000; Feng, 2003). Fourth, autocrats’ military spending needs to be relatively more excessive to sustain control over their population. Consequently, they need to extract more resources from society through tax or pillage, which hinders economic development. In addition, as autocrats allocate most of their energy to

96

Democracy and Economic Openness in an Interconnected System

maintaining social order and guarding against possible coups that might be staged by their opponents or the public, they are less likely to be effective at modernizing their economies (Nordlinger, 1970; Bienen, 1971). Finally, the autocrat, who requires the support of the ruling elite, may be subject to rent seeking, corruption, and patron–client relationships, all of which undermine development. Whereas autocrats have more decisionmaking power, they still make bad decisions. For example, capital-intensive development strategies, which autocrats have often chosen, may produce dual economies in LDCs, which is detrimental for development. In contrast, rent-seeking in democracy need not be as severe as in autocracy, and democratic deliberation does not necessarily lead to bad public policy (Goodin, 1979; King, 1981; Nafziger, 2006). THE EMPIRICAL LITERATURE

The previous section discussed various theoretical conjectures on the effects of democracy and development on each other. Taken together, these arguments suggest that democracy and development affect each other simultaneously. Moreover, because all the preceding arguments are theoretically plausible, the net effects of development and democracy on each other could be positive, negative, or insignificant. To the extent that theory cannot definitively calibrate the strengths of the causal forces, the exact nature and strength of the reciprocal relationship between democracy and development is an empirical issue. This section overviews empirical findings in the literature and discusses a number of critical design issues. In general, some stylized facts support Lipset’s (1959) modernization thesis. For example, today the developed countries (DCs) are all democratic, and those DCs with the most recent history of autocratic rule – Greece, Portugal, Spain, and Turkey – are also the least developed among the DCs. In Taiwan, South Korea, and several other newly industrialized countries, a move toward greater democracy followed economic development, whereas some of the worst democratic records are found in LDCs. That said, stylized facts rejecting the modernization thesis also exist (e.g., China and Russia) and, more generally, stylized facts may not generalize across many countries. Statistical studies of democracy typically employ single-equation models. Development is typically measured by real gross domestic product (GDP) per capita (RGDPPC) and democracy is typically measured as a continuous variable. Most of these studies find that a rise in RGDPPC increases

Democracy and Development

97

democracy.2 Other studies model democracy as a discrete event where democracy grows above some predefined threshold. The results of these studies are somewhat mixed.3 Turning to the reverse causality, analysts study the net effect of democracy on the growth rate of RGDPPC or GDP. The findings are mixed. In Borner et al.’s (1995) review, 10 studies find an insignificant or mixed effect of democracy on growth, 3 find a negative effect, and 3 find a positive effect. In Kurzman et al.’s (2002) survey, 19 studies report a positive effect, 6 report negative effects, 10 report insignificant effects, and 9 present mixed results. In Feng’s (2003) review, 4 studies find insignificant effects, 6 find positive effects, and 6 find negative effects. More recently, Krieckhaus (2004) also reports mixed results. Consider next the issue of simultaneity. If development and democracy indeed affect each other, the single-equation analysis is problematic and likely produces biased results. So far, only two studies have addressed this issue using a SEM, but their findings and designs differ. Helliwell (1994) found that a rise in real income per capita promotes democracy, whereas a rise in democracy does not affect economic growth. Feng (2003) found that a rise in economic growth does not affect democracy, whereas a rise in democracy promotes economic growth. The models of Helliwell and Feng raise three design issues. First, Helliwell uses the 1960 values as instruments for endogenous variables in 1985, but a SEM requires instruments based on all the exogenous variables in the model. Second, both models are cross-sectional, where variables are averages over more than two decades. However, as Feng notes, this design cannot study change over time, which is important in the relationship between democracy and development. Third, both studies focus on the economic growth rate, not the level of development, but the modernization thesis stresses the importance of the level of development in democracy, not growth. If one accepts the premise that democracy and development may affect each other, the democracy equation ought to include the level of development on the right-hand side, and the development equation should employ this variable as the dependent variable to maintain simultaneity 2

3

The findings of Lipset et al. (1993), Burkhart and Lewis-Beck (1994), Gasiorowski (1995), Londregan and Poole (1996), and Li and Reuveny (2003) support the modernization thesis. For contrary evidence, see Gonick and Rosh (1988) and Colaresi and Thompson (2003). For example, Przeworski et al. (2000) find that development does not affect democratic transition but promotes democratic consolidation post-transition. Boix and Stokes (2003) find that development promotes transition and consolidation.

98

Democracy and Economic Openness in an Interconnected System

between the two equations.4 Also, we note that growth and development tell us different things. An economy can grow fast and be underdeveloped (e.g., China), whereas DCs can exhibit slow growth (e.g., Japan). Growth may change erratically, but development changes slowly. Furthermore, growth and change in development also differ. Using RGDPPCt for development in time t, the change in development is RGDPPCt − RGDPPCt−1 , where t − 1 is the previous period. Growth is (RGDPPCt − RGDPPCt−1 )/RGDPPCt−1 . Economies may display the same change in RGDPPC but grow at different rates due to different values of RGDPPCt−1 in the denominator of the growth expression. As we demonstrate clearly in the chapter appendix, inferences based on a growth–democracy SEM differ from those based on a development–democracy SEM. We believe that any SEM model of the development–democracy nexus needs to seriously address these design issues. EMPIRICAL MODEL AND ANALYSIS

Although theoretical expectations indicate a possible simultaneous relationship between democracy and development, the empirical evidence obtained so far is mixed. The designs of previous studies that attempted to address the issue of simultaneity raise concerns, but we think their logic in recognizing the need for a SEM is sound. This section presents our statistical SEM, related research design issues, and main empirical findings. As in previous chapters, technical details of the model, measures, and data sources are presented in the chapter appendix.

Empirical Model We first need to consider the measurement of democracy as it affects our modeling strategy. The issue is not which features of democracy or autocracy should be recorded – which is important in its own right – but whether to model democracy as a discrete event indicating that democracy exists only above some threshold or as a continuous variable that ranges from full autocracy to full democracy. The discrete measure leads to models of the probability of democratic transition, whereas the continuous measure 4

If Y and Z are endogenous, the Y equation in a SEM needs to include Z on the right-hand side and the Z equation needs to include Y on the right-hand side. Specifying a democracy equation as a function of development and an economic growth equation as a function of democracy does not create a SEM.

Democracy and Development

99

leads to models of the level of democracy. We seek to study the reciprocal relationship between the level of development and the level of democracy rather than the event of meeting a predefined threshold for democracy, because development may interact with the whole spectrum of democracy, not just with the discrete event of democratic transition.5 We specify and estimate a SEM to test the net effect of democracy on development and the net effect of development on democracy. The model distinguishes between endogenous and exogenous (or control) variables. An endogenous variable appears on the left-hand side of one equation and the right-hand side of the other. The exogenous variables appear only on the right-hand side of each equation. As in the previous chapters, small capital letters denote variables, and Greek letters denote coefficients to be estimated empirically. A coefficient indicates the effect of the independent variable on the dependent variable. The notations ␣c and ␣t denote vectors of coefficients. The Greek letters εt and ut represent the random errors in the two equations, respectively. The subscripts t and t − 1 denote the current and previous time periods (or years), respectively. We lag the exogenous variables for reasons discussed in the appendix. To simplify presentation, we refer to variables in the text without the time subscripts. Equations (4.1) and (4.2) present the development and democracy equations of the SEM, respectively, specified based on the democratization and development literature. RGDPPCt = ␣0 + ␣1 democracyt + ␣2 prior rgdppct−1 + ␣3 investmentt−1 + ␣4 population growtht−1 + ␣5 educationt−1 + ␣6 instabilityt−1 + ␣7 tradet−1 + ␣c country fixed effects + ␣t yearly fixed effects + ut ,

(4.1)

democracyt = ␤0 + ␤1 rgdppct + ␤2 prior democracyt−1 + ␤3 tradet−1 + ␤4 diffusiont−1 + ␤5 inflationt−1 + ␤6 economic growtht−1 + ␤7 year + εt .

(4.2)

The dependent variable in Equation (4.1), rgdppc, is the level of economic development, measured as the logarithm of real GDPPC in 5

For the alternative discrete event approach, see Przeworski et al. (2000) and Boix and Stokes (2003).

100

Democracy and Economic Openness in an Interconnected System Table 4.1. Variables and expected effects in the simultaneous equations

Democracy (DEMOCRACY) equation Variable Endogenous RGDPPC

Sign

Development (RGDPPC) equation Variable

Sign

Positive or negative

Endogenous DEMOCRACY

Positive or negative

Positive

Exogenous PRIOR RGDPPC

Positive

Exogenous PRIOR DEMOCRACY YEAR TRADE DIFFUSION

Positive Positive or negative Positive

GROWTH INFLATION

Positive or negative Positive or negative

INVESTMENT TRADE POPULATION GROWTH EDUCATION INSTABILITY FIXED EFFECTS

Positive Positive or negative Negative Positive Negative Positive or negative

purchasing power parity (PPP) adjusted terms. The dependent variable in Equation (4.2), democracy, is the level of democracy of a country, measured as in the previous two chapters. The specification of the development equation follows the economic development literature, and the specification of the democracy equation follows Chapter 2 (and the comparative politics literature). Table 4.1 summarizes the relationships among the endogenous variables and lists the expected effects of the exogenous variables. The variable prior rgdppc is gdppc from the previous period; investment is a measure of the amount of physical capital in the country; trade is a measure of economic openness, given by the share of a country’s total trade in the national economy; population is the population growth rate of a country; education is the level of human capital in a country; instability is a binary variable indicating the occurrence or absence of revolution or coup d’etat in a country; and country fixed effects is a group of country indicators that capture country-specific heterogeneity, whereas yearly fixed effects is a group of year indicators that capture year-specific heterogeneity. In the democracy equation, prior democracy is the first lag of the level of democracy; year is a year counter that captures the possible linear trend in democracy; trade is the same variable appearing in the development equation; diffusion is the average democracy score of countries within the region around each country; economic growth is the yearly percent growth rate of real GDP, PPP adjusted; and inflation is the yearly inflation rate in a country.

Democracy and Development

101

Research Design Issues Several research design issues require clarification; we discuss their details in the chapter appendix. First, to assess and generalize the overall relationship between democracy and development, we employ a pooled time-series cross-sectional (TSCS) design. The unit of analysis is the country-year, and the empirical sample includes 127 countries from 1961 to 1996, including both DCs and LDCs. Second, stylized facts suggest that LDCs have generally experienced wide variations in their democracy levels during our time period, whereas DCs generally had consistently higher levels of democracy that were quite stable over time. To evaluate the effect of this difference on inference, we conduct analyses for two samples. One sample includes all the countries for which we have data, both DCs and LDCs. A second sample contains only LDCs, excluding members of the Organization for Economic Cooperation and Development, which we identify as DCs. Third, to assess whether any statistical basis exists for suspecting a simultaneous relationship between development and democracy, we employ a so-called exogeneity test. This test evaluates if democracy is not affected by development (thus, exogenous in the development equation) and if development is not affected by democracy (thus, exogenous in the democracy equation). The results of this test are not meant to replace a full SEM analysis but to provide some indication of the simultaneity and whether a SEM is necessary from a statistical perspective. Fourth, to estimate a SEM, we need to consider the mathematical issue of identification and the statistical issue of inference. We handle the issue of identification by using the familiar exclusion criterion. We handle the issue of inference in the presence of simultaneity between development and democracy by employing the two-stage least squares (2SLS) estimator. Fifth, we conduct an additional statistical analysis to evaluate the robustness of our results when we use a different statistical estimator – the three-stage least squares (3SLS) estimator, which is more efficient. Sixth, the economic development equation controls for cross-country heterogeneity for theoretical reasons. Empirically, one may model such heterogeneity as random or fixed effects. The choice between the two methods is based on the so-called Hausman test. Seventh, the economic development equation also includes the lagged dependent variable for theoretical reasons, which could lead to biased statistical results when used in conjunction with the country fixed effects. We check the robustness of our findings by using a variant of the 2SLS estimator, which we refer to as the 2SLS-Kiviet estimator that addresses this problem.

102

Democracy and Economic Openness in an Interconnected System

Eighth, the error term of a statistical model needs to satisfy certain assumptions for the inferences to be valid. We address the related assumption violations by using appropriate econometric techniques. Finally, it is possible that other right-hand-side variables in the model will be affected by democracy or development. We handle the possible endogeneity of the other variables by lagging them one year, a widely employed practice in the literature.

Empirical Findings We first conduct the exogeneity test, seeking to establish a statistical basis for our SEM. The results indicate we should reject that democracy is exogenous in the development equation or that development is exogenous in the democracy equation. They demonstrate the need for a SEM setup if one intends to model correctly the relationship between democracy and development. Table 4.2 reports the 2SLS estimation results for two samples: one for all countries and the other for LDCs only. We begin with the results for the democracy equation. In column 1, for the sample of all countries, the coefficient of economic development, rgdppc, is positive and significantly different from zero. A rise in the level of economic development promotes the level of democracy, which supports Lipset’s modernization thesis. Does this hold in the LDCs? In column 3, for the sample of LDCs only, the effect of a rise in the level of economic development on the level of democracy remains positive and significantly different from zero. Thus, the positive impact of development on democracy is not an artifact of sample choice. How large is the effect of economic development on the level of democracy in a country? The coefficient of rgdppc (the logarithm of real GDP per capita) for the sample of all countries indicates that when the real GDP per capita rises from the estimation sample minimum of 416 PPP-adjusted real international dollars (a very-low-income country) to the estimation sample mean of 3,994 PPP-adjusted real international dollars (a low-to mid-income country), the level of democracy increases by 0.48 unit on the 20-point democracy scale (from −10 to +10). This change in the level of democracy seems small, but it only represents the immediate impact of a rise in economic development on democracy. In our SEM, a rise in economic development in the current period not only affects the level of democracy in the current period but also continues to affect the next period democracy via prior democracy in the model. The long-run effect of some change in development on democracy

Democracy and Development

103

Table 4.2. Democracy and development, 2SLS Democracy Development DCs and LDCs DCs and LDCs RGDPPC PRIOR DEMOCRACY DIFFUSION INFLATION ECONOMIC GROWTH YEAR TRADE

0.2127∗∗∗ (0.0497) 0.9237∗∗∗ (0.0094) 0.2863∗∗∗ (0.0888) 0.00002 (0.00001) −0.0058 (0.0088) 0.0195∗∗∗ (0.0040) −0.0014∗∗ (0.0008)

PRIOR RGDPPC DEMOCRACY INVESTMENT POPULATION GROWTH EDUCATION INSTABILITY Constant Observations

−40.6822∗∗∗ (7.6661) 2914

0.0003∗∗∗ (0.00007) 0.9537∗∗∗ (0.0096) −0.0006∗∗ (0.0003) 0.0016∗∗∗ (0.0003) −0.0064∗∗∗ (0.0023) 0.0024 (0.0023) −0.0099∗∗∗ (0.0044) 0.3516∗∗∗ (0.0626) 2914

Democracy LDCs 0.1901∗∗∗ (0.0624) 0.9150∗∗∗ (0.0011) 0.3586∗∗∗ (0.107) 8.58E-6 (0.00002) −0.0060 (0.0097) 0.0286∗∗∗ (0.0054) −0.0009 (0.0010)

−58.5994∗∗∗ (10.1188) 2230

Development LDCs

0.0003∗∗∗ (0.00008) 0.9558∗∗∗ (0.0065) −0.00059∗ (0.0004) 0.0019∗∗∗ (0.0004) −0.0068∗∗∗ (0.0026) 0.0021 (0.0033) −0.0103∗∗∗ (0.0046) 0.0000 (0.0000) 2230

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

accumulates over time; we explain this computation in detail in the appendix. Suppose real GDP per capita rises from 416 to 3,994; the longrun change in democracy amounts to 6.30 units on the 20-point democracy. This increase in the level of democracy is much larger than the immediate impact of development, covering about one-third of the autocracy– democracy range of the democracy scale. Does a rise in the level of democracy in a country influence its level of development? In Table 4.2, the coefficient of democracy in the development

104

Democracy and Economic Openness in an Interconnected System

equation is negative and significantly different from zero for both the allcountries sample and the LDC sample. Hence, a rise in the level of democracy in a country reduces its level of economic development, all other things being equal. The negative effect of democracy on development is not sensitive to the sample choice. How large is the effect of democracy on economic development? Consider the following scenario. Suppose an autocratic country, whose level of democracy is −10, experiences a rise in democracy to +6, a level that scholars often use to indicate a country’s transition into a democracy. How much does this democratization affect economic development? Since democracy is not logged and rgdppc is logged, the percent change in real GDP per capita due to a change of 16 units in democracy is given by 100 × (Change in democracy) × (democracy Coefficient), which is −0.96% (i.e., 100 × 16 × −0.0006 in percent). Hence, an autocracy whose level of democracy rises from −10 by 16 units can expect to see its real GDP per capita fall by almost 1% in real terms in that year, all other things being equal. This immediate change is substantively not small. How large is the change in the long run? The change in democracy in the development equation affects real GDP per capita immediately (computed earlier) and continues to affect it in the next period through its effect on prior rgdppc in the model. The long-run effect of some change in democracy accumulates over time. Based on the computation illustrated in the appendix, the long-run percent change in GDP per capita due to a change of 16 units of the democracy score is −20.73%. That is, a country whose democracy rises from −10 to 16 will see its real GDP per capita fall by almost 21% in the long run, all other things being equal. The results for the control variables in both the development equation and the democracy equation are largely consistent with the theoretical expectations stated in Table 4.1. These expectations, in turn, are based on previous studies. This increases our confidence in the results from our SEM. We discuss the results of the control variables in the appendix.

Additional Analysis As planned, we conduct two additional analyses. First, we employ the 3SLS estimator. Second, we employ the 2SLS-Kiviet estimator. Overall, the results from these analyses are consistent with those in Table 4.2, both in terms of the reciprocal effects between democracy and development and in terms of the control variable in the two equations. The additional analyses suggest that development promotes democracy, and yet democracy reduces the level

Democracy and Development

105

of development. The robustness of our findings gives us more confidence in our analysis. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Scholarship on democracy and development has called for study of the reciprocal relationship between democracy and economic development. So far, almost all studies have employed single-equation models. The two important exceptions to this practice did not evaluate the possible reciprocal relationship within the context of a structural SEM that endogenizes the level of economic development and the level of democracy in a pooled TSCS large-N design. This chapter is the first to study the reciprocal relationship between the levels of economic development and democracy in a pooled TSCS research design using a SEM. The structure of the development equation follows directly from neoclassical economic growth theory, and the specification of the democracy equation resembles the one used in many studies of democratization. The large-N TSCS design captures both cross-sectional and time-series patterns. The results are robust across subsamples of countries and alternative SEM estimators. To summarize, in both the all-countries sample and the sample of LDCs only, we find that a rise in the level of economic development in a country leads to a rise in its level of democracy, whereas a rise in the level of democracy in a country reduces its level of economic development. The democracy–development causality goes two ways. Our use of the SEM methodology reveals a tension in the democracy– development relationship, which has important policy implications. Our results suggest that the wealth level of a country influences the creation and sustenance of democratic governance. In other words, the establishment and maintenance of democratic institutions are likely to be more successful in richer countries. If democracy is a desirable political system, promoting economic development should be as important, if not more important, a task as promoting democracy directly. Efforts to promote democracy in poor countries without lifting them out of poverty are likely to fail. The idea that promoting democracy will lift poor countries out of poverty receives little empirical support. In fact, democratization presents a dilemma for policymakers. A rise in the level of democracy reduces the level of economic development, which in turn undermines democracy. Countries that set development as their priority may need to delay democratization, at least for a while. Authoritarianism may provide conditions that are conducive to economic development. Nondemocratic countries can more swiftly choose

106

Democracy and Economic Openness in an Interconnected System

prodevelopment policies, avoiding the need to choose a middle-of-theroad approach that would placate and keep all constituencies on board. The drawback of this approach is that the public may have no say at all in the development process, leading to skewed economic development that favors the interests of a narrow group in society at the expense of the masses. The current policy approach in countries such as China, Singapore, Russia, and perhaps to a lesser degree Malaysia, and the policy approach of newly industrialized countries such as South Korea and Taiwan in the 1970s–1990s, suggest that when faced with a choice between democracy and development, many leaders and nations seem to choose development first. Our findings also suggest a significant implication about the outcome of democratic and economic reforms in LDCs. A strategy of development first, which postpones democratization until some level of development is reached, can eventually bring about a rise in democracy as development promotes democracy. In contrast, a strategy of democracy first may fail as a rise in democracy reduces economic development, which in turn is bad for democracy. Hence, the former strategy is more likely to improve both development and democracy. Ultimately, choosing between the two goals (i.e., which one should take the driver’s seat) is a normative decision of leaders and nations, but in doing so the leaders should be aware that they are walking a very fine line. Putting too much emphasis on the goal of economic development at the expense of democracy may backfire. As the country develops economically, pressures toward democratization in the country will likely rise. If these pressures are not released in an orderly manner, they may explode in the form of revolution or coup d’etat. Our results suggest that such political instability in turn reduces the level of economic development in a country, possibly wiping out the economic gains the society worked hard to achieve. SUMMARY AND OUTLOOK

The relationship between democracy and economic development involves two causal questions. One question asks whether economic development promotes democracy; the other asks whether democracy influences development. Statistical analyses typically focus on one of the two questions, relying on single-equation models. Taken together, however, these two questions imply a possible reciprocal relationship between democracy and development. The few statistical studies that attempted to address this possible simultaneous relationship suffer several methodological weaknesses.

Democracy and Development

107

In this chapter, we revisit this important issue regarding the possible reciprocal relationship between democracy and development. We specify and estimate a structural equations model of the simultaneity between the level of development and the level of democracy. This approach allows us to rigorously evaluate the nature of their relationship; that is, whether democracy affects development only, development affects democracy only, they affect each other, or they are not related statistically. The empirical results indicate that the level of democracy reduces the level of economic development, and development promotes democracy. This chapter concludes Part I of our book. In this part, we modeled the effects of economic openness on democracy, the effects of democracy and economic openness on income distribution, and the reciprocal relationship between economic development and democracy. As demonstrated in the different chapters, these are important issues. But these analyses ignored the possible influences of democracy and economic openness on interstate political relations. Yet we observe every day how interstate political events have influenced and been influenced by such significant transformations as democratization and economic liberalization. Part II of our book explicitly tackles the relevance of democracy (Chapter 5) and economic openness (Chapter 6) to interstate military conflict. In Chapter 5, we focus on the relationship between democracy and military conflict, which speaks to two separate bodies of literature: one on democratic peace and the other on the causes of democracy. We seek to highlight the previously overlooked reciprocal causal link between democracy and conflict.

APPENDIX MODELS OF DEVELOPMENT, GROWTH, AND DEMOCRACY

In this part of the appendix, we show in detail that a structural equations model of democracy and economic growth differs substantially from a structural equations model of democracy and economic development; hence, inferences based on the two models are not equivalent. Let us assume that theory expects the following SEM to be the “true” model: democracyt = b1 rgdppct + c1 Z1(t−1)

(4.A1)

rgdppct = b2 democracyt + b3 rgdppct−1 + c2 Z2(t−1), (4.A2)

108

Democracy and Economic Openness in an Interconnected System

where, as in the main text, rgdppct is real GDP per capita at time t; rgdppct−1 is real GDP per capita at time t − 1; democracyt is the level of democracy at time t; Z1 and Z2 are vectors of relevant control variables; b1 , b2 , and b3 are coefficients; and c1 and c2 are vectors of coefficients. Is a model of the change in rgdppc equivalent to the model of the level of rgdppc shown in equation (4.A2)? The answer is yes. To see this point, subtract rgdppct−1 from both sides of equation (4.A2), generating equation (4.A3): rgdppct − rgdppct−1 = b2 democracyt + (b3 − 1)rgdppct−1 + c2 Z2(t−1) .

(4.A3)

Estimation of Equation (4.A3) captures coefficient b2 for democracy and coefficients vector c2 for Z2 , as in Equation (4.A2). And coefficient (b3−1 ) for rgdppct−1 is a simple transformation of b3 . But is a model of the growth rate in rgdppc equivalent to Equation (4.A2)? The answer is no. To see this, rework Equation (4.A3) to have the growth rate of gdppc as the dependent variable, denoted g_rgdppc, by dividing both sides of Equation (4.A3) with rgdppct−1 and producing Equation (4.A4): [rgdppct − rgdppct−1 ]/rgdppct−1 = b2 democracyt /rgdppct−1 + (b3−1 )rgdppct−1 /rgdppct−1 + c2 Z2(t−1) /rgdppct−1 .

(4.A4)

This equation can be rewritten as G rgdppct = b2 democracyt /rgdppct−1 + (b3−1 ) + c2 Z2(t−1) /rgdppct−1 .

(4.A5)

For the rgdppc growth rate model (i.e., Equation (4.A5)) to be equivalent in parameters to the rgdppc level model (i.e., equation (4.A2)), the right-hand-side variables in the level model need to be transformed following Equation (4.A5). Previous scholars, in their typically estimated single-equation model of the rgdppc growth rate, do not transform their democracy and control variables as in Equation (4.A5). Therefore, the coefficients of democracy and the control variables in previous models of rgdppc growth rate are not the same as those in the rgdppc level model in Equation (4.A2). Similarly, for the democracy model in Equation (4.A1), the coefficient b1 for rgdppct is not the same as the coefficient of g_rgdppct . To see this, one

Democracy and Development

109

divides both sides of Equation (4.A1) by rgdppct−1 : democracyt /rgdppct−1 = b1 rgdppct /rgdppct−1 + c1 Z1(t) /rgdppct−1 .

(4.A6)

Next, we rewrite Equation (4.A6) to get the growth rate of GDPPC on the right-hand side: democracyt /rgdppct−1 = b1 [rgdppct /rgdppct−1 − 1] + b1 + c1 Z1(t) /rgdppct−1 .

(4.A7)

The expression in the square brackets in Equation (4.A7) is the growth rate of GDPPC. Thus, democracyt /rgdppct−1 = b1 G rgdppct + b1 + c1 Z1(t) /rgdppct−1 .

(4.A8)

Hence, for the coefficient of rgdppc in Equation (4.A1) to be the same as the coefficient of g_rgdppc, one needs to estimate Equation (4.A8). In other words, the coefficient for g_rgdppc in a regression of democracy on g_rgdppct and Z1 is not the same as that for rgdppc in Equation (4.A1) of the true model. EMPIRICAL MODEL AND ANALYSIS

Empirical Model We have two endogenous variables: rgdppc and democracy. rgdppc is the real GDPPC adjusted for PPP.6 Data are from the Penn World Table 6.1. democracy is measured from the POLITY IV database (Marshall and Jaggers, 2000), as in other chapters. The widely used POLITY data register various democratic and autocratic attributes for many countries on an annual basis from 1800 to 1999. Our measure of democracy is the composite indicator POLITY from POLITY IV. It is constructed as the difference between the democracy and autocracy scores in the POLITY IV data set, ranging from −10 (fully autocratic) to +10 (fully democratic).7 The effect 6

7

Studies of democracy typically use real GDPPC to measure development (e.g., Diamond, 1992a, 1992b; Gasiorowski, 1995; Przeworski et al., 2000; Boix and Stokes, 2003; Feng, 2003; Li and Reuveny, 2003). Among others, Londregan and Poole (1996) and Li and Reuveny (2003) also use this measure of democracy. As Marshall and Jaggers (2000) recommend, we code the “standardized authority codes” of −66, −77, and −88, which appear in the POLITY data set, into missing for −66, averages of the scores before and after the transition for −88, and zero for −77.

110

Democracy and Economic Openness in an Interconnected System

of rgdppc on democracy could be positive or negative, whereas the effect of democracy on rgdppc can be positive, negative, or statistically insignificant, in light of the earlier arguments. The specification of the exogenous variables in the development equation follows Solow’s (1956) theory, which Mankiw et al. (1992) extended to include human capital in addition to physical capital. This theory predicts that the steady-state level of development rises with education and investment in physical capital, falls with population growth, and depends on country-specific determinants. The country-specific effects “reflect not just technology but [also] resource endowments, climate, geography, institutions, and so on; it may differ therefore across countries” (Mankiw et al., 1992: 6). In statistical estimation, the country effects can be modeled as either fixed effects or random effects, an issue to which we will return. Mankiw et al. (1992) assumed that countries are in steady state and therefore employ a cross-sectional design. As they note, however, at any given point in time in the Solow model, the economy is somewhere on the transition path leading to the steady state, not necessarily in the steady state. Hence, one needs to model development dynamics. Islam (1995) employs this insight in his dynamic extension of the Mankiw et al. setup, where development adjusts over time. Islam’s extension includes the lagged dependent variable on the right-hand side, as well as country- and periodspecific effects.8 The specification of our development equation builds on the setups of Mankiw et al. (1992) and Islam (1995). The right-hand-side variables in our development equation include investment in physical capital, population growth rate, and average level of education, following the Solow model. We include, as called for by Mankiw et al. and Islam, country effects and, as called for by the Islam extension, the lagged level of development and yearly effects. We also control for political instability, as suggested by Feng (2003). The expected effects of the right-hand-side variables on rgdppc follow these studies. prior rgdppc is the lagged value of rgdppc. It models the tendency of economic processes to exhibit inertia due to frictions such as transportation delays, evolution of tastes, and the time it takes to negotiate agreements and produce goods. The effect of prior rgdppc on rgdppc is expected to be positive. investment is the logarithm of the percentage ratio of private and public investment in physical capital to real GDP, lagged one year. It is worth noting 8

The Solow/Islam model works in GDPPC level, not in GDPPC growth rate. Thus, it differs from the Barro (1998) type of economic growth model. As shown, moving the lagged dependent variable in the level model to the left-hand side does not lead to an economic growth model.

Democracy and Development

111

that foreign direct investment is part of this variable. These data come from the Penn World Table 6.1. The effect of investment on rgdppc is expected to be positive. population growth is the rate of population growth of a country, lagged one year. Data come from the Penn World Table 6.1. The effect of a rise in population growth on rgdppc is expected to be negative. education denotes the level of national education, measured by the average number of years of education in the overall population. Data come from Barro and Lee (2000) and are lagged one year. The effect of this variable on rgdppc is expected to be positive. instability is set to 1 if a country experiences at least one revolution or coup d’etat in a given year, and 0 otherwise. The data come from CNTS (2008).9 An occurrence of instability is expected to reduce rgdppc (Feng, 2003). trade is the sum of the values of imports and exports of a country with the world, divided by the country’s GDP, lagged one year. The data come from the Penn World Table 6.1, where the division by GDP controls for the importance of trade to the country. The effect of a rise in trade on development is debated. Neoclassical economics expects a positive effect, whereas a neo-Marxist/dependency interpretation expects a negative effect, particularly for LDCs. country fixed effects and yearly fixed effects are country and yearly specific effects, which, as noted, are called for by economic theory (Mankiw et al., 1992; Islam, 1995). In general, the yearly effects capture economic dynamics. The country effects capture structural attributes of development such as climate and topography. Since these effects can be correlated with other independent variables, we might need to model them as fixed effects, an issue to which we will return. The specification of the democracy equation follows studies in comparative politics. The endogenous variable on the right-hand side is rgdppc. The exogenous variables include three types of variables: economic performance, slow-moving structural variables, and external influences. The economic performance variables include economic growth rate and inflation.10 Structural variables (e.g., ethnic and religious compositions, party fractionalization, and institutional qualities) are also sometimes included. These 9

10

CNTS (2008) defines revolution as an illegal or forced government change, an attempt to generate such a change, or an armed rebellion whose aim is independence from the government. A coup d’etat is an extraconstitutional or forced change of the government or its effective control. For the use of inflation and economic growth in democracy models, see Gasiorowski (1995), Londregan and Poole (1996), Feng (2003), and Li and Reuveny (2003).

112

Democracy and Economic Openness in an Interconnected System

theoretical factors tend to be stable over time, suggesting that democracy may exhibit path dependence or inertia. To model this inertia, we include the lagged level of democracy on the right-hand side, which, in addition to its theoretical role, assists in modeling the effects of variables that are not present in the model.11 As one study notes, “with such a pervasive control in place, it is more difficult for spurious effects to be reported” (Burkhart and Lewis-Beck, 1994: 905).12 A third set of control variables involves international forces: trade, and diffusion of democratic ideas and norms.13 To these variables, we add a yearly trend, as democracy generally has risen in the period we study. prior democracy is the lagged value of democracy. As noted, democracy should exhibit inertia. A high level of democracy in the previous period is likely to promote democracy in the current period. The effect of prior democracy on democracy is expected to be positive. year controls for the possibility that democracy has a linear trend in our period. Casual observation suggests that democracy increased after 1945 in many countries, so we expected the effect of year on democracy to be positive. trade is the same variable defined for the development equation. As discussed in Gasiorowski (1995) and Li and Reuveny (2003), the effect of a rise in trade on democracy is under debate and could theoretically be positive, negative, or negligible. diffusion denotes the average democracy score of countries within a region around each country, lagged one year. The regions are Europe, the Middle East, Africa, Asia, and America. The diffusion of democratic norms is captured through channels such as economic relations, tourism, and communication networks (Starr, 1991; Gasiorowski, 1995; Li and Reuveny, 2003). The effect of diffusion on democracy is expected to be positive. growth is the yearly percent growth rate of real GDP, PPP adjusted and lagged one year, computed using data on real GDP from the World Development Indicators (World Bank, 2002). Several studies use negative economic growth as a proxy for economic crises; however, the effect of 11 12

13

See, e.g., Burkhart and Lewis-Beck (1994), Muller and Seligson (1994), Muller (1995), and Li and Reuveny (2003). The inclusion of the lagged dependent variable might absorb the variations in the dependent variable that could be explained by other independent variables, making it harder to find statistically significant results. Hence, our approach can be described as being conservative. For a study focusing on international forces, see Li and Reuveny (2003). See also Starr (1991) for the effect of diffusion, and Gasiorowksi (1995) for the effect of trade openness.

Democracy and Development

113

economic crisis on democracy is debated. Some analysts claim the effect is positive; others argue the effect is negative (for discussions, see Gasiorowski, 1995; Haggard and Kaufman, 1995; and Li and Reuveny, 2003). inflation is the yearly inflation rate in percent from the consumer price index, lagged one year. Data come from the World Development Indicators (World Bank, 2002). As discussed by Haggard and Kaufman (1995) and Li and Reuveny (2003), the effect of a rise in inflation on democracy is also debated. Although there are reasons for including country fixed effects in the development equation (as noted), no similar reasons exist for including them in the democracy equation. Moreover, because democracy often changes relatively slowly over time, fixed effects are likely to absorb most of the variations in democracy, leaving little to be explained by other substantive variables.

Research Design Issues Modeling Simultaneity For our SEM of democracy and development, we first need to consider the issue of identification. The model is clearly overidentified; the number of exogenous variables excluded from each of our two equations far exceeds the required number of one in our case, based on the number of endogenous variables in a particular equation minus one. As our model is identified, we can proceed to the estimation. Applying ordinary least squares (OLS) to each equation yields biased estimates because the endogenous variables are correlated with the error terms. To deal with the simultaneity, one needs to use SEM estimators, which are instrumental variable (IV) estimators, with one difference. Whereas the instrument in IV estimation can be any variable that is correlated with the endogenous variable and not correlated with the error term, in SEM estimation the instruments are based on all the exogenous variables, which is the best approach to deal with simultaneity; other methods (e.g., Heckman selection estimation) are second-best approaches. Two types of SEM estimators exist: 2SLS and 3SLS. In 2SLS, in stage one, each endogenous variable is regressed on all the exogenous variables in the model. In stage two, the endogenous independent variables are replaced by fitted values from stage one. In 3SLS, stage one is 2SLS; stage two uses the estimates to compute the system’s variance–covariance matrix. Stage three uses the matrix in estimating the system from generalized least squares.

114

Democracy and Economic Openness in an Interconnected System

Both 2SLS and 3SLS estimators are consistent. The 3SLS estimator is more efficient than the 2SLS estimator, but it also is more sensitive to any specification error than the 2SLS estimator. Moreover, as summarized by Kennedy (2005: 165), “Monte Carlo studies have shown it [2SLS] to have small sample properties superior on most criteria to all other estimators. They have also shown it to be quite robust . . . to the presence of other estimating problems such as multicollinearity and specification errors.” Therefore, we employ 2SLS as the main estimator. While we model simultaneity between development and democracy, some of our control variables may be affected by democracy or development (e.g., traders may be affected by changes in democracy). Obviously we cannot model all the possible simultaneous effects in one study. Our model already is complex by considering the reciprocal effects between democracy and development. The literature typically deals with this issue by lagging the control variables.14 Although this is an imperfect solution, we also employ it here. Modeling Country Effects in the Development Equation As noted, for theoretical reasons, we include country effects in the development equation (but not in the democracy equation). This leads to two statistical issues. First, country effects can be modeled as random or fixed. Random effects are modeled as part of the error term and fixed effects are set as country dummies. The random-effects estimator is more efficient than the fixed-effects estimator, but it is consistent only if the country effects are uncorrelated with the independent variables. Consistency of the fixed-effects estimator, or the least squares dummy variable (LSDV) model, is not subject to this requirement. Because our country effects capture attributes such as climate or input endowments, they are likely correlated with our independent variables (e.g., investment, population growth). The Hausman test compares the random-effects and fixed-effects estimators. The results corroborate our suspicion, suggesting the fixed-effects estimator should be used to model the country effects in the development equation. Second, the presence of the lagged dependent variable, together with the country fixed effects, makes the fixed-effects or LSDV estimator inconsistent when the asymptotics are considered in the N direction (where N is the number of countries). Amemiya (1967) and others have shown that the

14

For this approach, see, e.g., Muller and Seligson (1994) and Li and Reuveny (2003).

Democracy and Development

115

LSDV estimator is consistent in the T direction. The size of the inconsistency in the N direction is about 1/T (Nickell, 1981).15 Beck and Katz (2004) compare the properties of the LSDV estimator in a dynamic panel setup such as ours (TSCS with a lagged dependent variable) to the properties of two alternative but consistent estimators of such a model: the Anderson and Hsiao (1982) (AH) estimator and the Kiviet (1995) estimator. The AH estimator removes the fixed effects by using first difference and uses the second lag of the dependent variable or the second lag of the first difference of the dependent variable as an instrument for the first difference of the lagged dependent variable. The Kiviet estimator employs LSDV estimates but corrects them with a formula that plugs in values from a consistent estimator such as the AH estimator. Since the finite sample properties of these estimators are unknown, Beck and Katz (2004) studied their bias and root mean square error in Monte Carlo simulations. On the basis of their findings, Beck and Katz recommend that “the AH estimator should not be used for TSCS [time series cross section] data. We see little reason, in general, not to prefer LSDV over the Kiviet estimator when T is twenty or more” (2004: 14). They add: “For typical comparative TSCS data, it does not appear that OLS with fixed effects and a lagged dependent variable (LSDV) is problematic” (2004: 29). In another Monte Carlo study, Judson and Owen (1999: 13) reach a similar conclusion: “When T > 30, LSDV performs just as well or better than the viable alternatives.” Based on these results and the number of years (35) in our sample period, we employ the LSDV estimator for the development equation.16 For robustness check, we also use the estimator of Bruno (2005), which extends the Kiviet (1995) estimator (that applied to balanced panels) to unbalanced panels. Because the Kiviet algorithm has not yet been extended to 2SLS, we apply the two-step estimation. We refer to this estimator as the 2SLS-Kiviet estimator. Performing 2SLS step by step produces biased standard errors in the second stage because the procedure uses the predicted values for the endogenous variable in computing the standard errors instead of the required actual values (Greene, 2003). But if the R2 in the first stage is high, say 0.8 and above as in our case, the size of the bias is expected to be quite small.17 The Kiviet estimator requires one to choose a consistent estimator to provide initial values, and the order of the bias to adjust for. 15 16 17

For detailed discussion, see Beck and Katz (2004). To be exact, the size of the asymptotic bias is only about 1/35, or 2.8%. We verified this point in personal communication with William H. Greene, the author of the popular Econometric Analysis (e-mails dated April 30 and May 2, 2005, and May 11–13, 2008).

116

Democracy and Economic Openness in an Interconnected System

We use the estimator developed by Arellano and Bond (1991) for initial values, and 1/T for the bias (see also Bun and Kiviet, 2003). Heteroskedasticity, Serial Correlation, and Nonstationarity Since our sample involves panel data, we need to consider the possibilities of serial correlation and heteroskedasticity in the error terms of both equations. We deal with these issues in two ways. First, the lagged dependent variable in each equation helps to address serial correlation and mitigate the risks associated with this potential problem (Beck and Katz, 1995a, 1995b, 2004). Second, we estimate the standard errors of parameter estimates using a variant of the White (1980) estimator of robust standard errors that adjusts for clustering over countries, which is typically referred to as the Huber– White sandwich estimator. This estimator yields consistent estimates of the covariance matrix under very general conditions of heteroskedasticity and serial correlation (Wiggins, 1999). One may also be concerned that, if the variables in the model are nonstationary, the estimates generated by the 2SLS (or 3SLS) may not be valid. As it turns out, this is not the case. As shown mathematically by the important study of Hsiao (1997) – see also Johnston and DiNardo (1997) – the 2SLS estimator is valid when the dependent variables are nonstationary or co-integrated. “In structural equation applications, what one needs to worry about are the classical issues of identification and estimation, not non-stationarity” (Hsiao, 1997: 385).

Empirical Findings In this part of the appendix, we first discuss the results of the exogeneity test and the Hausman test, then explain the computation of the long-run effects of development and democracy, next present the results for the control variables, and finally discuss briefly the results of the alternative estimators from robustness tests. Exogeneity Test Before we evaluate our SEM, we need to establish whether there is an empirical basis for such a model. Burkhart and Lewis-Beck (1994) employ the bivariate Granger causality test. They find that economic development Granger-causes democracy, whereas democracy does not Granger-cause development. The pioneering result of Burkhart and Lewis-Beck is important but does not suffice for our case, which involves a multivariate model.

Democracy and Development

117

We therefore employ the Davidson–MacKinnon test of exogeneity for regression estimated via instrumental variables. The null hypothesis is that an OLS fixed-effects estimator of the same equation would yield consistent estimates. Rejection of the null indicates that the instrumental variables fixed-effects estimator should be employed (Davidson and MacKinnon, 1993; Wooldridge, 2002). In other words, the null hypotheses are that democracy is exogenous in a single-equation model of development and that development is exogenous in a single-equation model of democracy, respectively, whereas the alternative hypotheses are that the two affect each other in some direction (according to the theories discussed earlier). Turning to the results, the test statistic has a p-value of 0.010 for democracy in the development equation and a p-value of 0.035 for development in the democracy equation. Hence, we reject the null hypotheses of exogeneity in both cases: treating development as exogenous in a model of democracy is not supported by the data. Similarly, treating democracy as exogenous in a model of development also is not supported by the data. Ignoring the simultaneity would lead to inconsistent estimates. Fixed-Effects–Random-Effects Test As noted in the main text, the development model calls for the use of country effects, which can be modeled as fixed or random. We conduct the Hausman test, which compares the random-effects and fixed-effects estimators for the development equation. The test statistic has a p-value of 0.007, rejecting the null hypothesis that the coefficients from the two estimators are indistinguishable. This result supports the use of the country fixed-effects estimator. Computing Long-Run Effects of Development and Democracy on Each Other In our SEM, a rise in economic development in the current period affects the level of democracy in the current period and continues to affect the next period of democracy via prior democracy in the model. The long-run effect of some change in development on democracy accumulates over time and is computed by the following formula: long-run change in democracy = Change in rgdppc × [rgdppc Coefficient/(1 − prior democracy Coefficient)].

118

Democracy and Economic Openness in an Interconnected System

The expression in the square brackets of the formula is given by [0.2127/(1 − 0.9237)], which is 2.787. Thus, when real GDP per capita rises from 416 to 3,994, the long-run change in democracy is 6.30 units on the 20-point democracy scale, which is based on 2.787×[log(3,994) − log(416)]. How large is the long-run effect of democracy on development? The long-term effect of a rise in the level of democracy is given by the following formula: long-run percent change in gdp per capita = 100 × Change in democracy × [democracy coefficient/(1 − prior rgdppc coefficient)]. The expression in the square brackets of the formula is given by [−0.0006/(1 − 0.9537)], which is −0.013. Thus, the long-run percent change in gdp per capita due to a change of 16 units of democracy score is −20.73%, which is based on [100 × 16 × (−0.013)]. Results for Control Variables Next, we discuss the results for the control variables from the 2SLS estimator. Beginning with the development equation, the effect of a rise in prior rgdppc on development is statistically significant and positive for both samples. Economic development exhibits positive inertia, as expected. Also as theoretically expected according to the core Solow framework, the effect of a rise in investment on the level of economic development is positive and significant, and the effect of a rise in population on the level of economic development is negative and statistically significant. We obtain these results for both the all-countries sample and the LDC sample. The effect of a rise in trade on the level of economic development is statistically significant and positive. Trade openness promotes economic development, supporting the neoclassical position in this particular debate. The effect of a rise in the level of education on economic development is positive, as expected by the extended Solow framework, but is not statistically significant in our case. This result is sometimes reported in an economic development model, reflecting the correlation between this variable and the first lag of rgdppc. Note, however, that the lagged value of rgdppc is included in the model due to theoretical reasons called for by the Islam (1995) extension of the Solow model in a dynamic panel data setup. It should also be noted that the coefficient of education for the LDC sample is considerably smaller than this coefficient for the all-countries sample,

Democracy and Development

119

reflecting the fact that the level of education in LDCs is relatively lower in general and, therefore, has a weaker effect on development. Finally, the effect of a rise of instability from 0 to 1, indicating the presence of coup or revolution or both, on the level of economic development is negative and statistically significant in both the all-countries and LDC samples. Consistent with Feng’s (2003) findings, political instability in the form of coup and revolution reduces the level of economic development. In the democracy equation, for the sample of all countries, the effect of a rise in prior democracy on democracy is significant and positive, demonstrating the importance of democratic inertia. The effect of a rise in year is significant and positive, reflecting the general positive trend of democracy in our sample. The effect of a rise in diffusion of democracy ideas on the level of democracy in a country is positive and significant. As the number of democratic countries in a country’s geographical region rises, so does the country’s own level of democracy. All these reported effects are robust in the sample of LDCs only. As we found in Chapter 2, a rise in trade is found to reduce the level of democracy in a country. One explanation of this result is that trade widens the social cleavages between winners and losers, leading to more social unrest and repressive regime. Further discussions are presented in Chapter 2. The effect of a rise in trade also is negative for the LDC sample but is not statistically significant at standard levels. The effects of a rise in the rates of inflation and economic growth on democracy are statistically insignificant for both samples, again as reported by Feng (2003), Li and Reuveny (2003), Gasiorowski (1995), and others. Results from Alternative Estimators Tables 4.A1 and 4.A2 present the results from the 3SLS and 2SLS-Kiviet estimators. The effect of a rise in the level of economic development on the level of democracy remains positive and significantly different from zero across the two alternative estimators and between the two samples. Likewise, the effect of a rise in the level of democracy on the level of economic development is negative and significantly different from zero across the two alternative estimators and between the two samples.

120

Democracy and Economic Openness in an Interconnected System Table 4.A1. Democracy and development, 3SLS Democracy Development DCs and LDCs DCs and LDCs

RGDPPC PRIOR DEMOCRACY DIFFUSION INFLATION ECONOMIC GROWTH YEAR TRADE

0.2127∗∗∗ (0.0477) 0.9236∗∗∗ (0.0067) 0.2882∗∗∗ (0.0744) 0.00002 (0.0001) −0.0073 (0.0075) 0.0194∗∗∗ (0.0039) −0.0014∗∗ (0.0009)

PRIOR RGDPPC DEMOCRACY INVESTMENT POPULATION GROWTH EDUCATION INSTABILITY Constant Observations

−40.3984∗∗∗ (7.6554) 2914

0.0003∗∗∗ (0.0001) 0.9543∗∗∗ (0.0052) −0.00051∗∗ (0.0003) 0.0016∗∗∗ (0.0002) −0.0066∗∗∗ (0.0015) 0.0024 (0.0022) −0.0102∗∗∗ (0.0029) 0.3507∗∗∗ (0.0598) 2914

Democracy LDCs 0.1904∗∗∗ (0.0654) 0.9148∗∗∗ (0.0081) 0.3612∗∗∗ (0.0943) 0.00001 (0.0001) −0.0076 (0.0088) 0.0284∗∗∗ (0.0051) −0.0009 (0.0010)

−58.3035∗∗∗ (10.1004) 2230

Development LDCs

0.0003∗∗∗ (0.0001) 0.9562∗∗∗ (0.0061) −0.00047∗ (0.0003) 0.0019∗∗∗ (0.0003) −0.0070∗∗∗ (0.0018) 0.0023 (0.0031) −0.0107∗∗∗ (0.0034) 0.0000 (0.0000) 2230

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Democracy and Development

121

Table 4.A2. Democracy and development, 2SLS-Kiviet Democracy Development DCs and LDCs DCs and LDCs RGDPPC PRIOR DEMOCRACY DIFFUSION INFLATION ECONOMIC GROWTH YEAR TRADE

0.2237∗∗∗ (0.0510) 0.9233∗∗∗ (0.0095) 0.2837∗∗∗ (0.0875) 0.00002 (0.00002) −0.0050 (0.0081) 0.0189∗∗∗ (0.0040) −0.0015∗∗ (0.0009)

PRIOR RGDPPC

DEMOCRACY INVESTMENT POPULATION GROWTH EDUCATION INSTABILITY Constant Observations

−39.7159∗∗∗ (8.1152) 2917

0.0003∗∗∗ (0.0001) 0.9793∗∗∗ (0.0049)

Democracy LDCs 0.2010∗∗∗ (0.0641) 0.9148∗∗∗ (0.0113) 0.3549∗∗∗ (0.1068) 0.00001 (0.00002) −0.0053 (0.0088) 0.0277∗∗∗ (0.0055) −0.0010 (0.0010)

−0.00054∗∗ (0.00029) 0.00149∗∗∗ (0.00024) −0.0062∗∗∗ (0.0016) 0.0011 (0.0028) −0.0102∗∗∗ (0.0029) 0.0000 −57.1632∗∗∗ (0.0000) (10.9474) 2858 2233

Development LDCs

0.0002∗∗∗ (0.00008) 0.9825∗∗∗ (0.0063) −0.00055∗ (0.0003) 0.0017∗∗∗ (0.00023) −0.0066∗∗∗ (0.0019) −0.00001 (0.0033) −0.0108∗∗∗ (0.0039) 0.0000 (0.0000) 2193

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

PART II

BRINGING IN CONFLICT

FIVE

Democracy and Conflict

INTRODUCTION

Interstate military conflict as an extreme form of political exchange is a central feature of the international system and has received the enormous attention it deserves from scholars and policymakers. Currently state sovereignty and anarchy make world peace seem elusive. Chapters 2–4 in Part I modeled the effects of economic openness on democracy, the effects of democracy and economic openness on income inequality, and the reciprocal effects of economic development and democracy, respectively. These analyses ignored the possible impact of democracy and economic openness on interstate political relations. In reality, democracy and economic openness develop in and around nation-states such that they inevitably influence political relations. In fact, scholars have argued and sought to demonstrate that both democracy and economic openness affect interstate military conflict, even though they debate the nature of their effects. Part II of our book explicitly tackles the relevance of democracy (Chapter 5) and economic openness (Chapter 6) to interstate military conflict. In Chapter 5, we are particularly interested in the relationship between democracy and military conflict. This chapter speaks to two separate bodies of literature: one on democratic peace and the other on the causes of democracy. Both topics have been the subjects of voluminous research, generating two large bodies of literature. Our contribution is not to overthrow but to build on and connect these two separate bodies of literature, highlighting and analyzing the previously overlooked reciprocal causal link between democracy and conflict. The claim that democracies do not fight one another has received wide attention. Most statistical studies investigating this claim share similar designs, employing a dyadic level of analysis and estimating single-equation 125

126

Democracy and Economic Openness in an Interconnected System

models. In these studies, the dependent variable is typically dichotomous, measuring the presence or absence of a militarized interstate dispute (MID). The independent variables include dyadic, joint (dyadic) democracy, and more or less the same collection of control variables. The vast majority of studies report that the probability of military dispute between two countries declines as their joint democracy level increases, although a number of studies reject this argument. We note that a much smaller group of studies argues that democracies are more peaceful in general, not only toward other democracies. This claim has received less empirical support. In this chapter, we focus on the dyadic version of the democratic peace argument. Side by side with the democratic-peace literature is a growing number of international relations scholars who argue that military conflict affects democracy. But they debate about the direction of the effect of conflict on democracy. Conceptually, these studies form a subset of the large monadic literature within comparative politics that studies the determinants of democratization. This chapter is motivated by the observation that the democratic-peace literature and the literature on the effect of conflict on democracy have basically ignored the insights offered by each other. In particular, most international relations scholars have treated the effect of joint democracy on conflict as unidirectional, even though conflict has been demonstrated to influence democracy. Once the two separate bodies of literature have been linked, one can reasonably argue that conflict and democracy affect each other simultaneously. Our goal in this chapter is to address the simultaneity between conflict and democracy within a unified statistical framework. We stress at the outset that our agenda is not to prove or disprove the existence of democratic peace. Rather, we believe that democracy and dyadic conflict affect one another. Thus, researchers need to model the reciprocal effects explicitly, which is not an easy task; it requires connecting monadic and dyadic levels of analysis. We develop a novel way to link the two levels of analysis while maintaining consistency with both the dyadic democratic-peace literature and the literature on the monadic determinants of democracy. In our simultaneous equations model, the endogenous variables include dyadic conflict measured by the presence or absence of a MID, the higher level of democracy in a dyad, and the lower level of democracy in a dyad. As in studies by Dixon (1994) and Oneal and Russett (1997), the higher level of democracy in a dyad measures political regime dissimilarity and the lower level of democracy reflects dyadic, joint democracy. Our model usefully integrates

Democracy and Conflict

127

the dyadic determinants of dyadic conflict with the monadic determinants of democracy within one system of simultaneous equations. Some may view the simultaneity of dyadic conflict and democracy merely as a methodological nuisance. It is worth noting that, in contrast to this view, we consider the issue to be both theoretical and methodological. In fact, various scholars share this view. For example, Chan (1997a) argues in his comprehensive review that the democratic-peace literature needs to take more seriously the possibility of reverse causality from conflict to democracy. Gleditsch and Ward (2000) also note briefly that conflict often affects democratization. As discussed in detail later, in the first study that employs a partial simultaneous setup to examine the reciprocal relationship between conflict and democracy, James et al. (1999) argue that the effect of joint democracy on dyadic conflict is generally insignificant. Their study has led to a lively debate, but the conflict–democracy simultaneity issue remains unresolved.1 We believe that further investigation of the relationship between dyadic conflict and democracy within a simultaneous framework is warranted. As noted by Alvarez and Glasgow (2000: 163), “if there are theoretical reasons to suspect endogeneity in the model, it must be modeled. Ignoring endogeneity will lead to biased estimates.” Of course, the actual effect of simultaneity on the dyadic conflict–democracy nexus could be large or small; but unless one models the reciprocal interaction between democracy and conflict, one simply cannot evaluate the effect of ignoring the simultaneity, which is a common practice in the conflict community. Our work in this chapter assesses empirically the question of simultaneity between democracy and dyadic conflict across a sample of politically relevant dyads from 1950 to 1992.2 Our empirical findings can be summarized as follows. First, dyadic conflict reduces both lower and higher levels of democracy in a dyad. Second, joint democracy reduces the probability of dyadic MID involvement. When compared with the single-equation estimate of Oneal and Russett (1997), the absolute magnitude of the effect in our study is smaller but in relative terms, the effect is similar in size. The effect of joint democracy on MID involvement is considerably smaller for noncontiguous countries than for contiguous countries. Third, the level of democracy in a country is affected by a number of economic and social 1 2

See Oneal and Russett (2000) and the response of James et al. (2000). For an earlier exposition of the position stated by James et al. (1999, 2000), see Wolfson et al. (1998). Politically relevant dyads are dyads involving at least one major power (United States, United Kingdom, France, USSR/Russia, and China) or geographically contiguous countries.

128

Democracy and Economic Openness in an Interconnected System

variables that have commonly received attention in the democratization literature. Fourth, several control variables in the MID equation that prior studies have found to be significant in the single-equation framework are shown to lack statistical significance in our simultaneous equation model design. Notably, we find that preference similarity between two countries, based on the United Nations (UN) voting patterns, does not have a significant effect on dyadic MID involvement. We organize the chapter as follows. The next section reviews the democratic-peace literature as well as the literature on the effect of conflict on democracy, and the section that follows discusses the theoretical expectations. We then discuss the few empirical studies that have investigated some aspects of the simultaneity between democracy and conflict. The fifth section of this chapter presents our conceptual model. The sixth section discusses the statistical model, research design issues, and key findings and is followed by a conclusion section. TWO BODIES OF LITERATURE ON DEMOCRACY AND MILITARY CONFLICT

This section discusses the two bodies of literature that focus on the effect of democracy on military conflict and the effect of conflict on democracy, respectively.

Effect of Democracy on Conflict The democratic-peace literature has been reviewed extensively elsewhere.3 In this section, we briefly summarize this literature. Most scholars argue that democracies do not fight one another and that they are unlikely to engage in militarized disputes with one another.4 The monadic argument that democracies are less likely to be involved in MIDs with all countries also has been made but is relatively more controversial.5 As noted, we focus on the dyadic version of the argument. The democratic-peace literature is based on the empirical observation that democracies have almost never fought wars against each other. 3 4 5

For example, see Levy (1989), Russett (1993), Gleditsch (1995), Ray (1995), Chan (1997), and Russett and Oneal (2001). For example, see Bremer (1993), Maoz and Russett (1993), Dixon (1994), and Oneal and Russett (1997). Rummel argues that “the more a nation is democratic, the less severe its overall foreign violence” (1997: 5). However, he admits that this argument is controversial.

Democracy and Conflict

129

Although a few cases are said to disprove this empirical regularity, a few studies explain why these cases are questionable.6 Given this empirical regularity, the thrust of the literature has focused on explaining why democracies do not fight each other, whether this regularity can be attributed to democracy or other forces, and whether it is simply due to random historical occurrence. The proponents of the democratic-peace proposition offer several explanations for their claims. The first, so-called institutional, explanation traces back to Kant (1795). The decision to wage war faces more checks and balances in democracies than in autocracies (Russett, 1993). Democratic governments confront more constraints by the rule of law. The slowing down of the decision-making process leaves more time for negotiation to take place and reduces the risk of errors and misunderstandings, both of which reduce the probability of war (Maoz and Russett, 1993). Yet another institutional explanation concerns the informational role of democratic institutions (Fearon, 1994; Schultz, 1998; Bueno de Mesquita et al., 1999). Because democratic leaders are held accountable and compete for reelection, military defeats may lead to electoral defeats by political opponents. When democracies fight, they mobilize their resources and tend to win; therefore, they prefer to mediate disputes and, in the event of war, choose their fights carefully to increase the likelihood of victory. Democracies do not fight each other because they are unattractive targets for other democracies. So the fact that a democracy mobilizes and escalates a crisis sends a clear signal regarding its resolve to the other party. Because all democracies behave in this manner and their leaders know it, democracies are less aggressive toward one another. According to a second so-called normative argument, democratic governments tend to tolerate their domestic opposition and resolve domestic conflicts peacefully. Analogously, they also behave in this way toward other countries (Weart, 1998). Although these effects are monadic, they are allegedly stronger when democracies interact with one another. Since the domestic politics of democracies is transparent, democracies can more readily trust each other (Maoz and Russett, 1993; Russett, 1993). It follows that conflicts among democracies are more likely to be mediated by third parties than they are to escalate (Dixon, 1994). 6

For a discussion of these cases see, e.g., Doyle (1997). Russett and Oneal (2001) emphasize that the effect of joint democracy on dyadic conflicts is probabilistic. Even if a few democratic dyads exhibit militarized disputes, these cases do not refute the claim that democratic dyads are generally less likely to exhibit militarized disputes than other dyads.

130

Democracy and Economic Openness in an Interconnected System

A number of studies reject the democratic-peace proposition. Farber and Gowa (1995), for example, argue that the purported negative effect of joint democracy on conflict is an artifact of the common interests that were shared by democracies when facing the former USSR during the Cold War. Layne (1994) argues that the American Civil War is a strong disconfirming case of the democratic-peace argument. Polachek (1997) and Polachek and Robst (1998) argue that democracies do not fight each other because of their extensive trade relations, since conflict reduces trade.7 Lemke and Reed (1996) account for this apparent regularity by arguing that democracies tend to be content with the status quo, whereas autocracies tend to be dissatisfied with the status quo and, therefore, are more prone to conflict. This theory explains both the lack of wars between democracies and the higher incidence of wars between democracies and autocracies. From a different angle, Gartzke (1998) argues that democracies are less likely to disagree and, thus, have no need to fight each other. When one includes a measure of affinity based on the UN General Assembly voting patterns in the typical model, the effect of joint democracy on the probability of MIDs becomes weak and insignificant.8 Rosato (2003) does not challenge the empirical regularity of democratic peace but argues that extant evidence does not substantiate the different causal mechanisms offered by proponents to rationalize the existence of democratic peace.

Effect of Conflict on Democracy Whereas most scholars seem to agree that democracies do not fight each other, the effect of military conflict on democracy is a subject of controversy. According to one view, conflict reduces democracy. This view originated from the writings of de Tocqueville (1835: 650): “All those who seek to destroy the freedom of the democratic nations must know that war is the surest and the shortest means to accomplish this.” War may not immediately result in a military government, but it will eventually concentrate power in the hands of the government, leading to despotism and the decline of democracy. Elaborating on the negative effect of conflict on democracy, Lasswell (1977) argues that because some groups may not want to join the 7 8

For additional discussion of the link between democratic peace and economic relations, see Mousseau and Shi (1999). Another argument that weakens the democratic-peace proposition is offered by Mansfield and Snyder (1996), who argue that young democracies are more prone to engage in war with each other than are established democracies.

Democracy and Conflict

131

war effort, the state may become increasingly autocratic to reassert power. Moreover, conflict may require resources that the state does not have. When people do not agree to give up wealth for the war effort, the state may have to become oppressive to pay for the war and subsequently force the people to do so.9 Layne (1994) and Thompson (1996) suggest a different mechanism. Building democracy requires peace. States facing external threats need to allocate resources to national defense and maintain a large governmental apparatus; both factors reduce the level of resources available for democratic consolidation. A competing view posits that conflict leads to democracy (Tilly, 1992). Historically, many leaders, when mobilizing their populations, granted political rights or land in exchange for their support of the war efforts. Such redistribution of wealth enlarges the number of people that are able to affect governance and, thus, facilitates democratization. Moreover, wars often provide the impetus for social reforms, including democratization (Higgs, 1988; Porter, 1994; Kasza, 1996). Skocpol et al. (2001) demonstrate this effect in a case study and argue that the U.S. Civil War and U.S. participation in World War I exerted a strong positive effect on the U.S. democratic civil society.10 A third group of studies, mostly in the field of international relations (e.g., Modelski and Perry, 1991; Bueno de Mesquita et al., 1999; Mitchell et al., 1999), offers a conditional argument for the expected effect of military conflict on the level of democracy. When an interstate war ends in a victory by a democratic country, the winner often imposes democratization on the loser, replicating its own political system design. In contrast, when the winner of a war is autocratic, the level of democracy tends to decline in the losing country.11 Despite competing expectations about the direction of the effect, all three views agree that military conflict does influence democracy. The overall or net effect of conflict on democracy thus appears to warrant empirical investigation. However, these studies do not take into account the converse possibility: democracy may affect conflict behaviors. Again, as noted, ignoring such reciprocal effects may lead to erroneous empirical findings. 9 10 11

For a similar argument, see Mintz (1985), Almond (1990), Martin (1994), Segal (1994), Midlarsky (1995), and Gates et al. (1996). For other case studies that demonstrate the positive effect of acute conflict on democracy forces, see, e.g., Marwick (1988), Porter (1994), and Reuveny and Prakash (1999). For example, compare the cases of Germany and Japan after World War II, where some European countries were under Nazi or Soviet control.

132

Democracy and Economic Openness in an Interconnected System

IMPLICATIONS AND THEORETICAL EXPECTATIONS

We have reviewed two bodies of literature, one on the effects of democracy on conflict and the other on the effects of conflict on democracy. By synthesizing the two bodies of literature, we hypothesize that the relationship between dyadic conflict and democracy is reciprocal; in other words, the two forces simultaneously affect one another. Again, some may dismiss the simultaneity as a methodological nuisance; however, we believe this issue has important theoretical, methodological, and policy implications. The democratic-peace literature builds on results from single-equation models. However, such results rely on the regression of dyadic conflict on joint democracy – a technique that cannot distinguish whether joint democracy affects or is affected by MIDs. In contrast, our simultaneous-equations framework may yield one of three possible empirical outcomes. First, conflict influences joint democracy, but joint democracy does not affect conflict. Second, dyadic conflict does not affect joint democracy, but joint democracy influences conflict. Third, joint democracy and dyadic conflict affect each other. We cannot determine the empirical validity of these three distinct outcomes until we model the democracy–conflict simultaneity explicitly. That said, we believe the arguments that attribute peace to joint democracy are theoretically plausible. Furthermore, we cannot ignore the fact that the democratic-peace argument has received wide empirical support in many single-equation studies. We therefore state the following hypothesis: H1: Joint democracy reduces the incidence of MIDs between two countries in a dyad. In terms of the effect of conflict on democracy, although the arguments in the literature are made in the context of a monadic level of analysis, they can be readily applied to the dyadic level, as dyads are merely pairs of countries. If conflict affects the level of democracy of any of the two countries in a dyad, it also could, in principle, affect the joint democracy of that dyad. Building on our discussion in the previous section, we categorize the effects of conflict on democracy into four types. The absence of conflict involvement by a country enables the provision of resources for the nurturing of democratic institutions. In contrast, the presence of conflict reduces the level of resources available to support democratization. The preparation for conflict may increase autocracy as the state coerces the people to participate in, and contribute resources to, an unpopular war. This channel, however, may also promote democracy: the state may grant political rights

Democracy and Conflict

133

to groups in return for their support of the war. The occurrence of conflict generates a sense of crisis, which legitimizes power centralization. However, the state may also grant democratic rights during conflict to win support for the war. The postconflict effect comes in two variants. In one variant, conflict is a social shock, stimulating reforms that increase democracy. A second variant involves the winners of a war imposing their own political regime on the losers.12 In summary, the existing theories in the literature on the effect of conflict on democracy offer conflicting expectations, implying that the overall (net) effect of conflict on democracy is theoretically unclear and leading to the following expectation: H2: Conflict may raise or reduce the level of democracy. PREVIOUS EMPIRICAL STUDIES OF CONFLICT–DEMOCRACY SIMULTANEITY

In general, the issue of conflict–democracy simultaneity has not received much attention in the empirical literature. Before we turn to our analysis, however, we believe it is beneficial to review the few studies that have dealt with some aspects of this simultaneity. Investigating the reverse causality, Mousseau and Shi (1999) focus on the proposition that democracy declines as countries prepare for war. However, their study does not find statistical support for this claim. Crescenzi and Enterline (1999), on the other hand, argue that the relationship between war and democracy is reciprocal, as reflected by the portion of democratic states in the system and the frequencies of democratization and war in the system. They found, using vector autoregression analysis, that these variables do indeed influence one another, but the strengths of their effects change across time and space. Rasler and Thompson (2000) argue both that states perceiving high external threats will be less democratic and that democracies are less likely to have MIDs with one another. Focusing on nine major powers from 1816 to 1992, they estimate two equations separately. In one equation, the monadic level of democracy is regressed on the change in democracy, the level of external threat, the change in the level of external threat, a dummy variable measuring war participation, and the lagged level of democracy. In a second equation, a dyadic MID dummy is regressed on a joint democracy dummy and the level of external threat. They conclude that a high level of external 12

Mousseau and Shi (1999) suggest a similar typology involving war’s anterior, concurrent, and posterior effects.

134

Democracy and Economic Openness in an Interconnected System

threat reduces democracy and that democratic dyads are less likely to experience MIDs. None of these studies, however, has employed a simultaneous equations model. James et al. (1999) were the first to study the simultaneous relationship between joint democracy and dyadic conflict. They estimated the following structural equations model: hostility = F (regime, growth, proximity, ally),

(5.1)

regime = G (hostility, growth, stability).

(5.2)

where hostility is the highest level of dyadic hostility in a year, with five possible values (0 = no dispute, 1 = threat of force, 2 = display of force, 3 = use of force, and 4 = war); regime [(rH + rL )/(rH − rL +1)] is a measure of joint democracy from Maoz and Russett (1993), where rH and rL are the higher and lower regime scores in a dyad, respectively; r = c(d − a) is the regime index of a country, where autocracy, democracy, and concentration of power are represented by a, d, and c, respectively. growth is the average of the growth rates of the two countries, A and B, in the past three years. proximity captures if countries A and B are contiguous or if they can project force beyond their contiguous neighbors (i.e., United States, United Kingdom, France, China, and the Soviet Union). ally reflects the alliance ties between two countries. stability measures joint regime stability within a dyad. In their estimation, James et al. (1999) ignored the simultaneity for Equation (5.1) and estimated it as a single-equation model. Equation (5.2) is estimated with a two-stage procedure, taking into account the simultaneity of regime and hostility. The authors find that the effect of joint democracy on the probability of MIDs is generally not significant; the hypothesis that peace induces democracy receives stronger support. Oneal and Russett (2000) criticize James et al.’s (1999) study on several grounds. First, they criticize the use of a dated measure of joint democracy. Second, they argue that the multinomial approach to hostility is destined not to find significant effects for democracy on conflict since MIDs are rare events – with the five categories of MIDs representing even rarer events. Third, Equation (5.1) omits the capability ratio of the two countries in the dyad. Oneal and Russett obtained different results for Equation (5.2) than did James et al. when this variable was included in Equation (5.1). Fourth, Equation (5.2) omits variables routinely used in the comparative politics literature on the determinants of democracy, such as GDP per capita, inflation, or economic growth (Gasiorowski, 1995). Fifth, Equation (5.2) assumes that dyadic conflict affects joint democracy. However, because it

Democracy and Conflict

135

is a monadic quality, democracy ought to be affected by all the conflicts a country engages in, not only by dyadic conflict. James et al. (2000) replied that it is not clear why one empirical measure of joint democracy is superior to others and that the results of Oneal and Russett are sensitive to the measure used. Second, they argue that the aggregation of all MID types into one measure can introduce bias. Diplomatic protest, for example, is an action different from war. Third, they claim that capability ratio ought to be considered an endogenous variable, not an exogenous variable as Oneal and Russett assume. Moreover, including capability ratio in the equation of hostility still produces a weak effect for regime on hostility. Fourth, the sensitivity of the results from Equation (5.2) to the inclusion of a capability ratio in Equation (5.1) could reflect the sensitivity of the model to the practice of temporal aggregation. In the end, James et al. argued that whereas their simultaneous equations model is not meant to be the last word on the reciprocal relationship between military conflict and democracy, studies ignoring this issue do not offer reliable guidance for public policy. CONCEPTUAL MODEL

In putting together our statistical model in this chapter, we need to specify two types of equations: one for dyadic conflict and a second for democracy. The level of democracy of a country is a monadic attribute. Dyadic conflict is obviously a dyadic attribute. In the democratic-peace literature, dyadic conflict is assumed to decline with joint democracy. In our conceptual framework, joint democracy and dyadic conflict depend on each other. But democracy depends not only on dyadic conflict – it also depends on various monadic attributes, including the conflict a country has with other third-party countries. We include these factors in our model. Our measure of joint democracy follows the weak link assumption (Dixon, 1994). In this approach, joint democracy is given by the lower democracy score of the two countries in a dyad. This assumption formalizes the expectation that the likelihood of a dyadic militarized dispute depends on the regime score of the politically less constrained dyad member (i.e., the country that is less democratic). Authors of many statistical studies employ the weak-link assumption-based measure of joint democracy, including Dixon (1994), Oneal and Russett (1997), and Russett et al. (1998). Some studies also use the higher democracy score in a dyad as an independent variable in their single-equation model of MIDs. In fact, one would expect that the democracy scores of both countries in a dyad would interact with conflict; after all, it takes two states to engage in dyadic conflict.

136

Democracy and Economic Openness in an Interconnected System

Therefore, if democracy affects conflict, each of the democracy scores in the dyad should affect its likelihood. Consequently, we also include the higher democracy score in the dyad in our analysis and treat it as a third endogenous variable. As suggested by Oneal and Russett (1997) and Russett et al. (1998), by holding the other regime score in a dyad constant, one may interpret the higher democracy score in a dyad as representing the regime dissimilarity between two countries. These authors and others expect conflict to decline with the lower democracy score in a dyad but to rise with the higher democracy score in a dyad. We formally develop the conceptual model in the appendix. Here we present our set of estimated equations, Equations (5.3), (5.4), and (5.5), where F indicates functional dependency: demL = F (ML , midABU , midLRU ),

(5.3)

demH = F (MH , midABU , midHRU ),

(5.4)

midABU = F (XAB , demL , demH ).

(5.5)

In these three equations, demL is the lower one of the two democracy scores (one for each country) in dyad AB, where A and B are the two dyadmember countries; demH is the higher democracy score in dyad AB; midABU is the unobserved propensity of dyad AB to engage in a militarized dispute; ML indicates the monadic attributes that explain the level of democracy for the demL country (the one with the lower democracy score in a dyad); and MH indicates the attributes that explain the level of democracy for the demH country (the one with the higher democracy score). For midLRU in Equation (5.3) and midHRU in Equation (5.4), we first need to define the concept of politically relevant international environment (PRIE). The PRIE of some country A includes all the countries that share a border with A, as well as those that are classified as major powers in a time period (Maoz, 2001a). midLRU is the propensity of the low democracy country in a dyad to engage in conflicts with third parties in its politically relevant international environment, a variable that is unobserved (the variable subscript LRU is from the italicized words). Similarly, midHRU is the unobserved propensity of the demH country to engage in military disputes with third parties in its PRIE. EMPIRICAL MODEL AND ANALYSIS

Similar to the previous chapters, this section first presents our statistical model for the empirical analysis and then discusses several research design issues. Then the section presents the key results from the empirical analysis.

Democracy and Conflict

137

As in the other chapters, the discussion in this section is self-contained and does not require any specific statistical expertise. The technical details of the statistical model, measure construction, and data sources are in the chapter appendix and follow the same order of presentation as in the main text.

Empirical Model The next three equations state our statistical model and follow the logic in Equations (5.3), (5.4), and (5.5). The statistical model provides a structure for us to guide statistically uninitiated readers through the empirical exercise. The model distinguishes endogenous and exogenous (or control) variables. The endogenous variables appear on the left-hand side of one of the equations and the right-hand sides of the other two equations, and exogenous variables appear only on the right-hand sides of the equations. Small capital letters denote variables, and the Greek letters ␤, ␥ , and ␦ denote their coefficients, which we will estimate. The Greek letters εt , ␮t , and ␯ t represent the error terms in the three equations, respectively. The subscripts L and H refer to the lower and higher democracy countries in a dyad, respectively. The subscripts t and t − 1 denote the time period of a variable, where t denotes the current period and t − 1 denotes the previous period (or a lagged variable). Equation (5.6) operationalizes Equation (5.3) in the conceptual model, which explains the causes of democracy in the lower-democracy country in a dyad: demLt = ␤0 + ␤1 midABUt + ␤2 prior demLt−1 + ␤3 yeart−1 + ␤4 inflationLt−1 + ␤5 inflationLt−1 ∗ yeart−1 + ␤6 gdppcLt−1 + ␤7 gdppcLt−1 ∗ yeart−1 + ␤8 growthLt−1 + ␤9 tradeLt−1 + ␤10 diffusionLt−1 + ␤11 third-party midt−1 + εt .

(5.6)

Equation (5.7) operationalizes Equation (5.4) in the conceptual model, which explains the causes of democracy in the higher democracy country in a dyad: demLt = ␥0 + ␥1 midabut + ␥2 prior demLt−1 + ␥3 yeart−1 + ␥4 inflationLt−1 + ␥5 inflationLt−1 ∗ yeart−1 + ␥6 gdppcLt−1 + ␥7 gdppcLt−1 ∗ yeart−1 + ␥8 growthLt−1 + ␥9 tradeLt−1 + ␥10 diffusionLt−1 + ␥11 third-party midt−1 + ␮t .

(5.7)

138

Democracy and Economic Openness in an Interconnected System

Finally, Equation (5.8) operationalizes Equation (5.5) in the conceptual model, which explains the causes of dyadic military conflict: midABUt = ␦0 + ␦1 demLt + ␦2 demHt + ␦3 contiguityt−1 + ␦4 alliancet−1 + ␦5 capability ratiot−1 + ␦6 trade dependencet−1 + ␦7 growtht−1 + ␦8 affinityt−1 + ␯t .

(5.8)

The endogenous variables in this system are the unobserved propensity for conflict in dyad AB at time t (midABUt ), the lower level of democracy in dyad AB at time t (demLt ), and the higher level of democracy in dyad AB at time t (demHt ). Recall that the lower democracy score in a dyad represents the joint level of democracy shared by the countries in a dyad, and the higher democracy score in a dyad reflects the extent of regime dissimilarity between two countries. Also, we expect that demL reduces midABU , demH increases midABU , and the effects of midABU on demL and demH could be negative or positive. In empirical analysis, because we do not observe midABU , we replace it with an observed dichotomous indicator, midAB . It takes on a value of 1 when two countries are engaged in a dyadic militarized dispute (MID) and 0 otherwise. The involvement in a MID implies the threat of force, the display of force, the use of military force, or war. Table 5.1 summarizes not only the relationships among the three endogenous variables but also the expected effects of the exogenous variables. The exogenous variables ML and MH in the two democracy equations closely follow the model from the democratization literature, as in Chapter 2. To simplify the notation, we drop subscripts t and t − 1. In terms of the economic determinants of democracy, gdppcL and gdppcH indicate the levels of economic development for the low- and high-democracy countries in the dyad; growthL and growthH measure the yearly growth rates of the two countries; inflationL and inflationH indicate the annual inflation rates for the low- and high-democracy countries; tradeL and tradeH represent the countries’ levels of trade openness, where trade openness is the ratio of total trade over GDP; and year is a linear trend variable. The terms inflationL ∗ year and inflationH ∗ year are interaction terms between inflation and year in the low- and high-democracy countries, and the terms gdppcL ∗ year and gdppcH ∗ year are interaction terms between year and GDP per capita in the two countries. These interaction terms capture the time-varying effects of these factors.

Democracy and Conflict

139

Table 5.1. Expectation of direction of effects in the simultaneous equations DEML equation endogenous

Sign

MIDAB

+, −

EXOGENOUS PRIOR DEMOCRACYL YEAR INFLATIONL

DEMH equation endogenous MIDAB

Sign +, −

+ +, −

EXOGENOUS PRIOR DEMOCRACYH YEAR INFLATIONH

+ +, −

INFLATIONL ∗ YEAR

+, −

INFLATIONH ∗ YEAR

+, −

GDPPCL GDPPCL ∗ YEAR GROWTHL TRADEL DIFFUSIONL THIRD-PARTY MIDL

+ +, − +, − +, − + +, −

GDPPCH GDPPCH ∗ YEAR GROWTHH TRADEH DIFFUSIONH THIRD-PARTY MIDH

+ +, − +, − +, − + +, −

+

+

MIDAB equation endogenous

Sign

DEML DEMH EXOGENOUS CONTIGUITY

− +

ALLIANCE CAPABILITY RATIO TRADE DEPENDENCE GROWTH AFFINITY

− −

+

− − −

In terms of the noneconomic determinants of democracy, prior demL and prior demH are the one-year lagged democracy scores of the lowand high-democracy countries; diffusionL and diffusionH indicate the average democracy scores of the countries in the PRIEs of the low- and high-democracy countries in a dyad; and third-party midL and thirdparty midH are dichotomous variables, which indicate whether the lowand high-democracy countries, respectively, are involved in a militarized dispute with at least one third-party country in their respective PRIEs. The exogenous (or control) variables in the MID equation are the usual suspects in the dyadic studies of democratic peace: contiguity indicates whether the two states in a dyad are geographically contiguous (or separated by up to 150 miles of water); alliance indicates whether the two states in a dyad are members in a military alliance; capability ratio reflects the differences between two countries in terms of material and military capabilities; trade dependence indicates the level of trade interdependence between two countries; growth captures the shared economic growth rate in a dyad; and affinity measures the level of preference similarity between two states, based on the congruence of their votes in the UN General Assembly.

140

Democracy and Economic Openness in an Interconnected System

Research Design Issues To implement our statistical model requires us to address several special design issues. Although we discuss their technical details in the appendix, we offer a brief overview here for general readers. First, the simultaneous equations model includes equations for two observed continuous endogenous variables (demL and demH ) and one unobserved continuous endogenous variable (midABU ). As is typical in such cases, we replace the unobserved variable with an observed conflict indicator. Such models need to satisfy some econometric requirement so that their estimation is meaningful. In the chapter appendix, we show that this condition, which was developed by Heckman (1978), holds in our case. Second, in the empirical estimation of the model’s coefficients, we use a time-series cross-sectional data set (also known as panel data). The data are yearly, and the sample pools all the politically relevant dyads from 1950 to 1992. Scholars in the literature define politically relevant dyads as pairs of countries that are either geographically contiguous or that include at least one major power (United States, United Kingdom, France, USSR/Russia, and China). The politically relevant dyads are allegedly most likely to experience dyadic political conflict. Third, the estimation of any simultaneous equations model differs from the estimation of a single-equation model in that it needs to take into account the simultaneity relationships between variables or their reciprocal effects on one another. We estimate our model using a variant of the twostage least squares method suggested by Maddala (1983), which is designed for a system of simultaneous equations of the type developed in this chapter and includes both dichotomous and continuous endogenous variables. We use the word “variant” since, as we discuss in the appendix and in the next subsection, the Maddala method is designed for the case with one dichotomous and one continuous variable, but our model includes one dichotomous and two continuous variables. Finally, as noted in previous chapters, the error term in the statistical model needs to satisfy some assumptions for statistical inferences to be valid. We address the related assumption violations using appropriate econometric techniques. We also discuss other complications in our case of the simultaneous equations model in the appendix.

Empirical Findings This section presents our key empirical findings for the relationship between military interstate conflict and democracy. We first discuss our main results,

Democracy and Conflict

141

followed by results from additional analysis. In the appendix of this chapter, we fully present the findings of the analysis and provide all the technical details for interested readers. Tables 5.2 presents the empirical results from the three equations of the simultaneous equations system. General observations on the performance of our simultaneous equations model suggest that the model performs well. The model’s goodness of fit between the predicted values and the data is high. The empirical results for the control variables, which we fully discuss in the appendix, are generally consistent with our expectations and previous findings from the literature. Starting with the results for the demL equation, the coefficient of midAB is negative and statistically different from zero. This means that dyadic conflict reduces the lower level of democracy in a dyad. In terms of the size of effect, a dyadic military dispute reduces the lower level of democracy in a dyad by about 0.16 – a decline of about 38% in the average lower democracy value in the sample (0.425). Now, because dyadic conflict also affects the lower level of democracy in the next period through its effect on the lagged lower level of democracy in the model, we can compute the effect of MID involvement on the lower level of democracy in the long run, based on the formula described in the appendix. By employing the coefficients of dyadic MID and the lagged lower level of democracy, we find that dyadic MID involvement reduces the lower level of democracy by 2.5 points over the long run – a 590% decline in the average value of the variable in the sample. Therefore, both the short- and long-run effects of dyadic military dispute on joint democracy are large in size. In terms of the results for the demH equation, the coefficient of midAB is negative and statistically different from zero. Thus, dyadic conflict also reduces the higher level of democracy in a dyad. With respect to the size of effect, a dyadic military dispute reduces the higher level of democracy in a dyad by about 0.13 – a decline of about 1.7% in the average higher democracy value in the sample (7.9). Similarly, because dyadic conflict also affects the higher level of democracy in the next period through its effect on the lagged higher democracy in the model, we compute the effect of MID involvement on the higher level of democracy in the long run. By employing the coefficients of dyadic MID dispute and the lagged higher democracy, we find that dyadic MID involvement reduces the higher level of democracy by about 1.6 points over the long run – a 20% decline in the average value of the variable in the sample. These short- and long-run effects of dyadic conflict on regime dissimilarity are not as large as their effects on joint democracy. Based on the results for the MID equation, the effect of demL on the likelihood of midAB is negative and statistically different from zero. Hence,

142

0.9364∗∗∗ (0.0053) 0.0619∗∗∗ (0.0125) −0.0385∗∗ (0.0150) 0.00002∗∗ (7.52e-06) 13.4305∗∗∗ (2.9239) −0.0066∗∗∗ (0.0015) −0.0071∗∗ (0.0032) −0.4276∗∗∗ (0.0844) 0.0091∗∗∗ (0.0011) 0.0845 (0.0535) −0.1597∗∗∗ (0.0418) −124.9733∗∗∗ (24.7767) 11644 0.95 N Adjusted R2

Constant

MIDAB

THIRD-PARTY MIDH

DIFFUSIONH

TRADEH

GROWTHH

GDPPCH ∗ YEAR

GDPPCH

INFLATIONH ∗ YEAR

INFLATIONH

YEAR

PRIOR DEMOCRACYH

Equation of DEMH 0.9163∗∗∗ (0.0088) 0.0547∗∗∗ (0.0193) −0.0872 (0.0647) 0.00004 (0.00003) 12.1750∗∗∗ (4.2280) −0.0061∗∗∗ (0.0021) −0.0053 (0.0046) −0.1059 (0.0685) 0.0047∗∗∗ (0.0010) 0.0701∗∗ (0.0348) −0.1327∗∗∗ (0.0310) −108.7286∗∗∗ (38.0668) 11644 0.94

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

N Adjusted R2

Constant

MIDAB

THIRD-PARTY MIDL

DIFFUSIONL

TRADEL

GROWTHL

GDPPCL ∗ YEAR

GDPPCL

INFLATIONL ∗ YEAR

INFLATIONL

YEAR

PRIOR DEMOCRACYL

Equation of DEML

N ␹2

Constant

AFFINITY

DEMH

DEML

GROWTH

TRADE DEPENDENCE

CAPABILITY RATIO

ALLIANCE

CONTIGUITY

Equation of MID

Table 5.2. Interactions among joint democracy, regime dissimilarity, and dyadic conflict

−2.1944∗∗∗ (0.1493) 11644 96.73∗∗∗

0.7829∗∗∗ (0.1375) −0.0907 (0.1388) −0.0285 (0.0409) −20.3836 (14.5072) −0.0287∗∗ (0.0116) −0.0207∗∗∗ (0.0067) 0.0027 (0.0106)

Democracy and Conflict

143

Table 5.3. Probabilities and relative risks of MID involvement

Scenario 1: Contiguous, nonallied dyads Baseline: all continuous variables are set at mean values Increase in DEML by one standard deviation; other variables at baseline values Scenario 2: Noncontiguous, nonallied dyads Baseline: all continuous variables are set at their mean values Increase in DEML by one standard deviation; other variables at baseline values

Probability

Relative risk

0.0546 0.0402

73.6%

0.0086 0.0057

66.3%

Note: Based on coefficients in MID equation in Table 5.2.

democratic countries are less likely to engage in MIDs against each other. This result agrees with those reported by many single-equation studies in support of the democratic peace proposition. The effect of demH on midAB is not statistically different from zero. Russett et al. (1998) and Oneal and Russett (1997) argue that demH is an important part of the link between political regimes and conflict. For example, interpreting demH as political distance, Russett et al. (1998: 457) argue that “greater political distance makes a dyad more prone to conflict.” The result in Table 5.2 does not support this claim, however; the difference is attributable to the fact that we model demH –conflict simultaneity, whereas Oneal and Russett (1997) and Russett et al. (1998) do not. Table 5.3 illustrates the substantive effects in the MID equation by computing the probabilities of MID for contiguous and noncontiguous politically relevant dyads. Scenario 1 sets all the continuous variables at their mean values, contiguity at 1 (contiguous countries), and alliance at 0 (nonallied countries). Scenario 2 repeats the experiment but sets contiguity at 0 (noncontiguous countries). In Scenario 1, the baseline probability of dyadic MID involvement is 5.46%. When demL rises by one standard deviation (6.48 units), the probability of midAB drops to 4.02%, a decline of 1.44%. Because conflicts are rare events, it makes more sense to talk about the relative risk between the two scenarios (i.e., dividing the new probability by the baseline probability). The computed relative risk is 73.6%, which means that relative to the baseline contiguous dyad, the risk of MID involvement is 26.4% lower if the dyad’s joint democracy rises by one standard deviation. How do these effects compare with those found by Oneal and Russett (1997)? We observe

144

Democracy and Economic Openness in an Interconnected System

a smaller absolute effect of joint democracy on the likelihood of dyadic MID (a decline of 1.44% in our results vs. a decline of 2.1% from 7.8% to 5.7% in their results). But the changes in the relative risk are similar between the two studies (26.4% vs. 26.9% from Oneal and Russett). In Scenario 2 (noncontiguous countries), the substantive effect of demL (joint democracy) on dyadic MID involvement is smaller by an order of magnitude. A rise of one standard deviation in demL reduces the probability of midAB (dyadic MID involvement) from 0.86% (in the baseline case) to 0.57%. The results of this experiment, which are not reported by Oneal and Russett (1997), lend support to the argument of James et al. (1999) that the effect of joint democracy on the probability of dyadic MID may be small. However, the relative risk is 66.3%. Relative to the baseline (0.86%), the reduction of 0.29% amounts to a decrease of 33.7% in the relative risk of MID involvement. That is, relative to the baseline noncontiguous dyad, the risk of MID involvement drops by 33.7% if the dyad’s joint democracy increases by one standard deviation. Overall, the empirical results support our theoretical expectation that the relationship between dyadic military conflict and democracy is reciprocal – they provide support for our argument that models that ignore the simultaneity of conflict and democracy miss something important about this structural relationship. Dyadic military conflict reduces both the lower and higher democracy levels in a dyad, although its short- and long-run effects on the lower democracy level are much stronger. At the same time, the lower democracy level in a dyad reduces the probability of dyadic military dispute, whereas the higher democracy level in a dyad does not.

Additional Analysis One additional analysis concerns the debate over the impact of affinity, which presumably measures preference similarity based on UN voting patterns. We present the additional results in Table 5.A1, based on a simultaneous equations model with affinity in the MID equation. The purpose is to check the sensitivity of our results in light of the debate between Erik Gartzke on one side and Bruce Russett and John Oneal on the other. This debate is not our main focus, but it provides an opportunity to evaluate the robustness of our results. For our purpose, we compare the results between Tables 5.2 and 5.A1 and find them quite similar across all three equations. The effects of demL and demH and midAB on each other are similar whether affinity values are included or not.

Democracy and Conflict

145

As we further explain in the appendix, Maddala developed a method to estimate a simultaneous equations model with one endogenous continuous variable and one endogenous dichotomous variable. However, his method only works in this particular case and we therefore could not use it for our three-equations model. Thus, we conduct another analysis to assess the sensitivity of our results to the Maddala correction of standard errors. In this additional analysis, we estimate a model with two endogenous variables, midAB and demL . In other words, we drop the demH equation. We do this estimation in two ways. In one way, we use the Maddala method per se, including the special features it includes. In the second way, we do not use the full Maddala method but rather we use its two-stage features and compute robust standard errors with cluster, the approach we employed earlier for the three-equations models. The results from these two tests are very similar. We therefore expect that had a Maddala method been available and applied to our three-equation setup, it would not have changed the inferences reported here. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Our simultaneous equations model of dyadic conflict and democracy is markedly distinct from the typical single-equation research designs in the democratic-peace literature. In previous studies, midAB was assumed to be the dependent variable and the simultaneity of conflict and democracy was ignored. We demonstrate that conflict and democracy do in fact influence one another. Our major innovation is the combining of the dyadic analysis of military conflict with the monadic determinants that explain democracy levels in a dyad. Our approach exploits the insights from two bodies of literature that have remained on two separate courses: the monadic literature on the causes of democratization and the dyadic democratic-peace literature. The empirical findings support our choice of a simultaneous modeling approach. The simultaneous modeling approach in this chapter brings several new things to the study of democracy and dyadic conflict. First, it shows democratic peace is not an artifact of the effect of conflict on democracy, which has been a topic of debate in the literature. Second, the analysis shows that the pacifying effect of joint democracy is smaller in the absolute size than in the work of Oneal and Russett once one considers the democracy– conflict simultaneity. Third, it shows that dyadic conflict reduces both joint democracy and dyadic regime dissimilarity, rejecting the view that

146

Democracy and Economic Openness in an Interconnected System

conflict promotes democracy. Finally, it shows that the congruence of interest between countries does not explain the lack of conflict between democracies in the context of a simultaneous equations framework, an issue debated among scholars who use single equations for the effect of democracy on conflict. Our findings suggest two policy implications that single-equation studies of MID fail to uncover. First, there is a negative relationship between dyadic conflict and joint democracy. Dyadic conflict reduces joint democracy, which makes future conflict more likely and moves the dyad away from peace. Hence, the preservation of peace requires early intervention in disputes before they cause the belligerents to become seriously less democratic. Second, there exists a positive relationship between dyadic peace and democracy. Peaceful dyads experience a rise in democracy for both nations. Thus, by brokering peace between adversaries, actors such as major powers or international organizations can help to facilitate the development of democracy, which in turn promotes peace. Although these effects may not be large, they are statistically significant and could well be important tools in conflict resolution. We need to stress that this chapter represents the beginning of a research program rather than a final analytic effort. We believe that the field of international relations could gain insight by repeating previous investigations of the relationship between conflict and democracy while using our simultaneous framework. In particular, we believe that the issue of the size of the effect of joint democracy on the likelihood of MIDs merits further investigation. For example, studies could use other control variables, other estimators, or other measures to gain additional insight. However, in our view, the simultaneity of conflict and democracy is too important to be ignored by future research. SUMMARY AND OUTLOOK

Many statistical studies in international relations investigate the claim that democracies do not fight one another. Virtually all these studies employ a single-equation design, where the dependent variable measures the presence or absence of a dyadic MID. A separate group of studies argues that conflict affects democracy and that its effects could be positive or negative. By and large, these two bodies of literature have not incorporated one another’s insights. We argue that democracy and dyadic conflict affect each other significantly and that statistical models that ignore the reciprocal nature of these effects may make incorrect inferences.

Democracy and Conflict

147

In this chapter, we developed a simultaneous equations model of democracy and dyadic conflict. The model not only connects the monadic literature on the causes of democratization with the dyadic democratic-peace literature, but it also includes, in an innovative manner, the monadic causes of democracy and the dyadic causes of military conflict in one system of equations. In a sample of all the politically relevant dyads from 1950 to 1992, we find that dyadic military disputes reduce joint democracy, and joint democracy reduces the probability of MIDs. The effect of military conflict on joint democracy is large. And the absolute effect of joint democracy on conflict in our study is smaller compared with the single-equation estimates in the literature, whereas, in relative terms, the effect is similar in size. The effect of joint democracy on MID involvement is considerably smaller for noncontiguous countries than for contiguous ones. In this chapter, international trade played a role in the analysis as one of our control variables in both the democracy and the military conflict equations. The effect of trade on military conflict, however, is one of those issues that in itself has generated enormous controversy in the fields of international relations and international political economy. Studies have continuously debated the merits of trade as a tool to achieve international peace, and the issue is certainly far from reaching closure. Part of the reason for this controversy probably has to do with some previously noted technical differences in terms of samples, measures, estimators, and control variables. In the next chapter, we argue that another important substantive source of differences, which so far has not received much scholarly attention, has to do with the composition of trade in terms of both sectoral types and the directions of trade flows (import and export). Our work focuses on these very issues in the following order. We first offer a formal model of the effects of trade, broken down along sectoral type and trade flow directions, on the propensity of nations to initiate conflict against their trading partners. The model is then tested empirically using statistical methods. As we shall demonstrate, our work will bring new insight into this relatively old debate.

APPENDIX CONCEPTUAL MODEL

The conceptual model focuses on linking the monadic causes of democracy and the dyadic causes of military conflict. Recall that demL is the lower

148

Democracy and Economic Openness in an Interconnected System

democracy score between demA and demB in dyad AB, where A and B are the dyad member countries; demH is the higher democracy score in dyad AB; midABU is the propensity of dyad AB to engage in a militarized dispute – a variable that is unobserved; MA is a vector of monadic attributes affecting the democracy of A (demA ); MB is a similarly defined vector for B (demB ); and F denotes a functional dependency. demA is a function of three components: MA , midABU , and midAJU . Similarly, demB is a function of MB , midABU , and midBJU . The elements of the vectors MA and MB are based on the literature of the determinants of democracy discussed in Chapter 2. The democracy score of country A is affected by any conflict that country A conducts with any country in its PRIE (see the text) and not only by conflict with country B. We divide the PRIE of country A into two parts: country B and other third-party countries (denoted J, where J = B). The propensity of conflict between A and B has already been denoted as midABU . The unobserved aggregated propensity for conflict that country A has with all third-party countries J in its PRIE is denoted by midAJU . Hence, in notation: demA = F(MA , midABU , midAJU ) demB = F(MB , midABU , midBJU ).

(5.A1)

Combining the definition of demL and Equation (5.A1), we get demL = F(MB , midABU , midBJU )

if demA > demB ;

demL = F(MA , midABU , midAJU )

if demA < demB .

(5.A2)

Similar to demL , the derivation of demH is given by demH = F(MA , midABU , midAJU ) if demA > demB ; demH = F(MB , midABU , midBJU )

if demA < demB .

(5.A3)

midABU depends on the democracy scores of both countries in the dyad. Hence, midABU = F(XAB , demL , demH ),

(5.A4)

where XAB is a vector of dyadic attributes that affect the likelihood of dyadic MID. The variables included in XAB are those used within the conflict literature. The democratic-peace literature expects that a rise in demL reduces midABU and that a rise in demH raises midABU .

Democracy and Conflict

149

The full simultaneous equations model is now given by Equations (5.A5), (5.A6), and (5.A7), which were presented in the text: demL = F(ML , midABU , midLRU ),

(5.A5)

demH = F(MH , midABU , midHRU ),

(5.A6)

midABU = F(XAB , demL , demH ).

(5.A7)

EMPIRICAL MODEL AND ANALYSIS

Empirical Model The elements of the vectors ML and MH in Equations (5.A3) and (5.A4) of the conceptual model include the following variables. We provide details on data sources, expected effects, and measure construction. The discussion of the democracy equations follows Chapter 2. gdppcL and gdppcH are the logged real GDP per capita values for the low- and high-democracy countries in the dyad, respectively. These data are taken from the Penn World Table 5.6. As noted in Chapter 2, many studies use GDP per capita as an indicator of economic development. We expect that gdppcL and gdppcH positively influence demL and demH , respectively, an expectation that is in line with the modernization theory – that economic development breeds democracy. growthL and growthH are the yearly growth rates of real GDP, computed using data from the Penn World Table 5.6, for the low- and highdemocracy countries in a dyad, respectively. As noted in Chapter 2, the democracy literature registers conflicting expectations regarding the effect of economic growth on democracy. Some studies argue the effect is negative, and other studies claim the effect is positive. inflationL and inflationH are the yearly inflation rates, based on the consumer price index, for the low- and high-democracy countries in a dyad, respectively. Several studies use inflation as a proxy for economic crisis, but scholars debate its effect on democracy (see Chapter 2). Data are from the International Financial Statistics CD-ROM (2000). tradeL and tradeH are the trade openness ratios for the low- and highdemocracy countries in a dyad, respectively. Trade openness is calculated as the sum of the values of imports and exports of a country divided by its GDP. Data are from the Penn World Table 5.6. As noted in Chapter 2, the debated effect of trade openness on democracy could be positive or negative.

150

Democracy and Economic Openness in an Interconnected System

year tests whether democracy has a linear trend. In addition, the effects of inflation and economic development on democracy may change over time. As we performed and discussed in Chapter 2, to capture this possibility we include interaction terms between year and inflation, and year and GDP per capita, in the equations for demL and demH , respectively. prior democracyL and prior democracyH are the prior democracy values for the low- and high-democracy countries in a dyad, respectively. We expect these variables to influence demL and demH positively. The strategy of including the lagged dependent variable in the model was explained in Chapter 2. diffusionL and diffusionH are the averages of the democracy scores of the countries in the PRIEs of the low- and high-democracy countries in a dyad, respectively. As in Chapter 2, they capture the diffusion of democratic norms due to contact-based mechanisms such as economic relations, communication networks, and influence by major powers, but in this chapter they are computed for the PRIEs of the low- and high-democracy countries, not their geographical regions. We expect the effect of diffusion on demL and demH to be positive. third-party midL and third-party midH are dichotomous variables. The value of 1 denotes a situation in which the low- or high-democracy country in a dyad is involved in conflict with at least one third-party country in its respective PRIE. These variables capture the notion that the democracy level of country A, for example, may be affected by its conflict with any third-party country, not just by its conflict with country B. Drawing on the literature behind Hypothesis H2, we expect that the effect of thirdparty conflict on democracy could be positive or negative. The third-party dummy variable has been coded based on Zeev Maoz’s (2001b) data on third-party conflict in the PRIE of each country in a dyad. Moving to the MID equation, unless otherwise specified we use data from Oneal and Russett (1999a). As noted in the empirical test, the unobserved midABU variable is replaced by its realization midAB , which is the presence or absence of a MID in dyad AB. Another possibility is to replace midABU with a multinomial representation, classifying several types of MIDs, as in James et al. (1999). However, this method is not employed here because the estimation of simultaneous equations models involving continuous and multinomial variables is not well developed in the literature and because we wish to compare our results to the bulk of the democratic-peace literature, which employs the dichotomous operationalization of MIDs. contiguity takes the value of 1 when the two states in a dyad are geographically contiguous (or are separated by up to 150 miles of water) and

Democracy and Conflict

151

zero otherwise. Contiguous states are expected to be more likely to experience conflict with one another. alliance equals 1 when both states within a dyad are members in a common military alliance, and zero otherwise. Alliance partners are expected to be less likely to experience conflict with one another. The capability ratio is the natural logarithm of the ratio of the more capable state’s composite national capabilities score to that of the less capable state in the dyad. For each country, this score rises as a state commands a larger resource base, comprising total population, urban population, energy consumption, iron and steel production, military manpower, and military expenditures. Dyads with an imbalance in capabilities are expected to be less likely to experience conflict with one another. trade dependence measures dyadic trade interdependence. Each state in a dyad has a bilateral-trade-over-GDP ratio, which represents the relative importance of trade with a dyadic partner. Following Oneal and Russett, we use the lower of the two trade dependence values in a dyad to measure the common level of dyadic trade dependence shared by the two states. Many scholars expect that dyadic trade dependence reduces conflict involvement, but the issue is debated. growth is the lower of the two economic growth rates in a dyad computed from the rate of change in the real GDP per capita for each state. The growth rates are obtained by using moving averages over a three-year period (substituted by one- or two-year-period averages when required values are missing). The literature offers conflicting expectations about the direction of this variable’s effect. This variable is computed using data from Penn World Table 5.6. affinity measures the level of preference similarity between two states in a dyad. This variable is based on the similarity of the states’ voting patterns at the UN General Assembly. Data are from Gartzke (1998). Several scholars debate whether it is affinity or joint democracy that causes peace in a dyad. Gartzke (1998) argues that preference similarity accounts for the hypothesized effect of democratic peace and reports that in the same model, affinity is statistically significant whereas joint democracy is not. In contrast, Oneal and Russett (1999b) argue that joint democracy indirectly contributes to preference similarity. Gartzke (2000) rejects this possibility.

Research Design Issues The relationship between some of our exogenous and endogenous variables may be simultaneous. Although this possibility applies to virtually all studies

152

Democracy and Economic Openness in an Interconnected System

in the literature, we nevertheless believe we ought to discuss the issue. Of course, we cannot model all potential simultaneous effects between our variables, because our model is already quite complicated. To deal with this issue, we lag each of our exogenous variables by a single year, as we did in previous chapters. The next issue concerns the Heckman (1978) condition for our model. Heckman shows that a consistency requirement needs to be satisfied when one estimates a simultaneous equations model in which an unobserved endogenous variable is replaced by an observed dichotomous variable. If this condition is not fulfilled, the estimation is not meaningful because the structural equations are not consistent with one another (Amemiya, 1978). Heckman considers a model with two equations: one for the observed variable, and the other for the unobserved. The dichotomous realization of the unobserved variable is an independent variable in each equation. Suppose the coefficient of the endogenous variable in the equation of the unobserved variable is ␥ 2 , and the coefficients of the dichotomous variable (linked to the unobserved variable) are ␦1 and ␦2 in each equation, respectively. Heckman’s condition requires ␦2 = ␥ 2 ␦1 . In our case, Heckman’s condition holds because the coefficients of midAB in all equations are zero. That is, the demL, demH , and midABU equations do not include midAB as an independent variable. In addition, ␥ 2 = 0 in our case, because the coefficients of demL and demH in the midABU equation are not zero. Hence, we can say that in our case the Heckman condition holds. We estimate our model using a variant of the two-stage least squares method suggested by Maddala (1983) and pooling the politically relevant dyad from 1950 to 1992. As in other studies, we do not deal with the timeseries dynamics of democracy and conflict. Dealing with issues of dynamics would require time-series techniques, which are still not well understood in the context of MIDs. We defer this issue to future research. In the first stage of Maddala estimator, each endogenous variable is regressed on all the exogenous variables in the model. The midAB equation is estimated using probit. The equations of demL and demH are estimated using ordinary least squares. In the second stage, the endogenous variables on the right-hand side of each equation are replaced by their predicted values from the first stage, respectively. In the MID equation, the linear predictor is used – not the predicted probability. Maddala (1983) assumes that the error terms are white noise. He employs the coefficients from the second stage, as we do, but he applies a correction for their standard errors. Our model consists of one dichotomous and two continuous endogenous variables and we do not assume that the error terms are necessarily white noise. We do not use Maddala’s correction because,

Democracy and Conflict

153

to the best of our knowledge, the literature has not extended it either to a model with more than two equations or to a case with possible nonspherical errors. Following the suggestions of Guilkey et al. (1992), Bollen et al. (1995), and Alvarez (1997), we use the standard errors computed in the second stage of the procedure to test our hypotheses. However, to assess the sensitivity of our results to Maddala’s method, we estimate a model with two endogenous variables (midAB and demL ) in two ways: one with the Maddala correction and the other with robust standard errors (that can control for potential nonspherical errors). We need to consider the potential existence of heteroskedasticity and serial correlation in the model’s error terms. One may suggest that we examine the correlation structure of the error terms in the equations. So far the literature has not developed a diagnostic test that is appropriate for our case. Developing such a test is difficult for at least two reasons. First, one of the equations in our model has a dichotomous endogenous variable in a time-series cross-sectional design. Second, we have a structural equations model with one dichotomous and two continuous endogenous variables. Conventional tests such as Durbin–Watson are not designed to diagnose such models. Hence, future research is called for in this area. That said, in the presence of heteroskedasticity and serial correlation, a model’s estimated coefficients are consistent, but its standard errors are inefficient and could be biased. To correct for these potential problems, we estimate our model using a variant of the White (1980) estimator of robust standard errors in the presence of heteroskedasticity and adjust for clustering over dyads to help account for possible temporal dependence of the error terms within each dyad. This estimator yields consistent estimation of the covariance matrix under very general conditions of heteroskedasticity and serial correlation (Wiggins, 1999). Finally, we demonstrate the importance of the results by examining the effect size of the key variable. We look at the direct effect of the key variables and their long-run effects through the lagged endogenous variables in the two democracy equations. The long-run impact of dyadic MID on democracy is based on the following formula: [coefficient of MID variable/(1 − coefficient of lagged endogenous democracy variable)].

Empirical Findings In this part of the appendix, we provide additional details on the sample and some diagnostics and discuss results for the control variables. The sample sizes for Models 1 and 2 in the tables shown in the text are 11,644

154

Democracy and Economic Openness in an Interconnected System

and 11,022 observations, respectively. These samples are somewhat smaller than the samples used in some single-equation-based studies of MIDs, which is due to the fact that our model includes more variables than does the typical single-equation design and the data on some of these variables are missing for some years. While this places a potential limitation on our empirical analysis, we believe that our results are applicable to other samples, because our theoretical argument is not sample-dependent and our sample is nevertheless large. Table 5.2 in the main text presents the statistical results for all three equations. As noted in the text, the model’s goodness of fit is high. The adjusted R2 for the democracy equations is 0.94 or 0.95, and the ␹2 model fit test in the MID equation is statistically significant. We now discuss the results for the control variables in the demL equation. The effect of prior democracyL on demL is significant and positive, demonstrating the importance of democratic inertia. The effect of diffusionL is positive. As the level of democracy in the less democratic country’s PRIE rises, so does its own democracy level. The effect of year is significant and positive, reflecting the tendency of democracy to increase over time in our sample. The effect of inflationL on demL is significant and negative, and its interaction with year is significant and positive. Hence, a rise in inflation reduces a country’s level of democracy. This effect is not stable over time, however, as is also reported by Gasiorowski (1995) and in Chapter 2. The effect of GDP per capita on demL is significant and positive, whereas its interaction term with year is significant and negative. This replicates a known result in the literature; that is, democracy rises with economic development. We also find, however, that this effect declines over time. The literature offers contradicting expectations regarding the effect of economic growth on democracy. We find that this effect is significant and negative. As in Chapter 2 and Gasiorowski (1995), we also find that greater trade openness reduces the lower democracy score in a dyad. One interpretation is that trade widens the social cleavages between winners and losers, thus reducing democracy. This finding is consistent with the result in Chapter 2. Next we turn to the results for the demH equation. As one could expect, the results for the control variables are generally consistent with those from the demL equation, which adds justification for our modeling approach. The only exceptions are the effects of inflation, growth, and trade on demH , which are no longer significant in spite of having the same signs as in the DEML equation. This is not surprising because the high-democracy

Democracy and Conflict

155

countries in a sample of politically relevant dyads tend to be developed countries, for which the levels of democracy tend to be high and stable. On the other hand, the low-democracy countries are typically less developed, exhibiting larger changes in democracy over time. In our sample, most of the MIDs are purely dyadic. Thus, we ought to expect that democracy levels would mostly be affected by the variable midAB . However, despite their relative rarity, third-party disputes also could affect democracy. We find that the effects of third-party mid on demH and (less so) on demL are positive and statistically significant. We can explain this empirical result by noting that the demH countries in a sample of politically relevant dyads tend to be countries such as the United States, the United Kingdom, France, Israel, and West Germany. Between 1950 and 1992, these countries were at times involved in multiple militarized disputes – a fact that has manifested itself in the positive correlation between demH and third-party mid (correlation = 0.136). The demL countries in the sample are typically less developed and tend to be involved in fewer MIDs with third parties, which is also demonstrated by a positive but weaker correlation between demL and third-party mid (correlation = 0.049). Finally, we discuss the results for the MID equation in Table 5.2. The effect of contiguity on the likelihood of MIDs is statistically significant and positive, as one would expect from the extant literature. The effect of growth on midAB is statistically significant and negative, which also is consistent with some prior results. The effects of alliance, capability ratio, and trade dependence, however, are not statistically significant in our results, though they have the same sign as in comparable singleequation studies. We attribute these differences to the fact that our model accounts for the simultaneity of conflict and democracy, whereas previous studies have not.

Additional Analyses Table 5.A1 reports the results for all three equations based on the additional analysis and concerns the debate over the impact of affinity. The threeequation model adds the affinity variable to the MID equation. As noted, these results shed light on a recent debate in the single-equation MID literature. Gartzke (1998, 2000) argues that democratic peace is driven by dyadic preference similarity. In his studies, when affinity is included, the effect of joint democracy on the likelihood of MID is not statistically significant. Oneal and Russett criticize Gartzke’s approach and reject his

156

0.9346∗∗∗ (0.0056) 0.0615∗∗∗ (0.0131) −0.0388∗∗ (0.0150) 0.00002∗∗ (7.56e-06) 13.1769∗∗∗ (3.0734) −0.0065∗∗∗ (0.0016) −0.0066∗∗ (0.0033) −0.4638∗∗∗ (0.0886) 0.0096∗∗∗ (0.0012) 0.0968∗ (0.0557) −0.1742∗∗∗ (0.0434) −124.3320∗∗∗ (25.9593) 11022 0.95 N Adjusted R2

Constant

MIDAB

THIRD-PARTY MIDH

DIFFUSIONH

TRADEH

GROWTHH

GDPPCH ∗ YEAR

GDPPCH

INFLATIONH ∗ YEAR

INFLATIONH

YEAR

PRIOR DEMOCRACYH

Equation of DEMH 0.9143∗∗∗ (0.0090) 0.0602∗∗∗ (0.0204) −0.0841 (0.0649) 0.00004 (0.00003) 13.4123∗∗∗ (4.4671) −0.0067∗∗∗ (0.0023) −0.0060 (0.0048) −0.1384∗ (0.0722) 0.0050∗∗∗ (0.0011) 0.0635∗ (0.0360) −0.1295∗∗∗ (0.0316) −119.7263∗∗∗ (40.2063) 11022 0.94

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

N Adjusted R2

Constant

MIDAB

THIRD-PARTY MIDL

DIFFUSIONL

TRADEL

GROWTHL

GDPPCL ∗ YEAR

GDPPCL

INFLATIONL ∗ YEAR

INFLATIONL

YEAR

PRIOR DEMOCRACYL

Equation of DEML

N ␹2

Constant

AFFINITY

DEMH

DEML

GROWTH

TRADE DEPENDENCE

CAPABILITY RATIO

ALLIANCE

CONTIGUITY

Equation of MID 0.8041∗∗∗ (0.1346) −0.0543 (0.1577) −0.0498 (0.0445) −17.0885 (13.7160) −0.0261∗∗ (0.0123) −0.0188∗∗∗ (0.0065) −0.0018 (0.0110) −0.2145 (0.2127) −2.0699∗∗∗ (0.1777) 11022 82.75∗∗∗

Table 5.A1. Interactions among joint democracy, regime dissimilarity, and dyadic conflict, controlling for affinity

Democracy and Conflict

157

findings. Our model facilitates a more accurate assessment of this debate by modeling the reciprocal relationship between demL and midAB . We find that the effect of affinity on the likelihood of dyadic MID involvement is negative but statistically insignificant. On the other hand, the effect of joint democracy on the likelihood of MID involvement remains negative and significant. Our results thus support the position of Oneal and Russett in this debate. However, it is worth noting that affinity, based on UN voting patterns, may be a weak proxy for dyadic preference similarity. Hence, the question of whether preference similarity affects dyadic MID involvement would benefit from the development of better measures of this concept in future research. Finally, as noted, to assess the sensitivity of our results to the Maddala correction of standard errors, we estimate a model with two endogenous variables (midAB and demL ) in two ways: using the Maddala correction and using robust standard errors. Keshk (2002) provides details on the implementation of this estimator. The results from these tests are very similar, which is consistent with the findings from Monte Carlo simulations reported by Guilkey et al. (1992) and Alvarez (1997). They find that the Maddala correction has little effect on statistical inferences. We therefore expect that had a Maddala correction been available and applied to our three-equation setup, it would not have changed the inferences reported here.

SIX

Economic Openness and Conflict

INTRODUCTION

In this chapter, we shift our attention to the interaction between military conflict and the other focus of our book – economic openness – which played a crucial part in Chapters 2 and 3. Economic openness is measured in various ways, but in this chapter we emphasize the trade dimension of economic openness, which connects the analytical framework of our book to an important debate in the related fields of international relations and international political economy: the trade and conflict controversy. The idea that trade promotes peace dates from at least the late eighteenth century,1 and has two contemporary explanations: the liberal argument and the bargaining argument. The antithesis that trade generates conflict also has a long intellectual history,2 which is often explained from a neoMarxist view or a neomercantilist derivative of realism. So far, scholars have mathematically demonstrated the logical consistency of the liberal and bargaining arguments.3 These formal models consider aggregated or total trade, ignoring variations of trade across economic sectors and flow directions (export and import). With a few exceptions, statistical studies have also used total bilateral trade. Most of these studies find that a rise in trade reduces the likelihood of conflict. It is a fair statement that most theoretical and empirical works on the effect of trade on conflict ignore both the composition of trade across economic sectors and the trade flow directions in terms of export and import. We challenge the prevalent approach and provide new insights on the effect of trade on conflict. Indeed, it does not seem likely that international 1 2 3

See, e.g., de Montesquieu (1748), Smith (1776), Kant (1795), and Angell (1912). See, e.g., Hobson (1902), Lenin (1916), Hirschman (1945), and Waltz (1970). See, e.g., Polachek (1980) and Gartzke et al. (2001).

158

Economic Openness and Conflict

159

exchanges of heterogeneous commodities and services (say, oil, bananas, corn, machinery, fish, and computers, to name just a few) should have identical effects on conflict. As one renowned scholar has repeatedly warned, “unduly extensive aggregation of trade data” is likely to produce “biased estimates of the ability of bilateral trade to diminish conflict” (Polachek, 1992: 113; 2000: 10).4 We suggest that the theoretical and empirical focus on total trade results in two aggregation biases: one from aggregation across economic sectors or goods and the other from aggregation across trade flow directions (import and export).5 To be sure, some studies do look at the effects of trade in different sectors on military conflict – but without formally providing a microfoundation – and almost all the existing statistical studies of this subject except that of Dorussen (2006) have used very limited samples. Meanwhile, to the best of our knowledge, the possibility that the effects of trade on military conflict may differ between that of export and import has not received attention in the literature. In this chapter, we offer a theoretical formal model to explain how export and import flows in specific economic sectors influence the decision to initiate military conflict, a question almost all previous theories have ignored. Like virtually any formal model, our model also relies on simplifying assumptions. As in neoclassical economics, we model trade as performed by economic agents in the marketplace. As in international relations, we model conflict as a choice made by leaders. We believe that these assumptions are generally reasonable and allow us to model and anticipate the individual effects on military dispute initiation of bilateral imports and exports in different sectors. Simply put, our theory suggests that leaders are less likely to initiate conflicts that cause trade losses. This idea is not new, but our first contribution here is to show that this consideration could vary across import, export, and traded sectors. And our theory goes beyond this argument: it predicts that leaders are more likely to initiate conflicts that increase profits to their trade. Formalizing these ideas, we show that the costs and the benefits of conflict can vary across trade sectors, exports and imports, and that they work through the expected effect of conflict on trade prices. Leaders initiate conflict if they conclude that it will be economically beneficial by 4

5

Polachek is not alone. Russett and Oneal (2001: 141) also note that “dyadic [total] trade, even when adjusted for the size of the overall economy, is an imperfect indicator of economic interdependence. For one thing, the composition of trade is not considered.” Surveys of the trade and conflict literature conclude that researchers should consider disaggregating total trade. See Sayrs (1990: 22), McMillan (1997: 53), and Reuveny (2000: 37). We use the terms “goods,” “economic sectors,” and “sectors” interchangeably.

160

Democracy and Economic Openness in an Interconnected System

reducing the price of their import and increasing the price of their export, ceteris paribus. The decision depends on certain good-, export- and importspecific parameters. Using estimates of these parameters, our model predicts the effects of changes in exports and imports in the agriculture/fishery, energy, chemical/ mineral, manufactured products, and miscellaneous-consumption sectors on military interstate dispute (MID) initiation. We test these predictions using a large-N sample of all the directed-dyads from 1970 to 1997 for which data are available. We find that the effects of bilateral trade on MIDinitiation vary across sectors and between flow directions. Increases in agriculture/fishery imports and energy imports reduce the probability of MID initiation, and increases in energy exports, manufacturing imports, and manufacturing exports increase this probability. Yet changes in agriculture/ fishery export, chemical/mineral trade, and miscellaneous consumption products trade generally do not affect MID initiation. Our research, we believe, offers important insights for both academics and policymakers. As we explain in detail in the concluding section, our analysis sheds new light on a long-standing debate in international relations and offers an opportunity to rethink the logic of how trade affects conflict. Our theory encompasses the liberal argument as a special case and offers an alternative explanation to the bargaining, neo-Marxist, neomercantilist realist, and classical realist arguments. It also has various important policy implications. The remainder of this chapter is organized as follows. The next section presents the motivations of our analysis and discusses previous studies, which is followed by three sections that develop the theoretical model, a section that discusses the model’s implications and formulates hypotheses, and a section that presents the empirical model and discusses several research design issues. With these preliminaries out of the way, we turn to the statistical results of our analysis and the public policy implications of these findings. The last section summarizes the analysis and links the discussion to the next chapter. The appendix of this chapter presents technical details pertaining to the design of our analysis and the empirical results. DISAGGREGATED TRADE IN THE CONFLICT LITERATURE

The contemporary literature on the effect of trade on conflict is too large to fully review here.6 Generally speaking, scholars offer five main theoretical 6

For extensive reviews of the literature, see, e.g., Sayrs (1990), McMillan (1997), Reuveny (2000), Mansfield and Pollins (2003), and Schneider et al. (2003).

Economic Openness and Conflict

161

explanations. The classical liberal view argues that trade promotes peace because military conflict reduces the trade that countries value (e.g., Rosecrance, 1986; Polachek, 1992, 2000; Russett and Oneal, 2001). The bargaining approach considers trade a tool that facilitates bargaining over some contested political issue, expecting that trade reduces the likelihood of military conflict over the issue (e.g., Morrow, 1999; Gartzke et al., 2001; Crescenzi, 2003). A neo-Marxist view (e.g., dos Santos, 1970; Choucri and North, 1975; Ashley, 1980) and a neomercantilist derivative of realism (e.g., Waltz, 1970, 1979; Gilpin, 1981; Borrus and Zysman, 1992; Barbieri, 2002) envision trade as a tool of influence and exploitation, which can cause conflict if countries resent it or to peace if they comply. Classical realism expects that trade has no systematic effect on military conflict, the causes of which are political and strategic instead (e.g., Buzan, 1984; Gilpin, 1987; Ripsman and Blanchard, 1997; Keshk et al., 2004). Most existing empirical evidence supports the expectation of the first two explanations, that bilateral trade reduces interstate militarized disputes; but evidence to the contrary also exists, which the classical liberal view and the bargaining approach cannot explain. Attempts to apply some, but not all, of the five explanations to account for the effects of trade in different sectors on peace have produced conflicting conjectures. One argument expects that trade in sectors in which nations have a decisive comparative advantage facilitate peace, because such trade brings more economic benefits that would be lost to conflict. Another channel to peace is said to involve imported goods (e.g., oil, machinery, minerals) that a country considers important, the loss of which hinders economic growth. A third channel argues that because imported consumer goods benefit consumers, they support leaders that promote peace with the exporter.7 Other studies focus on strategic goods, including minerals, chemicals, steel, fuels, and high-tech goods. When strategic goods have no readily available substitutes, their trade may generate conflict,8 because countries may feel threatened by their vulnerability and resort to conflict to mitigate the scarcity. Countries also may be tempted to use trade as a political tool to exploit vulnerabilities of and exercise influence on others, causing resentment and conflict.9 Yet trade in strategic goods may also generate friendliness as one country supports another to protect its trade. A related expectation is that trade in goods with fewer substitutes may generate peace 7 8 9

Gasiorowski and Polachek (1982), Arad et al. (1983), Domke (1988), Polachek (1980, 1992). See Hirschman (1945), Baldwin (1985), Gasiorowski (1986), Sayrs (1989), and Førland (1991). Reuveny (1999a, 1999b) shows how Israel uses its trade with the Palestinians to affect their actions. For general discussions, see Gilpin (1984), Sen (1984), Borrus and Zysman (1992), and Vogel (1992).

162

Democracy and Economic Openness in an Interconnected System

because countries tend to cooperate to secure the continuing flows of such goods.10 Some scholars debate the usefulness of the concept of strategic goods. Førland (1991: 197) quotes President Eisenhower, saying, “if our opponent needs something badly, then that something is strategic.” Schelling (1958: 500) notes that any good will be strategic to a country if it plans to consume it but does not produce it. Baldwin (1985: 214–215) explains that the strategic good can be “[any good] that is needed to pursue a given strategy and that is relatively inefficient to produce at home.” He warns that it is hard to identify a priori the “strategic quality of a good” since it depends on the extent that the good has substitutes and on whether the good is important for a certain case. Studies on international competitiveness argue that trade in high-tech goods creates positive externalities for economic growth and the production of weaponry. International competition over trade shares in these goods can lead to conflict. Moreover, states wishing to excel in economic growth, or states suspecting that others may translate growth into military power in the future, will maximize relative gains from trade in high-tech goods. The attempt of one country to gain more from trade than their trade partner can, in turn, lead to tension and ultimately conflict.11 Only a few empirical studies attempt to evaluate the effect of disaggregated trade on conflict, and none of them, with one exception, focuses on military conflict and their samples include only a few dyads. Bennett et al. (1992) studied the effect of U.S.–Japan trade on political relations from 1948 to 1978 and concluded that trade in steel, textiles, and electronics generates conflict. Park et al. (1976) found that, as oil exports rise, oil-exporting countries become hostile toward their trade partners. Gasiorowski and Polachek (1982) conclude that trade in agriculture and manufactured goods induces cooperation by the Warsaw Pact toward the United States. Polachek and McDonald (1992: 277) study the effect of disaggregated trade in raw materials and manufactured goods on conflict for 14 OECD countries in 1973. They report that trade reduces conflict but caution that these results are preliminary due to limited data coverage. Using a large-N sample, Dorussen (2006) finds that disaggregated trade reduces the likelihood of threats and use of force but not in all sectors; the effects of nonmanufactured goods, chemicals, metal products, building supplies, and electronics are not statistically significant. 10 11

On the possibility of trade in strategic goods leading to cooperation due to compliance, see Hirschman (1945), Baldwin (1985), Reuveny and Kang (1998), and Reuveny (2001b). This view often traces back to List (1856). See also Tyson (1992) and Mastanduno (1992).

Economic Openness and Conflict

163

In sum, the literature presents conflicting theoretical conjectures and mixed findings. In the following sections, we attempt to add clarity to the trade and conflict literature through a model of the effects of trade, disaggregated across sectors and flow directions, on the initiation of military conflict. THEORETICAL MODEL

To study the effects on conflict initiation of bilateral trade broken down across goods and between flow directions, we need to model the attributes of traded goods such as sensitivities to conflict and price changes, and market forces such as import demand, export supply, and equilibrium. Stated broadly, our model considers two countries that trade in various goods with one another and maintain some political relations. We conceptualize conflict initiation (or the lack thereof) as a change in bilateral relations. A change in trade affects conflict initiation because conflict is expected to affect the price of a bilaterally traded good. When the price rises, the exporter’s profit and the importer’s cost rise. A country initiates conflict to increase export profit or reduce import cost, ceteris paribus. In our model, trade provides the incentive or disincentive for one state to initiate conflict toward another. Our logic, attributing the effect of trade on conflict to considerations of profit and cost, resembles the classical liberal argument, whose microfoundation is best demonstrated in the decision-theoretic model of Polachek (1980, 1992, 2000). We employ the approach of Polachek, but our model differs from his in two major ways. First, Polachek assumes that an exporter country that initiates conflict against a target will be paid less for its export to that country, and that an importer country that initiates conflict against a target will have to pay more for its import from that target. We do not make such assumptions; rather we explicitly allow import and export prices to be determined by demand and supply in the market. Second, Polachek’s model predicts that a rise in trade always brings peace, regardless of its composition or pattern. Our model predicts that the effect of a rise in bilateral trade on conflict initiation changes across traded sectors and between trade flow directions, generating ex ante sector- and flow-specific hypotheses. Our model includes two countries, A and B. We conceptualize political leaders of countries A and B as unitary rational actors whose utility functions rise with national consumption and military conflict they initiate toward each other. The conflict that country A sends toward country B is denoted by conflictAB . The unitary actor assumption is a simplification intended to keep the math of the model tractable. The assumption that leaders’ utility

164

Democracy and Economic Openness in an Interconnected System

rises with conflict, ceteris paribus, may seem counterintuitive, but in fact it is implicit or explicit in all conflict models. Although conflict is costly, leaders may conclude that its benefits outweigh its costs. Assuming leaders are rational actors, when country A is observed to initiate conflict toward country B, it must be the case that country A’s leader believes this action will increase his or her expected utility.12 As usual, consumption equals national output plus import of A from B, importAB , minus the export of A to B, exportAB . When leaders choose conflict, they take consumption, import, export, and national output as given, as in Polachek’s model. Country A’s leader chooses the level of conflict against its trade partner, B, to maximize utility, constrained by the need to maintain some trade balance with country B. The mathematical details of the model are in the chapter appendix. Here we offer an accessible, verbal discussion of the key ideas. As the conflict that A initiates against B rises in our model, the prices of country A’s imports from country B and its exports to B change. These price changes are determined endogenously inside the model. This expectation differs from Polachek’s framework, which assumes that conflict always raises the price a country pays for its import and reduces the price it receives for its export. Although this is certainly a plausible possibility to consider, it is not the only possibility. The reason, as we shall demonstrate, is the fact that the prices of traded goods are determined in international markets rather than in a deterministic manner through model assumptions. Economic logic suggests that supply and demand determine the prices and quantities of traded goods between countries A and B, and this logic is fully embedded in our model. Each traded good involves a supply side and a demand side. The demand of an importer defines how much of a good to import at different prices, whereas the supply of an exporter defines how much to export at different prices. As the price of a good rises, the demand for this good falls and the supply of this good rises. The point at which supply equals demand determines the equilibrium price of each traded good in the dyad. More specifically, suppose in dyad A–B that A exports to B in some sector i (i.e., B imports from A), and A imports from B in another sector j (i.e., B exports to A). In equilibrium, country A’s demand for good j equals country B’s supply of good j to A. If B does not send enough of good j to A, country A’s demand for good j becomes larger than country B’s supply. In 12

Polachek (1980) employs the same leader’s utility function we employ here. For an exposition of the unitary actor model in international relations, and the expected utility approach to modeling the determinants of military conflict, see Bueno de Mesquita (1981).

Economic Openness and Conflict

165

this case, the price of good j rises, prompting B to increase its supply to A. Sometimes, country B’s supply of good j is larger than country A’s demand. In this case, the price of good j declines due to the lack of demand in A, and B will reduce its supply. In equilibrium, the demand for bilateral import of good j equals the supply of bilateral export of that good. The same logic also applies to the export flow in sector i from A to B. How do trade prices change in response to conflict? This question stands at the center of our model. To answer it, we need to start with the supply and demand of each traded good. The formal import demand model presented in the appendix embeds the usual economic expectations that country A’s import demand for a product from B rises with country A’s income and falls with the good’s price. To this setup, we add a political–economic feature. We assume that conflict reduces country A’s import demand, for which we provide justifications shortly. The exporter side also follows the mainstream approach. As is usual in economics, the formal model in the appendix assumes that export supply rises with a good’s price and with the exporter’s income. The supply equation in the appendix includes a similar political–economic feature as in the import demand equation; that is, country B’s export supply to A falls with dyadic conflict. Several reasons explain our assumptions that both the import demand and the export supply fall with conflict. Importers and exporters may want to punish hostile partners by reducing trade. The importer may seek to reduce dependence on a hostile supplier, which may stop selling, whereas the exporter may seek to reduce dependence on a hostile importer, which may stop buying. Additionally, conflict often raises costs for both sides (e.g., higher risk insurance premium, delays in transportation, damages to goods), reducing trade. Furthermore, governments often restrict trade with foes, seeking to influence their opponents by weakening their economic and military capabilities. Various empirical examples and statistical studies support this assumption.13 In equilibrium, bilateral import demand for good j equals its bilateral export supply. We employ this condition to solve for the equilibrium price of good j. The solution indicates that the price of good j from B to A rises with country A’s income and falls with country B’s income, as is usually the case in bilateral trade models. More important, the direction of the effect of conflict on the price of the imported good depends on the difference 13

For additional discussion of these ideas see, e.g., Pollins (1989), Morrow et al. (1998), Reuveny (2001a, 2001b), and Li and Sacko (2002).

166

Democracy and Economic Openness in an Interconnected System

between two parameters: ␣3 (the effect of conflict on country B’s export supply of good j to A) and ␤3 (the effect of conflict on country A’s import demand for good j from B). Specifically, if ␣3 < ␤3 , the import price, denoted as PM, falls with conflict. If ␣3 > ␤3 , the import price rises with conflict. If ␣3 = ␤3 , the import price does not change with conflict. Figure 6.1 illustrates how the equilibrium import price, PM, for good j is determined at the intersection of the bilateral import demand and export supply curves. In each panel, the system is initially at equilibrium E1 , where country A’s import demand (MD A1 ) and country B’s export supply (MSB1 ) intersect. The subscript 1 indicates the initial state of affairs, where conflict has not occurred. Now, as A initiates conflict against B, curve MD A1 (country A’s import demand) shifts inward to curve MD A2 , because country A demands less from country B. The curve MSB1 (country B’s export supply) shifts inward to curve MSB2 , because country B supplies less to country A. The subscript 2 indicates the new state of affairs due to conflict. The shifts in the supply and demand curves due to conflict in Figure 6.1 are determined by ␣3 and ␤3 , respectively, which, as noted, represent the strengths of the effects of conflict on supply and demand. The market adjusts and reaches the new equilibrium at E2 . A key point from Figure 6.1 is that the new equilibrium price at E2 can be higher than, lower than, or the same as the initial equilibrium price at E1 . In panel A, the export supply curve (MSB ) shifts inward less than the import demand curve (MD A ) (i.e., ␣3 < ␤ 3 ), so the price of good j (PM) falls. In panel B, MSB shifts inward more than MD A (that is, ␣3 > ␤ 3 ), so PM rises. In panel C, the two inward shifts are equal in size, so PM does not change. What determines the size of the shift of the demand or supply curve? If export supply is sensitive to conflict, the export’s effect on raising PM increases with ␣3 . This effect, however, declines if the export supply turns more elastic; that is, if the supply curve becomes flatter in Figure 6.1. Similarly, if import demand is sensitive to conflict, the import’s effect on reducing PM rises with ␤3 . And this effect decreases if import demand becomes more elastic or flatter. So far, we have discussed country A’s import of good j from B (i.e., country B’s export to A). We now turn to country A’s export of good i to B (i.e., country B’s import from A). Country B’s demand for country A’s export and country A’s supply of country B’s import are determined in manners similar to that discussed earlier for the opposing trade flow. The mathematical details in the chapter appendix demonstrate similar outcomes. In other words, the price changes of this trade flow depend on the sensitivity of the demand of good i to conflict (which we denote as ␦, which is comparable in meaning to ␤ in the opposing trade flow discussed earlier)

Economic Openness and Conflict

Figure 6.1. Disaggregated bilateral trade equilibrium.

167

168

Democracy and Economic Openness in an Interconnected System

and the sensitivity of the supply of good i to conflict (which we denote as ␥ , which is comparable to ␣ in the opposing trade flow discussed earlier). The shifts of these demand and supply curves for good i also depend on their price elasticities (or slopes). Having identified how prices would respond to conflict, we analyze how these anticipated price changes influence the effects of trade on conflict initiation. In the chapter appendix, we present the details of the mathematical derivations. Following are the expectations derived from the model:14

 i = 1, 2, . . . ; (6.1)
(conflictAB ) = KE ␥3i − ␦3i ·
exportiAB
j 
 j j
(conflictAB ) = KI ␣3 − ␤3 ·
importAB j = 1, 2, . . . . (6.2) According to Equations (6.1) and (6.2), where i and j indicate particular economic sectors, changes (denoted by the Greek letter
) in bilateral trade flows affect the changes in conflict that country A sends toward country B. Specifically, Equation (6.1) indicates that a change in country A’s export of good i to country B,
(export iAB ), influences country A’s decision to change the conflict it initiates against country B,
(conflictAB ). The term KE is a positively signed term determined by the math of the model, as shown in the appendix. Since KE is positive, the sign of the effect of country A’s export on conflict depends on the relative size between ␦3 and ␥ 3 . Therefore, for any given good exported by A to B, if ␥ 3 > ␦3 , a rise in exportAB (country A’s export to B) raises the conflict A initiates toward B, conflictAB . If ␥ 3 < ␦3 , a rise in exportAB reduces this conflict. If ␥ 3 = ␦3 , a rise in exportAB does not affect conflictAB . Equation (6.2) indicates that a change in country A’s import of good j j from B,
(import AB ), affects country A’s decision to send conflict against B,
(conflictAB ). The term KI is a positively signed term determined by the math of the model, as shown in the appendix. Since KI is positive, the sign of the effect of country A’s import on conflict depends on the relative size between ␤3 and ␣3 . Therefore, for any given good imported by A from B, if ␣3 < ␤3 , a rise in importAB (country A’s import from B) raises conflictAB . If ␣3 > ␤3 , a rise in importAB reduces this conflict. If ␣3 = ␤3 , a rise in importAB has no effect on conflictAB .

Implications of the Model and Hypotheses What are the theoretical implications of our model predictions? And how do we formulate testable hypotheses based on these model predictions? We address these questions in this section. The theoretical model generates two 14

In Equations 6.1 and 6.2, i and j are superscripts denoting good type, not mathematical exponents.

Economic Openness and Conflict

169

new implications for the trade and conflict literature. First, the effects of country A’s export to B and country A’s import from B on the conflict A initiates against B can differ, because they depend on different coefficients (␣3 and ␤3 for import, and ␥ 3 and ␦3 for export). Second, the effects of different traded goods on the conflict that A initiates against B can vary across sectors because ␣3, ␤3 , ␥ 3 , and ␦3 vary across sectors. These implications differ from the thrust of the literature and provide new insights into the effect of trade on conflict. Focusing on total bilateral trade, previous studies ignore, implicitly or explicitly, the disparities in the effects of different trade flows in heterogeneous economic sectors. What is the logic behind the model? The effects of trade on conflict in our model are driven by the economic costs and benefits one anticipates from the expected conflict-induced price movements. For example, exporters expect to benefit from conflict if a rise in conflict causes the export price to rise. This export price increase occurs if the bilateral supply falls with conflict (raising price) more than the bilateral demand falls (lowering price), but when does the bilateral supply of a traded good fall more than its bilateral demand? This situation depends on the extent of substitution of the traded good by alternative suppliers or buyers. Expecting higher costs due to conflict, exporters may pursue alternative buyers, and importers may seek other suppliers. Both actions divert trade from foes, but imported and exported goods have varying degrees of substitution, ranging from full to none. If the exporter has alternative buyers for its good, it supplies less to its foe and more to others, which implies an inward shift of the supply curve, raising the price of the traded good. Expecting conflict, the importer may also buy from other suppliers, which implies an inward shift in the demand curve, reducing the price of the traded good. The trader that has relatively more alternatives is better positioned to benefit from conflict-induced market changes, and vice versa. Figure 6.1 illustrates this logic graphically. Suppose country A has more difficulty shifting its demand for some good j to other suppliers than country B has in shifting its supply of good j to alternative buyers. Country A depends more on country B as a supplier of good j than B depends on A as an outlet for this good. Panel B of Figure 6.1 represents this scenario. As conflict rises, the price of good j rises, making it more expensive for A to import each unit of good j from B. Hence, A has an incentive to initiate less conflict against B if its import of good j from B is large, ceteris paribus. This scenario is similar to Polachek’s prediction but it is not the only possible scenario in our model. Now, suppose country A’s import demand of good j from B is more sensitive to conflict than country B’s export supply of good j to A. Country A has less difficulty in locating other suppliers of good j than country B has

170

Democracy and Economic Openness in an Interconnected System

in finding alternative buyers for good j. In this scenario, Panel A of Figure 6.1 shows that country A’s import demand shifts inward more than country B’s export supply. As conflict rises, the price of good j falls, making it less expensive for A to buy each unit of good j from B. If the amount of import is large, country A has an economic incentive to initiate more conflict against B, ceteris paribus. Polachek’s model as well as other previous research fail to consider this scenario. Another logical possibility implied by our model, which previous research also fails to anticipate, is that country A’s import demand and country B’s export supply of good j could be equally sensitive to conflict. In this case, when conflict rises, country B’s export supply and country A’s import demand shift equally in absolute terms. The import price remains unchanged. No additional economic incentive exists for A to initiate conflict against B even if country A’s import from B is large. Similar to the preceding three scenarios with respect to country A’s import of good j from B, country A’s export of good i can also generate three distinct effects on the conflict A sends toward B. Likewise, these effects depend on the sensitivities of country A’s export supply and country B’s import demand to the conflict. Finally, it is worth noting that the substitution of traded goods, which plays an important part in the theoretical model, can occur across countries, within sectors, or both. For example, the United States may substitute Russian wheat with Mexican wheat (across countries), or with American corn (within sector), or with Canadian corn (across countries and within sector). The ease of substitution within sectors may vary. For example, it may be easier to substitute some agricultural goods (e.g., wheat with rice) but harder to substitute some manufactured or energy goods (e.g., television with cars, oil with nuclear energy). So far, we have discussed the model’s implications for generic types of goods. To test these implications, we need to derive testable hypotheses for specific traded goods (e.g., agricultural or manufactured goods). Predicting the effects across specific exported and imported goods requires estimates of coefficients ␣3 , ␤3 , ␥ 3 , and ␦3 . The estimation of ␣3 , ␤3 , ␥ 3 , and ␦3 requires data on bilateral trade quantities and prices, which are not readily available for all goods. Reuveny (2001b) is the only study that has produced these coefficient estimates for five economic sectors: (1) agriculture/ fishery, (2) energy, (3) chemical/mineral, (4) manufactured products, and (5) miscellaneous consumption products. Table 6.1 presents the five sectors employed by Reuveny, the range of goods in each sector, the implied signs of (␣3 − ␤3 ) and (␥ 3 − ␦3 ) per sector, and the expected effects on

171

Sector content

Increasing conflict

Reducing conflict

>0

<0

Reducing conflict

>0

Increasing conflict

Reducing conflict

>0

<0

Expected effect of import

␣3 − ␤3

<0

>0

>0

>0

<0

␥3 − ␦3

Reducing conflict

Increasing conflict

Increasing conflict

Increasing conflict

Reducing conflict

Expected effect of export

Note: ␣3 = conflict’s effect on country B’s export supply of good j to A; ␤3 = conflict’s effect on country A’s import demand for good j from B; ␦3 = conflict’s effect on country B’s import demand for good i from A; ␥ 3 = conflict’s effect on country A’s export supply of good i to B.

Agriculture products (fruits, vegetables, cereals, trees, sugars), fish, fishery products (2) Energy Coal, lignite, other briquettes, products of cooking, crude petroleum, natural gas, refined petroleum and products, electric power, gas, steam, radioactive materials, ores (3) Chemical/mineral Ferrous ores and metals, nonferrous ores and metals, nonradioactive ores and metals, nonmetallic minerals, chemicals (4) Manufactured products Metals and minerals products, agricultural and industrial machinery, office and data processing machines, precision instruments, optical instruments, electrical products, motor vehicles, other transport equipment (5) Miscellaneous consumption Meats, meat preserves, milk products, food, products beverages, tobacco products, textiles, clothing and accessories, leathers, leather and skin goods, footwear, wooden products, furniture, paper products, printing products, rubber goods, plastic products, goods not classified

(1) Agriculture/fishery

Sector

Table 6.1. Trade categories and conflict expectations

172

Democracy and Economic Openness in an Interconnected System

conflict initiation. Here we employ his estimates for these five specific sectors and generate the following testable hypotheses. H1: An increase in country A’s imports from country B in the agriculture/ fishery, energy, or chemical/mineral sectors reduces the probability of A initiating military conflict against B. H2: An increase in country A’s imports from country B in the manufactured or miscellaneous consumption products sectors increases the probability of A initiating military conflict against B. H3: An increase in country A’s exports to country B in the agriculture/ fishery or miscellaneous consumption products sectors reduces the probability of A initiating military conflict against B. H4: An increase in country A’s exports to country B in the energy, chemical/mineral, or manufactured products sectors increases the probability of A initiating military conflict against B. EMPIRICAL MODEL AND ANALYSIS

This section first presents our statistical model for the empirical analysis and then discusses several research design issues, followed by key results from the empirical analysis. As in the other chapters, the discussion is self-contained and does not require any statistical expertise. The technical details of the statistical model, data sources, and measure construction are in the chapter appendix, which follows the same structure of presentation as this section.

Empirical Model To test the hypotheses suggested by the theoretical model, we specify and estimate the following statistical model of dyadic conflict initiation: midt = ␤0 + ␤1 agriculture importt−1 + ␤2 agriculture exportt−1 + ␤3 chemical-mineral importt−1 + ␤4 chemical-mineral exportt−1 + ␤5 energy importt−1 + ␤6 energy exportt−1 + ␤7 manufactured importt−1 + ␤8 manufactured exportt−1 + ␤9 miscellaneous consumption importt−1 + ␤10 miscellaneous consumption exportt−1 + ␤11 gdp initiatort−1 + ␤12 gdp targett−1 + ␤13 contiguityt−1

Economic Openness and Conflict

173

+ ␤14 distancet−1 + ␤15 target democracyt−1 + ␤16 initiator democracyt−1 + ␤17 regime dissimilarityt−1 + ␤18 power balancet−1 + ␤19 initiator capability ratiot−1 + ␤20 alliancet−1 + ␤21 minor powert−1 + εt .

(6.3)

The model specification follows our theoretical expectations to include dyadic conflict, indicators for import and export flows in five hypothesized economic sectors, as well as some control variables identified in the conflict literature. We denote variables with small capital letters and their coefficients with Greek notation. Each coefficient indicates the effect of the independent variable on the dependent variable. The notation εt denotes the unexplained random error in the model. The subscripts t and t − 1 indicate the time period of the variable, where t represents the current time period and t − 1 the previous time period (a lagged variable). The use of lagged variables is fully discussed in the appendix. To simplify the presentation, we refer to the variables without their time subscripts. The phenomenon we seek to explain in the theoretical model, conflictAB , is dyadic conflict. We measure it using the empirical indicator mid, which takes on the value of 1 if country A initiates a MID against country B, and zero otherwise. As usual, the event of a MID includes the threat of military force, the display of force, the use of force, or war. Our sample period of 1970–1997 includes 352 MIDs in our estimation sample. Our theory indicates that the empirical model ought to include measures of importAB and exportAB , that is, country A’s import from B in various economic sectors and country B’s export to A in these sectors. Based on the hypotheses, we include five import variables and five export variables, all measured in terms of the value of bilateral flow in real terms: agriculture import is the real dollar value of country A’s agricultural products imported from B; chemical-mineral import is the real dollar value of imported chemical and mineral products; energy import is the real dollar value of imported energy products; manufactured import is the real dollar value of imported manufactured products; and miscellaneous consumption import is the real dollar value of the import of all remaining consumption goods that are not included in the aforementioned groups. The variables agriculture export, chemical-mineral export, energy export, manufactured export, and miscellaneous consumption export are defined similarly for country A’s export to B. The decision whether to initiate military conflict, of course, does not depend only on trade variables. We need to control for possible confounding

174

Democracy and Economic Openness in an Interconnected System

factors to avoid spurious inferences. In recent years, the conflict literature has generally used a similar set of control variables, which we also use: gdp initiator and gdp target are real GDPs of the conflict initiator and the target, respectively; contiguity indicates whether two states in a dyad are contiguous on land or through up to 150 miles of water; initiator democracy is the level of democracy of the MID initiator; target democracy is the level of democracy of the target; regime dissimilarity is the absolute value of the difference between the democracy levels of the initiator and the target; and power balance reflects the relative balance in materials capabilities between the two states in a dyad. The initiator capability ratio represents the relative importance of the initiator’s capabilities in the dyad, alliance indicates the presence or absence of dyadic military alliance, and, finally, minor power indicates whether both states in a dyad are only minor powers.

Research Design Issues Several issues guide the design of our research in this chapter. The first three design issues concern the measurement of our dependent variable, MID initiation. First, one may argue whether it is appropriate to aggregate the different types of MID to a single, 1–0 measure. This argument follows many studies from the literature and enables the comparison of our results with previous findings, and we address this issue in an additional analysis. Second, the theoretical concept of conflict in the model, conflictAB , is a continuous variable, which leaders can raise or lower, but the mid variable in the statistical model is dichotomous, which seems to make for a relatively crude measure. This issue, however, does not pose a key problem in our case, because our statistical technique and results inform us of the continuous likelihood of conflict, not the dichotomous event of conflict or no conflict. Third, the MID data may have measurement error from identifying which side of a dyad started a dispute. Since scholars code the MID data retrospectively based on historical accounts, it may not be easy to know who started the dispute. In spite of this limitation, the mid variable does indicate which side crosses the threshold into the realm of military force. Fourth, because our theoretical framework and implied hypotheses anticipate the likelihood of conflict initiated by country A against country B, we use a directed dyad framework. In this framework, the dyad US (A) => USSR (B) (the United States initiates conflict against USSR), for example, differs from the dyad USSR (A) =>US (B) (USSR initiates conflict against the United States). Our sample includes all those directed dyads for which trade and conflict data are available from 1970 to 1997. Fifth, conflict as

Economic Openness and Conflict

175

the dependent variable may affect some right-hand-side variables; that is, the relationship between some of the right-hand-side variables and conflict may be reciprocal. For example, conflict may affect dyadic trade or the level of democracy in each side of a dyad. Ignoring the issue can produce wrong results, and we deal with this risk by applying the appropriate econometric technique. Sixth, for statistical inferences to be valid, the error term εt needs to satisfy some assumptions. If the error term does not have certain desired properties, our ability to make correct inferences is compromised. We address these issues with appropriate econometric techniques in the chapter appendix. Seventh, the right-hand-side variables should not be excessively correlated to one another. This issue warrants our attention because sectoral trade flows could be so highly correlated that they affect statistical inference. We discuss the diagnostic test results in appendix. Finally, we assess the effects of sectoral trade flows on conflict initiation in several ways. We test whether the results support the hypotheses and we compute the substantive effect of any key variable that has an effect statistically different from zero. We then test whether the effect of trade in one sector on military conflict initiation differs from that of trade in another sector, which allows us to evaluate whether a key assumption in previous studies is appropriate. Finally, we provide information on how well our model predicts the frequency of MIDs within our sample.

Empirical Findings We present the results of our main analysis in Table 6.2, where the dependent variable is coded as 1 whenever a country initiates a MID of any type in a year. Column 1 of Table 6.2 reports the results for hypothesis testing and column 2 presents the size of the effect of key variables. Starting with column 1, the coefficient of agriculture import is negative, as expected, and statistically different from zero. The coefficient of agriculture export is also negative, as expected, but the effect is statistically not different from zero. Hence, a rise in a country’s agriculture/fishery import reduces the likelihood of MID initiation toward the exporter, but the effect of a rise in a country’s agriculture/fishery export is very weak. When a country depends on the import of agriculture/fishery products from its partner more than its partner relies on the country as an export outlet, the former is less likely to initiate a conflict against the latter. But a country’s greater dependence on the other as an export outlet for agriculture/fishery products has little effect on its propensity to initiate military conflict.

Table 6.2. Effects of bilateral import and export in five sectors on MID initiation

All MIDs Agriculture-fishery import Agriculture-fishery export Energy import Energy export Chemical-mineral import Chemical-mineral export Manufactured import Manufactured export Miscellaneous consumption import Miscellaneous consumption export Initiator capability ratio Power balance Initiator democracy Target democracy Regime dissimilarity Alliance Log of distance Contiguity Minor power Initiator GDP Target GDP Observations Pseudo R 2

−0.00046 [4.27]∗∗∗ −0.00006 [0.71] −0.00004 [1.66]∗ 0.00004 [2.67]∗∗∗ −0.00004 [0.76] −0.00008 [1.19] 0.00003 [1.89]∗ 0.00003 [2.27]∗∗ −0.00005 [0.85] 0.00002 [0.56] 0.35938 [2.34]∗∗ −0.01451 [0.08] −0.00789 [2.26]∗∗ −0.00181 [0.50] 0.00941 [2.90]∗∗∗ −0.01330 [0.18] −0.08084 [4.93]∗∗∗ 0.80909 [6.15]∗∗∗ −0.32540 [1.89]∗ 0.06681 [2.83]∗∗∗ 0.09974 [4.40]∗∗∗ 213790 0.31

Relative risk 0.78

0.96 1.07

1.13 1.16

1.44

0.84

1.20

0.71

1.51 1.84

Note: Z statistics in brackets. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%. Constant, peace-year, and spline variables not reported.

Economic Openness and Conflict

177

The coefficient of energy import is negative, as expected, whereas the coefficient of energy export is positive, as expected. Both coefficients are statistically different from zero. A rise in a country’s energy import reduces the likelihood of MID initiation toward the exporter, but a rise in energy export increases the likelihood of MID initiation toward the importer. When a country depends on the energy import from its partner more than its partner relies on the country as an export outlet, the former is less likely to initiate a conflict against the latter. Energy dependence matters. In contrast, when a country’s dependence on its partner as a destination for its energy export is smaller than its partner’s dependence on its supply of energy export, the country is more likely to initiate a conflict. The coefficients of both manufactured import and manufactured export are positive, as expected, and significantly different from zero. A rise in a country’s import of manufactured products raises the likelihood of MID initiation toward the exporter, and a rise in a country’s export of manufactured products raises the likelihood of MID initiation toward the importer. When a country depends on the import of manufactured products from its partner less than its partner relies on it as an export outlet, the former is more likely to initiate a conflict against the latter. And when a country depends on its partner as an export outlet of its manufactured products less than its partner relies on its supply, the former is more likely to initiate a conflict against the latter. The coefficients of chemical-mineral import/export and miscellaneous consumption import/export are not statistically different from zero, which may be due to aggregations of very different products: chemicals and minerals in one case and various consumption goods in the other. The reader may recall that these particular aggregations are dictated by the availability of data, for which we provide further discussion in the appendix. At this point, one may ask whether those trade flows whose effects are statistically different from zero are large or small from the substantive point of view. Column 2 of Table 6.2 answers this question. As customary in the conflict literature, we address this question by computing the relative risk of MID. Relative risk is particularly useful when we examine the effect size of rare events like MID. It is the ratio of the predicted probability of MID when the trade variable of interest increases by one standard deviation above its mean over the predicted probability of MID at the baseline. We define the baseline scenario as one in which all continuous variables are at their sample means and the dichotomous variables are set to the contiguous, minor power, no-alliance dyad.

178

Democracy and Economic Openness in an Interconnected System

When country A’s agriculture/fishery import from B rises by one standard deviation ($187 million) above the mean, country A is 22% less likely to initiate a MID against B, relative to the baseline level. When country A’s energy import from B rises by $439 million above the mean, it is 4% less likely to initiate a MID. When country A’s energy export to B rises by $439 million above the mean, country A is 7% more likely to initiate a MID. When country A’s import or export of manufactured products with B rises by $1.5 billion, country A is 13% or 16% more likely to initiate a MID, respectively. These effects vary across sectors and trade flows, as our theory suggests, but how do they compare in size with those of the nontrade variables? Column 2 also reports the results for the control variables. As in other chapters, we delegate the discussion of the control variables to the appendix. Here, we provide a summary of the comparison in terms of the size of effect. The substantive effects of increases in energy import and export are smaller than those for the nontrade variables; the effects of increases in agriculture/ fishery import, manufactured imports, and manufactured exports are of the same order of magnitude as that for the initiator’s democracy, target GDP, regime dissimilarity, and distance; and the effects of increases in the initiator’s GDP and capability ratio are larger than those for the trade variables. Overall, sectoral trade flows have important effects on conflict initiation but not the most important effect.

Additional Analysis We conduct three additional analyses to address several interesting questions. First, previous studies of trade and conflict implicitly assume that the coefficient for the total bilateral trade variable is the same for each of its component trade flows. One way to test this assumption directly is to test whether the statistically significant effects of the trade variables in Table 6.2 are equal in size across sectors. The results of this analysis in Table 6.A1 in the appendix show that the effects of increases of trade flows in the statistically significant sectors are not equal in size. Hence, aggregating exports and imports in different sectors masks their differential effects. Second, one may wonder if our model has good in-sample predictive power. For each dyad in each year in the sample, we compute the predicted probability of a dispute based on the model. If the probability is larger than a certain threshold (see the appendix), it suggests that the model expects a military dispute for that dyad in the year; otherwise, the model predicts peace. By comparing the predictions with actual dyadic events, we compute

Economic Openness and Conflict

179

the predictive power of the model. As shown in Table 6.A2, the model predicts correctly the absence of MID initiation in 91% of the dyad-years and the presence of MID initiation in 81% of the cases. Thus, the model performs reasonably well in terms of in-sample prediction. Third, as noted, our dependent variable aggregates the different types of MIDs, ignoring their variations. We believe a focus on all MIDs also has merits. Leaders moving forces to the border or threatening to use force may be driven by the same considerations as leaders actually using force. It is also debatable whether the threats to use interballistic missiles or moving many divisions to the border are less hostile than border skirmishes or even clashes that cause killings. Nevertheless, to check robustness, we employ two additional dependent variables: one involves display of force, use of force, and war; the other includes use of force and war. The results obtained from using these two variables are in Table 6.A3 in the appendix. In the model for MIDs involving at least the display of force, the results for the trade variables are identical to those in Model 1. In the model for MIDs involving the use of force and war, the results for the trade variables are consistent with those in Model 1, except that the coefficient of chemical and mineral export is now negative and significant, rejecting the respective part of H4, and the coefficient of miscellaneous consumption export is negative and significant, supporting the respective part of H3.

Comparisons to the Literature and Stylized Examples Our empirical findings confirm several results reported in previous smallerscale empirical studies that did not employ the MID measure or our empirical model specification. The finding that a rise in energy exports promotes conflict is consistent with the result of Park et al. (1976) for a sample of oil-producing countries. Our result that a rise in agriculture/fishery imports reduces conflict initiation agrees with the finding of Gasiorowski and Polachek (1982) regarding the conflict behavior of the Warsaw Pact toward the United States. Our finding that a rise in manufactured products import or export promotes conflict confirms the Bennett et al. (1992) study of the U.S.–Japan dyad. In many ways, the study of Dorussen (2006) is the closest to ours, because it also employs a large-N sample and investigates the effect of disaggregated trade on the likelihood of MID. In addition, several of our findings are consistent with the spirit of Dorussen’s results. For example, chemical/mineral trade has a weak effect on conflict; some sectoral trade flows reduce the probability of MID.

180

Democracy and Economic Openness in an Interconnected System

Our analysis, however, differs from Dorussen (2006) in other ways. Although his theory is generic like ours, it is not formalized. He argues that trade is more pacifying if traded goods have greater opportunity costs. If goods can be easily appropriated, the pacifying effect is weaker, or may be reversed. Our theory works through changes in trade prices and substitution. Dorussen notes the effect of trade may change between export and import, but he does not predict the effects of export and import in specific sectors. We predict the effects of import and export in different sectors on conflict and test these predictions. Empirically, Dorussen looks at both MID involvement and onset in a sample of nondirected dyads but excludes minor disputes below the use of force, whereas we study the MID initiations measured in three ways (all MIDs, at least display of force, and at least use of force) in a sample of directed dyads. Finally, whereas Dorussen finds trade either reduces or has no effect on conflict, depending on sector, we find that trade in some sectors can also increase the likelihood of conflict. Given the complexity of our model, our exposition could benefit from a few real-world examples, but we must first offer a caveat. Statistical studies such as ours typically do not discuss real-world examples, and for good reasons. Our model predicts a tendency toward conflict or peace due to a change in some trade variable, holding all the other variables in the model constant. In the real world, one observes the net effect of all the relevant variables held at the levels they happen to be, not the partial effects obtained from regression. We cannot, and do not, argue that the presence or absence of conflict in these examples is only driven by trade. Rather, we employ our model to illustrate the role of sectoral trade in conflict, ceteris paribus.15 Limitations notwithstanding, we may utilize the implications of our model to better understand some stylized examples. Consider first the Iran– U.S. dyad in the period since 1979, which included several U.S.–Iran MIDs. Based on our model, we speculate Iran may intensify conflict with the United States to raise the price of its oil exports. In other words, our model suggests that Iran, which is likely still selling oil to the United States directly or indirectly, benefits economically from hostility. A second example involves Bulgaria, Romania, and Hungary in the 1930s. At that time, their trade flows became highly concentrated on Germany. They imported primarily manufactured goods and coal from Germany and exported agricultural and light manufactured goods to Germany.16 15 16

We also need to be careful when using nonlinear models such as probit since employing different control groups may affect the inferred sizes of effects. On Nazi Germany’s trade with these countries (and others) see, e.g., Hillmann (1940) and Leitz (2004).

Economic Openness and Conflict

181

Since they were much less able than Germany to substitute these flows, they realized that not cooperating with Germany would damage their economies (Hirschman, 1945; Arad et al., 1983). Our model suggests that to pacify these countries, Germany could increase coal and agricultural exports and reduce manufactured exports and light manufactured imports. Finally, consider the U.S.–Japan dyad in the period leading up to December 1941. The United States reduced energy export to Japan, hoping to curb its territorial expansion in Asia, but the policy backfired. Historians such as Kennedy (1987) and Keylor (2001) argue that the fall in Japan’s energy import from the United States resulted in economic losses and a growing sense of vulnerability, which contributed to Japan’s hostility toward the United States – an explanation consistent with our model prediction. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Studies of the effect of trade on conflict typically invoke the notion of total bilateral trade, implicitly or explicitly assuming the effect does not vary across economic sectors and between trade flow directions. We challenge these premises. Our theory considers two countries that trade with one another and maintain some bilateral relations. The effects of changes in bilateral trade on the decision to initiate conflict vary across economic sectors and between trade flow directions. We employ our theory to identify ex ante the effects on conflict initiation of increases in export and import in the sectors of agriculture/fishery, energy, manufactured products, chemical/ mineral, and miscellaneous consumption products. In the empirical analysis, we find significant variations in the effects of trade on MID initiation across economic sectors and between trade flow directions. As predicted, rises in agriculture/fishery imports and energy imports reduce the probability of MID initiation, and rises in energy exports, manufacturing imports, and manufacturing exports increase this probability. Yet changes in agriculture/fishery export, chemical/mineral trade, and miscellaneous consumption products trade generally do not affect MID initiation. One logical explanation of these insignificant effects is that these sectors and flows are not important enough to the leader to influence her decision. But one may also attribute these insignificant results to two empirical problems: (1) the limited data for the measurement of the sensitivity parameters of sectoral import and export to conflict and (2) inappropriate aggregation of, say, chemicals and minerals into one category and very different consumption goods into one category. Improvements in solving

182

Democracy and Economic Openness in an Interconnected System

these two problems should lead to more accurate empirical testing of our theory in future research. Like all regression results, these findings hold when other things do not change. However, the effects of most of our control variables are also found to be significant, suggesting that they also affect the likelihood of MID initiation. Moreover, although we have estimated the ceteris paribus effects of changes in all these variables on conflict initiation, only the net effect is observed in reality. Thus, a dyad could be in peace even if some of its sectoral trade flows promote MID, provided that the pacifying forces combined outweigh the antagonizing effects together. How does our model fit with the two primary formal models offered in the literature to explain the effect of trade on conflict, that is, Polachek’s liberal model and Gartzke et al.’s signaling argument? As discussed throughout the chapter, our model subsumes and extends the one of Polachek, predicting that a rise in trade may promote either peace or conflict and the effect may vary across goods and imports versus exports, depending on certain parameters. Consider next the signaling argument. In this theory, military conflict results from a bargaining failure over some contested issue. Acting under incomplete information, trading nations are less likely to fight each other only because they can credibly signal resolve during bargaining by resorting to trade sanctions or trade dissociation (Morrow, 1999; Gartzke et al., 2001). Within this logic, the classical liberal argument formalized by Polachek does not hold because the size of the benefit from bilateral trade is subsumed into the initiator’s demand and the target’s concession without reducing uncertainty–the main cause of bargaining failure and military conflict. One might argue that the same criticism against the classical liberal argument could be leveled against our model. That is, leaders factor expected economic costs into their reservation values when making demands of their opponents, and they make more concessions to avoid military conflict when the expected economic costs are high. We argue that, in fact, our research poses some important challenges for the signaling argument. In the signaling argument, the benefit of each actor from economic exchange is modeled by one opportunity cost parameter, which is common knowledge to both sides (e.g., Gartzke et al., 2001). The situation in our framework is more complicated because we allow n trade markets for there are n goods or sectors. As we have shown, each bilateral trade market does not necessarily suffer a loss due to conflict; some conflict may actually induce a rise in profit, a condition that is assumed away in the bargaining model. Now, the leader needs to aggregate the costs and benefits across the sectors

Economic Openness and Conflict

183

in some way to figure out the welfare implications of a conflict decision. To do so, one needs to know who supplies what to whom, who demands what from whom, and who are the alternative suppliers and buyers and at what prices. The amount of information required is daunting, and much of the information is likely to be trade secrets. The leader may have difficulty getting access to all this information; even if she does, she may not be willing to share this financially valuable private information with the other side. Hence, our analysis of disaggregated bilateral trade creates four major problems for the signaling argument. First, the size of the opportunity cost is more difficult to pin down in multiple trading markets. Second, the costs and benefits in particular trading markets are not common knowledge such that the manipulation of a trading relationship may misinform as to the value of a particular trading relationship. Third, if one finds gains in trade from conflict that are unknown to the other side, this factor contributes to asymmetric information and uncertainty. Finally, the empirical finding that some sectoral exports and imports lead to conflict initiation challenges the theoretical prediction of the signaling argument that, if trade affects conflict at all, it should reduce the likelihood of military conflict. Taking a broader perspective, this research advances our theoretical understanding of the effect of trade on conflict initiation. To illustrate this point, let us further compare our theory with the five key theories in the literature in terms of their expected effect of trade on conflict, policy objective, policy tool, causal logic, and treatment of markets. Assuming states maximize material welfare, liberalism argues that leaders pursue peace with trade partners because conflict reduces valuable trade gains. The causal logic stresses economic incentives but oversimplifies market forces, assuming conflict is always costly or conflict lowers export price for an exporter and raises import price for an importer. The bargaining approach assumes leaders manipulate trade to facilitate negotiation over a disputed issue or influence its outcome. Cutting trade ties functions either as a credible signal to demonstrate resolve, enabling states to avoid bargaining failure and conflict, or as a tool to influence the bargaining outcome, resulting in less high-level conflict when actors face high exit costs. This approach also stresses economic ties, but reduces the economics of the marketplace to one or two parameters that denote the importance of trade to both sides. The neo-Marxist and neomercantilist realist arguments postulate that states use trade as a policy tool to affect the outcome of a contested issue or to expand their economic gains and political power. The outcome can be conflict or peace, depending on whether the other side rebels against, or

184

Democracy and Economic Openness in an Interconnected System

complies with, the external demand. Economic incentives also play a role here, but market forces are not fully modeled – only loosely represented by the differential importance of trade to both sides. The classical realist argument assumes that states maximize power and security, using conflict as a policy tool. In this view, trade should have no systematic effect on conflict, because political relations trump economic relations. In our theory, states use conflict (or peace) as a policy instrument to maximize national welfare, which depends on wealth and security. Bilateral import demand and export supply in different sectors have different sensitivities to conflict, reflecting the underlying abilities of states to substitute these trade flows with others. Markets play an essential role in this logic, informing leaders about the costs and benefits of conflict via the expected changes in the bilateral trade prices. The costs and benefits, in turn, motivate or deter conflict initiation. In general, our theory subsumes the liberal logic as one special case and offers a distinct explanation alternative to the bargaining/signaling, neoMarxist, neomercantilist realist, and classical realist arguments. Our theory also specifies ex ante when and how export and import flows in specific economic sectors influence the likelihood of military conflict initiation, a question that has not been systematically addressed before. We believe the value of our analysis is not merely academic. Our results inform policymakers about how to manage bilateral trade flows to improve interstate relations. Previous studies have focused on the effect of total trade on conflict. To a certain extent, finding the net effect of total trade (i.e., aggregated across goods) has limited value for public policy and could be misleading, because countries have heterogeneous trade structures, and increasing trade in some goods could harm political relations. Public policy that seeks peace should promote bilateral trade in peace-promoting goods. Dispute resolution mechanisms should focus on tensions associated with conflict-promoting types of trade. Countries with a history of bilateral conflict that are interested in peace may expand trade in some goods, but not in others. SUMMARY AND OUTLOOK

The effect of bilateral trade on interstate military conflict is the subject of a large and growing body of literature in international relations. Almost all theoretical and empirical studies focus on total bilateral trade, implicitly or explicitly assuming that the effect of trade on interstate conflict is constant

Economic Openness and Conflict

185

across imports and exports in different economic sectors. We challenge the premise of most trade–conflict studies in international relations that the effect of trade on interstate conflict does not vary across imports and exports in different economic sectors. We analyze theoretically and empirically the effects of bilateral imports and exports in different economic sectors on conflict initiation. Our theory encompasses the liberal explanation as a special case and is a distinct alternative to the bargaining/signaling, neo-Marxist, neomercantilist realist, and classical realist arguments. It also specifies the different effects of export and import flows, an issue never systematically addressed before. Our theoretical model generates ex ante predictions of the effects of changes in imports and exports in the agriculture/fishery, energy, chemical/ mineral, manufactured products, and miscellaneous consumption products sectors on the likelihood of MID initiation. The empirical test employs a sample of directed dyads from 1970 to 1997. We find strong empirical evidence supporting our expectations for agriculture/fishery, energy, and manufactured products sectors, and little evidence supporting the expectations for chemical/mineral and miscellaneous consumption products. These findings lead to several conclusions. Rises in agriculture/fishery and energy imports discourage MID initiation. Rises in energy export and trade in manufactured products induce MID initiation. Changes in chemical/mineral trade, agriculture/fishery export, and miscellaneous consumption products trade generally do not affect MID initiation. These results hold for all MIDs and for MIDs involving at least display of force, and largely so for MIDs involving at least the use of force. The significant effects vary across sectors and trade flows, and their sizes are relatively substantial. The model predicts correctly the large majority of peace and MID events in the sample and sheds some light on several real-world cases. In this chapter, we have investigated the effect trade flows disaggregated across goods on the propensity of a state to initiate military conflict against a country with which it also trades. In our analysis of the results and their implications, we listed a number of examples referring to a specific conflict taking place, of course, on Earth. The previous chapters also refer, again of course, to interactions taking place on this very planet. We add the words “of course” naturally and almost in passing; but if so, the reader may argue, why have we not thus far introduced elements pertaining to the physical environment within which these interactions take place? We think this is a good question that needs to be addressed more often in the fields of international relations and international political economy.

186

Democracy and Economic Openness in an Interconnected System

In the last part of our book, we take an initial step toward bringing the environment into our analysis of the complex transformation taking place in recent decades. In particular, we ask about the effect of the very forces studied in the first and second parts of the book – democracy, conflict, and economic openness – on the physical environment. We think these questions are important because changes in the physical environment can drastically alter the quality of life on Earth.

APPENDIX THEORETICAL MODEL

The quantity of trade in good i that country A exports to B is Xi and its price is PXi . The volume of trade in good j that country B exports to A is Mj and its price is PMj . The number of goods country A exports to B is NX . The number of goods country A imports from B is NM . The utility of country A’s leader is denoted by U (EC, C): EC is economic consumption of country A; C is the conflict country A initiates against B, defined as a positive number. The signs of the partial derivatives of utility are assumed to be as in Polachek: UEC > 0 (the marginal utility of economic consumption is positive, or utility rises with consumption), UEC EC < 0 (the marginal utility of consumption falls with more consumption), UC > 0 (the marginal utility of conflict is positive), and UCC < 0 (the marginal utility of conflict falls with more conflict). The partial derivative of the price of good i (A exports to B) with respect to conflict is PXCi (i = 1, 2, . . . , NX ). The partial derivative of the price j of good j (A imports from B) with respect to conflict is PMC (j = 1, j i 2, . . . , NM ). Their second-order derivatives are PXCC and PMCC . Polachek assumes certain directions for the effects of conflict on prices (PXCi < 0, j j i PXCC > 0, PMC > 0, and PMCC > 0), but in our model prices and their derivatives are set endogenously in the marketplace, which is one of our key innovations. Country A chooses the level of conflict (C) against B, taking consumption, import, and export as given, to maximize utility subject to the need to maintain a zero trade balance. It should be noted using a nonzero trade balance does not change any of the theoretical results we derive, because the math we present next is not sensitive to this particular modeling detail. Turning to the maximization itself, the problem that a decision maker is

Economic Openness and Conflict

187

assumed to solve is formally stated as follows: max U (EC, C) by choosing C,

subject to 0 =

NX 

X i · PXi (C)

i =1

−

NM 

M j · PM j (C).

j =1

(6.1A) The solution of this problem involves maximizing the following Lagrangian with respect to conflict (C) and the Lagrange multiplier (␭). The first-order condition for C gives   NX NM   j UC (EC, C) + ␭  X i · PXiC (C) − M j · PMC (C) = 0. (6.2A) i =1

j =1

The effects of changes in country A’s export of good i and import of good j on the conflict that A sends toward B are given by computing the comparative statics of Equation (6.2A) with respect to X i and M j : −␭ · PXiC ∂C =   M i X ∂Xi UCC + ␭ · iN=1 X i · PXiCC − ␭ · Nj =1 M · PM iCC i = 1, 2, . . . , NX ;

(6.3A)

j ␭ · PMC ∂C =  X i  M j ∂M j UCC + ␭ · iN=1 X · PXiCC − ␭ · Nj =1 M j · PM CC

j = 1, 2, . . . , NM .

(6.4A)

Economic logic suggests that supply and demand determine the prices and quantities of traded goods i (i = 1, 2, 3, . . . , NX ) and j (j = 1, 2, 3, . . . , NM ) between A and B. Each good involves a supply side and a demand side. The demand of an importer defines how much of a good to import at different prices, whereas the supply of an exporter defines how much to export at different prices. As the price of a good rises, the demand for this good falls and the supply of this good rises. The equilibrium condition defines the price of each traded good in the dyad. More specifically, suppose in dyad A–B that country A exports to B in sector i (i.e., country B imports from A), and country A imports from B in sector j (i.e., country B exports to A). In equilibrium, country A’s demand

188

Democracy and Economic Openness in an Interconnected System

for good j equals country B’s supply of good j to A. If B does not send enough of good j to A, country A’s demand for good j becomes larger than country B’s supply. In this case, the price of good j rises, prompting B to increase the supply to A. Sometimes, country B’s supply of good j is larger than country A’s demand. In this case, the price of good j declines due to the lack of demand in country A, and country B will reduce the supply of that good. In equilibrium, the demand for bilateral import of good j equals the supply of bilateral export of that good. The same logic also applies to the export flow in sector i from country A to country B. How do trade prices change in response to conflict? Demand and supply are often modeled as linear functions.17 We model country A’s import demand for good j from B (MAD ) by MAD = ␤0 − ␤1 P M + ␤2 Y A − ␤3 C,

(6.5A)

where the superscript j denoting the good is dropped hereafter to simplify notation. We define all the ␤ parameters in Equation (6.5A) to be positive. YA is country A’s income, and PM is the price country A pays for imports from country B. Equation (6.5A) embeds the usual economic expectations that country A’s import demand for products from country B rise with A’s income and fall with the good’s price. The effect of conflict on import demand of country A in Equation (6.5A) is negative (C is defined as positive), a point to which we return shortly. Country B’s export supply of good j to country A is modeled as follows: MBS = ␣1 P M + ␣2 YB − ␣3 C,

(6.6A)

where all the ␣ parameters in Equation (6.6A) are positive, YB is the exporter B’s income, and PM and C are defined as in Equation (6.5A). As usual in economics, export supply rises with the good’s price and with the exporter’s income. Equation (6.6A) also indicates that the supply of goods by B to A falls with conflict (C). Several reasons exist why import and export fall with conflict in Equations (6.5A) and (6.6A). Importers and exporters may want to punish hostile partners by reducing trade. The importer may seek to reduce dependence on a hostile supplier, which may stop selling, whereas the exporter may seek to reduce dependence on a hostile importer, which may stop buying. Additionally, conflict often raises costs to both sides (e.g., higher risk insurance 17

See, e.g., Bond (1985), and Nicholson (2005) for similar formulations that do not include conflict.

Economic Openness and Conflict

189

premium, delays in transportation, damages to goods), reducing trade. Furthermore, governments often restrict trade with foes, seeking to influence their opponents and weaken their economic and military capabilities. Various empirical examples and statistical studies support this assumption.18 In equilibrium, bilateral import demand for good j equals its bilateral export supply. The equation MAD = MBS is solved for PM, the equilibrium price of good j, as follows: PM =

␤2 ␣2 ␣3 − ␤3 ␤0 + YA − YB + C. (6.7A) ␣1 + ␤1 ␣1 + ␤1 ␣1 + ␤1 ␣1 + ␤1

As usual in bilateral trade models, Equation (6.7A) indicates that the price of good j from country B to country A rises with country A’s income and falls with country B’s income. The denominator of the expression multiplying C in Equation (6.7A) is positive. Hence, the effect of C on PM depends on ␤3 (effect of conflict on country A’s import demand for good j from B) and ␣3 (effect of conflict on country B’s export supply of good j to A). If ␣3 < ␤3 , PM falls with conflict (the coefficient of C is negative). If ␣3 > ␤3 , PM rises with conflict. If ␣3 = ␤3 , PM does not change with conflict. Figure 6.1 illustrates the solution for PM in Equation (6.7A), as discussed in the main text. Equations (6.5A)–(6.7A) model the determinants of trade prices, leading to the question, what determines the effect of conflict on trade prices? Suppose A raises hostility toward B by one unit. The effect on PM is given by the expression in front of C in Equation (6.7A), (␣3 − ␤3 )/(␣1 + ␤1 ). The change in PM when C rises depends on the conflict sensitivities of demand (␤3 ) and supply (␣3 ) and on the price slopes of the demand and supply curves (␤1 and ␣1 , respectively). The shift in PM comes from two sources: country B’s export supply [␣3 /(␣1 + ␤1 )] and country A’s import demand [−␤3 /(␣1 + ␤1 )]. When conflict rises, the export’s effect on raising PM increases with ␣3 , indicating that export supply is sensitive to conflict. The effect declines with ␣1 , indicating that export supply is elastic (the supply curve becomes flatter in Figure 6.1). Similarly, when conflict rises, the effect of import on reducing PM rises with ␤3 , suggesting import demand is sensitive to conflict, and falls with ␤1 , suggesting import supply is elastic. So far, we have discussed country A’s import of good j from B (i.e., country B’s export to A). We now turn to country A’s export of good i to B (i.e., country B’s import from A). Country B’s demand for country A’s export and country A’s supply of country B’s import are determined 18

For additional discussion of these ideas, see, e.g., Pollins (1989), Morrow et al. (1998), Reuveny (2001a, 2001b), and Li and Sacko (2002).

190

Democracy and Economic Openness in an Interconnected System

in manners similar to those defined by Equations (6.5A) and (6.6A). The volume of this flow is denoted as X. We use ␦ for the coefficients in country B’s import demand in Equation (6.8A) (instead of ␤ for country A’s demand in Equation (6.5A)), and ␥ for the coefficients in country A’s export supply in Equation (6.9A) (instead of ␣ for country B’s supply in (6.6A)). Thus, country B’s import demand for good i from A is specified as X BD = ␦0 − ␦1 P M + ␦2 YB − ␦3 C

(6.8A)

and country A’s export supply of good i to B is specified as X SA = ␥1 P M + ␥2 Y A − ␥3 C.

(6.9A)

As before, all the ␦ and ␥ coefficients are defined to be positive. The condition X BD = X SA determines the equilibrium price PX for good i: PX =

␦2 ␥2 ␥3 − ␦3 ␦0 + YB − YA + C. ␥1 + ␦1 ␥1 + ␦1 ␥1 + ␦1 ␥1 + ␦1

(6.10A)

Next we turn to the effects of trade on conflict in Equations (6.3A) and (6.4A). These effects depend on the first and second derivatives of the trade prices with respect to conflict (C) in Equations (6.7A) and (6.10A): PXC =

␥3 − ␦3 ; ␥1 + ␦1

PXCC = 0;

PMC =

␣3 − ␤3 ; ␣1 + ␤1

PMCC = 0. (6.11A)

Substituting these derivatives from Equation (6.11A) into Equations (6.3A) and (6.4A), respectively, and reinstating the goods’ notation in the appropriate places, we get the following solutions: ␥ i −␦i

−␭ · ␥3i +␦3i ∂C 1 1 = ∂ Xi UCC ∂C = ∂Mj

␭·

j

(6.12A)

j = 1, 2, . . . , NM .

(6.13A)

j

␣3 −␤3 j j ␣1 +␤1

UCC

i = 1, 2, . . . , NX ;

Expressions (6.12A) and (6.13A) specify, respectively, the effects of changes in country A’s export of good i (i = 1, 2, 3, . . . , NX ) and country A’s import of good j (j = 1, 2, 3, . . . , NM ) on its decision to initiate conflict against B. The signs of these effects depend on the relative sizes of the ␦3

Economic Openness and Conflict

191

and ␥ 3 coefficients for country A’s export to B in Equation (6.12A) and on the relative sizes of the ␤3 and ␣3 coefficients for country A’s import from B in Equation (6.13A). Recall that Ucc < 0 in Equations (6.12A) and (6.13A). In Equation (6.12A), if ␥ 3 > ␦3 , a rise in country A’s export to B raises the conflict A initiates toward B. If ␥ 3 < ␦3 , a rise in country A’s export to B reduces this conflict. If ␥ 3 = ␦3 , a rise in country A’s export to B does not affect country A’s conflict against B. In Equation (6.13A), if ␣3 < ␤3 , a rise in country A’s import from B raises the conflict A initiates against B. If ␣3 > ␤3 , a rise in country A’s import from B reduces this conflict. If ␣3 = ␤3 , a rise in country A’s import from B has no effect on country A’s conflict against B. Finally, we simplify the presentation of Equations (6.12A) and (6.13A) to obtain Equations (6.1) and (6.2) in the main text. We define KE and KI , which we have also used in the main text of this chapter, as the following subexpressions in Equations (6.12A) and (6.13A): K Ei

=

−␭ ·

␥3i −␦3i ␥1i +␦1i

UCC
j j ␭ · ␣1 + ␤1 j Kl = UCC

i = 1, 2, . . . , NX ;

(6.14A)

j = 1, 2, . . . , NM .

(6.15A)

With these expression thus defined, we can rewrite ∂C/∂X and ∂C/∂M for the partial derivatives in Equations (6.12A) and (6.13A) as their mathematical approximation by replacing the partial differential ∂ with the change
, giving the ratios
C/
X and
C/
M instead of the derivatives on the left-hand side of Equations (6.12A) and (6.13A), which together with Equations (6.14A) and (6.15A) can be rewritten to give Equations (6.1) and (6.2) in the chapter’s main text:

(conflict AB ) = K E ␥3i − ␦3i ·
(exportiAB ) i = 1, 2, . . . ; (6.1)
j j j
(conflict AB ) = K I ␣3 − ␤3 ·
(import AB ) j = 1, 2, . . . . (6.2) IMPLICATIONS OF THE MODEL AND HYPOTHESES

We have seen in the text that the important issue is whether countries engaged in conflict can substitute the traded goods. This event would depend on the sensitivity of the particular traded good to conflict. Formally, if country A’s import demand over good j from B is less sensitive to conflict

192

Democracy and Economic Openness in an Interconnected System

than country B’s export supply of good j to A (␤3 < ␣3 ), Equation (6.13A) predicts that a rise in country A’s import from B in sector j lowers the conflict A initiates against B. In this case, as conflict rises, country A’s import demand shifts inward less than country B’s export supply, but only if A has more difficulty shifting its demand for good j to other suppliers than B has in shifting its supply of good j to alternative buyers. In other words, A depends more on B as a supplier of good j than B depends on A as an outlet for good j. As conflict rises, the equilibrium price of good j, PM, rises, as shown in panel B of Figure 6.1 and in Equation (6.7A), making it more expensive for A to buy each unit of good j from B. Hence, A has an incentive to initiate less conflict against B if its import of good j from B rises, ceteris paribus. This scenario is similar to the scenario in Polachek’s model (recall that he assumes import price rises with conflict). However, it is not the only possible scenario in our model. If country A’s import demand of good j from B is more sensitive to conflict than country B’s export supply of good j to A (␤3 > ␣3 ), Equation (6.13A) predicts that a rise in country A’s import from B in sector j raises the conflict A initiates against B. In this case, country A’s import demand shifts inward more than country B’s export supply, which occurs only if A has less difficulty in locating other suppliers of good j than B has in finding alternative buyers for good j. As conflict rises, the equilibrium price of good j, PM, falls, as in panel A of Figure 6.1 and in Equation (6.7A), making it less expensive for A to buy each unit of good j from B. Hence, A has an economic incentive to initiate more conflict against B, ceteris paribus. Yet another possibility implied by our model occurs when country A’s import demand and country B’s export supply of good j are equally sensitive to conflict (␤3 = ␣3 ). In this case, a rise in country A’s import from B in sector j does not affect the conflict initiated by A against B. When conflict rises and country B’s export supply and country A’s import demand shift equally in absolute terms, PM does not change. As a result, country A’s incentive to initiate conflict against B does not change when its import of good j from B rises. The reasoning behind the effect of country A’s export on the conflict A initiates toward B is similar. In Equation (6.12A), if country A’s export supply of good i to B is less sensitive to conflict than country B’s import demand for good i from A (␥ 3 < ␦3 ), a rise in country A’s export to B in sector i reduces the conflict A initiates against B. In this case, with conflict, country A’s supply of good i to B shifts inward less than country B’s demand for good i from A. Country A depends more on B as a buyer of good i than

Economic Openness and Conflict

193

country B depends on A as a supplier. As a result, the price of good i, PX, falls, as shown in Equation (6.10A). When PX falls with conflict, A gets less for each unit of good i exported to B. Hence, A has an incentive to reduce conflict if its export of good i to B rises.19 If country A’s export supply of good i to B is more sensitive to conflict than country B’s import demand for good i from A (␥ 3 > ␦3 ), a rise in country A’s export to B in sector i raises the conflict that A sends against B. In this case, as conflict rises, country A’s supply of good i to B shifts inward more than country B’s demand for good i from A. Country A depends less on B as an outlet than B depends on A as a supplier. The price of good i, PX, rises, as shown in Equation (6.10A). Country A gets more for each unit of good i exported to B. Hence, A has an incentive to raise conflict if its export of good i to B rises, ceteris paribus. If country A’s export supply and country B’s import demand of good i are equally sensitive to conflict (␥ 3 = ␦3 ), a rise in country A’s export to B in good i does not affect the conflict A initiates against B. In this case, country A’s export supply and country B’s import demand shift inward equally in absolute terms in response to a rise in conflict, so the price of good i does not change with conflict. As a result, there is no incentive for A to increase or decrease the conflict it initiates toward B. Turning to our hypotheses, to generate specific hypotheses on the effects of different types of traded goods on conflict, one has to know the values of ␣3 , ␤3 , ␥ 3 , and ␦3 . The reader may recall that these four parameters are the coefficients of conflict in the bilateral import demand and export supply equations of each good. As with all demand and supply equations, these parameters are not readily observable and have to be estimated empirically. In principle, we could estimate the parameters ourselves, using data on bilateral trade prices and quantities for dyadic trade flows broken down along economic sectors, but these data are rarely available. Alternatively, we could use existing estimates of these parameters in the literature. We choose the latter approach, which offers an initial test of our theoretical model. Only a few trade studies estimate bilateral import demand and export supply using bilateral trade prices and quantities. They use data (Italianer, 1986) collected for the European Commission, but they do not include conflict. To our knowledge, only Reuveny (2001b) includes conflict in estimating sectoral bilateral import demand and export supply equations. He employs Italianer’s data for five sectors: (1) agriculture/fishery, (2) energy, 19

This scenario is implied by Polachek’s model, but again it is not the only possible scenario.

194

Democracy and Economic Openness in an Interconnected System

(3) chemical/mineral, (4) manufactured products, and (5) miscellaneous consumption products. For each good traded in any dyad, he estimates a system of six equations with six endogenous variables: export of A to B, import of B from A, export of B to A, import of A from B, conflict from A to B, and conflict from B to A. The coefficients of conflict in the demand and supply equations provide our ␣3 , ␤3 , ␥ 3 , and ␦3 . We use the average estimates across dyads to compute (␣3 − ␤3 ) and (␥ 3 − ␦3 ) for each of Italianer’s five sectors, which Equations (6.12A) and (6.13A) employ to predict effects of the sectoral exports and imports, respectively, on conflict.20 Table 6.1 in the main text presents the five sectors, the signs of (␣3 − ␤3 ) and (␥ 3 − ␦3 ) per sector, and the expected effects of country A’s import from B and country A’s export to B in each sector on the conflict A initiates against B. We employ the signs of these expected effects to generate the four hypotheses listed in the main text. EMPIRICAL MODEL AND ANALYSIS

Empirical Model Our theory indicates the empirical model ought to include bilateral sectoral imports and exports, and national incomes of both states in the dyad. We include import to A from B and export from A to B in the five Italianer (1986) sectors, denoted as agriculture import, agriculture export, chemical-mineral import, chemical-mineral export, energy import, energy export, manufactured import, manufactured export, miscellaneous consumption import, and miscellaneous consumption export. The raw data are purchased from the World Trade Flows database, sponsored by the University of California, Davis, and the National Bureau of Economic Research (Feenstra, 2000). They are provided in current U.S. dollars across 10 Standard International Trade Classification (SITC) one-digit sectors and include all the countries that have data. We convert these data into millions of 1995 constant U.S. dollars and regroup them to conform to the Italianer (1986) sectors. The one-digit SITC sectors we use are denoted by numbers from 0 to 9 and their nature is as follows: (0) agriculture and fishery; (1) beverages and tobacco; (2) minerals and inedible crude materials; (3) energy, fuels, 20

Reuveny’s work has two limitations from our perspective. It employs events data to measure all types of conflict, not just MIDs, and its sample includes a few major powers in 1963–1994. Reuveny’s estimates are not perfect (as in all studies), but the implied issue is empirical, not theoretical.

Economic Openness and Conflict

195

and lubricants; (4) animals, vegetable oils, fats, and waxes; (5) chemicals; (6) basic manufactured products; (7) manufactured machinery and transport products; (8) miscellaneous consumption products; and (9) goods not classified by kind. The five Italianer sectors and the way they match the SITC sectors listed in parentheses are agriculture-fishery (SITC 0), chemicalmineral (SITC 2, 5), energy (SITC 3), manufactured products (SITC 6, 7), and miscellaneous consumption products (SITC 1, 4, 9, 8). We do not use the SITC sectors directly because we do not have the required supply and demand coefficient estimates to form ex ante expectations for each sector. The trade data record only nonzero values. So for dyads covered by the trade database, trade values are set to zero for years showing no values, per exchange with Robert Feenstra. Moving to the control variables, gdp initiator and gdp target are real GDPs of the conflict initiator and the target, respectively, in millions of 1995 constant U.S. dollars. The data are from the World Bank (2002). We do not use the trade-to-GDP ratio measure because it may misrepresent the effect of trade. For example, if trade does not change and GDP falls, the ratio rises. Suppose we find in this case that a rise in the ratio reduces conflict. One often interprets this to imply a rise in trade reduces conflict, but in our case it is the fall in GDP that reduces conflict, not a rise in trade. This potential problem is discussed by Mansfield and Pevehouse (2000) and Keshk et al. (2004). These studies include trade and GDP separately in the model, a practice we follow. By including the GDP, we also control the size of the economy, similar to what the trade-to-GDP ratio does. The remaining controls are standard in studies of directed dyads (e.g., Bennett and Stam, 2000a, 2004; Hegre, 2004). The variable contiguity is set to 1 when two states in a dyad are contiguous on land or are separated by up to 150 miles of water and zero otherwise. Contiguous states are more likely to fight one another. The distance is logged distance between the capitals of two states in a dyad. Some scholars (e.g., Russett and Oneal, 2001; Dorussen, 2006) argue that a rise in distance decreases the probability of MID initiation, whereas others (e.g., Ray, 2003) argue against including this variable. For completeness, we include both variables in our analysis. We construct three democracy-related variables to capture the effects of domestic political regimes. These variables are based on the POLITY2 variable from the POLITY IV data set, ranging from −10 to 10 (−10 = high autocracy; 10 = high democracy): initiator democracy is the level of democracy of the MID initiator, target democracy is the level of democracy of the target, and regime dissimilarity is the absolute value of the difference between their democracy levels. The initiator’s democracy

196

Democracy and Economic Openness in an Interconnected System

variable tests whether democratic countries are more likely to initiate MIDs. the target’s democracy variable tests whether democracies are more likely to be MID targets, and the difference variable measures the effect of political dissimilarity on MID initiation. Following Bennett and Stam (2000b, 2004), we include two capability variables for each dyad, which are computed based on the COW composite index of national capabilities. power balance is the ratio of the larger state’s capabilities divided by the total of the initiator’s and target’s capabilities, ranging between 0.5 (equality) and 1 (dominance). The ratio tests if MID initiation is less likely for dyads of relatively more balanced capabilities. initiator capability ratio is the ratio of the initiator’s capabilities over total dyad capabilities and tests whether a rise in the initiator’s capabilities encourages MID initiation. alliance is a dummy variable indicating the presence (1) or absence (0) of dyadic defense pacts, neutrality pacts, or ententes. minor power is coded 1 if both states in a dyad are minor powers and zero otherwise. In the EUGene computer program, which is used to generate these variables, the major powers are China, France, the United States, the United Kingdom, and USSR/Russia (Bennet and Stam, 2000b).

Research Design Issues Several issues guide the design of our research in this chapter. The first issue is that our dependent variable aggregates the different types of MIDs, ignoring their variations. We have addressed this issue in the additional analysis in the main text. A second issue is that conflict rises or falls in the theoretical model, but our dependent variable admits only values of 0 and 1. This issue applies to all studies using MID data and no one argues that conflict is really a 1–0 process. The logit/probit estimation, however, predicts the probability of a dispute, which is a continuous variable (like conflict in our theoretical model), preserving the model’s spirit. A third issue is that although many studies employ MID initiation (e.g., Bennett and Stam, 2000a, 2004; Reiter and Stam, 2003; Hegre, 2004; Lai and Slater, 2006), the MID data code initiators as the side that first takes military action. This practice ignores actions falling below the MID threshold, may not represent who started the quarrel, ignores that for actual conflict to occur the target is important, and overlooks the possibility that when one side moves, the other may essentially ignore it. That said, we believe that these limitations are not crucial here; because our theory focuses on the

Economic Openness and Conflict

197

decision of one side to initiate military action, the MID initiation measure is appropriate for our purpose.21 Fourth, since the dependent variable is dichotomous, we use the probit estimator. Our sample includes 213,790 observations but, as usual, only a very small number of directed dyad-years involve MIDs (352). Fifth, conflict as the dependent variable may affect some right-handside variables, including trade and democracy, causing simultaneity bias. This possibility applies to almost all statistical studies that model conflict as a function of various variables. Deviating from this norm, Keshk et al. (2004) model the simultaneous relationship between aggregate trade and MID. This method, however, is not feasible here. In our case, the system of equations would have to include 11 endogenous variables (5 imports, 5 exports, and 1 conflict variable), but the Maddala–Amemiya method used by Keshk et al. (2004) only works for a system of two endogenous variables (one continuous and one dichotomous). Therefore, we employ a suboptimal solution for addressing the simultaneity issue; we lag the right-hand-side variables, a strategy that is typically applied in the literature. Sixth, given the pooled nature of our data, the error term of our model may be subject to serial correlation, heteroskedasticity, and temporal dependence within dyads. To control for heteroskedasticity and serial correlation, we estimate robust standard errors clustered over dyads. We also include the peace-year counter (peace_years) and three cubic spline variables (splines) to control for the duration dependence of peace within dyads, as suggested by Beck, Katz, and Tucker (1998). Seventh, sectoral trade flows may be correlated with one another, reflecting economic interrelationships, and may also affect other independent variables, which may lead to multicollinearity. We investigate this possibility by inspecting bivariate correlations and variance inflation factors (VIFs). VIF = 1/(1 − RI2 ), where RI2 is R2 from regressing the ith independent variable on all the other independent variables. If VIF > 10, multicollinearity is a problem. In this case, estimated coefficients and their variances remain unbiased, but the variances may be large. If the results are significant even now, they are robust since the test is demanding (Achen, 1982; Gujarati, 2002; Kennedy, 2005). Finally, we assess the effects of trade variables on conflict by testing hypotheses on the signs of effects, computing the sizes of statistically significant effects, testing hypotheses on the equality of significant effects, and 21

Whereas most quantitative conflict studies use the MID data, some studies use events data. Several studies discuss advantages and disadvantages of events data versus MID data (e.g., Reuveny, 2002).

198

Democracy and Economic Openness in an Interconnected System

employing the model for in-sample prediction. Because our sample is large, we employ the relatively more conservative two-tailed tests for the statistical significance of the effects of our variables, even though our hypotheses are directional and one-tailed tests are appropriate. We employ significance tests because we analyze a sample, not the entire population. Even if our sample included all the countries in the 1970–1997 period, this would not have been the population since we would not have data for 2000, 1960, and so on, and we only have one collected data set. In evaluating the size of the effect, we compute the relative risk for each significant variable, which is the probability of an event in a treatment group divided by its probability in a control group (e.g., Gartzke et al., 2001; Russett and Oneal, 2001; Dorussen, 2006). Specifically, we compute the probability of MID initiation when a variable rises by one standard deviation above its sample mean, divided by the probability of MID initiation when this variable is set at its sample mean, ceteris paribus. Strictly speaking, the interpretation is as follows: Assume there are 100 MIDs and the relative risk rises 5% when variable X rises. If we had many samples for a model when X rises, the average number of predicted MIDs across these samples would be 105. The one-standard-deviation rise is widely used. We also test whether the statistically significant effects of our trade variables are equal in size (the insignificant effects are statistically indistinguishable from zero), which is an important test of the practice of using total trade data aggregated across sectors and flow directions and which is common in the literature. This practice is equivalent to assuming the same coefficient applies to imports and exports in different sectors or that all these trade flows have identical effects on conflict. We also evaluate the results with in-sample predictions. This evaluation computes the predicted probability of a MID for each directed dyad-year and compares it to a threshold, which is typically set to the proportion of MID initiation in the sample. The sample proportion of MID initiations is a reasonable cutoff because an uninformed observer can always use this value as a guide to provide an educated guess of the probability of MID initiation. If the predicted probability is larger than the threshold, the model is said to have predicted a MID, and vice versa. The proportion of correctly classified cases indicates the model’s predictive power.

Empirical Findings Table 6.2 in the main text presents the estimation results. The model has a pseudo R2 above 0.3, indicating good explanatory power. In terms of the control variables, the results largely resemble previous findings. The effect of

Economic Openness and Conflict

199

a rise in initiator capability ratio on the likelihood of MID initiation is positive and significant at the 5% level. A state is more likely to initiate MIDs when its share of the dyad’s total capability increases. The effect of a rise in power balance is insignificant. Conflict is not less likely between states with relatively equal capabilities. The coefficient of alliance is insignificant, suggesting that the presence of a military alliance has little effect on MID initiation. The effect of a rise in initiator democracy is negative and significant at the 5% level, the effect of a rise in target democracy is insignificant, and the effect of a rise in regime dissimilarity is positive and significant at the 1% level. More democratic countries are less likely to initiate MIDs, but not more likely to be targeted. Countries are more likely to initiate conflict against those whose regimes differ more from their own. The effects of contiguity and distance are significant at the 1% level and are positive and negative, respectively. Contiguous states are relatively more likely to initiate MIDs against each other, whereas a rise in distance between countries reduces the likelihood of MID initiation. The effect of minor power is negative and significant at the 10% level. Minor power dyads are less likely to experience MIDs. The effects of initiator gdp and target gdp are positive and significant at the 1% level. Larger economies are more likely to be initiators and targets of MIDs. The results for the trade variables are discussed in the text. Here we list again the obtained sign of the effect and note its statistical significance level. The coefficient of agriculture import is negative as expected and significant at the 1% level. The coefficient of agriculture export is negative as expected but not significant. The coefficient of energy import is negative as expected and significant at the 10% level. The coefficient of energy export is positive as expected and significant at the 1% level. The coefficients of manufactured import and manufactured export are both positive as expected and significant at the 10% and 5% levels, respectively. The coefficients of chemical-mineral import/export and miscellaneous consumption import/export are not significant. These effects vary across sectors and trade flows, as our theory suggests, but how do they compare with those of the nontrade variables? In Table 6.2, relative to the control group, the risk of MID initiation due to a one-standard-deviation rise above the sample mean is 44% higher for the initiator capability ratio, 16% lower for initiator democracy, 20% higher for regime dissimilarity, 29% lower for log distance, 51% higher for initiator GDP, and 29% higher for target GDP. Hence, the comparison in terms of substantive effects indicates that trade has an important effect on conflict initiation, but not the most important effect.

200

Democracy and Economic Openness in an Interconnected System Table 6.A1. Equality tests on significant positive and negative effects of sectoral trade flows Agriculture/fishery import

Energy import

21.39∗∗ 22.34∗∗ 21.15∗∗

9.30∗∗ 4.68∗ 5.35∗

Energy export Manufactured import Manufactured export Note:

∗

significant at 5% level; ∗∗ significant at 1% level.

We examine the diagnostic tests for multicollinearity. Even though the largest correlation between any two independent variables is quite high (0.87 for manufactured goods export and chemical/mineral export), all the individual VIFs are smaller than the threshold value of 10. In all cases, the correlation between any two export flows is the same as that between the import flows of the same sectors, which is one check of the validity of the data. The VIF diagnostics indicate that multicollinearity is not a serious concern.

Additional Analysis Table 6.A1 presents results for the coefficient equality tests. The first row denotes the two flows (agricultural import and energy import) that reduce the probability of MID initiation. The first column denotes the three flows (energy export, manufactured import, and manufactured export) that encourage MID initiation. The results show that all the pairwise equality tests are statistically significant at conventional significance levels for two-tailed tests. Hence, the effects of these sectors are not equal in size, suggesting (again) that aggregating exports and imports in different sectors is inappropriate. Table 6.A2 presents results for the in-sample prediction of MID initiation for Model 1. The cutoff probability is 0.0016 (the sample average Table 6.A2. In-sample prediction of MID initiation Predicted MID Actual MID

0

1

Total

0 1

193,647 67

19,791 285

213,438 352

Total

193,714

20,076

213,790

Note: Model predicts MID Initiation = 0 correctly in 91% of the cases; model predicts MID Initiation = 1 correctly in 81% of the cases.

Economic Openness and Conflict

201

Table 6.A3. Effects of bilateral sectoral flows on display of force and use of force (1) Display of force+ Agriculture-fishery import Agriculture-fishery export Energy import Energy export Chemical-mineral import Chemical-mineral export Manufactured import Manufactured export Miscellaneous consumption import Miscellaneous consumption export Initiator capability ratio Power balance Initiator democracy Target democracy Regime dissimilarity Alliance Log of distance Contiguity Minor power Initiator GDP Target GDP Observations

(2) Use of force+

−0.00046 [4.25]∗∗∗ −0.00007 [0.74] −0.00004 [1.76]∗ 0.00004 [2.61]∗∗∗ −0.00004 [0.69] −0.00008 [1.26] 0.00003 [1.94]∗ 0.00003 [2.32]∗∗ −0.00006 [0.88] 0.00002 [0.51] 0.36393 [2.35]∗∗ 0.04679 [0.26] −0.00708 [1.93]∗ −0.00019 [0.05] 0.01147 [3.40]∗∗∗ 0.00617 [0.08] −0.07961 [4.85]∗∗∗ 0.82614 [6.33]∗∗∗ −0.34438 [2.01]∗∗ 0.06672 [2.84]∗∗∗ 0.10066 [4.32]∗∗∗ 213790

Note: Z statistics in brackets. ∗ significant at 10%; ∗∗ significant at 5%; Constant, peace-year, and spline variables not reported.

−0.00035 [3.31]∗∗∗ −0.00007 [0.74] −0.00005 [1.83]∗ 0.00004 [2.26]∗∗ −0.00001 [0.28] −0.00012 [1.80]∗ 0.00003 [2.05]∗∗ 0.00005 [3.14]∗∗∗ −0.00003 [0.75] −0.00012 [1.82]∗ 0.49638 [2.87]∗∗∗ 0.08097 [0.40] −0.00618 [1.52] 0.00356 [0.83] 0.00898 [2.31]∗∗ 0.00714 [0.08] −0.07474 [4.13]∗∗∗ 0.81936 [5.63]∗∗∗ −0.39713 [1.91]∗ 0.03584 [1.41] 0.11342 [4.37]∗∗∗ 213790 ∗∗∗

significant at 1%.

202

Democracy and Economic Openness in an Interconnected System

probability of MID initiations). Model 1 correctly predicts the absence of MID initiation in 91% of the dyad-years and the presence of MID initiation in 81% of the cases. Thus, the model performs reasonably well in terms of in-sample prediction. Table 6.A3 reports the model results based on alternative indicators of dyadic conflict. For both models, it is clear that the results closely resemble those obtained for the model based on all MIDs. In the model for MIDs involving at least the display of force, the statistical results for these control variables are essentially identical to those based on all MIDs. For the model for MIDs involving the use of force and war, the results for the control variables are similar to those based on all MIDs, except that the effects of initiator democracy and initiator gdp are not significant.

PART III

BRINGING IN THE ENVIRONMENT

SEVEN

Democracy and the Environment

INTRODUCTION

In this chapter, we begin the third and last part of this book. So far, the inquiries we have conducted could in principle have taken place anywhere, be it on planet Earth or somewhere else in space, as long as there exist people who interact with one another, markets that enable people to attach monetary values to goods and services, and governments that decide to use force against other countries. The global political economy and the interactions it entails, however, invariably occur within the confines of the physical environment, the biosphere of planet Earth. Consequently, the political and economic forces we focus on in this book – democracy, economic openness, and military conflict, and the complex transformations they entail – can influence this very physical environment within which we all live. The questions we seek to answer in the third part of the book concern whether and how these effects on the environment occur. Is a rise in the level of democracy good or bad for the environment? How does a rise in the level of international trade affect the environment? What are the implications of being involved in military conflict for a country’s physical environment? Chapters 7, 8, and 9 in the third part of the book address these important questions. Does a rise in the level of democracy of a country reduce or increase its environmental degradation? The democracy–environment link has captivated policy makers and the media with conflicting arguments and evidence, for which we are not taking sides. For example, former U.S. Vice President Al Gore has asserted that greater political and civil freedoms promote environmental quality (Gore, 1992). Geographer and media figure Jared Diamond (2005b) noted that the Scandinavian democracies have the best environmental record in the world. The experience of Central and 205

206

Democracy and Economic Openness in an Interconnected System

Eastern Europe is also quite telling (Forest Watch Indonesia and Global Forest Watch, 2002). Under the Communist system, the environment seriously deteriorated; the arrival of democracy to the region resulted in much better environmental policies. On the other hand, Diamond argues that autocracy also can benefit the environment because not only can it legislate and implement laws by decree but it also may possess a long-run view due to its potentially long-lasting reign. In his recent bestseller Collapse: How Societies Choose to Fail or Survive, Diamond (2005a) cites a number of autocrats that protected forests, including seventeenth-century Japan’s Tokugawa Shogunate and the Dominican Republic’s Rafael Trujillo and Joaquin Balaguer. Diamond (2005b) also points to contemporary China, which phased out leaded gasoline over a period of only one year and ended logging at once by decree, demonstrating the beneficial impact of autocracy for the environment. Hence, Diamond claims that one cannot generalize the effect of democracy or autocracy on the environment. Scholars also debate the effect of democracy on environmental degradation, both theoretically and empirically. Some theorists claim that democracy reduces environmental degradation; others argue that democracy does not reduce environmental degradation and may even harm the environment. Despite the contentious nature of this issue, systematic empirical evidence is relatively scant and mixed, particularly in terms of large-N statistical analysis. The various theoretical mechanisms on the effect of democracy on environmental degradation, though often conflicting, are all plausible given their theoretical assumptions. In other words, it is not possible to adjudicate the competing claims based solely on the theoretical arguments. Because these causal mechanisms may operate at the same time, it is important to study their overall or net effect empirically, which is the focus of our analysis in this chapter. The dependent variable in our empirical/statistical model, environmental degradation, is measured with five salient types of human-induced degradation: carbon dioxide (CO2 ) emissions, nitrogen oxide (NOx ) emissions, land degradation, forest depletion in terms of the rate of deforestation and the share of forested area, and organic pollution in water. For each type of degradation, we estimate several regression-based statistical models employing various measures of democracy or autocracy. Our empirical analysis differs from previous studies in a number of ways. First, previous empirical analyses studied the effect of democracy on the signing of international environmental agreements, resource scarcity and access to environmental amenities, and human actions that directly harm the environment. We focus on human activities that directly affect

Democracy and the Environment

207

environmental quality. Although it is important to study resource scarcity, access to environmental amenities, and commitment to environmental protection, we believe it is also important to study human actions that directly damage the environment. After all, the best way to protect the environment is to minimize the damage to the environment in the first place. Second, our sample size is generally larger than that of previous studies. Depending on data availability for the environmental indicators, the number of countries included in the empirical sample varies from 105 for the land degradation analysis to 143 for the CO2 emissions per capita analysis. A larger sample size allows us to have more variations in the data to facilitate statistical inference and gives us more confidence in the generalizability of our findings. Third, the empirical results for the effect of democracy on the environment are consistent across all the aforementioned types of human-induced degradation, except for deforestation: a rise in democracy reduces environmental degradation and improves environmental performance. The effects on forest depletion are mixed. Our cross-sectional analysis for forests finds that democracies tend to have relatively larger shares of forested area composing their total land area, but they also have higher rates of deforestation relative to nondemocracies. The substantive effect of democracy on the environment is considerable, but it varies in size across different aspects of environmental degradation. We also find that democracy has nonmonotonic effects that vary across the environmental indicators. The remainder of this chapter proceeds as follows. The next section discusses the theoretical channels linking democracy to environmental degradation. The section that follows presents our empirical model in terms of the variables, data, measures, and expected effects. The fourth section of this chapter discusses several technical design issues pertaining to the empirical analysis, followed by a section presenting and discussing our empirical results. Finally, we summarize our key findings. EFFECT OF DEMOCRACY ON THE ENVIRONMENT

This section discusses the two opposing views regarding the effect of democracy on the environment and the associated empirical evidence. The debate in the literature turns on institutional attributes of political regimes affecting environmental problems: the role of public opinion in policymaking, interest groups’ aggregation and representation, state autonomy, social movement mobilization, and the flow of information. Scholars take positions in the debate by emphasizing some of these regime characteristics.

208

Democracy and Economic Openness in an Interconnected System

Democracy Improves Environmental Quality A rise in the level of democracy in a country improves its environmental quality. Scholars holding this view have offered five different supporting arguments. According to the first argument, political rights and freedom of information promote the cause of environmental interest groups, which in turn raise public awareness and encourage environmental legislation (see, e.g., Schultz and Crockett, 1990; Payne, 1995). This effect works through environmental groups and public opinion at large. Information on environmental issues flows more freely, and political rights are more numerous and better protected in a democracy than in an autocracy. Environmental groups, therefore, are often more successful at informing people and organizing them to act on environmental problems in a democracy than in an autocracy. Whereas the elite in an autocracy may be more educated than the public (as education tends to rise with income), the autocratic regime censors information flows, and its decision making is more autonomous than that of a democratic government. Environmental degradation may not be reported by the media to the people. In contrast, because democracy allows for free media, environmental problems are more likely to be reported in the news. People in a democracy, therefore, are more likely to be informed about the environment than are members of the elite in an autocracy. Better informed actors, in turn, are more likely to act on environmental problems, raising environmental quality. A second argument is that democracies are more responsive to the environmental needs of the public than are autocracies (Kotov and Nikitina, 1995). This argument works through electoral accountability and the ability of groups to mobilize socially, achieve political representation, and influence public policymaking. Democracies hold regular and free elections, which can bring to power new parties, including those friendly to the environment (e.g., the Green Party in Germany). In an autocracy, the distribution of political power is concentrated, reducing the likelihood that environmentalists will come to power. Thus, environmentalists stand a greater chance of affecting policymaking in a democracy than they do in an autocracy. Of course, this logic implies that people can also freely elect extreme antienvironmental parties. Casual observation, however, suggests that in reality such situations do not occur frequently. A third argument focuses on institutional and ideational features of democracy. According to this argument, democracies are more likely to comply with environmental agreements because they respect the rule of

Democracy and the Environment

209

law, which in turn raises environmental quality (Weiss and Jacobsen, 1999). Also, democracies respect economic freedom and, therefore, have market economies (Berge, 1994). The market, in turn, promotes environmental quality. Furthermore, because democracies respect human life more than autocracies, they are more responsive to life-threatening environmental degradation. And to the extent that democracies engage in fewer wars, they should have a higher level of environmental quality because wars destroy the environment (Gleditsch and Sverdlop, 2003). Finally, famines promote environmental degradation because they divert attention away from longrun environmental concerns. Famines tend not to occur in democracies because democratic governments are more responsive to the needs of the people (Sen, 1994). Hence, environmental degradation will be higher in autocracies than in democracies. A fourth argument expects that the elite in an autocracy will be less pro-environment than the masses or the public at large in a democracy (Congleton, 1992). The logic of this argument relies on environmental regulation that curtails pollution and waste. With prevailing technologies and materials, environmental regulation lowers production and consumption, which, in turn, imposes a higher cost on the elite in an autocracy than on the masses in a democracy, because the ruling elite in an autocracy hold a much larger share of national income than most people in a democracy and are therefore relatively less pro-environment than the masses in a democracy. A fifth argument begins with the observation that because environmental degradation typically develops slowly, the discount rate and time horizon of the government have important impacts on environmental regulation (Congleton, 1992). The masses in a democratic country should have less at stake over regime change than the elite in an autocracy. In an autocracy, the elite are tightly linked to the leader. If the leader loses power, the elite may suffer heavy losses or even lose their lives. Facing this possibility, the elite may wish to prevent regime change by force, and to this end more resources are allocated to oppression. The elite may also think that the change is inevitable, becoming hedonic. Both actions raise the discount rate and reduce the time horizon of the autocratic government. As a result, the ruling elite in an autocracy will ignore the environmental damage expected in the future. If they invest more today to suppress real or potential rebels, they allocate resources away from environmental issues. If they consume more today, they ignore environmental degradation that takes a long time to rectify or current activities that will cause damage in the future. In both cases, environmental quality will decline.

210

Democracy and Economic Openness in an Interconnected System

Democracy May Worsen Environmental Degradation Democracy does not reduce environmental degradation and may even increase it. Scholars have provided several explanations for this position. First, Hardin (1968) warns about the impending hazards of unchecked natural resource exploitation and environmental mismanagement by selfinterested individuals and groups. When private property rights of natural resources are not well defined, as is often the case with “the commons” (e.g., clean air, oceans, forests), free individuals or interest groups tend to overexploit such resources and ignore the damage their economic actions inflict on the environment. Gleditsch and Sverdlop (2003: 70) note that Hardin’s “Tragedy of the Commons” does not encourage confidence in the effect of economic and political freedom on environmental quality. Second, Paehlke (1996: 28) argues that “the great danger for both democracy and the environment is that, while economy and environment are now global in character, democracy functions on only national and local decision levels.” Thus, global environmental problems may not necessarily be attended to in a timely manner. Heilbronner (1974) argues that global population growth threatens global environmental quality. Being autonomous decision makers, autocracies can curtail human reproduction, but democracies are held accountable by the public and therefore respect citizens’ rights, including those involving human procreation. Third, Dryzek (1987) notes that democracies tend to be market economies, where business interest groups have considerable clout. His argument highlights the asymmetric influence of profit-oriented corporate interests in capitalist democracies. Dryzek (1987: 121) lists countries in which democracy is systematically skewed in favor of corporate interests, while “environmental groups have a hard time getting a foot in the door.” Corporate interests, in turn, seek to maximize profit, not necessarily to better environmental quality. Thus, democratic leaders accountable to business interests that support their coming to power may not necessarily value environmental quality. “Polyarchy,” Dryzek argues, “will normally yield to the imperatives of the market, if not always to the interests of large corporations. . . . At their corporatist worst, polyarchies degenerate into caricatures of the ideal, with some dire consequences for ecological rationality” (Dryzek, 1987: 125).1 1

It is possible to argue that Dryzek’s argument is not about democracy but rather about the effects of wealth in a democracy. Our reading of Dryzek differs somewhat: his argument focuses on the ability of business to affect government policies through the democratic process and, hence, is about democracy.

Democracy and the Environment

211

Fourth, Midlarsky (1998) argues that democracies often experience public policy inaction where environmental degradation is concerned. Democratic leaders have the tendency to please competing interest groups to win as many votes as possible. “Corporations and environmental groups can fight each other to a standstill, leaving a decision making vacuum instead of a direct impact of democracy on the environment. As a result of budget constraints, democracies may not be responsive to environmental imperatives but to more pressing issues of the economic subsistence of major portions of the voting public” (Midlarsky, 1998: 159). In addition, democratic governments may be reluctant to alleviate environmental degradation because some groups are expected to benefit (or lose) from environmental policies more than others (Midlarsky, 1998: 159).

Previous Empirical Studies Extant empirical studies may be categorized according to the nature of their dependent variables. One set of studies examines government commitment to environmental quality in terms of signing international agreements that protect the environment (e.g., Congleton, 1992; Neumayer, 2002). A second set of studies investigates resource scarcity and access to environmental amenities such as safe water or sanitation (e.g., Shafik, 1994; Torras and Boyce, 1998). A third set of studies explores human activities harmful to the environment, such as greenhouse gas emissions (e.g., Midlarsky, 1998; Gleditsch and Sverdlop, 2003). Focusing on international agreements assumes (implicitly or explicitly) that they promote environmental quality. These agreements, however, also reflect international negotiations and bargaining. The end product may not necessarily address the specific problems of any particular country. At times, these agreements also include “cheap talk” that is intended to appease environmentalists. Resource scarcity and access to environmental amenities can reflect nonenvironmental and structural conditions concerning wealth and resource endowment, which are not directly related to environmental quality. The reader may recall that we seek to study the effect of democracy on human actions that directly degrade the environment. Before we proceed, a word of caution is necessary. Although the distinction between signing agreements and human actions that hurt the environment is clear, the distinction between conditions of resource scarcity and actions that cause environmental degradation may be blurred. For example, land degradation can suggest soil pollution by humans, but it may also indicate the

212

Democracy and Economic Openness in an Interconnected System

scarcity of productive land that forces humans to overexploit land already in use. The empirical literature pertaining to our focus on human actions that directly damage the environment is relatively small. In a sample of 118 countries for the year 1989, Congleton (1992) found that democracies have higher levels of methane and chlorofluorocarbon emissions per capita than autocracies. Midlarsky (1998) reports several findings. A rise in the level of democracy increases CO2 emissions per capita in a sample of 98 countries in the year 1990, soil erosion by water in a sample of 97 countries for the 1980s, and the percentage of annual deforestation between 1981 and 1990 in a sample of 77 countries. The level of biodiversity, as measured by the percentage of protected land area, rises with democracy for a sample of 100 countries in 1993. But democracy does not affect fresh water availability in a sample of 97 countries in 1990 and the level of soil erosion by chemicals in a sample of 97 countries during the 1980s. Barrett and Graddy (2000) found that a rise in democracy lowers per capita sulfur dioxide (SO2 ) emissions in a pooled sample of countries for about 33 years and particulate emissions in a sample of 27 countries for about 38 years. The effect of democracy on water pollution is not statistically significant in a sample of 45 countries for about 29 years. Torras and Boyce (1998) found that the level of democracy, measured by the 1995 values of the Freedom House indicators of political rights and civil liberties, reduced air pollution (SO2 , smoke, particulate emissions) in pooled samples of 19–42 countries from 1977 to 1991 and decreased water pollution (dissolved oxygen, fecal coliform, access to safe water, and access to sanitation) from 1977 to 1991 in pooled samples of about 58 countries. These results, however, do not appear to be robust across samples. Based on a sample of the 148–185 sites in 24 countries across three different periods (1979–1982, 1983–1986, and 1987–1990), Scruggs (1998) found that democracy does not affect water pollution and particulate emissions, but it reduces SO2 emissions. Gleditsch and Sverdlop (2003) reported that democracy reduced CO2 emissions per capita for a sample of 108 countries in 1990. We believe that it is safe to conclude that extant empirical evidence in the literature on human actions that directly degrade the environment is mixed. It is also apparent that some studies employed relatively small samples in terms of the country, year coverage, or both. Previous studies also differed in terms of empirical model specification. For example, many empirical analyses did not control for the environmental Kuznets curve effect (see later discussion), and none of the studies cited earlier controlled for the possible effect of military conflict on the environment (except for Midlarsky, 1998) or the effect of trade openness. In addition, these studies focused on

Democracy and the Environment

213

different aspects of environmental degradation and their number of indicators ranged between one and seven. Some of the indicators used measure environmental amenities, as opposed to our focus on human actions that damage the environment. In short, room exists for additional systematic empirical analyses that investigate multiple indicators of environmental degradation in samples that are as large as possible while controlling for the Kuznets curve effect, trade openness, and military conflict. EMPIRICAL MODEL AND ANALYSIS

This section presents our empirical model for evaluating the effect of democracy on environmental degradation, related research design issues, and the statistical findings. Because the competing causal mechanisms of how democracy affects the environment may operate simultaneously, the purpose of our empirical analysis is to assess the net effect of these forces, finding out overall if democracy is good or bad for the environment. As in the other chapters, this discussion is self-contained and does not require statistical expertise. The technical details of the statistical model, data sources, and measure construction are in the chapter appendix, following the same structure of presentation as in the main text.

Empirical Model To assess empirically the effect of democracy on the environment, we specify and estimate a statistical model of environmental degradation. As in other chapters, we denote variables with small capital letters, and their coefficients with Greek notations. Each coefficient indicates the effect of the independent variable on the dependent variable. The notations ␤y and ␤c denote vectors of coefficients for year and country fixed-effects variables, respectively. The notation εt denotes the unexplained random error in the model. The variable subscripts t and t − 1 indicate the time period of the variable, where t represents the current period and t − 1 the previous time period (a lagged variable). To simplify the presentation, we refer to the variables without time subscripts t or t − 1. environmentt = ␤0 + B1 regime typet−1 + ␤2 lagged environmentt−1 + ␤3 real gdppct−1 + ␤4 real gdppc squaredt−1 + ␤5 trade opennesst−1 + ␤6 population densityt−1 + ␤7 wart−1 + ␤y year fixed effectst + ␤c country fixed effectst + εt .

(7.1)

214

Democracy and Economic Openness in an Interconnected System

The dependent variable, environment, is a multidimensional concept of environmental degradation. We choose to focus on five specific human activities that directly harm the environment: greenhouse gas emissions, air pollution, water pollution, forest depletion, and land degradation. We analyze one representative, salient indicator for each area, but two indicators for forest depletion. Hence, we have six dependent variable measures. Specifically, the first dependent variable focuses on CO2 emissions because this gas constitutes about 60% of greenhouse emissions. The second indicator is NOx , because this gas is a major health hazard and an important source of smog in urban areas. The third indicator relates to organic pollution in water. The fourth and fifth indicators are the rate of deforestation and the forested area share out of total country area, respectively. The sixth indicator concerns land degradation, which seriously undermines agricultural activities. We measure the key independent variable, regime type, in two different ways: continuously and dichotomously. The continuous or interval measure of regime type is the widely used indicator of the level of democracy from the POLITY IV data employed in previous chapters. In the dichotomous measurement approach, the variable democracy dummy is set to 1 if a country’s polity score is greater than or equal to 6, and zero otherwise (i.e., anocracies and autocracies); the variable autocracy dummy is set to 1 if a country’s polity score is smaller than or equal to −6, and zero otherwise (i.e., anocracies and democracies). The literature on environmental degradation motivates the specification of the control variables in the model. The variable real gdppc is the gross domestic product (GDP) of a country expressed in real terms, per capita; real gdppc squared is the square of the real GDP per capita variable. The two variables together help capture the nonlinear effect of development on the environment, a phenomenon referred to as the environmental Kuznets curve, which is discussed at length in the appendix. The trade openness variable is the share of a country’s total trade over its GDP, indicating the importance of trade for the national economy; population density is the population size of a country divided by its total area; and war indicates the presence or absence of a country’s involvement in an interstate or intrastate war in a year. For the CO2 emissions per capita and water pollution models, we have pooled time-series cross-sectional data. This type of data structure requires additional control variables: lagged environment is the value of the dependent variable from the previous year (for which the rationale is discussed in the appendix), country fixed effects are country

Democracy and the Environment

215

dummy variables that capture country-specific heterogeneity, and year fixed effects are year dummy variables that measure year-specific effects. These additional control variables do not appear in the other models because we only have cross-sectional data for those other environmental variables.

Research Design Issues This section discusses several design issues that are special to the empirical analysis in this chapter. Again, we refer the interested reader to the appendix of this chapter for further details. The first design issue concerns the sample information and whether we conduct a cross-sectional analysis or a cross-sectional time-series (pooled) analysis for each environmental indicator. Data availability determines this choice. For the CO2 emissions and water pollution, we have time-series cross-sectional data and conduct pooled analyses. The sample of CO2 emissions per capita includes 143 countries from 1961 to 1997, and the sample of organic pollutants in water includes 112 countries from 1980 to 1998. For NOx , deforestation rate, forested land area, and land degradation, we have cross-sectional data and conduct cross-sectional analyses. The NOx sample includes 118 countries in 1990. The rate-of-deforestation sample covers 134 countries over two decades (1980s and 1990s). The share-of-forested-area sample also covers 134 countries but for three different years (1980, 1990, and 2000). The landdegradation sample includes 105 countries in the 1980s. It is worth noting that our sample coverages are larger than previous empirical studies on the topic. The second design issue concerns the possibility of endogeneity for our right-hand-side variables. In the appendix, we provide possible examples from our context. We address this possibility by lagging all the right-handside variables. Third, slow-changing national structural variables (e.g., climate, education) and global or local biophysical attributes (e.g., atmospheric integrity, existing damage) may also affect environmental degradation. Hence, environmental degradation is likely to exhibit inertia. We model this tendency with the variable lagged environment when we have time-series crosssectional data (for CO2 emissions and organic water pollution). Fourth, we employ appropriate econometric techniques to deal with possible assumption violations associated with the error term of the statistical model. Failure to do so may result in incorrect inferences regarding the effect of democracy on the environment. We also test for evidence of excessive correlation among the independent variables.

216

Democracy and Economic Openness in an Interconnected System

Finally, to give some substantive meaning to our findings, we discuss the size of democracy’s effect on the environment. We discuss the size of the estimated coefficient of democracy, provided that the coefficient is statistically different from zero. If the coefficient is not statistically different from zero, it is not necessary to discuss the size of the effect. In models that include a lagged dependent variable, we also compute the long-run impact of democracy.

Empirical Findings In discussing our findings, we first focus on the main analysis using the continuous measure of regime type, and then we discuss the results using the dichotomous measures of regime type. Findings for the control variables and additional technical details can be found in the appendix. Table 7.1 reports the results for the net effect of the level of democracy on environmental degradation. The columns of this table present the results for CO2 emissions per capita, NOx emissions per capita, organic pollutants in water, the rate of deforestation, the share of total land area that is forested, and the share of degraded land, respectively. Across the six columns in Table 7.1, the net effect of the level of democracy on six different environmental indicators is always statistically different from zero, consistently in the direction of reducing environmental degradation, except for the deforestation rate. Relative to their less-democratic counterparts, more democratic countries produce fewer CO2 emissions per capita, fewer NOx emissions per capita, fewer organic pollutants in water, and less land degradation. The effects of democracy on forest depletion, however, are mixed: democracies experience higher deforestation rates (recall that a negative value for the rate of deforestation implies faster deforestation whereas a positive value indicates faster afforestation) but have larger shares of forested land. It is worth noting that the effects of democracy on CO2 emissions and the level of organic pollution in water are significantly different from zero in spite of the taxing country and year dummies in the models. One may wonder how to reconcile the conflicting findings of the two indicators of forest depletion. On the one hand, a rise in democracy is associated with faster deforestation. On the other hand, the share of the forested land area rises with the democracy level. Since the forested land area measure is largely cross-sectional and the deforestation rate variable captures the temporal change from decade to decade in our sample, the results suggest that although democracies tend to have relatively more forests, they

217 1.9067∗∗∗ (0.1815) 108 0.41

−0.0198∗ (0.0140) 0.0043∗∗ (0.0025) 0.1102 (0.1891) 0.00014∗∗∗ (0.00004) −2.39e-09∗ (1.53e-09) −0.0005∗∗ (0.0002)

−0.0021∗∗ (0.0011) −0.0006 (0.0011) −0.0539∗∗∗ (0.0189) 0.00009∗∗∗ (0.00002) −2.80e-09∗∗∗ (5.33e-10) 0.0007∗∗∗ (0.0002) 0.8658∗∗∗ (0.0228) 0.1247∗ (0.0643) 3833 0.99

−0.0033∗ (0.0021) 0.0005 (0.0004) 0.0063 (0.0282) 0.00001 (0.00001) −4.83e-10∗ (3.13e-10) 0.0001∗∗∗ (0.0000) 0.7895∗∗∗ (0.0832) 1.4091∗∗∗ (0.5286) 1344 0.99

Organic water pollutants per km3

−2.8402∗∗∗ (0.5745) 204 0.28

−0.1187∗∗∗ (0.0362) 0.0144∗∗ (0.0070) 0.4832∗ (0.3418) 0.0004∗∗∗ (0.0001) −1.09e-08∗∗∗ (3.57e-09) −0.0011∗∗ (0.0006)

Annual deforestation rate

3.4046∗∗∗ (0.2879) 255 0.13

0.0807∗∗∗ (0.0236) −0.0030 (0.0035) −0.3983∗∗ (0.1749) −7.2033e-05 (6.8775e-05) 9.96e-10 (2.49e-09) −0.00005 (0.0002)

Forest area share of land area

−22.6986∗∗ (9.4663) 105 0.20

−0.0281∗ (0.0219) −0.3661∗ (0.2692) 0.3917∗ (0.3004) 6.3722∗∗∗ (2.3470) −0.384∗∗∗ (0.1.45) 0.2854∗∗∗ (0.0966)

Degraded area share of land area

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%. Coefficients for year and country dummy variables not shown for models of CO2 and organic water pollutants.

Observations R2

Constant

LAGGED ENVIRONMENT

POPULATION DENSITY

REAL GDPPC SQUARED

REAL GDPPC

WAR

TRADE OPENNESS

LEVEL OF DEMOCRACY

NO X per capita

CO2 per capita

Table 7.1. Effect of level of democracy on environmental degradation

218

Democracy and Economic Openness in an Interconnected System

also tend to deplete them faster. At the theoretical level, democracy may increase or decrease environmental degradation. Empirically, deforestation turns out to be the only aspect of environmental degradation on which the effect of democracy is positive. How large are the substantive effects of democracy on these six environmental indicators? In this computation, we raise the democracy measure one standard deviation above its mean in each sample while we consistently hold other continuous variables constant at their sample means and the war variable at zero. As discussed in the appendix, because of the differences in research design and data availability, we can show both the immediate and the long-run effects of democracy (accumulated via the lagged dependent variable) for CO2 emissions and water pollution. For the other indicators, we are only able to show the immediate effect of democracy. Based on Table 7.1, a one-standard-deviation increase in the level of democracy (7.6) above its sample mean (0.84) causes CO2 emissions per capita to fall by 0.47%. The immediate effect of democracy on CO2 emissions per capita appears small. The effect that cumulates over time through the lagged dependent variable is 0.12 metric tons per capita, which is about 7.5 times of the size of the immediate effect. This amounts to a decline of about 4% in CO2 emissions. It should be noted that these numbers pertain to per capita carbon emissions. Hence, if its democracy level rises one standard deviation above its sample mean, a country such as China, whose population is about 1.3 billion people, will see a decline of 156 million metric tons per year in the long run – a large drop that can make a difference for the global environment, considering that in 1998 the total carbon emissions of Asia and the Pacific region (2.5 billion people) were 2,167 million metric tons per year (UNEP, 2003: 215). In the second column of Table 7.1, the direct effect of democracy on NOx emissions per capita is much larger. An in-sample one-standard-deviation increase in democracy (7.60) above its sample mean (1.3) causes NOx emissions per capita to decline by 14%. This effect is larger than the longrun effect of democracy on CO2 discussed earlier, which in and of itself is not small. In the third column, an in-sample, one-standard-deviation increase in democracy (7.26) above its sample mean (3.44) causes the level of organic pollutants in water to decline by about 2.4%. The long-run net effect of a one-standard-deviation increase in democracy is much larger, reaching 11%. In the fourth column, an in-sample, one-standard-deviation increase in democracy (6.6) above its sample mean (1.25) causes the average annual

Democracy and the Environment

219

deforestation rate to rise by about 271%. This obviously is a very large effect. In the fifth column, an in-sample one-standard-deviation increase in democracy (6.9) above its mean (1.86) raises the share of forested land in a country by 75%, which also is a very large effect. Finally, in the sixth column, a one-standard-deviation rise in democracy (7.8) above its mean (0.2) causes the share of severely and very severely degraded land out of total land area to decline by 20%, and is also a relatively large effect.

Additional Analyses Seeking to further probe the robustness of our results, we conduct several additional analyses, based on (1) the dichotomous measure of democracy, (2) the dichotomous measure of autocracy, and (3) a composite environmental indicator that aggregates a number of environmental attributes. For these three analyses, we discuss the results for the key regime-type variables and discuss the control variables in the appendix. Tables 7.2 and 7.3 present results for the dichotomous measures of democracy and autocracy, respectively. Although these measures lose information contained in the level variable, they provide a sharper contrast of the difference in effect between democracy and nondemocracy (a group including anocracies and autocracies), or between autocracy and nonautocracy (a group including anocracies and democracies). We therefore view the dichotomous measures of democracy and autocracy as complementary to, rather than substitutes for, the level of democracy measure. In Table 7.2, we find that the net effect of the transition to democracy is not consistently significant across the five dimensions of environmental degradation. Democratic regimes do not appear to be different from nondemocratic regimes in terms of CO2 emissions per capita, NOx emissions per capita, or the level of organic pollutants in water. But they do have significantly lower levels of land degradation, larger forested land area, and higher deforestation rates compared with nondemocratic countries. In contrast, Table 7.3 shows that relative to nonautocratic countries, autocratic regimes experience higher CO2 emissions per capita, higher NOx emissions per capita, higher levels of organic pollutants in water, smaller forested areas, and lower deforestation rates. They also tend to exhibit more land degradation, an effect that is statistically somewhat weaker. Taken together, Tables 7.1, 7.2, and 7.3 demonstrate the patterns in the effects of political regime type on different dimensions of environmental degradation. As Table 7.1 shows, the continuous democracy measure

220

0.0119 (0.0153) −0.0006 (0.0011) −0.0541∗∗∗ (0.0189) 8.37e-05∗∗∗ (1.78e-05) −2.71e-09∗∗∗ (5.37e-10) 0.00072∗∗∗ (0.00020) 0.8660∗∗∗ (0.0228) 0.1251∗ (0.0650) 3833 0.99 2.0430∗∗∗ (0.1765) 108 0.40

−0.1651 (0.1785) 0.0044∗∗ (0.0025) 1.03e-01 (2.02e-01) 1.17e-04∗∗∗ (3.72e-05) −1.64e-09 (1.46e-09) −0.0005∗∗ (0.0002)

NOx per capita −0.0235 (0.0282) 0.0005 (0.0004) 0.0069 (0.0275) 7.10e-06 (8.92e-06) −4.47e-10∗ (3.09e-10) 0.00011∗∗∗ (0.00004) 0.7907∗∗∗ (0.0832) 1.3918∗∗∗ (0.5254) 1344 0.99

Organic water pollutants per km3

3.0809∗∗∗ (0.2658) 255 0.09

0.8884∗∗∗ (0.2907) −0.0028 (0.0037) −0.3827∗∗ (0.1845) −5.05e-05 (6.75e-05) 3.78e-10 (2.49e-09) −0.0001 (0.0002)

−0.8832∗∗ (0.4409) 0.0146∗∗ (0.0074) 0.4943∗ (0.3595) 0.0004∗∗∗ (0.0001) −8.96e-09∗∗∗ (3.68e-09) −0.0010∗ (0.0006) −2.4500∗∗∗ (0.5281) 204 0.22

Forest area share of land area

Annual deforestation rate

−21.6105∗∗ (9.4136) 105 0.21

−0.6729∗∗ (0.3475) −0.3785∗ (0.2672) 0.3272 (0.2994) 6.1217∗∗∗ (2.3207) −3.65e-01∗∗∗ (1.43e-01) 0.2794∗∗∗ (0.0971)

Degraded area share of land area

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%. Coefficients for year and country dummy variables not shown for models of CO2 and organic water pollutants.

Observations R2

Constant

LAGGED ENVIRONMENT

POPULATION DENSITY

REAL GDPPC SQUARED

REAL GDPPC

WAR

TRADE OPENNESS

DEMOCRACY DUMMY

CO2 per capita

Table 7.2. Effect of democracy on environmental degradation

221 1.6855∗∗∗ (0.1867) 108 0.43

0.4211∗∗∗ (0.1727) 0.0043∗∗ (0.0025) 9.09e-02 (1.81e-01) 1.58e-04∗∗∗ (3.59e-05) −3.11e-09∗∗ (1.48e-09) −0.0005∗∗ (0.0002)

0.0363∗∗∗ (0.0138) −0.0006 (0.0011) −0.0544∗∗∗ (0.0188) 8.62e-05∗∗∗ (1.77e-05) −2.80e-09∗∗∗ (5.31e-10) 0.00071∗∗∗ (0.00020) 0.8659∗∗∗ (0.0228) 0.1102∗ (0.0640) 3833 0.99

0.0498∗∗ (0.0296) 0.0005 (0.0004) 0.0070 (0.0284) 6.41e-06 (9.02e-06) −4.57e-10∗ (3.12e-10) 0.00012∗∗∗ (0.00004) 0.7932∗∗∗ (0.0840) 1.3578∗∗ (0.5390) 1344 1.00

Organic water pollutants per km3

−3.1578∗∗∗ (0.6532) 204 0.27

1.5445∗∗∗ (0.5144) 0.0147∗∗ (0.0069) 0.6424∗∗ (0.3607) 0.0004∗∗∗ (0.0001) −9.50e-09∗∗∗ (3.31e-09) −0.0008∗ (0.0006)

Annual deforestation rate

3.4179∗∗∗ (0.3434) 255 0.06

−0.6237∗∗ (0.3113) −0.0032 (0.0035) −0.4663∗∗∗ (0.1972) −2.27e-06 (6.03e-05) −5.19e-10 (2.36e-09) −0.0002 (0.0002)

Forest area share of land area

−23.8747∗∗ (9.2644) 105 0.19

0.3654† (0.2894) −0.3802∗ (0.2736) 0.4269∗ (0.3091) 6.6709∗∗∗ (2.2913) −4.05e-01∗∗∗ (1.40e-01) 0.2891∗∗∗ (0.0967)

Degraded area share of land area

Note: Standard errors in parentheses. † significant at 10.5%; ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%. Coefficients for year and country dummy variables not shown for models of CO2 and organic water pollutants.

Observations R2

Constant

LAGGED ENVIRONMENT

POPULATION DENSITY

REAL GDPPC SQUARED

REAL GDPPC

WAR

TRADE OPENNESS

AUTOCRACY DUMMY

NOx per capita

CO2 per capita

Table 7.3. Effect of autocracy on environmental degradation

222

Democracy and Economic Openness in an Interconnected System

exhibits significant effects on the five dimensions of environmental degradation. Tables 7.2 and 7.3, however, indicate that the effect of democracy is not monotonic across various segments of the continuous democracy measure. For CO2 , NOx , and organic pollution in water, the significant effects of the continuous measures are to some extent driven by the difference between autocratic and nonautocratic regimes. For land degradation, the effect of the continuous democracy measure is driven by the difference between democracy and nondemocracy. For forest depletion (deforestation and forested area), the effects of democracy appear to be monotonic, as they are significant regardless of how regime type is measured. Finally, although it is useful to study the effect of democracy on different dimensions of environmental degradation separately, one may wonder how democracy affects some composite environmental indicator. Recently, the Environmental Performance Measurement Project (EPMP) (2002) has embarked on a large-scale effort to construct composite environmental measures of several types, which we describe in the appendix. It is worth reiterating that these composites lump together different environmental indicators, just as gross domestic product does for the economy. But the environmental story is much more complex than the economic story. For one thing, whereas money value provides a common metric to aggregate economic activities, environmental forces do not have a similar, readily available common metric (i.e., they are each measured in specific physical, chemical, or geographical units). Thus, one should interpret the composite measure-based results with caution. With this in mind, we employ two core composite contributors to the Environmental Sustainability Index (ESI), put together by the EPMP. The environmental systems quality composite aggregates measures in the areas of air quality, water quantity, water quality, biodiversity, and terrestrial degradation. The reducing environmental stresses composite aggregates measures in the areas of reducing air pollution, reducing water stress, reducing ecosystem stress, and reducing waste, consumption, and population pressures on the environment. The data come from EPMP (2002) for the year 2002. Table 7.4 includes six columns. The first three columns focus on the environmental systems quality composite. The next three columns concern the reducing environmental stress composite. In each case, we report results using the three measures of political regime type, level of democracy, and the two dichotomous indicators of democracy and autocracy. Limitations notwithstanding, the results in Table 7.4 are in the spirit of those reported in Tables 7.1, 7.2, and 7.3. Table 7.4 shows that the effects of the level of democracy on both the environmental systems quality composite

223

−0.0226 (0.0324) 1.0001 (4.5546) 0.0004 (0.0006) −3.23e-10 (1.93e-08) −0.0511∗∗∗ (0.0208) 53.9768∗∗∗ (2.8579) 108 0.28

0.1474 (0.2753)

−0.0212 (0.0324) 0.9833 (4.4603) 0.0003 (0.0006) 2.54e-09 (2.02e-08) −0.0518∗∗∗ (0.0205) 53.3760∗∗∗ (2.8392) 108 0.28

2.5670 (2.9863)

Environmental systems quality

−6.8081∗∗ (3.4276) −0.0286 (0.0325) 0.6373 (4.4805) 0.0004 (0.0005) −2.63e-09 (1.81e-08) −0.0507∗∗∗ (0.0203) 55.2909∗∗∗ (2.9870) 108 0.29

Environmental systems quality

−0.0304 (0.0265) 0.7810 (2.7258) −0.0000 (0.0004) −3.04e-08∗∗ (1.44e-08) −0.0260∗ (0.0174) 61.0057∗∗∗ (2.5402) 108 0.45

0.1530 (0.1518)

Reducing environmental stress

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Observations R2

Constant

POPULATION DENSITY

REAL GDPPC SQUARED

REAL GDPPC

WAR

TRADE OPENNESS

AUTOCRACY DUMMY

DEMOCRACY DUMMY

LEVEL OF DEMOCRACY

Environmental systems quality

Table 7.4. Effect of political regime type on environmental composites

−0.0287 (0.0259) 0.8037 (2.6394) −0.0002 (0.0004) −2.65e-08∗∗ (1.44e-08) −0.0269∗ (0.0171) 60.2561∗∗∗ (2.5593) 108 0.46

3.1022∗∗ (1.8368)

Reducing environmental stress

−2.4238 (2.6298) −0.0327 (0.0269) 0.4803 (2.8041) 0.0001 (0.0004) −3.29e-08∗∗∗ (1.39e-08) −0.0254∗ (0.0172) 61.5675∗∗∗ (2.6550) 108 0.45

Reducing environmental stress

224

Democracy and Economic Openness in an Interconnected System

and the reducing environmental stress composite are positive, as in Table 7.1, but they are weak or not statistically different from zero. A transition from nondemocracy to democracy raises environmental system quality, but the effect is insignificant. On the other hand, this transition raises the reducing environmental stress composite, and the effect is statistically significant. Hence, democratic transition is good for reducing the environmental stress composite. A transition from nonautocracy to autocracy reduces environmental systems quality, and the effect is statistically significant – implying, once again, that democracy is good for the environment. This transition also lowers the reducing environmental stress composite, but this particular effect is weak. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Our research in this chapter shows that scholars debate the nature of the effect of democracy on the environment. Existing evidence is mixed and relatively scant. Our research in this chapter contributes to the democracy– environment literature by empirically testing the net effect of democracy on environmental degradation. The empirical scope of our data analysis in terms of sample size and the number of indicators is generally wider than in previous studies. We employ a wide array of important types of human activities that degrade the environment: CO2 emissions, NOx emissions, organic pollution in water, forest depletion in terms of the rate of deforestation and the share of forested area over land area, land degradation, and two composite environmental measures. For democracy, we employ a continuous measure of the level of democracy or autocracy and two dichotomous measures of democracy and autocracy. The results presented in this chapter have several important implications for public policy. Democracy is generally good for the environment and, as such, policies that expand democracy could reduce human activities that lead to environmental degradation. But this conclusion should be qualified in terms of which regime transitions are likely to cause what type of improvement. A transition from nondemocracy to democracy (across the +6 threshold on the level-of-democracy scale) does not significantly influence CO2 emissions per capita, NOx emissions per capita, or organic pollutants in water, but it lowers land degradation and yet raises the deforestation rates. In contrast, a transition from an autocracy to a nonautocracy (across the −6 threshold on the level-of-democracy scale) reduces CO2 emissions per capita, NOx emissions per capita, organic pollutants in water, and, somewhat less so, land degradation, but it also raises the deforestation rate.

Democracy and the Environment

225

These findings suggest that moving upward along the level-of-democracy scale always leads to faster deforestation but not always to a significant environmental improvement in terms of the other four measures; the effect depends on where a country is along the level-of-democracy scale. In fact, one could expect greater improvements of the environment when a transition occurs across the −6 threshold than when a transition occurs across the +6 threshold. In other words, increasing the democracy score of a highly autocratic country above the −6 threshold delivers more bang for the buck than increasing the democracy score of a less autocratic country above the +6 threshold. The findings concerning the size of effect also have important implications. Whereas, in all cases, a rise in democracy produces a noticeable effect on environmental degradation, the effect of the level of democracy on environmental degradation varies across environmental dimensions. We could expect considerable effects of political regime type in terms of deforestation rate, forested area, NOx emissions per capita, and land degradation. Furthermore, timing is also of the essence here; one should not expect to see the effects of a rise in democracy kick in full-force very quickly. The immediate (annual) effects of a rise in democracy on organic pollutants in water and CO2 emissions per capita appear to be small, but the cumulative beneficial effects of this rise in democracy are much larger over time. That said, these two effects are still smaller than the effects of democracy on NOx , deforestation rate, forested area, and land degradation. Hence, a rise in democracy reduces some types of environmental degradation more than other types and results in faster deforestation rate. Our results in this chapter also suggest that democratization could indirectly promote environmental degradation through its effect on national income. This effect is subtle and works through the environmental Kuznets curve discussed in the appendix of this chapter. Across the five aspects of environmental degradation, we find evidence supporting the existence of an environmental Kuznets curve for CO2 emissions per capita, NOx emissions per capita, and the level of land degradation. Hence, when income per capita is low, a rise in income per capita causes more degradation; once past a certain threshold, a rise in income per capita reduces degradation. Although existing evidence on the effect of democracy on economic growth is inconclusive – to the extent that a rise in democracy promotes economic growth – the environmental Kuznets curve effects we find suggest that democracy could indirectly cause more environmental degradation for the aforementioned indicators at the initial stage of development but could help to reduce the degradation with further development.

226

Democracy and Economic Openness in an Interconnected System

Finally, the reader may recall the examples with which we opened this chapter, referring to the works and evaluations of former U.S. Vice President Al Gore and the popular writer Jared Diamond. By and large, the insight of Al Gore that democracy benefits the environment is generally more on the mark than that of Jared Diamond. The latter suggests in one place that autocracy has its advantages when it comes to environmental quality and in another place argues that the effect of political regime type on the environment cannot be generalized. In light of our research, a caveat is in place for the arguments of both figures. Our analysis reveals the danger of making general statements regarding complex transformation using stylized facts – which is not to suggest that the stylized observations invoked by Al Gore and Jared Diamond are incorrect. But when it comes to public policy formulation, the more important issue is whether the anecdotal evidence translates into generalizable patterns. Our point is that stylized facts do not justify general claims. Al Gore is correct in observing that democracy is good for the environment, but he fails to identify that a rise in democracy leads to faster deforestation and that the effects of democracy vary in size across environmental indicators. These failures should not be taken lightly. For one thing, continued faster deforestation in democracy implies that Al Gore’s expectation that a rise in democracy would resolve our environmental problems may turn out to be a false prediction in the long run. Faster deforestation may elicit a rise in carbon dioxide emissions (because diminishing forests absorb less emissions), a decline in water resources due to global warming, and intensified land degradation and higher water pollution due to a decline in water resources. Furthermore, failing to recognize the varying effects of democracy, policymakers will not realize that the beneficial effects of democracy on the environment will be small and slow in terms of reducing water pollution and CO2 emissions. At the same time, Jared Diamond’s insights for deforestation receive support in our large-N sample: democracy induces faster deforestation. But our statistical findings suggest that overall Jared Diamond is “more wrong” than Al Gore. Whereas Diamond seems to signal that autocracy may not be a bad thing for the environment, his conclusion based on a few cases and deforestation per se does not hold for all environmental indicators, across many countries and over many years. Diamond’s second argument, that the effect of democracy on the environment cannot be generalized and that both democracy and autocracy can end up damaging the environment, is also wrong. Our results for carbon dioxide, nitrate dioxide, land degradation, pollution in water, and the

Democracy and the Environment

227

aggregate environmental indicators suggest that a rise in democracy reduces environmental damage. Hence, if all the countries in the world turned into democracies tomorrow, the expectation is that, all else being equal, carbon dioxide emissions per capita, for example, should decline as a result. This example, which assumes ceteris paribus conditions, however, also demonstrates the danger of employing the type of stylized analysis of Al Gore and Jared Diamond. Both thinkers failed to include the channel from democracy to the environment that works through the environmental Kuznets curve. This approach is typical of relying on stylized observations. Stylized analysis can neither sort out nor estimate the effects of various causal determinants. The statistical approach not only estimates the ceteris paribus effects of a large number of causal factors but also provides an opportunity of forecasting the near future using the estimated coefficients. SUMMARY AND OUTLOOK

This chapter evaluated the controversial effects of democracy on the environment. We found that a higher level of democracy leads to fewer CO2 emissions per capita, fewer NOx emissions per capita, less organic pollution in water, and less land degradation. Compared with nondemocratic countries, democracies have larger forested areas, but, going against the spirit of the preceding findings, they deplete their forests relatively faster. We also find that the effect of democracy on the environment is discontinuous along the continuous scale of democracy. Relative to nondemocratic countries, democratic regimes do not appear to be different in terms of CO2 emissions, NOx emissions, and pollution in water, but they have lower rates of land degradation and yet higher deforestation rates. In contrast, relative to nonautocracies, the autocratic regime experiences more CO2 emissions, greater NOx emissions, higher pollution in water, somewhat less land degradation, and yet lower deforestation rates. Hence, whereas the effects of democracy on the deforestation rate and the forested area are monotonic along the democracy scale, the difference between autocracy and nonautocracy significantly influences CO2 emissions, NOx emissions, and organic pollution in water, and the difference between democracy and nondemocracy affects land degradation. So far in Part III of our book, we have focused on the effect of democracy on human activities that hurt the environment, treating economic openness as a control variable. In the next chapter, we shift the focus of analysis from the level of democracy of a country to its level of economic openness and investigate its effect on the environment. Economic openness, as

228

Democracy and Economic Openness in an Interconnected System

we saw in Chapter 3, is a multidimensional concept that involves various aspects, including trade, foreign direct investments, portfolio investments, and so on. One aspect of economic openness that has attracted the most attention is the effect of international trade flows on the environment. The effect of trade on the environment has been controversial and unsettled. The subject also has important policy and international implications since trade is the only aspect of economic globalization that has led to the most international collaboration and the most aggressive development of international institutions – the formation of the General Agreement on Trade and Tariffs, followed by its successor regime, the World Trade Organization (WTO). Whether or not the WTO should consider environmental issues in the global trade talks and in the adjudication of trade disputes is highly debated. It is against the background of this debate that we turn our attention to the effect of international trade flows on the environment in the next chapter.

APPENDIX EMPIRICAL MODEL AND ANALYSIS

Empirical Model The dependent variable environment is a multifaceted concept. We focus on greenhouse gas emissions, air pollution, water pollution, forest depletion, and land degradation. We analyze separately one representative, salient indicator for each area, with the exception of forest depletion, for which we use two measures – the rate of deforestation and the share of total land area that is forested. Although these indicators are not exhaustive, they collectively provide a comprehensive picture of environmental degradation. They also have relatively more comprehensive data coverage. We conduct six different empirical tests. Each test employs several measures of democracy to be discussed later. The first empirical test focuses on carbon dioxide (CO2 ) emissions because this gas constitutes about 60% of all greenhouse emissions, generated by energy-related activities and sources including industry, burning of solid fuels (e.g., coal), liquid fuels (e.g., petroleum), gaseous fuels (e.g., natural gas), gas flaring (burning of gas released in petroleum extraction), cement manufacturing, and bunker fuels (stored fuels). We use per capita CO2 emissions to adjust for country-size differences. Data are from the World Development Indicators (World Bank, 2002).

Democracy and the Environment

229

The second test analyzes nitrogen oxide (NOx ) because this gas is a major health hazard and an important source of smog in urban areas. The NOx emissions are generated primarily from fossil fuel combustion in motor vehicle engines, and also from other processes including biofuel combustion, oil and gas production, solvent use in industry and other sectors, and waste burning. We employ the logged NOx emissions per capita. Data come from the GEO Data Portal (UNEP, 2006) maintained by the United Nations Environment Programme. The third test examines a major form of water pollution. The level of organic pollution in water is measured in terms of the amount of oxygen in kilograms, which bacteria living in the water consumes when breaking down organic matter (typically denoted as biochemical oxygen demand, or BOD). This standard measure is comparable across countries. Our measure is the logged ratio of BOD to the amount of internal renewable water resources in a country. Data on organic pollution in water come from the World Development Indicators (World Bank, 2002), and data on internal renewable water resources in a country come from the World Resources Institute (2001). The fourth and fifth tests focus on forest depletion because forests are important drivers of ecosystem health. Moreover, deforestation has been identified as one of the primary agents of climate change (IPCC, 2007a, 2007b). Forest depletion is measured here in two complementary ways. One measure is the rate of deforestation, where a positive value of this variable indicates a rise in forested area over time (afforestation) and a negative value implies a fall in forested area over time (deforestation). Available data measure deforestation in terms of permanent manmade conversion of natural forests into other uses (e.g., mining, ranching, agriculture). Areas logged with the intent of regeneration and areas degraded by acid rain and forest fires are not included. Data on average annual deforestation rates per decade are collected from the World Resources Institute (1999) and the State of the World’s Forest Report (2001). A second measure of forest depletion is a country’s share of total land area that is forested. Although the deforestation rate allows us to evaluate the effect of democracy on the change in forested area, this change does not necessarily reflect the size of forested area in a country.2 Hence, we also assess the effect of democracy on the size of the forested area. To that end, we employ the logged percentage share of the forested area in total land area. Data come from the Food and Agriculture Organization (FAO, 2000, 2002, 2006). 2

For example, a country that has destroyed a lot of forest area before the sample period may have a low deforestation rate in the sample period but may have only a small forested area.

230

Democracy and Economic Openness in an Interconnected System

The sixth test looks at severe land degradation, which greatly harms agricultural activities. Data come from the FAO (2000, 2002, 2006). As discussed in FAO (2000), the only available uniform global source of human-induced land degradation data comes from this report. These data measure land degradation from water erosion (e.g., loss of topsoil, deformation, sedimentation), wind erosion (e.g., loss of topsoil, deformation, overblowing), chemical deterioration (e.g., loss of nutrients, salinization, acidification), and physical deterioration (e.g., compaction, crusting, lowered water table). Data are from the late 1980s and are reported as the share of degraded land out of total land at five levels: not degraded, light degradation (reduced agricultural suitability), moderate degradation (greatly reduced productivity), severe degradation (largely destroyed, unreclaimable at farm level), and very severe degradation (fully destroyed biotic functions, unreclaimable). Our indicator is the logged share of severely and very severely degraded land out of total land area. The key independent variable is political regime type. We employ several measures for this concept: one continuous, level of democracy, and the other two dichotomous, democracy dummy and autocracy dummy. The continuous or interval measure of regime type is the widely used composite indicator of the level of democracy, ranging from −10 to +10, from the POLITY data set that we used in the previous chapters. Despite the popularity of this measure of democracy, one may question whether the POLITY score needs to be treated as interval or dichotomous. One may also question whether the effect is constant across a range of values along the scale. For example, the effect when the democracy score rises from −10 to −5 may not be the same as the effect when the score rises from 0 to +5. In light of these tensions over the measurement of democracy, some scholars (e.g., Dixon, 1994; Fearon and Laitin, 2003) often employ dichotomous measures of democracy and autocracy that are coded based on a continuous indicator. In these studies, a country is often defined as a democracy if the continuous measure of democracy is greater than or equal to 6 and as an autocracy if its score is smaller than or equal to −6. The dichotomous variables democracy and autocracy are coded 1 if a country is democratic or autocratic and zero otherwise. The chosen threshold of 6 is, of course, to some extent arbitrary in the sense that 7 or 5 could also be used (but probably not −7 or −1, etc.). Nevertheless, the threshold of 6 has been used in many studies. To maintain compatibility with this norm, we use the threshold level of 6. The considerations guiding our choice of the threshold level of −6 are similar.

Democracy and the Environment

231

The control variables we include are those used in a number of empirical studies in environmental economics and environmental politics. The first two variables are real income per capita, real gdppc, and its squared term, real gdppc squared. A large literature argues that economic growth has competing effects on environmental quality. A larger economy generates more output and, therefore, more pollution and waste. Some types of technological progress, which are associated with growth, also may damage the environment (e.g., greenhouse gases). The effect is typically referred to as the scale effect. As income per capita rises above some threshold, the importance of environmental quality for people is said to rise, and they begin to employ cleaner production techniques and fewer natural resources, thereby increasing investment in environmental regulation. This behavior is referred to as the income effect. The combined operation of the scale and the income effect generates an inverted U-shaped figure when environmental degradation is plotted against income per capita. The inverted U shape is known in environmental economics and environmental politics as the environmental Kuznets curve (EKC).3 Its empirical existence, however, is debated, an issue to which we later return. Relevant to our analysis, one may also frame the Kuznets curve debate in terms of whether it is an economic effect (environmental quality as a luxury good that is affordable at higher per capita incomes) or a political effect (the emerging middle class as a byproduct of industrialization asserts itself politically on issues of air and water quality). Since the economic effect would occur independent of the level of democracy but the political effect arguably would not, it is important to include in the Kuznets curve analyses both democracy and income per capita on the right-hand side of the statistical model, as we do. Although the empirical existence of the Kuznets curve is debated, most studies allow for it in statistical models by including income per capita and its squared term. If the EKC exists, the coefficient of income per capita should be positive, the coefficient of income per capita squared should be negative, and the coefficient of income per capita should be larger than the absolute value of the coefficient of income per capita squared. GDP per capita, in purchasing power parity–adjusted, constant 1996 international dollars, is from the Penn World Table 6.1 (Heston et al., 2002).4 3

4

The name EKC is used in the literature, because the original Kuznets curve hypothesizes the existence of an inverted U curve for income inequality as a function of income per capita. For review of these arguments, see Dinda (2004) and Panayotou (2000b). An alternative indicator of technology and knowledge is GDP per worker. But GDP per capita and GDP per worker are correlated very highly (correlation coefficient = 0.98).

232

Democracy and Economic Openness in an Interconnected System

The third control variable is trade openness, which is a country’s total trade over its GDP. The trade and environment literature argues that trade can affect the environment in two broadly defined ways. In one way, the pattern of domestic production and consumption, and the methods of production, change under trade openness as countries follow their comparative advantages and/or adopt certain more efficient, cleaner (or not) technologies to produce for other countries. For example, a country that trades environmentally clean goods would see its environmental quality rise, and vice versa. This channel also involves changes to environmental regulation, as some international trade treaties may require regulatory changes at home. In addition, trade may affect the environment by promoting economic growth and, hence, altering people’s behaviors over time. Although the combined empirical effect of trade on the environment is debated, it needs to be included in the model.5 We employ a popular measure of the importance of trade openness to a national economy, which is the sum of national exports and imports divided by GDP. Data are from the Penn World Table 6.1. The fourth control variable is population density (population divided by land area).6 The effect of population density on environmental degradation may change across indicators. A rise in population density is expected to generate more CO2 emissions as a larger population consumes and produces more. But it may generate fewer NOx emissions as denser areas tend to use more public transportation and fewer cars (the primary generator of this type of emissions). Water pollution is expected to rise with population density. As more people engage in consumption and production, organic water pollution should rise. Many densely populated nations tend to be more urbanized and depend less on the environment for livelihood (e.g., consider Western European countries such as the Netherlands or Belgium). As such, they may clear fewer forests. Greater population implies more pressure to use agricultural land for food and industry and, therefore, more land degradation. Data are from the World Development Indicators (World Bank, 2002). Finally, as discussed by Reuveny (2002), for example, military conflict also can affect the environment. Military conflict, however, may generate competing effects on environmental degradation. The variable war is coded 1 if a country is involved in an interstate or intrastate war in a given period, and 0 otherwise. According to Gleditsch et al. (2002), from which we obtain the conflict data, a militarized interstate or intrastate conflict is defined as a war if it involves at least 1,000 battle-related deaths per year. 5 6

For a review of these arguments, see, e.g., OECD (1994b), Pugel (2003), and Harris (2006). For examples of studies employing this indicator, see Panayotou (2000b).

Democracy and the Environment

233

Research Design Issues First, the choice between cross-sectional analyses or cross-sectional timeseries (pooled) analyses is dictated by data availability. As noted in the main text, for the CO2 emissions and water pollution, we have time-series cross-sectional data. For NOx , deforestation rate, forested land area, and land degradation, we have cross-sectional data. Thus, we conduct pooled analyses for the first two indicators and cross-sectional analyses for the last six indicators. Given the various data structures in these indicators, we ensure that the time periods of the right-hand-side variables match the time period of the dependent variable observations. The sample design for other indicators are clearly discussed in the main text, but the sample design for the forest depletion variables requires additional clarification. For the rate of deforestation, the sample covers 134 countries. For each country, the data include two average annual deforestation rates, one rate for the 1980s and the other for the 1990s. The annual rates during a decade are computed based on the total forest areas for 1980, 1990, and 2000. The right-hand-side variables take on their decade-average values to represent their values during the decade, capturing the cross-sectional patterns. For the share of total land area that is forested in a country, we have three data points for each country (1980, 1990, 2000). Our data source does not specify when exactly a measurement was taken for a country; we assume that these possible measurement problems are absorbed into the error term. The second design issue concerns the possibility of endogeneity for our right-hand-side variables. Our empirical framework treats level of democracy, democracy dummy, autocracy dummy, real gdppc, real gdppc squared, population density, war, and trade openness as exogenous variables. One may argue that environmental degradation can affect these variables. For example, land degradation may reduce the exports from an agrarian economy. Or environmental degradation may lead to conflict, a hotly debated issue in the literature. To mitigate the potential risk of simultaneity, the right-hand-side variables are lagged one year, as is done in many studies (e.g., Oneal and Russett, 1999a, 1999b, 1999c; Li and Reuveny, 2003). This popular practice is sufficient for addressing the possibility of endogeneity under the weak exogeneity assumption (Wooldridge, 2002). Third, national structural variables (e.g., climate, education) and global or local biophysical attributes (e.g., atmospheric integrity, existing damage) may also affect environmental degradation. Because these factors tend to change slowly, environmental degradation is likely to exhibit inertia, which is

234

Democracy and Economic Openness in an Interconnected System

modeled empirically through the inclusion of the lagged dependent variable on the right-hand side for the pooled analysis. We further guard against the potential risk of missing structural variables in our pooled analyses by using the two-way fixed-effects estimator. This estimator includes separate intercepts for each country and each year. In addition, the lagged dependent variable and the fixed-effects country and year dummies help to control for the fact that different types of environmental degradation may have different causal determinants. Unfortunately, due to data availability, we could only do so for CO2 emissions and organic water pollution, not for NOx emissions, deforestation, forested area, and land degradation. Including the lagged dependent variable, lagged environment, and using the two-way fixed-effects estimator, however, do not come without cost. It is well known that they soak up the variations in the dependent variable that could otherwise be explained by other right-hand-side variables, which should make it harder for us to find statistically significant results. Hence, this approach to empirical modeling can be considered conservative. As a caveat, this approach suggests that statistical results and inferences for CO2 emissions and organic water pollution are likely to be more reliable than results for the other indicators with fewer data. Future research with more data for the other indicators is certainly in order. Fourth, we consider the risks of heteroskedasticity and serial correlation. When error terms are not spherical, the estimated regression coefficients are consistent, but their standard errors are inefficient and biased. To deal with the risk of heteroskedasticity, we estimate Huber–White robust standard errors (White, 1980). Both the year dummies and, as Beck and Katz (1995a, 1995b) suggest, the lagged dependent variable capture the temporal dynamics in the pooled data, controlling for possible serial correlation. Fifth, we need to consider the issue of multicollinearity. This potential problem is a cause of concern when the effects of key variables are statistically insignificant in models of good fit (high R2 ). In this case, statistical insignificance may be an artifact of multicollinearity that increases the standard errors of the coefficient estimates. We assess the extent of this possible problem using the variance inflation factor (VIF) diagnostic. Sixth is the issue of which statistical significance level to use when one interprets the results. In evaluating the effect of democracy on the environment, we have discussed two types of theories. One type expects that democracy will promote environmental quality, whereas the other type expects the opposite. As noted, it is not possible to reject either set of theories based a priori on theoretical grounds. Given their assumptions, both

Democracy and the Environment

235

types of theories appear correct concerning the expected sign of the net effect. In each model, we therefore test the sign of this net effect of democracy on the environment against the null hypothesis of no effect – or rather the two competing effects are equal in size – by employing a one-tailed test in reporting the results. Many other studies have used this approach (see, e.g., Oneal and Russett, 1999a; Li and Reuveny, 2003; Reuveny and Li, 2003). In the interpretation of the results, we employ the 10% significance level in addition to the 1% and 5% levels. We take this approach because some of our samples are relatively small due to limits on data availability. Our approach is justifiable also for our pooled models since we include fixed effects and the lagged depended variables, which soak up the variations in the dependent variable. Finally, we need to measure the size of democracy’s effect on the environment. We discuss the size of the estimated coefficient of democracy in our models, provided that the coefficient is statistically significant. If the coefficient is statistically insignificant, the size of effect is so small that it is statistically zero. For the size of change in the independent variable (in our case democracy), we follow other studies by using its one-standarddeviation change in the sample. We proceed by first computing a base value for the dependent variable, holding all the continuous right-hand-side variables at their sample mean and setting the dichotomous war variable at zero, and then computing a new value for the dependent variable when the continuous democracy variable is increased by one standard deviation in the sample. The two values are then compared to each other, and the difference is expressed in terms of percent change. However, this practice does not tell the full story when the lagged dependent variable is included on the right-hand side in the pooled analysis because it only captures the immediate impact of democracy; it does not take into account the fact that the impact of democracy on the environment from previous periods is absorbed into the effect of the lagged environmental degradation variable, which is also on the right-hand side. Democracy affects the current environment via its direct effect and continues to affect the environment in the next period via the lagged environmental degradation. These effects accumulate over time. The long-run impact of a change in democracy produces the following change in environmental quality: [coefficient of democracy/(1 − coefficient of lagged environmental degradation)] × (change in democracy). We compute the long-run effect in the pooled analyses of CO2 and water pollution, because the approach is not applicable to cross-sectional data.

236

Democracy and Economic Openness in an Interconnected System

Empirical Findings Having discussed in the main text the effect of democracy in Table 7.1, we now focus on the control variables. The variable trade openness has no effect on CO2 emissions per capita, on organic water pollution, or on forested area. But the effect of trade openness on NOx emissions per capita is positive and statistically significant. The effect of this variable is negative and statistically significant on land degradation, but positive and significant on the deforestation rate. Trade is not necessarily a boon for the environment, but it appears to reduce the deforestation rate. The effect of war on CO2 emissions per capita is statistically significant and negative, reflecting the net effect of competing forces. CO2 emissions may rise during wars from greater production and operation of weaponry. But they may also fall during wars due to the reduced normal economic activity that generates emissions as the labor force is drafted and/or parts of the economy are destroyed. This logic applies to all our environmental indicators. For CO2 , the emission-reducing effect of war is larger than the increasing effect. The net effects of war on NOx emissions per capita and organic pollution in water are statistically insignificant. In these cases, the competing effects of war are of about the same size. In contrast, war significantly reduces the rate of deforestation and raises land degradation. Compared with countries at peace, countries involved in war in our crossnational analysis for forest depletion have smaller shares of forested area. The effects of the lagged CO2 emissions per capita and the lagged organic pollution in water on the contemporary values of these indicators, respectively, are both positive and statistically significant. Hence, these indicators exhibit inertia. As noted, population density may also exhibit competing effects on the environment. We find that the net effects of population density on CO2 per capita, organic pollution in water, and land degradation are statistically significant and positive. In these cases, a larger population density leads to more environmental damage. In contrast, the statistically significant effect of population density on NOx emissions per capita is negative. As for forest depletion, more densely populated countries exhibit significantly higher deforestation rates, but the effect of population density on the share of forested area is not statistically significant in our sample. Our findings for the signs and significance levels of real gdppc and real gdppc squared suggest that CO2 emissions per capita, NOx emissions per capita, and land degradation exhibit an EKC (for land degradation, income is logged). Stated in real 1996 international dollars, the turning points are

Democracy and the Environment

237

$16,071 for CO2 ; $25,951 for NOx ; and $4,012 for land degradation. Organic water pollution and the share of forested land do not exhibit a Kuznets curve. For the rate of deforestation, the coefficient of GDP per capita is positive and significant, and the coefficient of GDP per capita squared is negative and significant. Hence, deforestation declines as GDP per capita rises and then rises as GDP per capita rises above some level, which indicates a nonlinear effect that does not follow the shape of the EKC. Our results for the environmental Kuznets effect are within the range of results reported in the extensive literature on this topic. The location of the turning point and whether the Kuznets curve exists are empirical issues. Results vary among studies and depend on model specification, data, indicators, and estimators. For example, Cole et al. (1997) found turning points of $25,100 for CO2 and $15,100 for NOx ; Moomaw and Unruh (1997) found $18,333 for CO2 ; and de Bruyn et al. (1998) did not find Kuznets curves for CO2 or NOx . Many studies did not find a Kuznets curve for water pollution (Panayotou, 2000b). For forests, Shafik and Bandyopadhyay (1992) and Barbier (2001) did not find a Kuznets curve. We did not find any studies of the Kuznets curve for land degradation. Investigating the sources for differences in the Kuznets curve literature is a very large task and is beyond the scope of this chapter. It is apparent that our results support both economic and political effects discussed earlier. However, the income variables largely capture the economic effect, because the political effect (via the middle class) is controlled for by the democracy variable. To the extent that Seymour Lipset is correct about the modernization thesis of democracy, part of democracy’s effect on the Kuznets curve also traces to economic development, which is present in the model. Finally, we should discuss whether multicollinearity affects our results in Table 7.1. Multicollinearity becomes a concern when the diagnostic VIF statistic exceeds the threshold of 10. For the CO2 model in Table 7.1, the average VIF is 5.7. Hence, multicollinearity is not a concern. For the NOx model in Table 7.1, the average VIF is 8.7, suggesting that multicollinearity may be a concern. Using the matrix of variance decomposition, we find that the multicollinearity is caused by the high correlation between real income per capita and its squared term. The VIFs for other variables are smaller than 2.5, indicating that multicollinearity for them is not a concern. For the model involving organic pollution in water, the average VIF is 4.8, suggesting that multicollinearity is not a concern. For the deforestation model, the average VIF is 4.9, also suggesting that multicollinearity is not a concern. For the land degradation model, the average VIF is 95; hence, multicollinearity is

238

Democracy and Economic Openness in an Interconnected System

a concern. However, using the matrix of variance decomposition, we find that the source of the collinearity is the correlation between real income per capita and its squared term. The VIFs for other variables are all smaller than 2, indicating that multicollinearity is not a concern.

Additional Analyses Tables 7.2 and 7.3 present the results for the dichotomous measures of democracy and autocracy, respectively; these results are fully discussed in the text. The additional analysis section in the text also discusses the Environmental Performance Measurement Project’s (EPMP’s) (2002) composite environmental measures. The project aggregates environmental indicators by computing the average of their z-scores in their respective distribution. The z-score transformation subtracts the variable’s sample mean from its value in the sample and divides the result by its standard deviation. For details, see Appendix A in http://www.yale.edu/esi/. As noted in the main text, we employ two core composite contributors to the ESI, put together by the EPMP: the Environmental Systems Quality composite and the Reducing Environmental Stresses composite. The ESI aggregates five core composites, of which we use two.7 In addition, EPMP has released data for 2005. We do not use these data because we do not have data on other variables for 2005. The results are reported in Table 7.4. Among the control variables, only the effect of population density is consistently significant across the six columns; its sign is negative. This result is not surprising given that the control variables often have different effects on the various dimensions of environmental degradation, as shown in Tables 7.1–7.3 and in the literature, which should caution us against being overly confident about the results based on any composite environmental indicator; the environment, unlike the economy, cannot be easily aggregated because its various components lack a single metric. 7

The other three core contributors involve economic, social, and political indicators in areas such as science and technology, capacity for public debate, private sector responsiveness to environmental issues, and governance. Given our focus on human actions that harm the environment, we do not employ these core areas or the ESI.

EIGHT

Economic Openness and the Environment

INTRODUCTION

In Chapter 7, we studied the effect of democracy in a country on various types of environmental degradation, controlling for international trade merely as a secondary force. This chapter studies how international trade and its interaction with democratic governance influence the terrestrial environment. Specifically we examine land degradation and deforestation, two indicators that reflect the health of the terrestrial environment of a country. These two dimensions of the terrestrial environment have an important impact on food production and climate change – two issues that have posed serious challenges to the contemporary world. Land degradation significantly constrains a country’s ability to provide food for its people and animals, including animals to be consumed. A rise in land degradation often reduces the land’s productivity and, therefore, food yields. Because of growing populations and land degradation, many less developed countries (LDCs) find that their people have to confront declining food security. The problem is intensifying as rising oil prices have induced food crises in many poor countries. Deforestation has important environmental implications, including global warming: it speeds up land erosion, causes fresh water supplies to decline, reduces the level of biodiversity, raises the levels of sedimentation in rivers and lakes, intensifies sand and dust storms, increases the occurrence of mudslides, and deteriorates air quality. In fact, the Intergovernmental Panel on Climate Change has listed deforestation as a key driver of climate change. Shrinking forests fail to absorb carbon dioxide emissions, the accumulation and trapping of which is recognized as the most significant cause of global warming.

239

240

Democracy and Economic Openness in an Interconnected System

Many commodities that are traded internationally depend heavily on the endowment of land and forests in countries. Consequently, international trade produces important effects on the terrestrial environment. But the effect of international trade on the environment is controversial. In one view, trade reduces environmental degradation and promotes environmental quality; in another view, trade harms the environment, increasing environmental degradation. The controversy in and of itself suggests that further analysis is warranted and also justifies our focus on one main aspect of globalization to keep our analysis tractable. Meanwhile, the democracy– environment relationship is also debated, as we have demonstrated in the preceding chapter. In general, the democracy–environment debate and the trade–environment controversy have produced separate literatures that remain largely on separate courses. We argue that the effects of trade and democracy on environmental degradation need to be studied together. Whereas both trade and democracy have controversial effects on the environment, the literature also argues that international trade flows influence democracy, as we will see later. Empirical studies that focus on the effect of only trade or democracy on environmental degradation, excluding the other, produce incorrect statistical inferences by attributing the effect of one factor to the other. Furthermore, international trade and democratic governance may well interact with each other to amplify or mitigate the other’s impact on the environment – a possibility that has not been contemplated in the literature or in our previous chapter. In this chapter, we are interested in the interactions among trade, democracy, geography, and the environment. We discuss the various mechanisms through which trade and democracy influence the environment and how trade and democracy relate to each other. In our comprehensive empirical analysis, we address various possibilities: trade and democracy may interact with one another, and their effects may vary across development levels, between democracies and autocracies, for countries with low and high levels of trade, and across geographical regions. A country in Africa, for example, may be exposed to conditions different from those in the United States and may exhibit different rates of development and urbanization. Our key empirical findings suggest that a rise in trade openness reduces the rate of deforestation in autocracy but increases the rate of deforestation in democracy. These effects are similar between LDCs and developed countries (DCs). A rise in trade openness reduces land degradation, but the effect is not robust and does not depend on regime type. A rise in democracy increases deforestation and reduces land degradation, but these effects are

Economic Openness and the Environment

241

weaker in LDCs than in DCs. In addition, open-economy democracies experience faster deforestation than closed-economy democracies. The effect of democracy on land degradation does not depend on trade openness. THEORETICAL AND EMPIRICAL LITERATURES

This section summarizes the arguments on the effect of democracy on the environment presented in the preceding chapter, discusses the theories of the effect of trade on environmental degradation, highlights the link between trade and democracy, and goes deeper into the environmental Kuznets curve mentioned in the Chapter 7 appendix. Although our primary interest lies with trade and its interaction with democracy, the various topics we survey will assist us in putting together the statistical model. The relevant empirical literature pertaining to our primary interest is also summarized.

Theories on Trade and Environment International trade influences environmental degradation both directly and indirectly. The direct causal mechanisms involve microlevel economic activities that affect the environment, regardless of whether the economy is growing or not. The indirect causal mechanisms concern the positive effect of international trade on the economic growth of a country, which in turn affects the environment.1 The positive effect of international trade on economic growth is established in many studies and will be taken here as a given.2 International trade can directly increase or decrease the level of environmental degradation in a country. These direct effects occur as international trade changes the profitability of producing some products at home and shapes the patterns of domestic and foreign consumption and the allocation of resources across borders. The structural effect of trade on the environment involves changes in the patterns of domestic production, consumption, investment due to trade, and in the location of agricultural activities. The structural effect of trade can, for example, reduce the production of chemical-intensive crops, which promotes environmental quality. However, this effect also could promote 1

2

For reviews of the literature on the effect of trade on the environment, see, e.g., the discussions in OECD (1994b), Panayotou (2000b), Pugel (2003), and Harris (2006). Our discussion builds on and synthesizes these sources. For recent reviews of the body of literature on international trade and economic growth, see Pugel (2003) and Salvatore (2004).

242

Democracy and Economic Openness in an Interconnected System

activities that hurt the environment, such as increased drainage of wetlands to satisfy the demands for greater production for a larger international market. The composition effect of trade can damage the environment by changing the composition of the goods produced at home as countries follow their comparative advantages when they open their economies for international trade. For example, consider a country that produces two goods: a timberintensive good by harvesting forests and a labor-intensive good that does not affect the environment. According to the neoclassical trade theory, if the country is highly forested and abundant in timber, it will have comparative advantage in the timber-intensive good. Therefore, international trade in our example increases the production of the timber good and promotes deforestation, which degrades the environment. If the country is laborabundant, on the other hand, a rise in international trade increases the production of the labor-intensive good and reduces the production of the timber good, which slows down deforestation. The technology effect of trade can reduce environmental degradation by shifting domestic production to methods that are less environmentally damaging due to the requirements brought about by international trade. For example, a rise in the foreign demand for goods produced with farming methods that use fewer fertilizers may benefit the environment. Trade may also create more contacts and increase the diffusion of environmentally cleaner, new technologies and innovations. However, trade could also spread the use of “dirtier technologies,” whose use damages the environment. Such technologies could be cheaper to employ and make exporters more competitive in international markets while remaining legal within the bounds of existing domestic environmental laws. The result would be a fall in environmental quality. The regulatory effect of trade is expected to encourage the proenvironment policies. Some trade agreements require countries to limit environmental damage. Trade openness also may lead to more stringent environmental regulation across the board if an influential country that sets the pace for others is pro-environmental. The approach of this country could influence other countries that are interested in selling their products in the former’s market. Yet this effect also may work in the opposite direction: the influential country may not be pro-environmental in some or all areas. In addition, parochial trade interests may push for a relaxation of existing regulation to shift to cheaper production methods that may damage the environment but strengthen international competitiveness. Because

Economic Openness and the Environment

243

other countries may follow this course of action, the result could be a race to the bottom, increasing environmental damages. The existing multilateral trade accords are virtually all sector-specific. The Convention on International Trade on Endangered Species (CITES), for example, has regulated trade in endangered species since 1973 (e.g., Frankel, 2003). The Montreal Protocol on Substances that Deplete the Ozone has regulated and phased out the production and trade of most chlorofluorocarbons (e.g., Brack, 1996; Parson, 2002). Wider setups are either bilateral or regional. Regional and bilateral trade agreements (e.g., the European Union, Australia–New Zealand Closer Economic Relationship, North American Free Trade Agreement) have required members to harmonize environmental standards and regulate environmental damage due to traderelated activities (e.g., Hufbauer et al., 2000; Crutsinger, 2001; Frankel, 2003). Environmental activists have tried to bring their concerns into the World Trade Organization, but thus far have failed. The United Nations Environmental Programme, which is currently the only truly across-theboard multilateral environmental organization, thus far has not focused on trade (e.g., Frankel, 2003; Harris, 2006).

Arguments on Democracy and Environment In Chapter 7 we demonstrated the debate over the effect of democracy on the environment; here we summarize the arguments. Beginning with the view that democracy may not benefit the environment, the market argument postulates that various businesses such as corporations, investors, banks, and their subcontractors normally seek to maximize profits, not environmental quality. Because these business interests are influential, democratic governments may not necessarily pursue environmental quality. If environmental concerns jive with business concerns, the environment will be preserved; otherwise business considerations may trump environmental concerns. The policy inaction argument considers the inner workings of democratic governments, which confront checks and balances and respect the power of public opinion of competing voices. Democratic governments may exhibit policy inaction when it comes to the environment and its preservation, because deliberations on environmental issues often are not able to converge on an outcome acceptable to all parties involved. The global-commons thesis observes that democracy is a national phenomenon, whose effects stop at the border. As such, democracy may not alleviate global environmental problems, particularly when natural resources

244

Democracy and Economic Openness in an Interconnected System

do not have well-developed property rights or can be freely accessed by a collective. Actors left on their own may seek to maximize their interest, ignoring the damage their actions may inflict on the environment. Because all actors may behave in this manner, environmental degradation will rise. The population growth channel argues that population growth in a country tends to increase environmental degradation, all other things being equal. Because democratic governments respect free choice, including the freedom of humans to procreate, the spread of democracy may enable a higher global population growth rate, indirectly promoting environmental degradation. Now, the view that democracy benefits the environment also invokes multiple arguments. The responsiveness argument is that democracies are generally more responsive to the needs of their citizenries than are autocracies because of electoral accountability and political competition. These needs also include environmental preferences. If the public comes to value the environment, democratic governments are more likely than autocratic ones to translate the public preference into pro-environment policies. The rule of law argument observes that democratic regimes respect the rule of law, whereas autocracies often cater to the needs to their supporting elite, regardless of the law, and are subject to the whims of the autocrat, clientelism, and extreme rent-seeking. Thus, democratic governments are more likely to comply with international and domestic environmental agreements, regulations, and laws than autocratic leaders. The freedom-of-information channel suggests that greater freedom for information flow characterizes democracies and their market economies compared to autocracies (with or without market economies). This characteristic facilitates the pursuit of environmental groups, raises public awareness of environmental problems and possible solutions, and encourages environmental legislation that curtails degradation. The famines and human life argument suggests that famines lead to environmental degradation by diverting attention away from environmental concerns and pushing people to exploit natural resources excessively for survival. Environmental degradation itself can put lives at risk. Because democracies respect human life more than do autocracies, they would move more swiftly to alleviate famines and life-threatening environmental conditions. As a result, their environmental qualities should be higher than that in autocracies. The war channel reasons that countries with democratic regimes may engage in fewer international wars and experience fewer civil wars than countries with autocratic regimes. Meanwhile, interstate and intrastate wars

Economic Openness and the Environment

245

often damage the environment. To the extent that both tendencies hold, democracies should have better environmental quality than autocracies.

Environmental Kuznets Curve Effect In the appendix of Chapter 7, we discussed the basics of the environmental Kuznets curve (EKC) theory. The implication of this argument is that international trade flows may affect the environment through their effect on the rate of economic growth in a country. The argument that trade is the engine of growth can be traced to British economists Adam Smith and David Ricardo. Since then this argument has become one of the hallmarks of economics. In this argument, nations specialize in or focus on producing the goods of their comparative advantage, ending up with a surplus of these goods. They then export some of this production to other countries and import in return the goods that other countries can produce relatively more efficiently. Since, in this arrangement, each partner to the trade uses its factors of production most efficiently, all partners end up with more output and higher efficiency. Trade then, as noted, is the engine of economic growth, but what is the effect of economic growth on the environment? The economic growth effect of trade on the environment falls under EKC theory. International trade promotes economic growth, which in turn affects the level of environmental degradation in a country. The theory argues that up to some threshold of income per capita, environmental damages rise as the economy grows in scale. As income per capita grows above this level, environmental damages decline. The plot of environmental degradation as a function of income per capita is the EKC, which takes the shape of an inverted U. “EKC” is given by analogy with the original Kuznets curve, which plots income inequality as a function of income per capita and also takes the shape of an inverted U.3 What drives the inverted U shape of the EKC? EKC theory attributes the inverted U shape to two economic forces. They are known as the scale effect and the income effect, which affect the environment differently. Beginning with the scale effect, as the economy grows, income per capita grows; but as the economy produces and consumes more goods and services, it also requires more natural resources of various types. With the current type of energy sources, technologies and methods of production, and means 3

For recent reviews and discussions of the principles of the EKC see, e.g., Panayotou (2000a) and Dinda (2004).

246

Democracy and Economic Openness in an Interconnected System

of transportation, higher levels of production and consumption simply generate more pollution, waste, and other forms of environmental degradation. The impact of the scale effect on the level of environmental degradation in a country is positive (i.e., degradation grows with income per capita). However, this is not the only postulated effect of economic growth. As income per capita rises, human preferences are expected to shift toward consuming goods and products that generate less environmental damage in consumption, as well as goods whose production techniques generate less environmental damage. This gives rise to the income effect of economic growth. Richer people (with higher income per capita) also are argubly more willing to pay for cleaner goods produced with cleaner technologies and that generate less pollution. The same people are also more able to and arguably more willing to pay for and cover the consequences of more environmental protection. Therefore, the sign of the income effect of economic growth is expected to be negative. Putting the income and scale effects to work together, the EKC argument hypothesizes that as income per capita rises above some threshold – the level of which is unknown based on theoretical reasoning, but can be estimated empirically – the income effect of economic growth comes to dominate the scale effect of growth. The combined operation of the two forces generate the inverted U shape of the EKC. Because trade – both domestic and international exchanges of goods and services – is said to be the engine of economic growth, it follows that we should be able to observe an EKC in the presence of international trade flows.

Theoretical Link between Trade and Democracy This chapter investigates the effect of trade and democracy on environmental degradation, but international trade flows and the level of democracy are theoretically related. Political scientists acknowledge that international trade flows are determined by economic forces, but they argue that political factors also can be important determinants of trade, particularly in determining who trades with whom, in what, and how much. Scholars have long been aware that international trade can be used by one country as a tool of foreign policy and statecraft to influence another (see, e.g., Hirschman, 1980; Hufbauer, 2005). Stating this argument succinctly, Diaz-Alejandro (1975) writes: “which markets are allowed to operate and how, which are encouraged and which are repressed – these are political decisions, both nationally and internationally” (p. 214).

Economic Openness and the Environment

247

If international trade flows can be used to exert power at the service of meeting some foreign policy goal, who will trade with whom? Grieco (1988), Gowa (1994), and Gowa and Mansfield (1993), among others, expect states to avoid trade with those they consider their actual or potential adversaries. In such situations, the concern for relative gains – who gains more from trade – may reduce trade flows to trickles, as the side that gains less worries that the side that gains more may translate its larger economic benefits into greater military power. During the Cold War, for example, the United States regulated its international trade relations with the Soviet bloc based on this logic, and today the argument suggests that the United States should regulate its international trade with China based on similar principles. Indeed, bilateral cooperation and conflict have significantly influenced the level of bilateral trade flows (Pollins, 1989; Li and Sacko, 2002). Today, because democracies are less likely to engage in wars against each other, as we discussed and analyzed extensively in Chapter 5, democracies should feel more secure in their bilateral trade regardless of who gains more. Consequently, they should trade more with each other (e.g., Morrow et al., 1998; Russett and Oneal, 2001). Moreover, because democracy tends to respect the rule of law and private property, businesses should face less risk and higher economic returns when they trade and invest in democracies (e.g., Olson, 1993; Clauge et al., 1996; Li and Resnick, 2003; Souva et al., 2008). At the same time, scholars have argued that international trade openness itself can affect the level of democracy in a country. In fact, Chapter 3 focuses explicitly on this relationship. The theoretical picture we painted in Chapter 3 is one of a controversy, where scholars debate the effect of economic openness on democracy. On one side, some observers and writers expect that a rise in international trade should raise the level of democracy in a country because, for example, trade promotes economic growth and technological diffusion. The expanding middle class that benefits from trade is more educated. Its members push for democracy, which serves them best with its transparency and preference for the rule of law and stable property rights. On the other side, scholars expect that a rise in international trade should have a negative effect on democracy by creating winners and losers in a country, intensifying income inequality, weakening the state’s ability to govern autonomously when facing external economic forces, and pushing the governments to serve particular economic interests instead of the public interest.4 4

See the sources cited in Chapter 3, including Bhagwati (1994), Im (1996), Schmitter (1996), Martin and Schuman (1997), Diamond (1999), and Held et al. (1999).

248

Democracy and Economic Openness in an Interconnected System

Summary of Theoretical Forces Affecting the Environment The causal mechanisms through which trade and democracy influence the environment are many and complex. Tables 8.1A–C list these causal mechanisms and note the sign of each expected effect on the environment. Table 8.1A focuses on the effects of international trade, Table 8.1B emphasizes the effect of democracy, and Table 8.1C clarifies the effect of economic growth on the environment (i.e., the EKC effect). For each mechanism, a negative sign implies less environmental quality or higher environmental degradation, whereas a positive sign indicates increasing environmental quality or reduced environmental degradation.

Empirical Literatures Thus far we have discussed the arguments behind four distinct but related relationships: trade and environment, democracy and environment, growth and environment, and trade and democracy. These arguments have received various degrees of empirical attention in the literature. Here we discuss the empirical studies related to three of these four topics and refer the reader to the preceding chapter for the empirical literature on democracy and the environment. Beginning with trade and the environment, the number of studies that statistically investigate the effect of trade on environmental degradation in a large sample is relatively small. On the one hand, some evidence exists that trade may have some small, beneficial effect on the environment. Lucas et al. (1992) studied the toxic intensity of output produced in 80 countries from 1960–1988. They concluded that the growth rate of this toxic intensity was lower for economies open for trade that grow rapidly, relative to comparatively closed economies. Grossman and Krueger (1993) employed city data from 1977, 1982, and 1988 for emissions of sulfur dioxide (SO2 ; for 42 countries), fine smoke (for 19 countries), and suspended particulate matter (SPM; for 29 countries). They found that trade openness (measured as ratio of export plus import to gross domestic product [GDP]) reduced SO2 emissions. The effect of trade openness on smoke and SPM is not significant. Suri and Chapman (1998) studied the effect of the ratio of import over GDP on energy consumption per capita (used as a proxy for air pollution) for 33 industrialized countries in the period 1971–1990. They reported that the effect is negative. Antweiler et al. (2001) employed city data of SO2 for 40 nations in the 1970s and 1980s. They found that trade liberalization reduced SO2 , but the effect was small. De Soysa and Neumayer

Economic Openness and the Environment

249

Table 8.1A. Summary of causal mechanisms: Trade and the environment

Channel

Expected effect on environmental quality

Structural Composition Technology Regulatory

Positive or negative Positive or negative Positive or negative Positive or negative

Table 8.1B. Summary of causal mechanisms: Democracy and the environment

Channel Market Policy inaction Global commons Population growth Responsiveness Freedom of information Rule of law Famines Human life War

Expected effect on environmental quality Negative Negative Negative Negative Positive Positive Positive Positive Positive Positive

Table 8.1C. Summary of causal mechanisms: Environmental Kuznets curve

Channel

Expected effect on environmental quality

Scale Income

Negative Positive

(2005) showed that during the 1980–1999 period, trade, foreign direct investment, and economic freedom increased sustainable development as measured by the genuine savings rate at which investment in the total stock of manufactured, human, and natural capital exceeds its depreciation.

250

Democracy and Economic Openness in an Interconnected System

On the other hand, some evidence exists that trade may harm the environment. Barbier (2001) studied the effect of trade on agricultural land expansion for tropical countries in 1961–1994. He found that agricultural export promoted agricultural land expansion, concluding that trade intensifies pressure on the environment. Dean (2002) employed data on Chinese water pollution in 1987–1995. She found that trade liberalization, which is measured from reductions in black market premiums, promotes water pollution. The mixed results and relatively small number of studies suggest, as summarized by Panayotou (2000a), that room exists for more empirical analysis of the effect of trade on environmental degradation. Next, the body of empirical literature that investigates the presence of the EKC is considerably larger than the literature on the effects of trade and democracy on the environment. Since the seminal study of Grossman and Krueger (1993), many studies have attempted to verify or reject the presence of the EKC. For recent extensive reviews of these studies, see Panayotou (2000b) and Dinda (2004). The empirical studies use data from various sources for national cross sections, panel data, and pooled data, depending largely on the availability of data. Most studies focus on air pollution (particularly SO2 , but also carbon monoxide and CO2 , nitrous oxides, and SPM). Some studies investigate water quality (organic and nonorganic pollution), a few studies investigate deforestation, and, as far as we know, virtually no study investigates land degradation. In general, some studies have found that the EKC exists for local air pollutants, primarily SO2 , and considerably less so for water quality and other types of air pollution. Global direct indicators such as CO2 and municipal waste, and indirect indicators such as traffic volume and energy consumption, generally do not exhibit an inverted U-shaped pattern. The EKC results for deforestation are mixed and debated (e.g., Bhattarai and Hamming, 2001; Bulte and van Soest, 2001). Finally, EKC studies typically exclude trade, democracy, or both from their model specifications despite the empirical evidence that demonstrates the link between trade and democracy.5 As for the effects of trade and democracy on each other, several empirical studies find that democracy promotes trade, whereas the sign of the effect of trade on democracy is debated. In any case, these studies taken together suggest that trade and democracy affect each other, and we should therefore include both in our empirical model. For studies of the effect of trade 5

For other reviews and assessments of the empirical literature on EKC, see Stern (1998), Harbaugh et al. (2002), Pugel (2003), and Harris (2006).

Economic Openness and the Environment

251

openness on the level of democracy in a country, we refer the reader to Chapter 3, where we show that trade openness decreases the level of democracy in a country. With respect to the effect of democracy on trade, Morrow et al. (1998) found that joint democracy promotes trade among great powers, and Remmer (1998) found that democracies in South America signed more trade agreements with each other. Dixon and Moon (1993) reported that the United States exports more to democracies than to nondemocracies, while Bliss and Russett (1998) and Russett and Oneal (2001) found that this result holds for a larger sample. In contrast, Verdier (1998) concluded that democracy only promotes trade between DCs, whereas Mansfield and Bronson (1997) and Gowa and Mansfield (1993) reported that the effect of democracy on trade was not statistically significant. Souva et al. (2008) showed that joint democracy increases bilateral trade, but the effect largely works through the effect of democracy on property rights protection. EMPIRICAL MODEL AND ANALYSIS

Similar to the previous chapters, this section first presents our statistical model for the empirical analysis and then discusses several research design issues. Next, the section presents the key results from the empirical analysis. As in the other chapters, the discussion in this section is self-contained and does not require any specific statistical expertise. The technical details of the statistical model, measure construction, and data sources are in the chapter appendix, following the same order of presentation as in the main text. We have seen that trade may affect the environment through several channels that operate in parallel. Hence, as in the previous chapter, we do not seek to investigate the relative importance of each of these channels, but rather we study their net effect.

Empirical Model To assess statistically the effect of trade and democracy on the terrestrial environment, we specify and estimate the following statistical model. As in other chapters, we denote variables with small capital letters, and their coefficients with Greek notations. Each coefficient indicates the effect of the independent variable on the dependent variable – the phenomenon we seek to explain. The notation εt denotes the random error not explained by the statistical model. The variable subscript t indicates the current period. To simplify the presentation, we refer to the variables without the time

252

Democracy and Economic Openness in an Interconnected System

subscript t. The specification of the model enables democracy, trade, and their interaction to affect the environment: terrestrial environmentt = ␤0 + ␤1 democracyt + ␤2 trade opennesst + ␤3 democracyt · trade opennesst + ␤4 gdppct + ␤5 gdppc squaredt + ␤6 population densityt + εt .

(8.1)

The dependent variable, terrestrial environment, is measured in two ways: the level of land degradation and the rate of deforestation in a country. Among the key independent variables, democracy is dichotomous, taking the value of 1 if a country is democratic and zero if a country is autocratic; trade openness is the share of a country’s total trade in its national economy; democracy × trade openness represents the interaction between democracy and trade openness, testing whether democracy and trade openness mitigate or amplify each other’s effect on the environment. For the control variables in the model, gdppc is the real GDP per capita of a country, gdppc squared is the square of the gdppc, and population is the population density of a country.

Research Design Issues To implement our statistical model, we need to address several special design issues. Although we discuss the technical details in the appendix, we offer a brief general overview here. First, whereas the theoretical arguments suggest that multiple causal mechanisms may work at the same time between democracy and trade on the one hand and environmental degradation on the other, our statistical model estimates the net effect of these mechanisms. Even though it is useful to understand the role of each mechanism, assessing the net effect of democracy or trade on the environment is in our view more important in terms of both resolving the debates among scholars and grasping the consequences of policies that intend to expand trade or democracy. Second, our statistical model is distinct from those in extant studies. Previous empirical studies typically did not include both democracy and trade in one model. One could gain insight by studying the two forces together, because they influence not only the environment but also each other. Furthermore, we also include the interaction between democracy and trade in our model. Thus, in addressing environmental degradation, economically open democracies may differ qualitatively from economically

Economic Openness and the Environment

253

open autocracies, and open-economy democracies may differ from closedeconomy democracies. In democracies, leaders are held accountable to the electorate and their political supporters and may alienate pro-environment interests. In autocracies, the ruling elite that have adopted an open economic policy may be less responsive to trade-related environmental concerns, because they do not answer to the public. Third, the rate-of-deforestation sample covers 134 countries over two decades (1980s and 1990s). The land degradation sample includes 105 countries during the 1980s. We conduct our analysis in two samples: one for all the countries and the other for LDCs only. As in Chapter 2, we identify LDCs as those countries that are not OECD members. We are interested in identifying whether the patterns that exist in the all-countries sample also stay robust in LDCs alone. For the all-countries sample, both democracy and trade exhibit more variations because both developed and developing countries are included. These variations decline considerably in the LDC-only sample. Also, stylized observation suggests that many LDCs are relatively more dependent on trade than many DCs. The Kuznets hypothesis and stylized facts lead us to expect that LDCs should have weaker institutions to monitor the status of the terrestrial environment and promote its preservation. We report the results for the LDC sample in the additional analysis. Fourth, as environmental conditions vary dramatically across continents, we control for regional heterogeneity by including regional indicators: ASIA, SOUTH AMERICA, EUROPE, MIDDLE & AFRICA, SUBSAHARAN AFRICA, and OCEANIA. They are set to 1 when a country is in Asia, South America, North Africa and the Middle East, Sub-Saharan Africa, Europe, and Oceania, respectively. North America is used as the reference category, so all other regions in the model are compared with North America in their environmental outcomes (i.e., more than or less than the outcome of being in North America). We report the results for regional indicators in the additional analysis. Finally, as usual, we need to ensure that the model’s error term behaves as assumed. We address the related technical issues in the appendix.

Empirical Findings We first present the results for the all-countries sample and then discuss the results of several additional analyses. Findings for the control variables and additional technical details are in the chapter appendix. Table 8.2 presents the results for both deforestation and land degradation. For each indicator, we report the results of two models: one model tests the separate

254

Democracy and Economic Openness in an Interconnected System

Table 8.2. Effects of democracy and trade openness on deforestation and land degradation Deforestation

DEMOCRACY TRADE OPENNESS DEMOCRACY∗ OPENNESS GDPPC GDPPC SQUARED POPULATION DENSITY Constant Observations R2

Land degradation

(1) All countries

(2) All countries

(1) All countries

(2) All countries

−0.9327∗∗ (0.4415) 0.0130∗∗ (0.0070)

−0.8698∗∗ (0.4205) 0.0189∗∗ (0.0094) −0.0206∗∗ (0.0109) 0.0004∗∗∗ (0.0001) −9.04e-09∗∗ (3.58e-09) −0.0013∗∗ (0.0007) −1.2346∗∗∗ (0.3507) 204 0.23

−0.6834∗∗ (0.3427) −0.0068∗ (0.0051)

−0.7539∗∗ (0.3707) −0.0025 (0.0058) −0.0094 (0.0111) 5.8017∗∗ (2.4452) −0.3476∗∗ (0.1501) 0.3082∗∗∗ (0.1076) −21.7429∗∗ (9.8598) 105 0.21

0.0004∗∗∗ (0.0001) −8.55e-09∗∗ (3.58e-09) −0.0009∗ (0.0006) −1.2740∗∗∗ (0.3464) 204 0.21

6.1067∗∗∗ (2.3417) −0.3652∗∗∗ (0.1440) 0.2798∗∗∗ (0.0969) −22.9935∗∗ (9.4540) 105 0.20

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

effects of democracy and trade, and the other adds the democracy–trade interaction. Starting with deforestation, the coefficient of democracy in Model 1 is negative and statistically different from zero. Hence, the democratic country experiences faster deforestation than the nondemocratic country, because a negative value of the deforestation variable indicates deforestation and a positive value indicates afforestation, which is consistent with our finding in the preceding chapter. Now, the coefficient of trade openness in Model 1 is positive and statistically different from zero. Economies that are more open to international trade actually experience slower deforestation. The interpretation of Model 2 for the all-countries sample is more complicated; details are explained in the appendix. Holding trade openness at its mean (we center this variable around its mean), the effect of democracy on deforestation remains negative and significant. Hence, democracy still experiences faster deforestation in a country with average trade openness than does autocracy. Now the coefficient of democracy × trade openness in Model 2 is negative and significantly different from zero, suggesting that in democracy, rising trade openness should lead to even faster deforestation.

Economic Openness and the Environment

255

That is, trade openness amplifies the deforestation-increasing tendency in democracy. In terms of an autocracy (when democracy = 0), its deforestation rate would be slower than in a democracy, and trade openness does not increase the rate of deforestation in an autocracy. In fact, because the coefficient of trade openness is positive and significantly different from zero, rising trade openness in an autocracy increases its rate of afforestation or, in other words, reduces its rate of deforestation. On average, democracy has faster deforestation than autocracy, and trade openness speeds up deforestation in a democracy but reduces it in an autocracy. Next we turn to the results for the all-countries sample for land degradation. In Model 1 for land degradation, the coefficient of democracy is negative and significantly different from zero. Hence, democracy has a lower level of land degradation than autocracy. The coefficient of trade openness in Model 1 is negative and significantly different from zero. A rise in trade openness reduces the level of land degradation in a country. Is there an interactive effect between democracy and trade for land degradation? Model 2 for land degradation answers this question. When trade openness is held at its sample mean level, democracy still has a lower level of land degradation than autocracy. But the coefficient of democracy × trade openness is statistically not different from zero, suggesting that the democratic country does not have varying levels of land degradation when its level of trade openness is raised. In addition, the coefficient of trade openness is not statistically different from zero. Taken together, trade openness does not affect land degradation in a country, and its effect does not depend on regime type.

Additional Analyses We conduct two types of additional analysis. First, we add regional indicators to the models of Table 8.2 for the all-countries sample. Second, we reestimate all models for the LDC sample. The results for deforestation and land degradation are reported in Tables 8.A1 and 8.A2, respectively. Model details are introduced in the chapter appendix. Here we provide a summary of the key results. With deforestation in Table 8.A1, once we control for regional heterogeneity, democracy no longer has a faster deforestation rate than autocracy, but rising trade openness in democracy still leads to significantly higher deforestation rates. As we look at the countries around the world today,

256

Democracy and Economic Openness in an Interconnected System

newly democratizing countries also tend to embrace greater trade openness. We likely expect to see faster deforestation in these countries. As for trade openness, the results are similar to those presented in Table 8.2 even after we control for regional differences. Open-economy autocracies experience smaller deforestation rates than closed-economy autocracies, but open-economy democracies witness higher deforestation rates than closedeconomy democracies. Controlling for regional differences does not fundamentally change the results, despite the regional indicators that suggest regional differences do exist (see the appendix). For the LDC sample in Table 8.A1, the results for democracy are generally similar to those for the all-countries sample. Democracy is associated with faster deforestation than autocracy in the first two models for LDCs, but the effect of democracy vanishes once we control for regional differences. But democracies with more trade openness continue to experience higher deforestation rates than those with less trade openness, a result that remains robust in the LDC sample. The effect of trade openness remains the same in the LDC sample as in the all-countries sample. Trade openness reduces deforestation in autocracies but raises it in democracies. Table 8.A2 reports the results of six models for land degradation. Democracies have lower levels of land degradation than autocracies for the allcountry sample and when regional differences are controlled for. In the models for LDCs without the regional indicators, democracies still experience less land degradation, but the effect is not significantly different from zero when controlling for regional differences. Also as in Table 8.2, the effect of democracy on land degradation does not depend on the level of trade openness, regardless of whether it is the all-countries sample plus the regional indicators or across the LDC models. The effect of trade openness on land degradation is not significantly different from zero when we control for regional differences for the allcountries sample or across all the models for the LDC sample. Changes in trade openness generally do not affect land degradation in LDCs and the significant effect of trade openness in Model 1 for the all-countries sample in Table 8.2 seems to be driven by DCs or without controlling for regional heterogeneity. This outcome is intuitive if we recall that trade flows in DCs tend to be intra-industry, which generally do not degrade land. In contrast, LDCs tend to export primary commodities, minerals, and mined resources, the production of which tends to degrade land. The effect of trade openness on land degradation in three out of four models for the LDCs is indeed positive. A rise in trade might increase land degradation, but the effect is not large, probably due to the relatively smaller size of trade

Economic Openness and the Environment

257

flows in LDCs to begin with. Finally, the effect of trade openness does not appear to be contingent on the nature of a country’s political regime type. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

In this chapter, we study the controversial effect of trade openness on the terrestrial environment and further analyze whether the effect of trade on the environment differs between democracy and autocracy. Our findings demonstrate the complex interactions among trade, democracy, and the environment. A rise in trade openness reduces deforestation in autocracy and increases deforestation in democracy, and the effect is similar for LDCs and DCs. A rise in trade openness reduces land degradation, but the effect is not robust and does not depend on regime type. Now, a rise in democracy increases deforestation and reduces land degradation, but these effects are weaker in LDCs than in DCs. In addition, the effect of democracy on deforestation is stronger when trade openness is high. The effect of democracy on land degradation does not depend on trade openness. What are the broader implications for theory and public policy? One thing comes across clearly; that is, the effects of both trade and democracy on the terrestrial environment are not clear-cut. It appears that trade produces some beneficial effect on the quality of land, but the effect is generally weak. And although trade reduces deforestation, the beneficial effect tends to occur in autocracies, not in democracies. In fact, democracy increases deforestation. In contrast, democracy seems to reduce the level of land degradation. Our results suggest that the spread of democracy has a mixed effect on the terrestrial environment: it is good for land degradation but bad for deforestation. The evidence for land degradation is consistent with the notion that democratic governments have to care for the physical environment in which their citizens reside and that institutional checks and balances in democracy empower a broad range of public interests, including pro-environment groups. The evidence for deforestation, in contrast, is consistent with the notion that democratic regimes may ignore negative effects on the environment if strong constituencies can gain from it. In the case of deforestation, these interests often include logging firms, farms gaining cleared land, urbanized areas facing population pressures alleviated by clearing forested land, and landless people seeking to gain land by clearing forests. At the same time, the effect of international trade on the terrestrial environment is mixed. Trade appears to produce some beneficial effect on

258

Democracy and Economic Openness in an Interconnected System

the quality of land, but the effect is generally weak. And trade leads to opposite effects on deforestation between democracy and autocracy. The opening up of the Philippines or Brazil for trade led to increased logging, which in the case of the Philippines led to large-scale deforestation of hillsides. Hence, economic reforms may produce unintended, malignant consequences on the environment. The public policy problem is potentially serious for LDCs, because they face pressures from within and outside, which push them to both democratize and liberalize their economies. Thus, DCs push LDCs to democratize, and international organizations run by DCs push LDCs to liberalize their economies and remove their trade barriers. Facing these pressures, LDCs are neither well equipped in terms of environmental regulatory expertise, nor do they have the capabilities and incentives to deal with the negative externalities of these policies. The practical effect for the environment may not be beneficial, which underscores the need for a careful and fully monitored transition. SUMMARY AND OUTLOOK

In this chapter, we studied theoretically and empirically the effect of international trade openness on the quality of the terrestrial environment, in the context of countries with either democratic or autocratic regimes. Our goal is to better understand the interactions among trade, democracy, and the environment by linking together literature and debates that have largely been separate from each other. We find that a rise in trade openness reduces deforestation in an autocracy and increases deforestation in a democracy, and the effect is similar for LDCs and DCs. A rise in trade openness reduces land degradation, but the effect is not robust and does not depend on regime type. A rise in democracy increases deforestation and reduces land degradation, but these effects are weaker in LDCs than in DCs. In addition, the effect of democracy on deforestation is stronger when trade openness is high. The effect of democracy on land degradation does not depend on trade openness. In this and previous chapters, we have studied the effects of democracy and trade on the environment. In these analyses, an indicator of military conflict served as one of the control variables. The next chapter shifts the analytical focus to conflict. The role of the environment in international and intrastate violent conflict has received ample scrutiny in the literature. But the effect of violent conflict on the environment itself has received much less attention and has not been studied statistically in a large-N sample.

Economic Openness and the Environment

259

The channel from violent conflict to the environment stands at the center of the next chapter, which also provides the last empirical analysis in this book. We focus particularly on two aspects of the environment: the level of CO2 emissions and the rate of deforestation. These two environmental aspects are particularly interesting to study because they play a major role in climate change. In our next chapter, we turn to something interesting about how violent interstate and intrastate conflicts influence human activities that cause climate change.

APPENDIX EMPIRICAL MODEL AND ANALYSIS

Empirical Model The data sources and measure construction for deforestation, land degradation, and the level of democracy are discussed in the appendix of Chapter 7. The democracy variable used in the statistical model is coded 1 if a country’s POLITY score is larger than 6 and is coded 0 otherwise. We measure the effect of trade on the environment by the sum of the values of export and import of a country divided by its GDP. Data are from the Penn World Table 6.1 (Heston et al., 2002). We also construct an interactive term between democracy and openness to test their conditional effects on the dependent variables. We center the trade variable for ease of interpretation. We control for the indirect effect of trade on the environment through its effect on income per capita and the EKC hypothesis. To that effect, we include (on the right-hand side) the GDP per capita of a country and the square of the GDP per capita. If the EKC holds, the coefficient of GDP per capita should be positive, and the coefficient of GDP per capita squared should be negative. Data on real GDP per capita expressed in purchasing power parity–adjusted, constant 1996 international dollars come from Penn World Table 6.1 (Heston et al., 2002). Population pressures affect environmental degradation. To capture this effect we use population density (total population divided by land area). Previous studies (see, e.g., Panayotou, 2000b) have shown that population density generally increases the level of environmental degradation. These data come from the World Development Indicators (World Bank, 2002). Terrestrial environmental degradation may vary across continents. A country in Africa, for example, may be exposed to conditions different from those in the United States and may exhibit different rates of development

260

Democracy and Economic Openness in an Interconnected System

and urbanization (although this variation is controlled for by GDP per capita). To control for regional variations, we include regional dummy variables in the additional analysis models, set to 1 when a country is in Asia, South America, North Africa and the Middle East, the rest of Africa, Europe, and Oceania, respectively, with North America being the reference category. It is worth noting that these dummies are not theoretical and often tend to absorb much of the variations in the dependent variable that can be otherwise attributed to substantive variables. Hence, results from the models including these dummies need to be interpreted with caution.

Research Design Issues One empirical design issue we need to consider is the risk of heteroskedastic error variance. We address this risk by estimating Huber–White robust standard errors, which are consistent in the presence of heteroskedasticity (White, 1980). Serial correlation is not a concern in our particular case, because we only have cross-sectional data, not time-series data, for both environmental indicators. For the interpretation of the interactive term democracy × trade openness, recall that trade openness is centered around its mean. The coefficient of democracy measures the effect of a rise in democracy from 0 to 1, holding trade openness at its mean (for which the interaction term democracy × trade openness equals 0 as, again, trade openness is centered around its mean). Next, note that the coefficient of trade openness measures the effect of raising trade openness by one unit in autocracies (when democracy and the interaction term democracy × trade openness are both equal to zero). Hence, the coefficient of the interaction term democracy × trade openness has two interpretations: it indicates the additional effect in a democracy of raising trade openness (for in an autocracy the interaction term equals zero) or it indicates the additional effect of raising democracy from 0 to 1 when trade openness is not at its mean (making the interaction term democracy × trade openness nonzero).

Empirical Findings For the control variables in Table 8.2, in the all-countries sample, the coefficient of GDP per capita is positive and statistically significant, and the coefficient of GDP per capita squared is negative and significant. The rate

Economic Openness and the Environment

261

of deforestation falls (becomes less negative) as GDP per capita rises and then rises (becomes more negative) as GDP per capita rises above some level. Hence, we do not find evidence supporting the EKC hypothesis in the full sample. The nonlinear effect found in the all-countries sample largely disappears in the LDC sample. Next, the effect of population density on deforestation is negative and statistically significant for both samples. High population density is associated with more deforestation, as such countries tend to exhibit larger pressures to clear forest from needing more agricultural and urban areas, ceteris paribus. The regional dummy variables suggest that the rate of deforestation is relatively more intense in North America (the United States, Canada, Mexico, and Central America, the reference group), than in other regions. For the control variables in the land degradation models, the EKC is statistically significant for the all-countries sample, but not for the LDC sample. As before, this result is to be expected, because the DCs, which tend to exhibit the EKC in our results, are not included in the LDC sample. A rise in population density leads to more land degradation in both the all-countries sample and the LDC sample, which makes intuitive sense. A more dense country is expected to exert a larger pressure on land due to more intense agriculture and waste dumping, ceteris paribus. Finally, the regional dummy variables in this set of statistical models suggest that the land in the continents of North America is statistically significantly more degraded than land in Asia, South America, Africa, the Middle East and North Africa, and Oceania and is as degraded as the land found in Europe.

Additional Analyses We conduct two types of additional analyses. First, we add regional dichotomous variables to the models estimated using the all-countries sample. Second, we estimate the four models (basic, basic + interaction term, basic + regional dummy variables, and basic + interaction term + regional dummy variables) for a sample of LDCs. Beginning with deforestation, in Models 1 and 2 for the all-countries sample in Table 8.A1, the results are similar to those in Table 8.2, except that the coefficient of democracy, while negative as before, is statistically insignificant. The coefficients of trade openness and democracy × trade openness are similar to those obtained for the all-countries sample in Table 8.2. Hence, the addition of the regional dummy variables does not change the results, despite these dummies being statistically significant.

262

0.0003∗∗∗ (0.0001) −8.27e-09∗∗ (3.48e-09) −0.0010∗ (0.0006) 1.6408∗∗∗ (0.4275) 0.7692∗∗ (0.4550) 1.0592∗∗∗ (0.4420) 1.4341∗∗∗ (0.3551) 3.5943∗∗∗ (0.8068) 0.8443∗ (0.4617) −2.7703∗∗∗ (0.4267) 204 0.31

−0.2131 (0.2896) 0.0177∗∗ (0.0091) −0.0234∗∗ (0.0109) 0.0004∗∗∗ (0.0001) −9.48e-09∗∗∗ (3.41e-09) −0.0013∗∗ (0.0007) 1.4892∗∗∗ (0.4427) 0.5005 (0.4790) 1.2931∗∗∗ (0.4231) 1.4223∗∗∗ (0.3658) 3.5403∗∗∗ (0.8096) 0.7290∗ (0.5539) −2.6785∗∗∗ (0.4381) 204 0.33

−0.2732 (0.3095) 0.0119∗∗ (0.0070) (0.3512) 0.0189∗∗ (0.0093) −0.0262∗∗ (0.0112) 0.0001 (0.0001) 9.52e-09 (1.01e-08) −0.0015∗∗ (0.0007)

(0.3810) 0.0123∗∗ (0.0071)

−0.9598∗∗ (0.3987) 187 0.24

−0.9026∗∗ (0.4041) 187 0.27

−0.5207∗

−0.6185∗

0.0001 (0.0001) 9.31e-09 (9.77e-09) −0.0011∗∗ (0.0006)

(2) LDC

(1) LDC

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Observations R2

Constant

Oceania

Middle East and North Africa

Africa

Europe

South America

Asia

POPULATION DENSITY

GDPPC SQUARED

GDPPC

DEMOCRACY∗ OPENNESS

TRADE OPENNESS

DEMOCRACY

(2) All countries

(1) All countries

Table 8.A1. Effects of democracy and trade openness on deforestation

0.00020∗ (0.00015) 3.15e-09 (9.99e-09) −0.0010∗∗ (0.0006) 1.6251∗∗∗ (0.4332) 0.8168∗∗ (0.4579) 1.4495∗∗∗ (0.4762) 1.3477∗∗∗ (0.3793) 3.1479∗∗∗ (0.8480) 1.3831∗∗∗ (0.4691) −2.5099∗∗∗ (0.5580) 187 0.30

−0.2405 (0.3025) 0.0107∗ (0.0075)

(3) LDC −0.1517 (0.2810) 0.0170∗∗ (0.0091) −0.0310∗∗∗ (0.0114) 0.00021∗ (0.00015) 2.73e-09 (1.02e-08) −0.0014∗∗ (0.0007) 1.3964∗∗∗ (0.4586) 0.4135 (0.4952) 1.8372∗∗∗ (0.4537) 1.3159∗∗∗ (0.3997) 2.9772∗∗∗ (0.8350) 1.3570∗∗ (0.5898) −2.3765∗∗∗ (0.5719) 187 0.34

(4) LDC

Economic Openness and the Environment

263

For the LDC sample in Table 8.A1, in Models 1 and 3, the results for democracy are generally similar to the results for the all-countries sample. The effect of a rise in democracy from 0 to 1 on deforestation is negative and statistically significant in Model 1 and negative but not significant when including regional dummies in Model 3. Hence, the effect of democratization is weaker in size in the LDC sample than in the all-countries sample, reflecting the fewer democracies among the LDCs. In Models 3 and 4, with the regional dummy variables, the results from democracy are again similar to those from all-countries sample, but they are not statistically significant. The coefficients of the interaction term democracy × trade openness in Models 2 and 4 for the LDCs are similar to the results for the all-countries sample. The effects of rises in trade openness on deforestation in Models 1 and 3 for the LDC sample are positive and significant, as we obtained for the all-countries sample. The effect is similar in size across the samples, suggesting that it largely originates from the LDCs. The results for Models 2 and 4 (with the interaction terms) for the LDCs are also similar to the results from the all-countries sample. Table 8.A2 reports the results for land degradation. The effect of a rise in democracy from 0 to 1 on the level of land degradation is negative and significant in Models 1 and 2 for the all-countries sample when the regional dummies are included and for Models 1 and 2 from the LDC sample when the regional dummies are not included. The effect of a rise in democracy from 0 to 1 is also negative when the regional dummies are included for the LDC sample in Models 3 and 4, but it is not statistically significant. The effect of a rise in the level of trade openness on land degradation is not statistically significant when regional dummies are included in Models 1 and 2 for the all-countries sample and not statistically significant in any of the models estimated for the LDC sample. Hence, change in trade openness generally does not affect land degradation in LDCs and the result in Model 1 of Table 8.2 for the all-countries sample seems to be driven by DCs. As for the democracy × trade openness interaction term in the additional analyses for land degradation, Model 4 for the all-countries sample with regional dummies shows that a rise in democracy from 0 to 1 significantly reduces land degradation when trade openness is held at its mean, but a further rise in trade openness above its mean does not add to this effect (because the interaction term is not significant). Similar results hold for LDCs, but the negative effect of a rise in democracy from 0 to 1 is not significant when the regional dummies are added in Model 4. For the

264

(0.3584) 0.0012 (0.0065) −0.0105 (0.0104) 4.7158∗ (3.0702) −0.3070∗∗ (0.1842) 0.1775∗ (0.1308) −0.9758∗∗∗ (0.4002) −0.6179∗∗ (0.3599) −0.0769 (0.4565) −1.7181∗∗∗ (0.5868) −1.4338∗∗∗ (0.4528) −2.0820∗∗∗ (0.8152) −14.0756 (12.8749) 105 0.32

(0.3417) −0.0033 (0.0053)

DEMOCRACY

TRADE OPENNESS

−5.6673 (14.1314) 84 0.25

1.4095 (3.5583) −0.0591 (0.2217) 0.4358∗∗∗ (0.1046)

(0.3549) −0.0033 (0.0053)

−0.5807∗

(1) LDC

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Observations R2

Constant

Oceania

Middle East and North Africa

Africa

Europe

South America

Asia

POPULATION DENSITY

GDPPC SQUARED

GDPPC

4.7759∗ (3.0145) −0.3069∗∗ (0.1819) 0.1490 (0.1236) −0.8948∗∗ (0.4064) −0.5313∗ (0.3657) −0.1790 (0.4980) −1.6576∗∗∗ (0.5754) −1.4416∗∗∗ (0.4879) −2.1813∗∗∗ (0.8356) −14.5444 (12.5896) 105 0.31

−0.8557∗∗∗

−0.7830∗∗

DEMOCRACY∗ OPENNESS

(2) All countries

(1) All countries

−5.0725 (14.6453) 84 0.26

(0.4095) 0.0004 (0.0057) −0.0118 (0.0117) 1.2757 (3.6868) −0.0526 (0.2295) 0.4609∗∗∗ (0.1078)

−0.6824∗∗

(2) LDC

Table 8.A2. Effects of democracy and trade openness on land degradation

0.7217 (4.7716) −0.0419 (0.2946) 0.3033∗∗∗ (0.1203) −0.7592∗∗∗ (0.2603) −0.7377∗∗∗ (0.2508) −0.4155 (0.4172) −1.5231∗∗∗ (0.5260) −1.0329∗∗∗ (0.3600) −3.8804∗∗∗ (0.4677) 0.2205 (19.4282) 84 0.40

−0.3595 (0.3732) 0.0025 (0.0052)

(3) LDC −0.3829 (0.4230) 0.0032 (0.0062) −0.0027 (0.0100) 0.7665 (4.7993) −0.0452 (0.2960) 0.3101∗∗∗ (0.1207) −0.7784∗∗∗ (0.2688) −0.7578∗∗∗ (0.2497) −0.3938 (0.4182) −1.5260∗∗∗ (0.5324) −1.0389∗∗∗ (0.3660) −3.8190∗∗∗ (0.5646) 0.0659 (19.5502) 84 0.40

(4) LDC

Economic Openness and the Environment

265

LDC sample, the interaction term is not statistically significant across the board, suggesting that the effect of a rise in democracy does not become stronger when trade openness is above the mean, and the effect of trade openness does not depend on whether democracy is 1 or 0 (or on regime type).

NINE

Conflict and the Environment

INTRODUCTION

This chapter studies the effects of war on the environment in an innovative way and as part of the story told in this book. Thus far we have studied the effects of economic openness and democracy on conflict, inequality, development, and the environment, and how these forces interact. The concern of analysts and policy makers is growing, including those in the Pentagon, that the environment will play a role in conflict as climate change progresses. The effect of conflict on the environment, in the context of globalization, thus becomes an important complex transformation to study. The Intergovernmental Panel on Climate Change (IPCC) expects that climate change will intensify during this century, causing growing environmental degradation (IPCC, 2007a, 2007b). Consequently, we may see more wars as countries vie for degraded resources or quarrel over who should cover the costs of ameliorating or reducing environmental damages. Although the link from climate change to war is still only a possibility, many scholars have observed that environmental forces have already caused conflicts. In contrast, the effect of war on the environment has received little scholarly attention. Part of the reason for this neglect is perhaps the temptation to assume that wars always destroy the environment. One may recall the pollution caused by burning oil wells during the 1991 Gulf War or the destruction of forests during the Vietnam War. We argue that it is not the only possibility; war may benefit the environment, albeit perversely, by disrupting environmentally harmful economic activities. And it follows that war may have no net effect on the environment, for its positive and negative effects could cancel each other. If all these effects are theoretically possible, the net effect of war on the environment is an empirical issue.

266

Conflict and the Environment

267

What are the effects of war on the domestic environment of a country taking part in the war? Does war intensify or slow down environmental degradation? Perhaps war does not have any net effect on the environment. Perhaps the effects of war on the environment vary across environmental forces. Virtually all extant empirical studies of these questions either cite anecdotal evidence or employ in-depth case studies focusing on one or a few wars and types of environmental degradation. Overall, extant empirical findings paint an ambiguous picture, suggesting that war can either damage the environment or reduce environmental degradation. Naturally, one is tempted to ask what the general tendency is across many countries and wars. These research questions need to be addressed statistically in a largeN sample of countries and wars over time. This important task has thus far not been performed, which motivates this chapter. We begin our analysis by theorizing on the channels leading from war to the environment. This theoretical analysis leads us to expect that the net effect of wars on the environment may depend on the type of environmental degradation and on whether wars are fought abroad or at home. Our statistical models employ a large-N sample of countries and wars over time. The dependent variables are two salient aspects of environmental degradation: CO2 emissions per capita and the rate of deforestation. These environmental indicators received much attention in recent years and for good reasons. CO2 emission is the most important contributor to the global stock of greenhouse gases in the atmosphere, which drives climate change on the source side. Deforestation weakens nature’s ability to reduce the global stock of carbon emissions, which is the most important contributor to climate change on the sink side. The theoretical mechanisms discussed in the next section by which war can affect the environment are applicable for other environmental forces (e.g., water, land) but, given the scope of the current empirical analysis, we believe they are better evaluated in separate studies. In studying the effects of war on the environment, we obviously do not seek to investigate if warfare could be used as a tool to reduce environmental degradation; rather we seek to learn about how the world works, which can have important implications. For example, to the extent that warfare leads to faster deforestation, the effect of localized wars can in fact be much larger than realized thus far – also generating effects on the global climate system. Moreover, if environmental degradation causes warfare, as many scholars argue, a finding that warfare degrades the environment would suggest the presence of a negative feedback that can cause even more warfare and environmental degradation.

268

Democracy and Economic Openness in an Interconnected System

To summarize our primary empirical findings, war reduces the level of CO2 emissions in a country regardless of the location of the conflict. War fought at home increases the rate of deforestation at home, whereas war fought abroad reduces this rate. These results hold in a number of sensitivity analyses that employ different empirical model specifications, samples, and measures. When evaluated in the context of climate change, these findings imply that wars produce competing effects on climate change, promoting and alleviating environmental degradation that affects climate change. The remainder of the chapter proceeds as follows. The next section theorizes the effects of warfare on environmental degradation within countries. The section that follows presents an overview of the empirical research design and statistical model, followed by a discussion of the key empirical results. The last two sections discuss implications for global environmental policy and summarize the chapter, respectively. Technical details and results from additional analyses are provided in the appendix. THEORIZING THE EFFECTS OF WAR ON THE ENVIRONMENT

What are the effects of war on the environment of a country? Although it is tempting to assume that war always destroys the environment, there are two other logical possibilities: war benefits the environment, or war does not affect the environment. This section argues that all three possibilities are theoretically plausible. War may damage the environment through direct and indirect channels. Beginning with indirect channels, the movement of armies to the battlefield, for example, may destroy fields and vegetation and degrade land. Bombing campaigns may cause unintended fires and destroy grazing land, forests, and cropland and kill animals. Fighting may also discharge wastes that pollute rivers and lakes. War may also affect the composition and scope of production, which in turn affects the environment. For example, the economy may shift to produce more weapons and fewer consumption goods, which, in turn, may increase pollution and waste. Finally, Dasgupta (1995) reasons that war may weaken social norms that support environmental protection of the commons, increasing environmental degradation. Other than indirect effects caused by outright combat, peacetime military preparations for war can also indirectly damage the environment. For example, the need to sustain armies may intensify water extraction and livestock consumption, and munition storage may release toxic materials. As discussed in Westing’s (1990) edited volume, nuclear and chemical weapons production and storage may damage ecosystems. Additionally, as noted by

Conflict and the Environment

269

Singer and Keating (1999), standing armies may dump materials such as shells and chemicals into the environment. The buildup of army camps may destroy arable land and vegetation and increase logging, and training armies may degrade arable land and cause pollution. Another indirect channel involves war refugees. As observed by Allan (1987) for Afghani refugees, the temporary settlement of many refugees in hastily made-up camps may create mountains of waste and destroy the surrounding vegetation and forestry. Moving to the direct effects, war can destroy the environment. The warring parties may destroy the environments of each other as part of their strategies to win the war. For example, armies may set forests on fire or dump defoliates on them to deny the enemy timber or hiding places, as well as set oil wells on fire; destroy fresh water resources, crops, grazing fields, and vegetation; and kill domesticated animals to prevent their use by the enemy. Armies may also flood large areas by destroying dams or opening floodgates or destroy other natural resources to slow down or prevent enemy movements. Alternatively, the warring parties may intensify the exploitation of natural resources to sell them and finance their war efforts. A number of stylized facts support these expectations. McNeill (2001) lists wars that destroyed forests, including the French war in Morocco in the 1920s, World War II, the Greek Civil War in the late 1940s, the British war in Malaya in the 1950s, the Vietnam War, and the Russian war in Afghanistan in the 1980s. Articles in The Economist (1995, 2003) report that warring parties in Myanmar, Angola, and Sierra Leone caused intensified timber logging in forests to finance their war efforts. Deacon (1994) noted that wars in less developed countries (LDCs) eroded norms of preservation, increasing the rate of deforestation. Finally, during the 1991 Gulf War the Iraqis set oil wells on fire and spilled oil into the Persian Gulf, generating pollution and destroying natural habitats. Sometimes, however, wars may also benefit the environment. If the war effort reduces the scope of ordinary economic activities that destroy the environment at home, and the damaging effects of war are not too high, the net effect of war could be beneficial. For example, fewer fishing boats may go to sea during wars because fishermen are enlisted in the army or seek to avoid maritime combat locations. As a result, fish stocks may grow. A second example involves changes in industrial activities. Wars may destroy industrial plants and transportation infrastructures. As production declines, pollution and waste discharges also decline. Furthermore, wars may also alter the composition of production and consumption at home. For example, more fuel may be shipped to the front, creating fuel shortages

270

Democracy and Economic Openness in an Interconnected System

at home. Consequently, ordinary economic consumption and production may decline, reducing pollution and waste. Or, as more people are enlisted in the army, economic activities in sectors using them in peacetime may shrink, including timber logging, fishing, and industrial production. As a result, stocks of forests and fish may increase, and industrial pollution may decline. Turner (1990) and McNeill (2001) provide a number of examples of wars that benefited some environmental resources. For example, World War II helped fish stocks increase in the Atlantic Ocean because normal fishing decreased. In German-occupied Europe and Japan, industrial emissions fell during World War II because of coal shortages and the damages that resulted from the Allies’ bombing campaigns. Iraqi land mines kept people from going into the Kuwaiti dessert, allowing the stocks of formerly overexploited plants and animals to revive. Thus far we have discussed positive and negative effects of war on the environment. However, two reasons exist for why war may not affect the environment. First, the intensity or nature of the fighting may not suffice to damage the environment. For example, surgical air strikes may not destroy much of the economy and, therefore, may not largely affect emissions. Second, the negative and positive effects of wars on the environment may cancel each other. One could probably break down the positive and negative effects along sectors and forces, but this task is currently not feasible due to data unavailability. That said, one ought to note that theoretical ambiguity and competing effects are quite common in environmental studies, including the effects of democracy and trade on the environment as indicated in Chapters 7 and 8, as well as the effects of the environment on war. In such situations, the relative sizes of competing forces are not known theoretically, but their net effect is estimable. As in these and other social science areas, we can and should study the net effect of war on the environment empirically.1 Before we turn to our research design, we need to consider three additional theoretical issues. First, our discussion suggests that the effects of war on the domestic environment may vary across environmental features. It follows that our empirical analysis ought to examine various environmental aspects, and our statistical results may also differ across these disparate 1

Other areas that involve theoretical ambiguity and competing forces include the effect of trade on armed conflict, the effect of globalization on democracy, and the effect of economic growth on inequality.

Conflict and the Environment

271

dimensions. In other words, wars may benefit some environmental aspects and harm others. A second issue concerns the locations of wars. Often conflicts are categorized into two types: interstate and intrastate. Intrastate conflicts have been on the rise; interstate conflicts have declined but not disappeared. In the context of the environmental impact of war, the interstate–intrastate distinction is less useful than the location where a conflict actually occurs: at home or abroad. It is tempting to argue that only wars fought on a country’s own territory should matter for its environment; however, this is not the only possibility. Wars fought abroad may change the relative balance of sectors in the national economy and alter the scope and composition of production and consumption at home. For example, fuel may be sent to the front abroad while rations are imposed at home. People may be sent to fight abroad, reducing production in some domestic sectors. Production in other sectors may grow as war materials are sent abroad. All these war-induced economic changes can significantly affect the state of the environment at home. Finally, the effect of war on the environment may also differ between developed countries (DCs) and LDCs. In the case of CO2 emissions, we expect to see a weaker effect of war for LDCs than for DCs. Consider, for example, two economies X and Y that obtain 90% and 10% of their outputs, respectively, from emission-generating sectors (e.g., industry, utilities). Because Y has fewer emission-generating activities, the net effect of war on Y’s emissions should be weaker than that on X’s emissions. On the other hand, in the case of deforestation, we expect to see a stronger effect of war for LDCs than for DCs. Because LDCs are more likely to be agrarian economies and their output depends more on forests, the effects of war on the rate of deforestation should be larger in LDCs than in DCs. On a more applied level, in the time period for which our data are available (recent decades), LDCs generally have experienced more interstate and intrastate wars than DCs, further suggesting the need to distinguish LDCs in the empirical analysis. To summarize, wars produce competing theoretical effects on the domestic environments of the countries fighting them. These effects may change depending on the environmental features at issue, the locations of warfare, and the development levels of the countries. However, it is worth stressing that despite this complexity, it is feasible and desirable to estimate the net effect of wars on particular environmental features, an issue to which we turn next.

272

Democracy and Economic Openness in an Interconnected System

EMPIRICAL MODEL AND ANALYSIS

This section illustrates our empirical model, discusses related research design issues, and presents the main findings. As in the previous chapters, this discussion is self-contained and does not require statistical expertise. The technical details of the statistical model, data sources, and measure construction are shown in the chapter appendix, following the same order of presentation as in the main text.

Empirical Model To assess the effect of war on the environment, we specify and estimate the following statistical model of the environment. We denote variables with small capital letters and their coefficients with Greek notations. The notation ␤c denotes a vector of coefficients for country fixed-effects variables. Each coefficient indicates the effect of the independent variable on the dependent variable, the phenomenon we seek to explain. The notation εt denotes the random error not explained by the statistical model. The variable subscripts t and t − 1 indicate the time period of the variable, where t represents the current period and t − 1 the previous time period (a lagged variable). To simplify the presentation, we refer to the variables without their time subscripts t or t − 1. The model specification is based on the aforementioned theoretical discussions and the literature on democratization. It provides a structure for us to guide the statistically uninitiated readers through the empirical exercise. environmentt = ␤0 + ␤1 wart−1 + ␤2 democracyt−1 + ␤3 real gdppct−1 + ␤4 real gdppc squaredt−1 + ␤5 trade opennesst−1 + ␤6 population densityt−1 + ␤7 lagged environmentt−1 + ␤8 year + ␤c country fixed effects + εt .

(9.1)

The dependent variable, environment, represents two different aspects of environmental degradation: CO2 emissions per capita and the rate of deforestation. These two forces are important, if not the most important, aspects of global environmental degradation today, and they will become increasingly so in the coming decades. CO2 emissions are a major cause of climate change because they contribute to the greenhouse effect. Deforestation reduces the availability of fresh water because forests reduce losses

Conflict and the Environment

273

due to runoff and evaporation and absorb much of the world’s rainwater into underground aquifers. Deforestation also promotes land degradation because forests keep topsoil in place and prevent land erosion. Moreover, deforestation influences the global climate system by reducing nature’s ability to absorb and break down CO2 emissions. Hence, both indicators correlate with climate change, albeit in distinct manners. The CO2 emissions indicator reflects the outcome of human activities that produce emissions, whereas the rate-of-deforestation indicator measures the outcome of human activities that absorb CO2 emissions. The stock of emissions at any given point in time depends on both types of human activities (Harris, 2006; IPCC, 2007a, 2007b). The two indicators give us an opportunity to investigate how warfare relates to different processes that influence climate change. The key independent variable, war, is measured in several ways. The first indicator, war at home or abroad, reflects the presence or absence of a country’s involvement in at least one interstate, intrastate, or internationalized violent conflict that is identified as a war in our time period (yearly for CO2 emissions per capita, and decade for the rate of deforestation). This aggregate indicator does not distinguish wars in terms of their geographical locations. The reader may recall that our theory allocates importance to whether a war is fought at home or abroad. We therefore also create two more war-related indicators: war abroad reflects a country’s involvement in any war fought abroad and war at home represents a country’s involvement in any war fought at home (including a civil war). These indicators allow us to test whether the location of warfare matters in environmental outcomes. All these indicators of war denote armed conflict situations that cause at least 25 battle deaths. We describe the details about these variables in the chapter appendix. Like the other models in this book, this statistical model also controls for other forces that influence the environment. The empirical literature explaining environmental degradation motivated our inclusion of the control variables: real gdppc is the gross domestic product (GDP) of a country expressed in real terms, per capita; real gdppc squared is the square of the real GDP per capita variable. As discussed at length in the previous chapter, these two variables together represent the environmental Kuznets curve effect. democracy is the level of democracy in a country, measured (as in the other chapters) over a continuum ranging from complete democracy to complete autocracy; trade openness is the share of a country’s total trade in its national economy; and population density is the population of a country divided by the area of its territory.

274

Democracy and Economic Openness in an Interconnected System

Because the model of CO2 emissions per capita uses pooled time-series cross-sectional data, it requires some additional control variables to ensure the validity of statistical inference: year is a yearly counter that models the linear trend in the emissions, lagged environment represents the CO2 emissions from the previous year, and country fixed effects are country dummies that capture heterogeneity across different countries. These additional variables are not necessary for the deforestation model, because its data are largely cross-sectional.

Research Design Issues In designing our empirical analysis, we also need to consider several technical issues. We provide an overview of these issues here and delegate the details to the appendix. The first issue is whether we should conduct a pooled time-series cross-sectional analysis or a cross-sectional analysis. The sample of CO2 emissions per capita includes 143 countries from 1961 to 1997. The deforestation sample covers 134 countries over two decades (1980s and 1990s). Given the availability of data, we employed the pooled design for the CO2 model, and the cross-sectional design for deforestation. In the crosssectional design, the variables, except for the dichotomous war variables, take on decade-average values, and in the pooled design they take on yearly values. The second issue concerns the properties of the error terms and the possibility that the right-hand-side variables are excessively correlated. We apply appropriate econometric techniques to address any issue related to the error term and provide some diagnostic test for the possible high correlation among independent variables. Third, we model the possibility of path dependence for carbon dioxide by including the lagged level of CO2 emissions per capita as one of the independent variables, which also helps to control for the effect of any possibly omitted variable. For deforestation the lagged dependent variable is not included, given the cross-sectional nature of the sample. Fourth, we lag the right-hand-side variables in the CO2 models for two reasons. The effects of the independent variables on the environment may not be immediate and take some time to materialize. In addition, the righthand-side variables may themselves be affected by the environment. Finally, following our argument that the effect of war on environmental degradation may differ between DCs and LDCs, we estimate our models for both a sample of all countries and a sample of LDCs only.

Conflict and the Environment

275

Table 9.1. War and CO2 emissions per capita All countries (C1) WAR AT HOME OR ABROAD WAR AT HOME

−0.0375∗∗∗ (0.016)

WAR ABROAD −0.0016∗∗ (0.001) REAL GDPPC 0.00008∗∗∗ (0.00002) REAL GDPPC SQUARED −2.75e-09∗∗∗ (5.70e-10) TRADE OPENNESS −0.0009 (0.001) POPULATION DENSITY 0.0007∗∗∗ (0.0003) 0.866∗∗∗ LAGGED CO2 (0.023) YEAR 0.001 (0.001) Constant −1.794 (2.13) Observations 3830 0.99 R2

DEMOCRACY

(C2)

−0.027∗ (0.017) −0.057∗∗ (0.028) −0.0016∗∗ (0.001) 0.00009∗∗∗ (0.00001) −2.76e-09∗∗∗ (5.70e-10) −0.0009 (0.001) 0.0007∗∗∗ (0.0003) 0.866∗∗∗ (0.023) 0.001 (0.001) −1.597 (2.15) 3830 0.99

LDCs (C3)

(C4)

−0.0173 (0.016)

−0.0028∗∗∗ (0.001) 0.00008∗∗∗ (0.00003) −2.00e-09∗ (1.36e-09) −0.0002 (0.001) 0.0006∗∗ (0.0003) 0.850∗∗∗ (0.035) 0.001 (0.001) −1.474 (2.30) 3053 0.97

−0.021 (0.017) 0.004 (0.020) −0.0028∗∗∗ (0.001) 0.00008∗∗∗ (0.00003) −2.02e-09∗ (1.36e-09) −0.0002 (0.001) 0.0006∗∗ (0.0003) 0.850∗∗∗ (0.035) 0.001 (0.001) −1.516 (2.30) 3053 0.97

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

Empirical Findings This section presents our empirical findings. We first present the main results and then discuss results for several additional analyses. Findings for the control variables and additional technical details are presented in the appendix of this chapter. Table 9.1 presents the results for CO2 emissions per capita. The diagnostics used to ascertain the econometric performance of the model perform well, as discussed in the appendix. Models C1 and C2 are based on the sample of all the countries, while Models C3 and C4 focus on the sample of LDCs. The key independent variable in Models C1 and C3 is war at home or abroad, which does not distinguish between wars fought abroad and

276

Democracy and Economic Openness in an Interconnected System

at home. Models C2 and C4 distinguish between wars according to their locations. The coefficient of the variable war at home or abroad in Model C1 is negative and significantly different from zero. War involvement reduces CO2 emissions per capita, regardless of its location. The effects of war abroad and war at home in Model C2 are also negative and significantly different from zero. In the full sample, both war at home and war abroad reduce CO2 emissions per capita. What do these results mean? Recall that the effect of war on the environment may occur through changing output, resource allocation and sectoral breakdown, and destroying the economy. After controlling for the output effect (i.e., through real GDP per capita in the model), the net effect of war on the environment is negative. How large are these effects? We compute the sizes of the effects of war on CO2 emissions per capita in Models C1 and C2. We assume that the war variable changes from 0 to 1, and all the other variables in the model are held at their sample mean values. An involvement in war at home or abroad reduces CO2 emissions per capita by 1.14%, an involvement in war at home reduces CO2 emissions per capita by 0.76%, and an involvement in war abroad reduces CO2 emissions per capita by 1.61%. These substantive effects represent statistically significant outcomes but seem relatively small in size. At the same time, as in other models in the book that include lagged dependent variables on the right-hand side, these immediate effects do not tell the whole story of the effect of war on CO2 emissions per capita – war affects the current level of CO2 emissions per capita via its direct effect, as computed earlier, and continues to affect the CO2 emissions per capita in the next period via its effect on the lagged emissions, which accumulates over time. With some additional computation, we find that in the long run, an involvement in war at home or abroad reduces CO2 emissions per capita by 8.5%. Similarly, an involvement in war at home reduces CO2 emissions per capita by 5.67%, and an involvement in war abroad reduces CO2 emissions per capita by 12.01%. These long-run effects are not only substantial but also considerably larger than the short-run effects. We have argued that the effect in LDCs may differ from that observed in a sample of all countries. Turning to Models C3 and C4 for the LDCs, we find that an involvement in war at home or abroad does not affect CO2 emissions per capita. We get the same results for the effects of involvements in war at home and war abroad. Recalling our theoretical discussion, one may interpret this result in two ways. First, it is possible to argue that the competing effects of war on the environment cancel out in the case

Conflict and the Environment

277

Table 9.2. War and deforestation All countries (D1) WAR AT HOME OR ABROAD WAR AT HOME

−0.075 (0.326)

WAR ABROAD −0.121∗∗∗ (0.038) REAL GDPPC 0.0004∗∗∗ (0.0001) REAL GDPPC SQUARED −1.03e-08∗∗∗ (3.75e-09) TRADE OPENNESS 0.013∗∗ (0.007) POPULATION DENSITY −0.0010∗ (0.0006) Constant −2.524∗∗∗ (0.657) Observations 204 R2 0.249

DEMOCRACY

(D2)

−0.509∗∗ (0.281) 0.840∗∗ (0.414) −0.122∗∗∗ (0.037) 0.0004∗∗∗ (0.0001) −1.02504e-08∗∗∗ (3.58e-09) 0.012∗∗ (0.007) −0.0009∗ (0.0006) −2.328∗∗∗ (0.549) 204 0.272

LDCs (D3) −0.188 (0.347)

−0.099∗∗∗ (0.036) 0.0002∗ (0.0002) 3.83e-09 (9.43e-09) 0.012∗ (0.007) −0.0011∗∗ (0.0006) −2.057∗∗∗ (0.713) 187 0.253

(D4)

−0.696∗∗∗ (0.281) 1.065∗∗∗ (0.472) −0.097∗∗∗ (0.034) 0.0002∗ (0.0001) 5.16519e-09 (7.50e-09) 0.011∗ (0.007) −0.0010∗ (0.0006) −1.791∗∗∗ (0.579) 187 0.292

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

of LDCs, resulting in a statistically insignificant result. Alternatively, an interpretation we prefer, the effect of involvement in war at home or abroad, war abroad, or war at home on CO2 emissions per capita in LDCs is so small as to be statistically insignificant. This interpretation is in line with the observation that the extent of CO2 -promoting activities per capita in LDCs is very small. As such, involvement in wars simply does not affect CO2 -promoting activities much, driving a statistically insignificant outcome. Table 9.2 reports the results for deforestation. As in the case of CO2 emissions per capita, Models D1 and D2 are based on the sample of all countries, whereas Models D3 and D4 are based on the sample of LDCs. Models D1 and D3 do not distinguish the locations of wars (home or abroad), whereas Models D2 and D4 make this distinction. In Model D1, the effect of war at home or abroad – war involvement regardless of its location – on deforestation is not statistically different from zero. But in Model D2, the coefficient of war at home is negative and significantly different from zero, whereas the coefficient of war abroad is

278

Democracy and Economic Openness in an Interconnected System

positive and significantly different from zero. These results suggest that war involvement and the location of the war matter for deforestation. Involvement in wars fought at home increases the rate of deforestation, whereas involvement in wars fought abroad reduces it. Our results suggest that wars fought at home destroy forests and/or intensify timber logging and forest clearing at a faster-than-usual speed. Consequently, the rate of deforestation at home increases. In contrast, wars fought abroad alleviate these pressures. As a result, wars fought abroad reduce the rate of deforestation at home. It is plausible that wars fought abroad divert labor and financial resources from timber logging and forest clearing. In addition, wars fought abroad inflict less damage on forests at home than do wars fought at home. The opposite effects of wars fought at home and abroad on deforestation also help to explain the lack of effect of war at home or abroad. The competing effects cancel out when the war variable does not distinguish the locations of wars. Once wars are separated according to their geographical locations, their effects become statistically significant. For the LDCs sample, Models D3 and D4 yield results consistent with those from the sample of all countries. Wars do not affect deforestation in the LDCs if we do not distinguish their locations. Once we separate wars based on their locations, their effects start to emerge in patterns consistent with those from the full sample. Wars abroad slow down deforestation whereas wars at home lead to faster deforestation. In addition, as expected, the coefficients of the war variables in Model D4 for the LDCs are considerably larger in size than those in Model D2 for the sample of all countries. We now compute the sizes of the effects of the statistically significant war variables, implementing the same procedure as in the CO2 case (i.e., change a war variable from 0 to 1 while holding all the other variables at their mean and computing the change in the effect in percent). Based on Model D2 (the sample of all countries) in Table 9.2, we find that war at home increases the rate of deforestation rate by 179.8%, and war abroad reduces the rate of deforestation by 296.8%. Wars fought abroad or at home produce very large effects on the rate of deforestation at home, albeit in opposite directions.

Additional Analyses We have presented a relatively large number of results on the effect of warfare on CO2 emissions and deforestation. To establish the robustness of our results, we conduct two additional analyses: (1) we include a cubic real

Conflict and the Environment

279

GDP per capita term for both the CO2 and the deforestation models and (2) we exclude the lagged dependent variable (LDV) from the CO2 models. The results of these analyses are fully presented and discussed in the appendix. To summarize, we first reestimate our statistical models while assuming that the dependence of CO2 emissions per capita on real GDP per capita is cubic, not quadratic. The results for the effects of war at home or abroad, war at home, and war abroad are similar to those reported in Table 9.1 in terms of their signs, significance, and sizes, indicating the robustness of the CO2 models. The coefficients of some of the GDP per capita terms differ from those in Table 9.1, as could be expected, and they are interpreted in the appendix. In general, the dependence of CO2 emissions per capita on real GDP per capita may be cubic, but the effects of war remain robust, which is the important issue here. We then include a cubic real GDP per capita term for the rate of deforestation. The coefficients for the war and the control variables are similar to those in Table 9.2. The effects of war on deforestation also remain robust. Finally, we estimate the models of CO2 emissions per capita presented in Table 9.1 without the LDV. The results, presented in the appendix, are again in line with Table 9.1 in terms of signs and significance, with one exception. In the new analysis, the effect of war at home on CO2 emissions per capita is negative and significantly different from zero in the LDC sample, as in the full sample. The effects of war on CO2 emissions per capita are now larger in size than those in the presence of the LDV. We offer an explanation for this outcome in the appendix. Hence, the results in Table 9.1 are not an artifact of the presence of the LDV. IMPLICATIONS FOR THEORY AND PUBLIC POLICY

Our theoretical discussion suggests that war may slow down or speed up environmental degradation of a country participating in the war. The overall effect may change on the basis of the location of the war (at home or abroad), the particular environmental indicator inspected, and the level of the country’s economic development. Our empirical analysis, focusing on the net effect of war on CO2 emissions and deforestation, employs a statistical model for many countries and years, which to our knowledge is the first of its type in the conflict–environment literature. Summarizing our finding, a country’s involvement in interstate or intrastate warfare reduces its CO2 emissions per capita, but the effect is very weak in LDCs. Wars fought at home increase the rate of deforestation in a country, whereas wars fought abroad slow deforestation. The effects of

280

Democracy and Economic Openness in an Interconnected System

war on the rate of deforestation (either positive or negative) are larger in LDCs than in DCs. These findings have implications for climate change because our environmental indicators are important contributors to this phenomenon. In general, some of the effects of climate change have already been observed today, including rising sea levels, retreating glaciers, melting ice poles, and more frequent and intense extreme weather occurrences such as storms and droughts. Although exact numbers are still unknown, climate change is expected to be the largest source of environmental degradation in the coming decades, assuming business as usual (IPCC, 2007a, 2007b). How will people respond to these expected changes? In principle, they could try to mitigate the expected effects, adapt to the changes by altering practices, or do nothing and accept the decline in their quality of life. Although adaptation and mitigation may seem reasonable measures, they face formidable obstacles. For example, adapting to the sea-level rise by building coastal defenses may cost tens of billions of dollars and take many years to complete (Allen et al., 1998; IPCC, 2001). Currently, a number of U.S. and European policymakers appear to be preparing for the possibility that climate change will increasingly induce violent conflict (New York Times, February 29, 2004). As Schwartz and Randall (2003: 22) write in a report commissioned by the U.S. Department of Defense, “Disruption and conflict will be endemic features of life.” The issue is not without debate, but many scholars agree that growing environmental degradation and natural resource problems already make violent conflict significantly more likely. The possibility of climate change–induced warfare recasts our findings in a new light. Will the increased incidence of warfare accelerate climate change, or slow it down? Since climate change is an evolving force whose attributes are not yet fully manifested and understood, answering these questions requires prediction. In social sciences, prediction typically is based on historical data. Caution is needed when employing this method, but the working assumption is that, when analyses can explain historical patterns, they give us useful information about the future. We believe that our findings help us gain insight on our question. Our findings suggest that wars can slow down the pace of climate change, but, of course, we do not argue that warfare is a good policy instrument for combating climate change. On the contrary, we believe that there are far better, more peaceful ways for reducing CO2 emissions and deforestation, including developing alternative sources of energy, appropriate institutions, technological measures, and perhaps reducing the growth rate of production

Conflict and the Environment

281

and consumption while fossil fuel remains our main source of energy. However, our analysis demonstrates that although war is generally believed to be damaging, it also can perversely benefit the environment. SUMMARY AND OUTLOOK

The role of environmental forces in violent intrastate and interstate warfare has received much scholarly attention in recent studies, but the reverse causal arrow from warfare to the environment has been largely overlooked. In this chapter, we study the effects of war on the domestic physical environment. Does war have any effect on the domestic environment of a country taking part in the war? Does it intensify or slow down environmental degradation? Do the effects of war on the environment vary across environmental forces? Only a few empirical studies have investigated the effects of war on the physical environment, and virtually none of these studies has employed statistical methods for a large sample of countries and years. This chapter develops a statistical model of the effects of war on two important types of environmental degradation: CO2 emissions and deforestation. We find that war reduces CO2 emissions at home regardless of whether the war is fought at home or abroad. War fought at home increases the rate of deforestation at home, whereas war fought abroad reduces this rate. These findings have implications for understanding the sources of climate change as well as the environmental consequences of warfare. This chapter contains the last theoretical and empirical analysis we conduct in this book. Throughout Chapters 2–9, we examined different types of interactions among democratic governance, economic openness, income inequality, national economic development, involvement and initiation of interstate and intrastate military conflicts, and various indicators of the quality or degradation of the physical environment within which all these socioeconomic and political interactions take place. Guided by various theoretical arguments, debates, and empirical studies on a wide range of topics, we have demonstrated that these forces interact with one another along various channels over time. The reader may recall that we started our analytical theoretical and empirical journey in Chapter 1 with a stylized graphical model of the various interactions among the economic, political, social, and environmental forces we study in this book. This graphical model offers merely a scheme of how our inquiry would unfold, providing little specifics as to the nature of the various interactions in our conceptual framework. In the last chapter of the book, we revisit this graphical model. In Chapter 10, we flesh out the

282

Democracy and Economic Openness in an Interconnected System

graphical model with the various central findings we obtained in the various chapters. The revised graphical model offers a relatively complete view of the complex transformation the global system is undergoing in our times. It also uncovers for us the kinds of trade-offs and mutually reinforcing effects the various important processes generate for us.

APPENDIX EMPIRICAL MODEL AND ANALYSIS

Empirical Model The dependent variable, environment, is measured in two ways: carbon dioxide (CO2 ) emissions per capita and the rate of deforestation. The indicator of CO2 emissions per capita is expressed in metric tons per capita in a country in a year, adjusting for cross-national differences in population. This indicator captures emissions generated by various human activities, including industrial production, burning of fossil fuels, burning of gas released in petroleum extraction, cement manufacturing, and gases released from stored fuels. The data come from the World Development Indicators (World Bank, 2002). The indicator of deforestation measures the rate at which forests decrease or increase in a country due to activities such as building settlements, mining, ranching, farming, and reforestation. For this variable, a positive value indicates a rise in forested area over time (afforestation) and a negative value implies a fall in forested area over time (deforestation). Areas harvested with the intent of natural regeneration and areas degraded by gathering wood for fuel, by acid rain, or by natural fires are not included. Data on average annual deforestation rates per decade per country are collected from the World Resources Institute (1999) and the State of the World’s Forest Report (2001). The key independent variable, war, is also measured in several ways. The variable war at home or abroad is dichotomous, which is set to 1 if a country participated in at least one interstate, intrastate, or internationalized violent conflict that is identified as a war in our time period. For the empirical analysis, we need to decide how to identify a violent conflict as a war, which requires some threshold. Many studies employ the threshold of 1,000 battle deaths for that purpose. For our purpose the threshold of 1,000 battle deaths may be too high, because it results in classifying many

Conflict and the Environment

283

periods involving violent conflicts in recent decades as peaceful, thus omitting relevant information. Therefore, we decided to employ a lower threshold in identifying periods of warfare in this study. We use data compiled by the Department of Peace and Conflict Research at the University of Uppsala, Sweden and the International Peace Research Institute, Oslo, Norway. In these data, one can identify an armed conflict as one causing at least 25 battle deaths and involving the use of armed force between two parties, where at least one party is the government of a state (Gleditsch et al., 2002; Strand et al., 2005); we employ this approach. To evaluate the effects of wars according to whether they are fought at home or abroad, we create two additional dichotomous variables based on the Conflict Sites data set compiled by Raleigh and Gleditsch (2005). The dichotomous variable war abroad is set to 1 when a war is fought abroad, and 0 otherwise. The dichotomous variable war at home is set to 1 when a war is fought at home (including a civil war), and 0 otherwise. The empirical literature explaining environmental degradation motivates our inclusion of the following control variables: real gdppc, real gdppc squared, democracy, trade openness, population density, and, in some models, real gdppc cubed, and lagged environment (e.g., Panayotou, 2000a; Li and Reuveny, 2006; Dinda, 2004). real gdppc denotes real GDP per capita. As discussed in Chapter 7, the literature suggests that environmental degradation follows an inverted U with respect to real gdppc, known as the EKC. The empirical validity of the EKC, we recall from previous chapters, is debated. To model the EKC, we include real gdppc and its squared term, real gdppc squared, in the model. If the EKC exists, the coefficient of real GDP per capita should be positive whereas the coefficient of real GDP per capita squared is negative, with the former coefficient larger in size than the absolute value of the latter coefficient. The data, expressed in constant 1996 international dollars adjusted for purchasing power parity, come from the Penn World Table 6.1 (Heston et al., 2002). The level of national democracy, democracy, is measured using POLITY IV data (Marshall and Jaggers, 2006). The construction of this variable follows the one in previous chapters. The measure ranges between −10 (most autocratic) and 10 (most democratic). The effect of democracy on the environment was discussed in Chapter 7. The level of trade openness of a country, trade openness, is the sum of exports and imports of a country divided by its GDP, reflecting the importance of trade in the country’s economy. The effect of trade openness on the environment was discussed in Chapter 8.

284

Democracy and Economic Openness in an Interconnected System

A rise in population density is expected to increase CO2 emissions per capita because it leads countries to consume and produce disproportionally more (e.g., China, India, the United States, Western Europe). Densely populated nations may require more land for agriculture and industry, raising the rate of deforestation. Yet these countries may also be more urbanized and depend less on the environment for livelihood, thereby clearing fewer forests (e.g., Western Europe). Templeton and Scherr (1999) argue that densely populated countries may strengthen property rights or privatize forests, reducing deforestation, and Panayotou (2000a) concludes that a rise in population density creates competing theoretical effects on forests. Data for population density come from the World Development Indicators (World Bank, 2002). For the model of CO2 emissions per capita, we also include the first lag of the emission variable (lagged environmentt−1 in Equation (9.1)). The role of the lagged dependent variable is similar to its role in the other models in this book: it addresses path dependence in CO2 emissions, ameliorates the undesired effects of nonspherical regression errors, and accounts for possible omitted variables. Finally, for the model of CO2 emissions per capita, we also include year and country fixed effects; year is a year counter that captures the possibility of trending in the CO2 emissions data, as discussed in the next section, and country fixed effects are dichotomous country dummy variables. Each variable is set to 1 for all the years for some country, and is set to zero otherwise. These variables capture nonchanging structural variables of a country that may affect CO2 emissions per capita, as discussed next.

Research Design Issues We consider several design issues. The first issue is whether we should conduct a pooled time-series cross-sectional analysis or a purely crosssectional analysis. Although the pooled design is superior in capturing both temporal and cross-sectional patterns, our choice is dictated by data availability. Our sample for the CO2 models covers 134 countries with data available annually from 1961 to 1997, whereas the sample for the rate of deforestation covers 134 countries with data available for two decades (1980s and 1990s). Thus, the unit of analysis in the test is the country-year for CO2 emissions per capita and the country-decade for the rate of deforestation. For deforestation, the right-hand-side variables, except for the dichotomous

Conflict and the Environment

285

war variables, take on their decade-average values to capture the crosssectional patterns. The second issue concerns the presence of possible heteroskedastic error variance. As in previous chapters, we deal with this issue by estimating Huber–White robust standard errors for both models of the CO2 emissions per capita and models of the rate of deforestation. Third, we consider the possibility of path dependence for CO2 , for we have pooled data for this variable. For deforestation, as noted, we have a crosssectional sample, which makes path dependence a nonissue on technical grounds. CO2 emissions per capita are expected to exhibit path dependence because the CO2 -generating industrial facilities, energy equipment, and transportation devices depreciate slowly. In the prevailing technological paradigm, even when old units are replaced, new units generate emissions. Thus, a high level of CO2 emissions per capita in one period is likely to be followed by high emissions in the next period. As discussed in Chapter 2, if not modeled explicitly, path dependence in the dependent variable is likely to cause serial correlation in the error term. A more refined modeling of this inertia is outside the scope of this chapter. We return to this issue in Chapter 10. The fourth issue concerns the possibility that our models omit some global and national structural variables that might affect the environment, subjecting our results to an omitted variable bias. The inclusion of the LDV accounts for some of the omitted variable bias. We further guard against this possibility in the CO2 models by using the country fixedeffects estimator and the year variable. The country fixed effects control for excluded country-specific characteristics (e.g., climate, regional atmospheric integrity, institutional setup, geography); year accounts for the possibility that CO2 emissions rise over time due to economic growth or a declining ability of the atmosphere to break up emissions over time (as the stock of emissions rises). Although these extra features make it harder for us to find statistically significant effects, we prefer to err on the side of caution. For deforestation, because our sample includes only two cross sections, we employ ordinary least squares without having to apply these additional controls. Fifth, we lag the right-hand-side variables in the CO2 models for two reasons. The effects of our covariates on the environment may take some time to materialize. Also, whereas our design treats the right-hand-side variables as exogenous, they may be affected by environmental degradation. For example, CO2 emissions may correlate with environmental laws that affect

286

Democracy and Economic Openness in an Interconnected System

the real GDP per capita, and environmental degradation, as noted, may cause violent conflict. As we explained in previous chapters, many scholars often lag the right-hand-side variables to control for possible simultaneity, and we do so as well. Sixth, we need to consider the risks of multicollinearity and, for CO2 , nonstationarity. As noted, when the effects of key right-hand-side variables are insignificant and the model R2 value is high, the insignificance can reflect multicollinearity, which inflates the standard errors of the coefficient estimates. Nonstationary dependent variables and error terms may cause spurious results. We employ the variance inflation factor (VIF) statistic to diagnose the problem of multicollinearity, and we conduct the Levin– Linard and Im–Pesaran–Shin panel unit root tests to diagnose the problem of nonstationarity.

Empirical Findings We first evaluate if the model suffers from the presence of multicollinearity and if the error term has good properties. As noted, multicollinearity is a cause of concern when effects of key variables are statistically insignificant but the model goodness of fit is very high. The average VIF is very high (28.7), but it is caused by the country fixed-effects dummy variables, which is to be anticipated. The individual VIFs for the variables are below 3.5, much lower than the conventional threshold of 10, with the exceptions of real GDP per capita and its squared term. This result is expected and in fact cannot be helped if one wants to model the EKC. The results obtained from Levin–Linard and Im–Pesaran–Shin panel unit root tests indicate that the issue of nonstationary is not a cause of concern in our case. In Table 9.1, the coefficients of war at home or abroad in Model C1 and of war abroad and war at home in Model C2 are negative and significant. These results are discussed in the main text. Note that the effect is statistically significant even in the presence of the empirically taxing country fixed effects. One may wonder why war reduces CO2 emissions per capita if real GDP per capita is included in the model. Recall that the effect of war may work through changing output, affecting resource allocation and sectoral breakdown, and destroying the economy. Controlling for the output effect, we find that the net effect of war due to the second and third channels is negative. A sectoral analysis seems desirable but is outside the scope of this chapter. We will return to this point. For the sizes of the effects of war in Models C1 and C2 we report changes relative to a base case. We compute the size of the effect as the percent

Conflict and the Environment

287

change in CO2 emissions per capita when the war variable changes from 0 to 1, with all the other variables in the model set at their sample mean values. On the basis of all estimated coefficients for the base case, we find that an involvement in a war reduces emissions by 1.14%, an involvement in war fought at home reduces emissions by 0.76%, and an involvement in wars fought abroad reduces emissions by 1.61%. On the basis of only the statistically significant coefficients for the base case (i.e., setting insignificant coefficients to zero), we find larger effects: an involvement in a war reduces the CO2 emissions per capita by 2.77%, an involvement in a war fought at home reduces emissions by 1.68%, and an involvement in a war fought abroad reduces emissions by 3.56%. As discussed in the main text, these effects do not tell the whole story. In Chapter 2, we computed long-term impact via the LDV. Using that approach, the long-run impact of war on CO2 emissions per capita is given by [coefficient of war/(1 – coefficient of lagged emissions)] × change in war, where the change in war is 1 (a change from 0 to 1). This computation indicates that the long-run effect on CO2 emissions per capita is 7.46 times larger than the short-run effect. In the long run, involvement in a war reduces emissions by 8.5%, an involvement in war fought at home reduces emissions by 5.67%, and an involvement in wars fought abroad reduces emissions by 12.01%. If we only use the significant coefficients in the computation, we get 20.66% reduction in emissions for any war, 12.53% for a war fought at home, and 26.56% for a war fought abroad. In Models C3 and C4, war at home or abroad, war at home, and war abroad do not affect CO2 emissions per capita for the LDC sample. These results are discussed in the main text. The results for the control variables in Table 9.1 are the same across all four models. These results are also consistent with theoretical expectations and the empirical results in previous studies of the determinants of CO2 emissions per capita. A rise in democracy reduces CO2 emissions per capita. The effect of the LDV is positive, indicating that CO2 emissions exhibit inertia. Rises in trade openness and year have no effect on CO2 emissions. A rise in population density raises CO2 emissions per capita. The coefficients of real gdppc and real gdppc squared indicate the existence of an EKC that turns down at 14,545 constant 1996 international dollars for the all-countries sample, and at 20,000 constant 1996 international dollars for the LDC sample. These turning points are in the range of the results reported in the literature. For example, Cole et al. (1997) reported a turning point of $25,100 for CO2 emissions per capita, and Moomaw and Unruh (1997) found a turning point of $18,333 for these emissions.

288

Democracy and Economic Openness in an Interconnected System

As for the deforestation models in Table 9.2, the R2 values are in the range 0.25–0.29, which indicates a reasonably good model fit for a cross-sectional design such as ours. This R2 value is smaller than that for CO2 because the deforestation model is cross-sectional due to data availability and does not include country fixed effects, year, and lagged environmentt−1 . The average VIF is 6.3 and individual VIFs are much below the threshold value of 10. Hence, multicollinearity is not a concern in this model. In Model D1, the effect of war at home or abroad on deforestation is statistically insignificant. In Model D2, the coefficient of war at home is negative and significant, whereas the coefficient of war abroad is positive and significant. We have discussed the interpretations in the main text. For the LDCs, Models D3 and D4 yield results that are similar to those for the full sample. As expected, the war coefficients appear considerably larger in size in Models D3 and D4 compared with those values in Models D1 and D2. To compute the sizes of effects for the war variables, we implement the same procedure as in the CO2 case and we discuss the results in the main text. The effects of the control variables fall within the range of possibilities discussed in the literature. A rise in democracy raises the rate of deforestation for all four models in Table 9.2. For the all-countries sample, the rate of deforestation exhibits a significant nonlinear pattern due to real GDP per capita, but not an EKC. The rate of deforestation falls as real gdppc rises; it then grows as real gdppc rises above a threshold. For the LDC sample, we also do not find an EKC for deforestation. Our results support those studies that did not find the EKC (e.g., Shafik and Bandyopadhyay, 1992; Barbier, 2001) and disagree with those that found the EKC effect (Bhattarai and Hammig, 2001). A rise in trade openness reduces the rate of deforestation, whereas a rise in population density increases the rate of deforestation.

Additional Analyses We first reestimate our models while assuming that the dependence of environmental degradation on real GDP per capita is cubic, not quadratic (as hypothesized by the EKC theory). Table 9.A1 presents results for CO2 emissions per capita (Models C5 and C6) and deforestation (Models D5 and D6). For CO2 emissions per capita, the results for the effects of war and control variables are similar to those in Table 9.1 in terms of signs, significance, and sizes, indicating the robustness of our CO2 models. The coefficients of some of the real GDP per capita terms differ from Table 9.1, as could be expected.

Conflict and the Environment

289

Table 9.A1. War, cubic GDP per capita, CO2 emissions per capita, and deforestation All countries

LDCs

All countries

(C5)

(C6)

(D5)

(D6)

−0.532∗∗ (0.278) 0.798∗∗ (0.412) −0.117∗∗∗ (0.035) 0.00018 (0.0001) 1.417e-08 (1.45e-08) −6.8145e-13∗∗ (4.07e-13) 0.012∗∗ (0.0067) −0.001∗ (0.0006)

−0.633∗∗ (0.291) 1.022∗∗∗ (0.452) −0.095∗∗∗ (0.033) −0.00008 (0.0002) 4.524e-08 (3.61e-08) −1.3530e-12 (1.25e-12) 0.011∗ (0.007) −0.001∗∗ (0.0006)

−1.975∗∗∗ (0.574) 204 0.277

−1.478∗∗∗ (0.489) 187 0.294

−0.027∗ (0.017) WAR ABROAD −0.051∗∗ (0.028) DEMOCRACY −0.0015∗ (0.001) REAL GDPPC 0.00012∗∗∗ (0.00002) REAL GDPPC SQUARED −5.82e-09∗∗∗ (1.57e-09) REAL GDPPC CUBED 7.85e-14∗∗ (4.16e-14) TRADE OPENNESS −0.0010 (0.001) POPULATION DENSITY 0.001∗∗∗ (0.0003) 0.866∗∗∗ LAGGED CO2 (0.023) YEAR 0.001 (0.001) Constant −1.163 (2.129) Observations 3830 0.99 R2 WAR AT HOME

−0.021 (0.017) 0.005 (0.020) −0.0027∗∗∗ (0.001) 0.00010∗∗∗ (0.00004) −4.87e-09 (5.25e-09) 9.69e-14 (2.04e-13) −0.0002 (0.001) 0.001∗∗ (0.0003) 0.850∗∗∗ (0.034) 0.001 (0.001) −1.345 (2.214) 3053 0.99

LDCs

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

For the all-countries sample (Model C5), the coefficient of real GDP per capita is positive and statistically significant and the coefficient of squared real GDP per capita is negative and statistically significant, as in Table 9.1. For the LDCs sample (Model C6), the coefficient of squared real GDP per capita is also negative, but insignificant. The coefficient of the cubic real GDP per capita term is positive in both samples but is significant only for the all-countries sample. Model C5 shows that the dependence of CO2 emissions per capita on real GDP per capita may be cubic for the all-countries sample. The emissions increase with real GDP per capita up to a local maximum attained at $14,481, after which they decline with real GDP per capita to a local minimum

290

Democracy and Economic Openness in an Interconnected System

Table 9.A2. War and CO2 emissions per capita, omitting the lagged dependent variable All countries (C7) WAR AT HOME OR ABROAD WAR AT HOME

−0.124∗∗∗ (0.035)

WAR ABROAD −0.0102∗∗∗ (0.003) REAL GDPPC 0.0008∗∗∗ (0.00003) REAL GDPPC SQUARED −1.89e-08∗∗∗ (1.13e-09) TRADE OPENNESS −0.0032∗∗ (0.002) POPULATION DENSITY 0.0033∗∗∗ (0.0004) YEAR 0.005∗∗ (0.002) Constant −9.426∗∗ (3.79) Observations 3838 0.964 R2

DEMOCRACY

(C8)

−0.084∗∗ (0.036) −0.161∗∗ (0.064) −0.0103∗∗∗ (0.003) 0.0008∗∗∗ (0.00003) −1.89e-08∗∗∗ (1.13e-09) −0.0031∗ (0.0016) 0.0033∗∗∗ (0.0004) 0.004∗∗ (0.002) −8.779∗∗ (3.81) 3838 0.964

LDCs (D7) −0.047 (0.028)

−0.0166∗∗∗ (0.0028) 0.00059∗∗∗ (0.00005) −6.55e-09∗∗ (2.62e-09) −0.0003 (0.0017) 0.0019∗∗∗ (0.0006) 0.007∗∗∗ (0.002) −13.91∗∗ (4.07) 3061 0.951

(D8)

−0.060∗∗ (0.031) −0.043 (0.041) −0.0165∗∗∗ (0.0028) 0.0006∗∗∗ (0.00005) −6.64e-09∗∗ (2.63e-09) −0.0003 (0.0017) 0.0019∗∗∗ (0.0006) 0.007∗∗∗ (0.002) −14.07∗∗∗ (4.08) 3061 0.951

Note: Standard errors in parentheses. ∗ significant at 10%; ∗∗ significant at 5%; ∗∗∗ significant at 1%.

attained at $35,481, and then they rise again with real GDP per capita ad infinitum. Model C6 suggests that the cubic relationship does not hold for the CO2 emissions per capita in the LDC sample. Models D5 and D6 present the results from including a cubic real GDP per capita term for deforestation. The coefficients of war and control variables are similar to those in Table 9.2, indicating (again) the robustness of our models. The coefficients of the real GDP per capita variables change, which is expected. The coefficients of real GDP per capita and real GDP per capita squared are not statistically significant for either sample. The coefficient of the cubic term is only statistically significant for the all-countries sample. Models D5 and D6 indicate that the dependence of the rate of deforestation on real GDP per capita does not follow a cubic relationship. Finally, Table 9.A2 presents results from estimating the models of CO2 emissions per capita in Table 9.1 without the LDV. We find that the effects

Conflict and the Environment

291

of the war variables agree with those in Table 9.1 in terms of their signs and significance, except that now the effect of war fought at home is statistically significant for the LDC sample. The effects of war are now larger in size than the effects in the presence of the LDV. This result is expected because the LDV absorbs some of the variations to be explained by other variables in the model; when it is excluded, this variance is allocated to other variables on the right-hand side. The results for the effects of the control variables also agree with those in Table 9.2, except that now a rise in year has a significant and positive effect across the board.

TEN

Conclusion

We are approaching the end of the intellectual journey we mapped out at the beginning of our book. Looking back, we have addressed a number of important questions, crossed various disciplinary boundaries, engaged numerous academic and policymaking controversies, and uncovered theoretical and empirical linkages among various literatures and topics, some of which have previously received little attention. The preceding eight chapters have delved into a wide range of issues that pertain to the causes or effects of democratic governance, economic openness, income inequality, economic development, military violence, and environmental degradation. In each chapter, we have focused on specific parts of the big picture, holding other things in the system constant, and have generated new knowledge, which we believe should be very useful to both academics and policy makers. In this chapter, our goal is to bring together our key findings and draw an elaborate big picture of some of the complex transformations faced in our world today. This approach forces us to ask broader questions and cross chapter boundaries. The insights we thus generate go beyond our findings from the individual chapters. As we relax the ceteris paribus assumption underlying each chapter’s analysis and connect the dots across the different chapters, we see that more processes interact and more disciplinary boundaries break down, producing new policy trade-offs and solutions. In this chapter, we seek to accomplish three tasks. Our first task is to take stock of key findings from the various analyses we have conducted. Instead of simply listing these findings, we fit them into the big picture we employed in Chapter 1 to describe our conceptual framework. This approach provides a visual display of the connections of the key findings in Chapters 2–9. Our second task in this chapter is to offer caveats regarding our analyses and suggest general directions for future research. Taking a broad view, this manuscript is based on a certain epistemology, research methodology, 292

Conclusion

293

and theoretical thinking. Our investigation is essentially part of an infinite process of pursuing new knowledge. Thus, instead of treating our findings as the end point, we view our research as part of a larger, perpetual human endeavor to acquire broader and deeper understanding of sociopolitical– economic interactions between people and between a human being and nature. In this spirit, we highlight possible themes for future research. Our third task in this chapter is to integrate and synthesize our key findings and distill broad policy implications that transcend those identified in Chapters 2–9. Our goal is to reveal new linkages, trade-offs, and solutions. This undertaking is only possible if we place the findings from the different chapters within a larger, unified conceptual framework. TAKING STOCK OF KEY FINDINGS

In Chapter 1, we developed Figure 1.1 as an intellectual device to motivate and organize the social scientific inquiry that was about to unfold in this book. In taking stock of the key findings from our many analyses, we return to this useful device but flesh it out to create Figure 10.1, which documents our findings and illustrates their interconnectedness. In doing so, Figure 10.1 distinguishes among different types of economic openness to the world economy – trade, foreign direct investment (FDI), and portfolio investments – and different types of environmental degradation – CO2 emissions, NOx emissions, land degradation, water pollution, and deforestation. Figure 10.1 also marks the sign of each causal arrow to indicate that a particular cause–effect relationship is statistically different from no effect and its direction. The figure also takes note of the statistically significant nonlinear and reciprocal effects identified by our analyses. In short, whereas Figure 1.1 provided us with an analytical roadmap at the beginning of our long intellectual journey, Figure 10.1 showcases the social scientific insights we have collected along our way. Let us first take a look at the role of six forces included in Figure 10.1. Five of these forces were dependent variables in different chapters (i.e., we studied their determinants). Democracy was a dependent variable in Chapters 2, 4, and 5; income inequality in Chapter 3; economic development in Chapter 4; military conflict in Chapters 5 and 6; different types of environmental degradation were the dependent variables in Chapters 7, 8, and 9; and economic openness was one of the key independent variables in Chapters 2, 3, 4, 6, and 8. Within this context, we can group our key findings according to the dependent variables studied, integrate results obtained from the different

294

+

–

Democracy

+

Economic Development

–

de Tra

– for trade, portfolio + for FDI

– –

– on deforestation by war abroad

– for trade + for FDI x for portfolio

Income Inequality

Military Conflict +

+ or –, depending on trade sector

Economic Openness

+ on deforestation by war at home

– on CO2 by all wars

+ on forest in democracy

– on land – on deforestation in autocracy

–

+

Inverted U for LDCs

Figure 10.1. Graphical layout of key findings.

Environmental Degradation

+ on deforestation and rising in open economy democracies

– on CO2, NOx, water, land

+

Inverted U for CO2, NOx, Land Degration

+ +

Conclusion

295

parts of the book, and point out the chapter in which they are based. In doing so, we focus on the key determinants, not on all the variables included in each of our models. As we shall see, our key findings suggest a number of uneasy trade-offs for policymakers who seek to meet multiple objectives.

Determinants of Democracy Figure 10.1 demonstrates that the level of democracy in a country is affected by a number of forces that act in different directions. FDI, economic development, and the diffusion of democratic ideas promote democracy. A rise in FDI flows into a country increases its level of democracy, but this positive effect declines over time and approaches zero (Chapter 2). A rise in economic development increases the level of democracy, supporting the modernization thesis (Chapters 2 and 4). The flow of democratic ideas into a country continuously increases its level of democracy (Chapters 2 and 4). Democracy, however, faces challenges posed by the forces of trade, portfolio investment, income inequality, and military conflict, as indicated in Figure 10.1. Increasing trade and portfolio investments reduce the level of democracy in a country. The negative effect of portfolio investments on democracy grows over time (i.e., further reduces the level of democracy in a country as we progress with time; see Chapter 2). Countries embroiled in interstate military disputes can expect to see their levels of democracy decrease (Chapter 6). As the level of income inequality in a country rises, its level of democracy declines (Chapter 2). Finally, the level of democracy in a country exhibits path dependence; the current level of democracy is positively affected by its own previous value. In other words, democracy is a relatively slow-changing force, suggesting that attempts to change its level “overnight” may not be successful due to entrenched interests seeking to sustain the political status quo (Chapters 2, 4, and 5).

Determinants of Income Inequality Like democracy, the level of income inequality in a country is influenced by a number of factors (Chapter 3). As shown in Figure 10.1, a rise in the level of trade openness of a country reduces its income inequality. Income inequality also falls as the level of democracy in a country rises, suggesting a positive role for democracy if the goal is to increase egalitarianism. In contrast, a rise in FDI flows into a country raises its income inequality, implying an uneasy trade-off between attracting FDI and achieving greater income equality.

296

Democracy and Economic Openness in an Interconnected System

The positive effect of FDI on inequality, however, is attenuated in more democratic countries. The effect of economic development on inequality is complex; it follows a nonlinear, inverted U-shaped curve. As a country develops economically and its income per capita rises, the level of its income inequality rises until its income per capita reaches a fairly high threshold of about $6,500. After this income per capita threshold is crossed, further increases in the level of economic development in a country reduce its level of income inequality. Like democracy, income inequality also exhibits path dependence, where the current value of inequality is positively associated with its previous value. This tendency reflects two primary forces. Rich and powerful actors tend to reject public policies that redistribute wealth from the rich to the poor and tend to support policies that increase their own share of the national pie. In addition, poor people face economic, educational, social, and political obstacles when they try to rise out of poverty; these forces create income inequality inertia, working against abrupt changes.

Determinants of Economic Development As shown in Figure 10.1, the level of economic development in a country is influenced by competing forces, including political regime type, trade openness, and path dependence (Chapter 4). Starting with negative influences, the level of economic development declines as the level of democracy in a country rises. This finding suggests that democracy is not optimal for promoting economic development, which challenges the belief of many policymakers. In contrast, the level of economic development rises with trade openness. This finding supports the claim of neoclassical economics that trade is the engine of economic growth. However, as we shall see in the next subsection, this positive effect also entails a cost. Like democracy and income inequality, development also exhibits inertia or persistence over time. Economic development changes slowly over time due to two types of determinants that tend to change slowly over time. One type of determinant includes class structure and institutional characteristics such as the rule of law and the extent and enforcement of private property rights. A second type includes determinants that essentially cannot be drastically changed unless some extraordinary events such as natural disasters or conquests by other countries occur (e.g., geography, climate, resource endowments). Thus, the finding that economic development exhibits inertia suggests that poor and rich nations are likely to maintain their current identities well into the future.

Conclusion

297

Determinants of Interstate Military Conflict Figure 10.1 demonstrates that interstate military conflict is subject to competing influences from democracy (Chapter 5) and trade (Chapter 6). A rise in the level of democracy shared by the two countries in a dyad reduces the probability of dyadic conflict, supporting the so-called democratic-peace hypothesis. With this result in mind, two things are worth noting. First, the finding is dyadic, not monadic. Although the United States and Denmark, for example, are both democracies, the former clearly engages in more military conflicts than the latter. Yet our findings suggest that the United States and Denmark are less likely to engage in military conflict than the United States and Iran. Second, this finding is probabilistic, not deterministic, indicating that countries are less likely to exhibit military disputes when their level of joint democracy rises, not that they will never engage in military interstate disputes. Chapter 6 suggests that the effect of bilateral trade on interstate military conflict is much more complex than has been realized thus far in the literature. Figure 10.1 shows that the effect of increases in bilateral trade on the likelihood of dyadic military conflict varies across bilateral imports, exports, and economic sectors. This finding refutes general statements, which are often made in the literature in international relations, that “trade causes peace” or “trade causes conflict.” We find that rises in the flows of agriculture/fishery and energy imports reduce the probability of military conflict initiation in a dyad, whereas rises in the flows of energy export and manufactured products increase the probability of conflict. Military conflict, like our other dependent variables, exhibits path dependence. Dyads that have been at peace more years are less likely to exhibit military conflict in the present. This finding suggests that countries may develop trust in one another following some long period of peace and may then learn to resolve their conflicts without resorting to military force.

Determinants of Environmental Degradation Analyses of the determinants of environmental degradation have produced our most complex findings, as Figure 10.1 demonstrates. Democracy, trade, military conflict, and economic development produce varying effects on environmental degradation, depending on the type of environmental degradation studied. As we shall see, this finding implies that the policy trade-offs surrounding environmental issues are also going to be complex. Specifically, a rise in democracy reduces the levels of CO2 and NOx emissions per capita in a country, the share of organic pollution in water,

298

Democracy and Economic Openness in an Interconnected System

and the share of land degraded in a country. However, a rise in democracy also speeds up deforestation in the country (Chapter 7). If one expects the spread of democracy to benefit the environment, it is important to take into account the trade-off between deforestation and the other aspects of environmental degradation, because forests play an important role in sustaining the environmental integrity of the biosphere. A rise in trade openness generates even more complicated effects on environmental degradation than democracy, depending on the type of degradation and the country’s political regime. A rise in trade reduces deforestation in autocracies and land degradation in both democracies and autocracies. However, a rise in trade leads to faster deforestation in democracies and causes higher NOx emissions per capita in both democracies and autocracies (Chapters 7 and 8). Involvement in interstate and civil wars, regardless of whether the interstate war is fought at home or abroad, lowers CO2 emissions per capita and intensifies land degradation. A war fought at home raises the rate of deforestation at home, whereas a war fought abroad reduces this rate (Chapter 9). Of course, we do not argue that war should be used as a policy instrument for promoting environmental quality in a country, but the fact that war may perversely slow these types of environmental degradation demonstrates the complexity of our empirical findings. The effect of an increase in the level of economic development on environmental degradation is often nonlinear. CO2 and NOx emissions per capita and land degradation exhibit a nonlinear, inverted U-shaped curve that rises with income per capita until reaching a turning point and then falls. The turning points are about $18,000 for CO2 , $25,000 for NOx , and $4,000 for land degradation. Because most LDCs are far below these income levels, our findings imply that environmental degradation will continue to rise for years to come as these countries continue to develop. Finally, CO2 emissions per capita and water pollution exhibit inertia, suggesting that these types of environmental degradation do not decline abruptly (Chapter 7).

Caveats and Future Research We have uncovered a number of important and interesting interconnections among forces of broad interest, but the quality of our analysis invariably depends on the conditions related to our current epistemology, research methodology, data availability, and theoretical thinking. Therefore, we regard the journey we have undertaken in this book as part of an ongoing research enterprise.

Conclusion

299

In our view, our research can be fruitfully extended in three directions. First, one could expand the empirical coverage of our samples, construct alternative empirical indicators, collect new data, and apply new estimation techniques. Our empirical analyses may benefit from getting data for more years and countries, even though our theoretical arguments and findings ought not to be sample-dependent. For example, one might collect more data and redo our environmental analyses that employed a crosssectional design. It is also worthwhile to study other indicators of globalization such as migration flows and information technology and flows. Future research on income inequality may benefit from employing alternative development indicators, such as the Human Development Indicator, or alternative inequality indicators such as access to medical services and food. Further research on trade and conflict could benefit from a more refined sectoral trade breakdown, as well as from better bilateral trade price and quantity data used for estimating parameters in our theoretical model. Finally, more refined modeling of democracy and conflict dynamics in a context of a simultaneous equations model could also be beneficial, which requires techniques that are currently unavailable in the case of our model. Second, because many of the relationships we study involve multiple causal mechanisms that operate in parallel, future research may seek to assess the validity of each specific mechanism. For example, whereas we modeled the net effect of economic openness on democracy, future research may attempt to calibrate the relative importance of the channels driving this net effect. The same idea applies to the net effect of democracy on the environment, the net impact of trade on the environment, and the net effects of democracy and development on each other. These assessments should help us evaluate the strength of the competing arguments by facilitating the accumulation of theoretical and empirical knowledge. Third, we have connected many topics that so far have been largely studied separately, but we have not considered all the possible interconnections. For example, we did not analyze the impact of FDI on the environment. We treated economic openness as exogenous, but extant research suggests that it is affected by political forces such as democratic governance, military conflict, economic forces such as development, and environmental forces such as resource endowments. Future research may also go beyond the topics we covered in this book. For example, one may study the impact of labor mobility across countries on income inequality. Finally, new linkages may emerge from unpacking aggregate concepts. For example, one could study whether different types of democracies, such as presidential or

300

Democracy and Economic Openness in an Interconnected System

parliamentary systems, or different categories of autocracies, perform differently in promoting environmental quality. We believe future research in these three general directions could produce interesting and useful insights for both academics and policymakers. More important, the suggested research extensions reflect the spirit running throughout our book. That is, to understand the impact of complex transformations in our world today, we have to cross disciplinary boundaries, integrate previously ignored connections, and create a relatively more realistic, albeit more complex, bigger picture. As existing scholarly pursuits become more and more specialized, the lessons they generate tend to become increasingly narrow, because scholars tend to view their own relationship of interest as the first-order effect and to brush aside other interconnections as negligible second-order effects. The danger we all face in taking this approach is that we fail to see the possibility that a useful lesson based on a narrow context may imply costly trade-offs in a broader context. We believe our efforts in this book help reveal some of the inherent tensions, complexities, and interconnectedness that operate among our forces. These interactions, illustrated in Figure 10.1, involve direct, indirect, competing, and nonlinear interconnections; they demonstrate various tensions and trade-offs among competing objectives such as economic development versus environmental quality, economic development versus equitable income distribution, prodevelopment economic liberalization versus prodemocracy political liberalization, and pro-openness economic liberalization versus environmental quality. Further demonstrating and explaining these tensions, complexities, and interconnectedness is our final task in this book. TENSIONS, COMPLEXITIES, AND INTERCONNECTEDNESS

The preceding two sections summarized our key findings and suggested future research. This section integrates our findings from a public policy perspective. The reader may recall that the last section of each chapter discussed implications for public policy, under the assumption that policymakers sought to achieve only one or two goals. For example, in Chapter 2 we assumed that policymakers seek to promote democracy, and in Chapter 4 we assumed they merely seek to promote both democracy and economic development. In reality, policymakers often seek to pursue multiple goals at the same time. In the parlance of Figure 10.1, we assume that these concurrent goals include raising the level of democracy, encouraging economic

Conclusion

301

development, promoting a more equitable income distribution, reducing various types of environmental degradation, and strengthening national security. In general, the competing effects included in Figure 10.1 suggest that policymakers who seek to meet several goals at the same time will experience policy tensions and have to confront trade-offs due to the interconnectedness among various forces in the international system of polity, economy, and environment. A cursory look at our system indicates that as the number of goals increases, a discussion of the policy implications often becomes very complicated. For our purpose, we can illustrate the gist of this argument by considering several relatively simple examples in which policymakers seek to pursue only two goals at a time. However, before we discuss these examples is a word of caution: As we will see, even these seemingly simple examples generate a fair amount of complexity, which cannot be helped. In other words, complexity is perhaps the most pronounced feature of the international system of polity, economy, and environment studied in this book. DEMOCRACY AND EQUITY

Figure 10.1 shows that increasing democracy reduces income inequality, and reducing income inequality increases democracy. Because the two goals support each other, there seems to be no tension, but democracy and equity also interact indirectly with other forces in our system. Increasing democracy reduces economic development. A rise in economic development, in turn, increases income inequality if countries are located to the left of the turning point of the inverted U-shaped curve linking development to inequality. Consider next the effect of economic openness on the relationship between democracy and income inequality. A rise in trade openness reduces income inequality and democracy. However, a rise in FDI openness increases both these forces. Thus, increasing both trade openness and FDI openness at the same time presents tension. Increasing portfolio investment openness seems benign, because reducing it promotes democracy without raising inequality. However, portfolio investment can finance the accumulation of physical capital, facilitating economic development. Thus, reducing portfolio investment, which promotes democracy, may reduce economic development. Reducing economic development, in turn, undermines democracy and raises income inequality if the country is located to the left of the turning point in the inverted U-shaped curve linking development to inequality.

302

Democracy and Economic Openness in an Interconnected System

DEMOCRACY AND DEVELOPMENT

We find that a rise in development promotes democracy, whereas a rise in democracy reduces development. Hence, leaders seeking development may have to delay democratization, at least for a while. Promoting development, however, can also conflict with the goal of increasing democracy through the channel of income inequality. If a country is located to the left of the turning point on the inverted U-shaped curve that links development and inequality, a rise in development will raise income inequality, which, in turn, will reduce democracy. Consider now the effect of economic openness on this relationship. Neoclassical economists have long hailed free trade as an engine of development. Our results support this argument, but we also find that a rise in trade reduces democracy, our other goal in this example. Thus, again, we face an unintended policy trade-off. Next, recall our findings that FDI liberalization promotes democracy, whereas a rise in democracy reduces development. This combination suggests that FDI liberalization indirectly impedes our ability to meet the goals of democracy and development at the same time. Moreover, although not studied here (and itself a subject of debate), some studies argue that FDI promotes economic development. The rise in development, in turn, raises democracy, but the rise in democracy may reduce development – suggesting, once again, a policy trade-off. DEVELOPMENT AND EQUITY

The goals of development and equity are also at tension. Leaders seeking equity and development face a dilemma: inequality rises with development until an income per capita of about $6,500 is reached and then falls with further increases in income per capita. Because the per capita income of LDCs is typically much smaller than $6,500, their income inequality is expected to rise as these countries develop. Democracy further complicates the relationship. As inequality rises, democracy falls. This decline may, in turn, increase development. However, a rise in development will increase inequality if income per capita is less than $6,500. Once income per capita grows greater than $6,500, a rise in development reduces inequality. As inequality falls, democracy rises, but policymakers are still not entirely out of the woods. As democracy rises, the level of development falls, which may reduce income per capita below the $6,500 threshold and may return the nation back to the income range

Conclusion

303

in which development increases inequality, which in turn reduces the level of democracy. Turning to openness in Figure 10.1, a rise in trade increases development and reduces inequality, supporting both goals. However, trade liberalization can indirectly increase income inequality through its positive effect on development. The effect of development on inequality, as noted, is nonlinear, increasing inequality until an income per capita of about $6,500 is attained. Thus, development will increase inequality in most LDCs (as in China, for example). Moreover, if trade liberalization promotes a rise in FDI, which is often observed to be the case, the net effect of the increases in the levels of trade and FDI openness on inequality becomes unclear, because a rise in FDI increases inequality and a rise in trade reduces it. ENVIRONMENTAL QUALITY AND DEMOCRACY

In this example, we consider leaders who pursue democracy and environmental quality. Figure 10.1 suggests that a rise in democracy reduces all types of environmental degradation that we studied, other than the rate of deforestation. Let us assume, for the moment, that leaders ignore deforestation, coveting more the goals of democracy and environmental quality in those other areas. Our analysis suggests that this approach is likely to fail. Deforestation, we know, promotes climate change (forests absorb CO2 emissions), damages fresh water supply (forests absorb rainwater), and degrades lands (forests prevent land erosion). These damages may progress slowly for a while, but they do accumulate and eventually grow large. Thus, ignoring the detrimental effect of democracy on deforestation is to postpone the need to deal with the issue. This strategy, however, carries the risk of generating nonreversible outcomes such as degrading land to the point of becoming permanently unsuitable for agriculture. To further complicate things, we find that growing trade openness increases deforestation in a democracy, but reduces it an autocracy. Thus, one might be tempted to delay democratization in countries facing severe deforestation, but our results suggest that this strategy would increase other types of environmental degradations (because a rise in democracy reduces all the other types of environmental degradation we studied). ENVIRONMENTAL QUALITY AND DEVELOPMENT

Countries often pursue both development and environmental quality. However, the two goals often conflict with each other. As shown in Figure 10.1,

304

Democracy and Economic Openness in an Interconnected System

the effect of development on environmental degradation follows an inverted U-shaped curve. As a country develops, environmental degradation first rises and only later falls. We find the turning-point income levels for the different environmental attributes that we studied are relatively high and are much larger than the incomes of most, if not all, of the LDCs, where the large majority of the world’s population resides. Thus, given the current technological paradigm, policymakers need to decide which goal is more important – promoting development or improving environmental quality. If they choose development first, they will likely see their environment degrade for years. Next, we find that trade liberalization promotes economic development. However, this may not benefit the environment. The effect of a rise in trade on environmental degradation depends on the environmental aspect examined and the political regime type. In democracies, trade liberalization intensifies deforestation, which degrades the environment in the long run by reducing fresh water supplies, increasing land erosion, and increasing CO2 stocks, which accelerate global warming and climate change. DEMOCRACY AND NATIONAL SECURITY

The link between democracy and national security is also not simple. If leaders choose to use military force (assuming leaders are rational), we can assume they deem it to be beneficial for national security. Under this reading, Figure 10.1 suggests tension in attaining the goals of promoting both democracy and national security. We find that a rise in joint democracy reduces the likelihood of military conflict, whereas military conflict reduces democracy on both sides of a dyad. With this in mind, consider the scenario in which leaders seek to promote democracy while pursuing national security by resorting to force. The goals of national security and democracy will clash as conflict reduces democracy. This decline, in turn, may reduce the level of joint democracy, making conflict more likely and further undermining democracy. TRADE AND SECURITY

One may also consider the decision to liberalize trade to bring about peace, but this approach may fail as well. Our results suggest that leaders should scrutinize the composition and pattern of their trade with potential foes, because a rise in some sectoral trade flows makes military conflict more likely. If peace is a desirable component of national security, bilateral trade

Conclusion

305

in some economic sectors, with some countries, may need to be regulated and even reduced. Meanwhile, as trade openness rises, democracy declines. This process may cause the level of joint democracy to decline and make military conflict more likely. ENVIRONMENTAL QUALITY AND NATIONAL SECURITY

Let us now consider the interaction between environmental quality and national security. In this example, we assume that leaders seek to promote both national security and environmental quality. The pursuit of these goals can generate policy tension. Depending on where the war is fought and the particular type of environmental degradation considered, war may harm or benefit a country’s environment. Military conflict, however, can also harm or benefit the environment indirectly through its effect on democracy. Military conflict, we recall, reduces democracy, which, in turn, affects the environment. Thus, environmental quality and national security may become competing goals, suggesting a potential policy trade-off. DEMOCRACY AND ECONOMIC OPENNESS REVISITED

Finally, consider tensions that arise from the desire to pursue goals pertaining to the two forces that stand at the center of our book: democratic governance and economic openness. In recent decades, many policymakers have sought to promote both economic and political liberalization, assuming the two activities improve national welfare that is broadly defined to rise with increases in development, equality, environmental quality, and national security. Figure 10.1 suggests the risks associated with promoting democracy and economic openness at the same time. Democracy is good for the environment, except for deforestation. Democracy also favors equitable income distribution. These two components increase national welfare. However, a rise in democracy also reduces development – another component of social welfare. Moreover, if countries are on the right side of the turning points on the inverted U-shaped curve linking development to inequality and on the inverted U-shaped curve linking development to environmental degradation, a fall in development will increase income inequality and environmental degradation, both of which reduce national welfare. Promoting economic openness as a goal also has competing effects on welfare. Part of the problem is due to the fact that different components of economic openness (e.g., export, import, sectoral trade, FDI, portfolio

306

Democracy and Economic Openness in an Interconnected System

investments) exert competing effects on income equality, national security, and environmental quality. Moreover, economic openness affects development (the fourth component of welfare) both positively, as in neoclassical economics, and negatively, working through its negative effect on democracy. The development component, in turn, has implications for both the equality and environmental quality components of welfare. In all, pursuing both economic liberalization and political liberalization may not necessarily increase a multidimensional metric of national welfare. The overall costs of pursuing the two goals together may not outweigh their benefits. Casual observations suggest that political and economic liberalizations have benefited DCs, but many LDCs have not been as fortunate. In fact, some countries have given up pursuing at least one of these goals. For example, during the Cold War, a number of countries rejected both goals, though their experiment failed. China has chosen to promote development but to delay democratization. Russia has considerably scaled back its democracy in recent years and has concentrated on economic development. Malaysia has decided to reduce its economic openness by imposing capital flow restrictions. South Korea and Taiwan developed quickly but did so behind high trade and currency barriers and a fair amount of autocracy. Taken together, these stylized facts, as well as some of the two-goal examples discussed earlier, suggest that a one-size-fits-all policy of simultaneously promoting political and economic liberalizations, which is often widely advocated, may not always be optimal. One should note, however, that we do not intend to suggest that the goals of promoting democracy and economic openness themselves are not worthy of pursuing. Rather, we argue that the inherent complexities and tensions we demonstrated in this book may produce some unintended dismal outcome for the simultaneous pursuit of political and economic liberalization. FINAL WORDS

This book has demonstrated the growing complexity of formulating public policies to pursue multiple economic, social, political, and environmental goals in an interconnected system. However, the overall picture is even more complicated than the one we painted. Because all countries face these very choices, and they may not all value the various conflicting goals in the same way, policy coordination among countries adds yet another layer of complexity – one we have not considered in this book. The collective-action

Conclusion

307

problem entrenched in the anarchy of the international system makes the national-level policymaking endeavor all the more demanding and challenging, requiring continuous adjustments in the process of meeting one’s own goals. Leaders are often hard pressed to come up with solutions, and they often claim to have found solutions. However, the truth of the matter is there really is no silver bullet. Leaders interested in simultaneously pursuing several goals in the system depicted by Figure 10.1 essentially have two options. First, they may have to prioritize and make some tough choices – deciding to pursue goal X at the expense of sacrificing goal Y. Second, in pursuing a particular policy goal, leaders may have to adopt certain policy measures to minimize the unintended side effects of their effort. For example, when they open up the economy to FDI, leaders may want to compensate firms and people hurt by the foreign entry to reduce income inequality that rises with FDI. Similarly, if they remove barriers to financial capital flows, they may have to strengthen regulation to prevent excessive speculation that leads to financial crises, social strife, state repression, and a decline in democracy. Taking a broader view, one may wonder whether the Westphalian state system is up to the task of formulating effective public policies in an interconnected world. Even if policymakers are well informed about all the competing considerations and trade-offs, they may not necessarily be able to achieve a first best solution, for they may each prioritize conflicting goals differently. This observation suggests that as the world system becomes more and more interconnected, the pressures for coordinated global governance or even for forming some sort of a world government may grow; but we believe that this outcome is unlikely anytime soon. In the end, current policymakers have to increasingly confront the daunting challenges and trade-offs we have demonstrated and those others we have not analyzed here, which their predecessors only two or three decades ago did not have to grapple with. We have arrived at the end of our analytical journey. We hope our readers have benefited from the many messages offered by this book. But if we had to choose only one message we would like our readers to remember about our book, we think it would have to be the following. A tendency exists among many social scientists to emphasize specialization in a certain discipline and to strive for parsimonious theories to explain reality. As we have demonstrated in this book, however, a disciplinary focus leads to partial, and often erroneous, understandings of how the world works because it ignores potentially important processes emphasized by other disciplines.

308

Democracy and Economic Openness in an Interconnected System

As the global system of polity, economy, and environment continues to become increasingly interconnected, the demands on national leaders and their advisors will only increase, requiring them to cross disciplinary boundaries in order to understand and address the policy trade-offs of even more complex and interrelated processes than the ones studied here. This is the time of Complex Transformations.

References

Achen, Christopher. 1982. Interpreting and Using Regression. Beverly Hills, CA: Sage. Ahluwalia, Montek S., Nicholas G. Carter, and Hollis B. Chinaware. 1979. “Growth and Poverty in Developing Countries.” Journal of Development Economics 6(3): 299–341. Aitken, Brian, Ann Harrison, and Robert E. Lipsey. 1996. “Wages and Foreign Ownership: A Comparative Study of Mexico, Venezuela, and the United States.” Journal of International Economics 40(3): 345–371. Alderson, Arthur S., and Francois Nielsen. 1999. “Income Inequality, Development, and Dependence: A Reconsideration.” American Sociological Review 64(4): 606–631. Alesina, Alberto, and Dani Rodrik. 1994. “Distributive Politics and Economic Growth.” The Quarterly Journal of Economics 109(2): 465–490. Allan, Nigel J. R. 1987. “Impact of Afghan Refugees on the Vegetation Resources of Pakistan’s Hindukush-Himalaya.” Mountain Research and Development 7(3): 200– 204. Allen, Leslie, Stephen L. Harris, Catherine Herbert Howell, K. M. Kostyal, and Rick Sammon. 1998. The Wonder of the World. Washington, DC: National Geographic Society. Almond, Gabriel Abraham. 1990. A Discipline Divided: Schools and Sects in Political Science. Beverly Hills, CA: Sage. Alvarez, R. Michael. 1997. Endogeneity in Choice Models. Division of the Humanities and Social Sciences. California Institute of Technology. http://www.hss.caltech.edu/ rma/home.html (June 3, 2000). Alvarez, R. Michael, and Garrett Glasgow. 2000. “Two-Stage Estimation of Nonrecursive Choice Models.” Political Analysis 8(2): 147–165. Amemiya, Takeshi. 1967. “The Estimation of Balestra-Nerlove Models.” Technical Report No. 4, Institute for Mathematical Studies in the Social Sciences, Stanford University. Amemiya, Takeshi. 1978. “The Estimation of a Simultaneous Equation Generalized Probit Model.” Econometrica 46(5): 1193–1205. Amin, Samir. 1996. “The Challenge of Globalization.” Review of International Political Economy 3(2): 216–259. Anderson, T. W, and Cheng Hsiao. 1982. “Formulation and Estimation of Dynamic Models Using Panel Data.” Journal of Econometrics 18(1): 47–82. 309

310

References

Angell, Norman. 1912. The Great Illusion: A Study of the Relation of Military Power in Nations to Their Economic and Social Advantages. London: Heinemann. Antweiler, Werner, Brian R. Copeland, and M. Scott Taylor. 2001. “Is Free Trade Good for the Environment?” American Economic Review 91(4): 877–908. Arad, Ruth W., Seev Hirsch, and Alfred Tovias. 1983. The Economics of Peacemaking: A Focus on the Egyptian-Israeli Situation. New York: St. Martin’s. Arellano, Manuel, and Stephen Roy Bond. 1991. “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations.” Review of Economic Studies 58(2): 277–297. Armijo, Leslie Elliott. 1998. “Mixed Blessing: Foreign Capital Flows and Democracy in Emerging Markets.” Presented at the Annual Meeting of International Studies Association, Minneapolis, MN, March. Ashley, Richard. 1980. The Political Economy of War and Peace: The Sino-Soviet-American Triangle and the Modern Security Problematique. London: Frances Pinter. Baldwin, David. 1985. Economic Statecraft. Princeton, NJ: Princeton University Press. Baran, Paul A. 1973. The Political Economy of Growth. Harmondsworth, UK: Penguin. Barbier, Edward B. 2001. “The Economics of Tropical Deforestation and Land Use: An Introduction to the Special Issue.” Land Economics 77: 155–171. Barbieri, Katherine. 2002. The Liberal Illusion: Does Trade Promote Peace? Ann Arbor, MI: University of Michigan Press. Barnet, Richard J., and John Cavanagh. 1994. Global Dreams: Imperial Corporations and the New World Order. New York: Simon & Schuster. Barrett, Scott, and Kathryn Graddy. 2000. “Freedom, Growth, and the Environment.” Environment and Development Economics 5: 433–456. Barro, Robert J. 1996. “Democracy and Growth.” Journal of Democracy 1: 1–27. Barro, Robert J. 1998. Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge, MA: MIT Press. Barro, Robert J. 2000. “Recent Developments in Endogenous Growth Theory.” In The Determinants of Economic Growth, edited by M. S. Oosterbaan, Thijs De Buyter Van Stevenick, and N. Van Der Windt, 13–38. Boston: Kluwer Academic. Barro, Robert J., and Jong-Wha Lee. 2000. “International Data on Educational Attainment: Updates and Implications.” Manuscript, Harvard University. Batra, Geeta, and Hong Tan. 1997. Malaysia: Enterprise Training, Technology and Productivity. Washington, DC: World Bank. Beck, Nathaniel, and Jonathan N. Katz. 1995a. “Nuisance vs. Substance: Specifying and Estimating Time-Series-Cross-Section Models.” Political Analysis 6: 1–34. Beck, Nathaniel, and Jonathan N. Katz. 1995b. “What to Do (and Not to Do) with Time-Series Cross-Section Data.” American Political Science Review 89: 634– 647. Beck, Nathaniel, and Jonathan N. Katz. 2004. “Time Series-Cross Section Issues: Dynamics.” Manuscript, Department of Political Science, New York University, and Division of the Humanities and Social Sciences, California Institute of Technology. Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. 1998. “Taking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable.” American Journal of Political Science 42: 1260–1288.

References

311

Beck, Ulrich. 1996. “Kapitalismus Ohne Arbeit.” Der Spiegel 20: 140–146. Bennett, D. Scott, and Allan Stam. 2000a. “Research Design and Estimator Choices in the Analysis of Interstate Dyads: When Decisions Matter.” Journal of Conflict Resolution 44(October): 653–685. Bennett, D. Scott, and Allan Stam. 2000b. “EUGene: A Conceptual Manual.” International Interactions 26: 179–204. Bennett, D. Scott, and Allan Stam. 2004. The Behavioral Origins of War. Ann Arbor, MI: University of Michigan Press. Bennett, Marlin R., Alan Rosenblatt, and Jianxin Wang. 1992. “The Visible Hand: The United States, Japan, and the Management of Trade Disputes.” International Interactions 17: 191–213. Berge, Erling. 1994. “Democracy and Human Rights: Conditions for Sustainable Resource Utilization.” In Who Pays the Price? The Sociocultural Context of Environmental Crisis, edited by B. R. Johnson, 187–193. Covelo, CA: Island Press. Bhagwati, Jagdish. 1994. “Globalization, Sovereignty and Democracy.” In Democracy’s Victory and Crisis: Nobel Symposium, edited by A. Hadenius, 263–281. Cambridge, UK: Cambridge University Press. Bhagwati, Jagdish. 2000. The Wind of the Hundred Days: How Washington Mismanaged Globalization. Cambridge, MA: MIT Press. Bhattarai, Madhusudan, and Michael Hammig. 2001. “Institutions and the Environmental Kuznets Curve for Deforestation: A Cross-Country Analysis for Latin America, Africa and Asia.” World Development 29(6): 995–1010. Bienen, Henry. 1971. The Military Modernization. Chicago: University of Chicago Press. Birdsall, Nancy. 1998. “Life is Unfair: Inequality in the World.” Foreign Policy 112(Summer): 76–83. Blanchard, Olivier. 2000. Macroeconomics. Upper Saddle River, NJ: Prentice Hall. Bliss, Harry, and Bruce Russett. 1998. “Democracy Trading Partners: The Liberal Connection, 1962–1989.” Journal of Politics 60: 1126–1147. Blomstrom, Magnus, and Ari Kokko. 1996. “The Impact of Foreign Investment on Host Countries: A Review of the Empirical Evidence.” Policy Research Paper 1745, World Bank, Washington, DC. Blomstrom, Magnus, Robert E. Lipsey, and Mario Zejan. 1992. “What Explains Developing Country Growth.” NBER Paper 4132, National Bureau of Economic Research, Cambridge, MA. Boix, Carles. 1998. Political Parties, Growth and Equality: Conservative and Social Democratic Economic Strategies in the World Economy. Cambridge, UK: Cambridge University Press. Boix, Carles, and Susan C. Stokes. 2003. “Endogenous Democratization.” World Politics 55: 517–549. Boli, John, and George M. Thomas. 1999. “INGOs and the Organization of World Culture.” In Constructing World Culture: International Nongovernmental Organizations Since 1875, edited by J. Boli and G. M. Thomas, 13–50. Stanford, CA: Stanford University Press. Bollen, Kenneth A. 1979. “Political Democracy and the Timing of Development.” American Sociological Review 44: 572–587. Bollen, Kenneth A. 1983. “World System Position, Dependency, and Democracy: The Cross-National Evidence.” American Sociological Review 48: 468–479.

312

References

Bollen, Kenneth A. 1991. “Political Democracy: Conceptual and Measurement Traps.” In On Measuring Democracy, edited by Alex Inkeles, 3–20. New Brunswick, NJ: Transaction. Bollen, Kenneth A., David K. Guilkey, and Thomas A. Mroz. 1995. “Binary Outcomes and Endogenous Explanatory Variables: Tests and Solutions with an Application to the Demand for Contraceptive Use in Tunisia.” Demography 32(1): 111–131. Bollen, Kenneth A., and Robert W. Jackman. 1985. “Political Democracy and the Size Distribution of Income.” American Sociological Review 50(4): 438–457. Bond, Marian E. 1985. “Export Demand and Supply for Groups of Non-Oil Developing Countries.” International Monetary Stuff Papers 32: 56–77. Borenszstein, Eduardo, Jose DeGregorio, and Jong-Wha Lee. 1994. “How Does Foreign Direct Investment Affect Economic Growth?” IMF Paper WP/94/110, International Monetary Fund, Washington, DC. Borjas, George J. 1987. “Immigrants, Minorities and Labour Market Competition.” Industrial and Labour Relations Review 40(3): 382–392. Borjas, George J. 1990. Friends or Strangers: The Impact of Immigration on the US Economy. New York: Basic Books. Borjas, George J. 1993. “The Impact of Immigrants on Employment Opportunities of Natives.” In The Changing Course of International Migration, 191–198. Paris: Organisation for Economic Cooperation and Development. Borner, Silvio, Aymo Brunetti, and Beatrice Weder. 1995. Political Credibility and Economic Development. New York: Macmillan. Bornschier, Volker, and Christopher Chase-Dunn. 1985. Transnational Corporations and Underdevelopment. New York: Praeger. Boron, Atilio. 1998. “Faulty Democracies? A Reflection on the Capitalist ‘Fault Lines’ in Latin America.” In Fault Lines of Democracy in Post-Transition Latin America, edited by Felipe Aguero and Jeffrey Stark, 41–66. Miami: North-South Centre Press at the University of Miami. Borrus, Michael, and John Zysman. 1992.“Industrial Competitiveness and American National Security.” In The Highest Stakes: The Economic Foundations of the Next Security System, edited by Wayne Sandholtz, Michael Borrus, John Zysman, Ken Conca, Jay Stowsky, Steven Vogel, and Steve Weber, 7–54. New York: Oxford University Press. Brack, Duncan. 1996. International Trade and the Montreal Protocol. London: Royal Institute of International Affairs and Earthscan Publications. Bremer, Stuart A. 1993. “Democracy and Militarized Interstate Conflict, 1816–1965.” International Interactions 18: 231–249. Brunn, Stanley D., and Thomas R. Leinback. 1991. Collapsing Space and Time: Geographic Aspects of Communications and Information. London: HarperCollins Academic. Bruno, Giovanni S. F. 2005. “Approximating the Bias of the LSDV Estimator for Dynamic Unbalanced Panel Data Models.” Economics Letters 87(3): 361–366. Bryan, Lowell, and Diana Farrell. 1996. Market Unbound. New York: John Wiley. Bueno de Mesquita, Bruce. 1981. The War Trap. New Haven, CT: Yale University Press. Bueno de Mesquita, Bruce, and George W. Downs. 2005. “Development and Democracy.” Foreign Affairs 84: 77–86.

References

313

Bueno de Mesquita, Bruce, James D. Morrow, Randolph M. Siverson, and Alastair Smith. 1999. “An Institutional Explanation of the Democratic Peace.” American Political Science Review 93: 920–941. Bulte, Erwin H., and Daan P. Van Soest. 2001. “Environmental Degradation in Developing Countries: Households and the (Reverse) Environmental Kuznets Curve.” Journal of Development Economics 65: 225–235. Bun, Maurice, and Jan F. Kiviet. 2003. “On the Diminishing Returns of Higher Order Terms in Asymptotic Expansions of Bias.” Economics Letters 79: 145–152. Burkhart, Ross E., and Michael S. Lewis-Beck. 1994. “Comparative Democracy: The Economic Development Thesis.” American Political Science Review 88: 903–910. Bush, George W. 2001. Remarks by the President to the Council of the Americas. Washington, D.C.: The White House. May 7. http://www.whitehouse.gov/news/ releases/2001/05/20010507–6.html. Bussmann, Margit, Indra de Soysa, and John R. Oneal. 2005. “The Effect of Globalization on National Income Inequality.” Comparative Sociology 4 (3–4): 285–312. Buzan, Barry. 1984. “Economic Structure and International Security: The Limit of the Liberal Case,” International Organization 20: 5–41. Cameron, Rondo. 1997. A Concise Economic History of the World: From Paleolithic Times to the Present. New York: Oxford University Press. Cammack, Paul. 1998. “Globalization and the Death of Liberal Democracy.” European Review 6: 249–263. Chan, Steve. 1997a. “In Search of Democratic Peace: Problems and Promise.” Mershon International Studies Review 41(1): 59–91. Chan, Steve. 1997b. “Democracy and Inequality: Tracking Welfare Spending in Singapore, Taiwan, and South Korea.” In Inequality, Democracy and Economic Development, edited by Manus I. Midlarski, 227–243. Cambridge, UK: Cambridge University Press. Chan, Sylvia. 2002. Liberalism, Democracy and Development. Cambridge, UK: Cambridge University Press. Chirot, Daniel. 1977. Social Change in the Twentieth Century. New York: Harcourt Brace Jovanovich. Choucri, Nazli, and Robert North. 1975. Nations in Conflict: National Growth and International Violence. New York: Freeman. Clauge, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson. 1996. “Property and Contract Rights in Autocracies and Democracies.” Journal of Economic Growth 1(2): 243–276. Clinton, William J. 1996. Between Hope and History. New York: Random House. CNTS. 2008. Cross-National Time Series Data Archive, 1815–2003. Binghamton, NY: Databanks International. http://www.databanksinternational.com/. Coe, David T., Elhanan Helpman, and Alexander W. Hoffmaister. 1994. “North-South Research and Development Spillover.” IMF Paper WP/94/144. International Monetary Fund, Washington, DC. Cohen, Patricia. 2007. “Does Capitalism Lead to Democracy, and How? ” International Herald Tribune, June 13, London. Cohen, Yousseff. 1985. “The Impact of Bureaucratic-Authoritarian Rule on Economic Growth.” Comparative Political Studies 18: 123–136.

314

References

Colaresi, Michael, and William R. Thompson. 2003. “The Economic DevelopmentDemocratization Relationship: Does the Outside World Matter?” Comparative Political Studies 36: 381–403. Cole, Matthew A., Anthony J. Rayner, and John M. Bates. 1997. “The Environmental Kuznets Curve: An Empirical Analysis.” Environment and Development Economics 2: 401–416. Congleton, Roger D. 1992. “Political Institutions and Pollution Control.” Review of Economics and Statistics 74: 412–421. Cox, Robert W. 1996. “Globalization, Multilateralism, and Democracy.” In Approaches to World Order, edited by Robert W. Cox and Timothy J. Sinclair, 524–536. New York: Cambridge University Press. Cox, Robert W. 1997. “Democracy in Hard Times: Economic Globalization and the Limits to Liberal Democracy.” In The Transformation of Democracy? Globalization and Territorial Democracy, edited by Anthony McGrew, 49–72. Malden, MA: Blackwell. Crescenzi, Mark J. 2003. “Economic Exit, Interdependence, and Conflict: An Empirical Analysis.” Journal of Politics 65(3): 809–832. Crescenzi, Mark J., and Andrew J. Enterline. 1999. “Ripples from the Waves? A Systemic, Time Series Analysis of Democracy, Democratization and Interstate War.” Journal of Peace Research 36(1): 75–94. Crutsinger, Martin. 2001. “Environmental Review OK-ed.” Associated Press, April 21. Dahl, Robert A. 1989. Democracy and Its Critics. New Haven, CT: Yale University Press. Dahl, Robert A. 1994. “From Immigrants to Citizens: A New Yet Old Challenge to Democracies.” Paper presented at the International Conference on the 21st Century and Democracy, Seoul, South Korea. Dailami, Mansoor. 2000. “Financial Openness, Democracy, and Redistributive Policy.” World Bank Policy Research Paper #2372, World Bank, Washington, DC. Dasgupta, Partha. 1993. An Inquiry into Well Being and Destitution. Oxford: Clarendon Press. Dasgupta, Partha S. 1995. “Population Poverty and the Local Environment.” Scientific American 272(2): 40–46. Databanks International. 2008. Cross-National Time Series Data Archive, 1815–2003. Binghamton, NY: Databanks International. http://www.databanks.sitehosting.net/. Davidson, Russell, and James G. MacKinnon. 1993. Estimation and Inference in Econometrics. New York: Oxford University Press. de Bruyn, S. M., J. C. J. M. Van Den Bergh, and J. B. Opschoor. 1998. “Economic Growth and Emissions: Reconsidering the Empirical Basis of Environmental Kuznets Curves.” Ecological Economics 25: 161–175. de Montesquieu, Baron [1748] 1900. The Spirit of Laws. New York: Collier. de Soysa, Indra, and Eric Neumayer. 2005. “False Prophet, or Genuine Savior? Assessing the Effects of Economic Openness on Sustainable Development, 1980–1999.” International Organization 59(3): 731–772. de Tocqueville, A. [1835] 1969. Democracy in America, translated by G. Lawrence. Vol. 1–2, edited by J. P. Mayer. Garden City, NY: Anchor Books. Deacon, Robert T. 1994. “Deforestation and the Rule of Law in a Cross Section of Countries.” Land Economics 70: 414–430. Dean, Judy. 2002. “Does Trade Liberalization Harm the Environment? A New Test.” Canadian Journal of Economics 35: 819–842.

References

315

Deininger, Klaus, and Lyn Squire. 1996. “A New Dataset Measuring Income Inequality.” World Bank Economic Review 10: 565–591. Deininger, Klaus, and Lyn Squire. 1998. “New Ways of Looking at Old Issues: Inequality and Growth.” Journal of Development Economics 57(2): 259–288. Diamond, Jared. 2005a. Collapse: How Societies Choose to Fail and Survive. London: Penguin. Diamond, Jared. 2005b. “Tokugawa Shoguns vs. Consumer Democracy.” New Perspective Quarterly 22: 1–6. Diamond, Larry. 1992a. “Economic Development and Democracy Reconsidered.” In Reexamining Democracy, edited by Gary Marks and Larry Diamond, 93–139. Beverly Hills, CA: Sage. Diamond, Larry. 1992b. “Economic Development and Democracy Revisited.” American Behavioral Scientist 35: 450–499. Diamond, Larry. 1995. “Democracy and Economic Reform: Tensions, Compatibilities, and Strategies for Reconciliation.” In Economic Transition in Eastern Europe and Russia: Realities of Reform, edited by Edward P. Lazear, 107–158. Stanford, CA: Hoover Institution Press. Diamond, Larry. 1999. Developing Democracy: Toward Consolidation. Baltimore: Johns Hopkins University Press. Diamond, Larry, Juan J. Linz, and Seymour M. Lipset. 1987. “Building and Sustaining Democratic Government in Developing Countries: Some Tentative Findings.” World Affairs 150: 5–19. Diamond, Larry, and Marc F. Plattner, eds. 1994. “Economic Reform and Democracy: Special Edition.” Journal of Democracy 5. Diaz-Alejandro, Carlos. 1975. “North-South Relations: The Economic Component.” International Organization 29: 213–241. Dinda, Soumyananda. 2004. “Environmental Kuznets Curve Hypothesis: A Survey.” Ecological Economics 49: 431–455. Dixon, William J. 1994. “Democracy and the Peaceful Settlement of International Conflict.” American Political Science Review 88(1): 14–32. Dixon, William J., and Terry Boswell. 1996. “Dependency, Disarticulation, and Denominator Effects: Another Look at Foreign Capital Penetration.” American Journal of Sociology 102(2): 543–562. Dixon, William, and Bruce Moon. 1993. “Political Similarity and American Foreign Trade Partners.” Political Research Quarterly 46: 5–25. Dollar, David, and Aart Kraay. 2000. “Growth is Good for the Poor.” Development Research Group Papers, World Bank, Washington, DC. Domke, William K. 1988. War and the Changing Global System. New Haven, CT: Yale University Press. Dorussen, Han. 2006. “Heterogenous Trade Interests and Conflict: What You Trade Matters.” Journal of Conflict Resolution 50(1): 87–107. dos Santos, Theotonio. 1970. “The Structure of Dependence,” American Economic Review 80: 1151–1161. Doyle, Michael. 1997. Ways of War and Peace: Realism, Liberalism and Socialism. New York: Norton. Drake, Paul W. 1998. “The International Causes of Democratization, 1974–1990.” In The Origins of Liberty: Political and Economic Liberalization in the Modern World, edited

316

References

by Paul W. Drake and Mathew D. McCubbins, 70–91. Princeton: Princeton University Press. Drucker, Peter F. 1994. “The Age of Social Transformation.” Atlantic Monthly November, 53–80. Dryzek, John S. 1987. Rational Ecology: Environment and Political Economy. Oxford: Blackwell. Easterly, William. 1999. “Life During Growth.” Journal of Economic Growth 4(3): 239– 275. The Economist. 1994. “Democracy and Growth: Why Voting is Good For You. ” August 27. The Economist. 1995. “Southeast Asian Wars and Timber.” June 17: 35. The Economist. 2003. “The Global Menace of Local Strife,” May 24: 23–25. Environmental Performance Measurement Project. 2002. Environmental Sustainability Index. http://sedac.ciesin.columbia.edu/es/esi/ESI2002˙21MAR02tot.pdf. FAO (Food and Agriculture Organization). 2000. World Soil Resource Report: Land Resource Potential and Constraints at Regional and Country Levels. Land and Water Development Division, Food and Agriculture Organization, United Nations, Rome. FAO. 2002. TERRASTAT. Land and Plant Nutrition Management Service, Food and Agriculture Organization, United Nations, Rome. http://www.fao.org/ag/agl/agll/terrastat (April 5, 2009). FAO. 2006. TERRASTAT. Land and Plant Nutrition Management Service. Food and Agriculture Organization, United Nations, Rome. http://www.fao.org/ag/agl/agll/ index.stm (May 3, 2006). Farber, Henry S., and Joanne Gowa. 1995. “Polities and Peace.” International Security 23: 123–146. Fearon, James D. 1994. “Domestic Political Audiences and the Escalation of International Disputes.” American Political Science Review 88(September): 577–594. Fearon, James, and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” American Political Science Review 97: 75–90. Feenstra, Robert C. 2000. “World Trade Flows.” Center for International Data, Institute of Governmental Affairs, University of California, Davis, and National Bureau of Economic Research, Cambridge, MA. Feenstra, Robert C., and Hanson Gordon. 1995. “Foreign Direct Investment and Relative Wages: Evidence from Mexico’s Maquiladoras.” NBER Paper 5122. Cambridge, MA: National Bureau of Economic Research. Feng, Yi. 1997. “Democracy, Political Stability and Economic Growth.” British Journal of Political Science 27: 391–418. Feng, Yi. 2003. Democracy, Governance, and Economic Performance: Theory and Evidence. Cambridge, MA: MIT Press. Feng, Yi, and Paul J. Zak. 1999. “The Determinants of Democratic Transitions.” Journal of Conflict Resolution 43(2): 162–177. Ferreira, Francisco. 1999. Inequality and Economic Performance: A Brief Overview to Theories of Growth and Distribution. Washington, D.C.: World Bank. http://www. worldbank.org/poverty/inequal/index.htm (November 7, 2000). Findlay, Ronald. 1990. “The New Political Economy: Its Explanatory Power for the LDCs.” Economics and Politics 2: 193–221.

References

317

Flandreau, Mar. 2007. “Do Democracy and Globalization Go Hand in Hand? The Bayesian Heresy.” http://www.bayesianheresy.blogspot.com/2007/01/doesdemocracy-and-globalization-go.html (January 3, 2007). Forest Watch Indonesia and Global Forest Watch. 2002. The State of the Forest: Indonesia. Bogot, Indonesia, and Washington DC: Forest Watch Indonesia and Global Forest Watch. Førland, Tor Egil. 1991. “Economic Warfare and Strategic Goods: A Conceptual Framework for Analyzing COCOM.” Journal of Peace Research 28: 191–204. Frankel, Jeffrey A. 2003. “The Environment and Globalization.” NBER Working Paper 10090. Cambridge, MA: National Bureau of Economic Research. Freedom House. 1999. Freedom in the World. New York: Freedom House. Frieden, Jeffry A., and Ronard Rogowski. 1996. “The Impact of the International Economy on National Policies: An Analytical Overview.” In Internationalization and Domestic Politics, edited by Robert O. Keohane and Helen V. Milner, 25–47. Cambridge, UK: Cambridge University Press. Friedman, Milton. 1962. Capitalism and Freedom. Chicago: University of Chicago Press. Fruhling, Hugo. 1998. “Judicial Reform and Democratization in Latin America.” In Fault Lines of Democracy in Post-Transition Latin America, edited by Felipe Aguero and Jeffrey Stark, 237–262. Miami: North South Centre Press at the University of Miami. Fukuyama, Francis. 1992. The End of History and the Last Man. New York: Free Press. Garrett, Geoffrey. 1999. “Global Markets and National Politics: Collision Course or Virtuous Circle?” International Organization 52(4): 787–824. Gartzke, Erik. 1998. “Kant We All Just Get Along?: Motive, Opportunity, and the Origins of the Democratic Peace.” American Journal of Political Science 42: 1–27. Gartzke, Erik. 2000. “Preferences and the Democratic Peace.” International Studies Quarterly 44: 191–210. Gartzke, Erik, and Quan Li. 2003a. “Measure for Measure: Concept Operationalization and the Trade Interdependence-Conflict Debate.” Journal of Peace Research 40: 555– 573. Gartzke, Erik, and Quan Li. 2003b. “All is Well that Ends Well: A Reply to Oneal, Barbieri and Peters,” Journal of Peace Research 40: 727–732. Gartzke, Erik, and Quan Li. 2005. “Mistaken Identity: A Reply to Hegre.” Conflict Management and Peace Science 22: 225–234. Gartzke, Erik, Quan Li, and Charles Boehmer. 2001. “Investing in the Peace: Economic Interdependence and International Conflict.” International Organization 55: 339–438. Gasiorowski, Mark J. 1986. “Economic Interdependence and International Conflict: Some Cross National Evidence.” International Studies Quarterly 30: 23–38. Gasiorowski, Mark J. 1995. “Economic Crisis and Political Regime Change: An Event History Analysis.” American Political Science Review 89: 882–897. Gasiorowski, Mark J. 1997. “Political Regimes and Industrial Wages: A Cross-National Analysis.” In Inequality, Democracy and Economic Development, edited by Manus Midlarski, 244–267. Cambridge, UK: Cambridge University Press. Gasiorowski, Mark J., and Solomon W. Polachek. 1982. “Conflict and Interdependence: East-West Trade and Linkages in the Era of Detente.” Journal of Conflict Resolution 26: 709–729.

318

References

Gates, Scott. 2002. “Empirically Assessing the Causes of Civil War.” Paper presented at the Annual Meeting of the International Studies Association, New Orleans, March 24–27. Gates, Scott, Torbjørn L. Knutsen, and Jonathon Moses. 1996. “Democracy and Peace: A More Skeptical View.” Journal of Peace Research 33: 1–10. Germain, Randall. 1997. The International Organization of Credit. Cambridge, UK: Cambridge University Press. Gill, Stephen. 1995. “Globalization, Market Civilization, and Disciplinary Neoliberalism.” Journal of International Studies 24: 399–423. Gilpin, Robert. 1981. War and Change in World Politics. New York: Cambridge University Press. Gilpin, Robert. 1984. “Structural Constraints on Economic Leverage: Market-Type Systems.” In Strategic Dimension of Economic Behavior, edited by Gordon H. McCormick and Richard E. Bissel, 105–128. New York: Praeger. Gilpin, Robert. 1987. The Political Economy of International Relations. Princeton, NJ: Princeton University Press. Gleditsch, Kristian S., and Michael D. Ward. 2000. “War and Peace in Space and Time: The Role of Democratization.” International Studies Quarterly 44: 1–29. Gleditsch, Nils Petter. 1995. “Geography, Democracy and Peace.” International Security 20: 297–323. Gleditsch, Nils Petter. 1998. “Armed Conflict and the Environment: A Critique of the Literature.” Journal of Peace Research 35: 381–400. Gleditsch, Nils Petter, and Bjorn O. Sverdlop. 2003. “Democracy and the Environment.” In Human Security and the Environment: International Comparisons, edited by Edward Paper and Michael Redclift, 45–70. London: Elgar. Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and H˚avard Strand. 2002. “Armed Conflict 1946–2001: A New Dataset.” Journal of Peace Research 39: 615–637. Gonick, Lev S., and Robert M. Rosh. 1988. “The Structural Constraints of the World Economy on National Political Development.” Comparative Political Studies 21: 171– 199. Goodell, Grace. 1985. “The Importance of Political Participation for Sustained Capitalist Development.” Archives Europeennes de Sociologie 26: 93–127. Goodell, Grace, and John P. Powelson. 1982. “The Democratic Prerequisites of Development.” In Freedom in the World: Political Rights and Civil Liberties, edited by Raymond Gastil, 167–176. New York: Freedom House. Goodin, Robert E. 1979. “The Development-Rights Trade-off: Some Unwarranted Economic and Political Assumptions.” Universal Human Rights 1: 31–42. Gore, Albert. 1992. Earth in the Balance: Ecology and the Human Spirit. New York: Houghton Mifflin. Gowa, Joanne. 1994. Allies, Adversaries and International Trade. Princeton, NJ: Princeton University Press. Gowa, Joanne, and Edward D. Mansfield. 1993. “Power Politics and International Trade.” American Political Science Review 87: 408–420. Graham, Edward, and Erika Wada. 2000. “Foreign Direct Investment in Mexico.” World Economy 20(6), 777–797. Gray, John. 1996. After Social Democracy. London: Demos.

References

319

Green, William H. 2007. Econometric Analysis. Upper Saddle River, NJ: Prentice Hall. Grieco, Joseph M. 1988. “Realist Theory and the Problem of International Cooperation: Analysis with Amended Prisoner Dilemma Model.” Journal of Politics 50: 600–624. Griffiths, William E., R. Carter Hill, and George G. Judge. 1993. Learning and Practicing Econometrics. New York: Wiley. Grossman, Gene, and Alan B. Krueger. 1993. “Environmental Impacts of a North American Free Trade Agreement.” In The Mexico-U.S. Free Trade Agreement, edited by Peter M. Garber, 13–56. Cambridge, MA: MIT Press. Guilkey, David K., Thomas A. Mroz, and David Taylor. 1992. “Estimation and Testing in Simultaneous Equations Models with Discrete Outcomes Using Cross Section Data.” Unpublished manuscript, University of North Carolina, Chapel Hill. Gujarati, Damodar N. 2002. Basic Econometrics. New York: McGraw-Hill/Irwin. Gurr, Ted Robert, and Keith Jaggers. 1999. Polity98 Project: Regime Characteristics 1800–1998. http://www.systemicpeace.org/polity/polity4.htm (April 5, 2009). Haddad, Mona, and Ann Harrison. 1993. “Are There Dynamic Externalities from Foreign Direct Investment?” In Market Structure, and Industrial Performance, edited by Claudio R. Frischtak and Richard S. Newfarmer, 160–272. London: Routledge. Haggard, Stephan. 1990. Pathways from Periphery: The Politics of Growth in the Newly Industrialized Countries. New York: Cornell University Press. Haggard, Stephan, and Robert R. Kaufman. 1995. The Political Economy of Democratic Transitions. Princeton: Princeton University Press. Harbaugh, William T., Arik Levinson, and David Molloy Wilson. 2002. “Reexamining Empirical Evidence for an Environmental Kuznets Curve.” NBER Working Paper 7711, Cambridge, MA: National Bureau of Economic Research. Hardin, Garrett. 1968. “The Tragedy of the Commons.” Science 162: 1243–1248. Harris, Jonathan M. 2006. Environmental and Natural Resource Economics: A Contemporary Approach. Boston: Houghton Mifflin. Hatzius, Jan. 1997. “Foreign Direct Investment, Capital Formation and Labor Costs: Evidence from Britain and Germany.” London School of Economics, Paper 336. London: Center for Economic Performance, London School of Economics. Hayek, Friedrich. 1944. The Road to Serfdom. Chicago: University of Chicago Press. Heckman, James J. 1978. “Dummy Endogenous Variables in Simultaneous Equation System.” Econometrica 46: 931–959. Heerink, Nico. 1994. Population Growth, Distribution, and Economic Development: Theory, Methodology and Empirical Results. Berlin: Springer-Verlag. Hegre, H˚avard. 2004. “Size Asymmetry, Trade, and Militarized Conflict.” Journal of Conflict Resolution 48: 403–429. Hegre, H˚avard. 2005. “Identifying How Trade Matters in Empirical Studies of Interstate Conflict.” Conflict Management and Peace Science 22(3): 217–224. Heilbronner, Robert L. 1974. An Inquiry into the Human Prospect. New York: Norton. Held, David. 1991. “Democracy, the Nation State and the Global System.” In Political Theory Today, edited by David Held, 197–235. Stanford, CA: Stanford University Press. Held, David. 1992. “Democracy: From City States to a Cosmopolitan Order?” Political Studies 40: 13–40. Held, David, and Anthony McGrew, eds. 2007. Globalization Theory: Approaches and Controversies. Cambridge, UK: Polity Press.

320

References

Held, David, Anthony McGrew, David Goldblatt, and Jonathan Perraton. 1999. Global Transformations: Politics, Economics, Culture. Cambridge, UK: Polity Press. Held, David, Anthony McGrew, David Goldblatt, and Jonathan Perraton. 2009. “Researching Globalization,” Global Transformations. http://www.polity.co.uk/global/ research.asp (March 29, 2009). Helliwell, John F. 1994. “Empirical Linkages Between Democracy and Economic Growth.” British Journal of Political Science 24: 225–248. Heston, Alan, Robert Summers, and Bettina Aten. 2002. Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP), October. Hewlett, Sylvia Ann. 1979. “Human Rights and Economic Realities: Tradeoffs in Historical Perspective.” Political Science Quarterly 94: 453–473. Hewlett, Sylvia Ann. 1980. The Cruel Dilemmas of Development: Twentieth Century Brazil. New York: Basic Books. Higgins, Matthew, and Jeffrey G. Williamson. 1999. “Explaining Inequality the World Round: Cohort Size, Kuznets Curves, and Trade Openness.” NBER Paper 7224. Cambridge, MA: National Bureau of Economic Research. Higgs, Robert. 1988. Crisis and Leviathan: Critical Episodes in the Growth of American Government. New York: Oxford University Press. Hillmann, H. C. 1940. “Analysis of Germany’s Foreign Trade and the War,” Economica 7: 66–88. Hirschman, Albert. [1945] 1980. National Power and the Structure of Foreign Trade. Berkeley, CA: University of California Press. Hirst, Paul. 1997. “The Global Economy: Myth and Realities.” International Affairs 73: 409–425. Hirst, Paul, and Grahame Thompson. 1996. Globalization in Question: The International Economy and the Possibilities of Governance. Cambridge, UK: Polity Press. Hobson, John A. 1902[1948]. Imperialism, A Study. London: Allen and Unwin. Hsiao, Cheng. 1997. “Statistical Properties of the Two Stage Least Squares Estimator Under Co-integration.” Review of Economic Studies 64: 385–398. Huber, Evelyne, Dietrich Rueschemeyer, and John D. Stephens. 1993. “The Impact of Economic Development on Democracy.” Journal of Economic Perspective 7: 71–85. Hufbauer, Gary. 2005. Economic Sanctions Reconsidered. Washington, DC: Institute for International Economics. Hufbauer, Gary, Daniel Esty, Diana Orejas, Luis Rubio, and Jeffrey Schott. 2000. “NAFTA and the Environment: Seven Years Later.” Policy Analysis in International Economics, No. 61, Institute for International Economics, Washington, DC. Hughes, Barry. 2000. Continuity and Change in World Politics: Competing Perspectives. Upper Saddle River, NJ: Prentice Hall. Human Development Report. 1999. United Nations Development Program. New York: Oxford University Press. Human Development Report. 2000. United Nations Development Program. New York: Oxford University Press. Huntington, Samuel. 1968. Political Order in Changing Societies. New Haven, CT: Yale University Press. Huntington, Samuel. 1984. “Will More Countries Become Democratic?” Political Science Quarterly 99: 193–218.

References

321

Huntington, Samuel. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman, OK: University of Oklahoma Press. Ichino, Andrea. 1993. “The Economic Impact of Immigration on the Host Country.” In Migration Policies in Europe and the United States, edited by Giacomo Luciani, 145–162. Dordrecht, Netherlands: Kluwer. Im, Hyug Baeg. 1987. “The Rise of Bureaucratic Authoritarianism in South Korea.” World Politics 38: 231–257. Im, Hyug Baeg. 1996. “Globalization and Democratization: Boon Companions or Strange Bedfellows?” Australian Journal of International Affairs 50: 279–291. Inter-American Development Bank. 1998. Facing Up to Inequality in Latin America. Washington, DC: Inter-American Development Bank. International Financial Statistics CD-Rom. 2000. Washington, DC: International Monetary Fund. IPCC. (Intergovernmental Panel on Climate Change). 2001. Special Report on The Regional Impacts of Climate Change: An Assessment of Vulnerability, Intergovernmental Panel on Climate Change. United Nations Environmental Programme and World Meteorological Organization. http://www.ipcc.ch/ (April 5, 2009). IPCC. 2007a. Climate Change 2007: The Physical Science Basis, Summary for Policymakers, Intergovernmental Panel on Climate Change. United Nations Environmental Programme and World Meteorological Organization. http://www.ipcc.ch/ (April 5, 2009). IPCC. 2007b. Climate Change 2007: Climate Change Impacts, Adaptation and Vulnerability, Summary for Policymakers, Intergovernmental Panel on Climate Change. United Nations Environmental Programme and World Meteorological Organization. http://www.ipcc.ch/ (April 5, 2009). Islam, Nazrul. 1995. “Growth Empirics: A Panel Data Approach.” Quarterly Journal of Economics 110: 1127–1170. Italianer, Alexander. 1986. Theory and Practice of International Trade Linkage Models. Boston: Kluwer. Jackman, Robert W. 1993. Power Without Force: The Political Capacity of Nation States. Ann Arbor, MI: University of Michigan Press. Jaggers, Keith, and Ted Robert Gurr. 1995. “Tracking Democracy’s Third Wave with the Polity III Data.” Journal of Peace Research 32: 469–483. James, P., E. Solberg, and M. Wolfson. 1999. “An Identified Systemic Model of the Democracy-Peace Nexus.” Defense and Peace Economics 10: 1–37. James, P., E. Solberg, and M. Wolfson. 2000. “Democracy and Peace: A Reply to Oneal and Russett.” Defense and Peace Economics 11: 215–229. Jenkins, Rhys. 1996. “Theoretical Perspectives on the Transnational Corporation.” In International Political Economy: State-Market Relations in the Changing Global Order, edited by C. R. Goddard, J. T. Passe-Smith, and J. G. Conklin, 439–459. Boulder, CO: Lynne Rienner. Johnston, Jack, and John DiNardo. 1997. Econometric Methods. New York: McGraw-Hill Companies. Jones, R. J. Barry. 1995. Globalization and Interdependence in the International Political Economy. London: Frances Pinter. Judge, George G., R. Carter Hill, William E. Griffiths, Helmut Lutkepohl, and TsoungChao Lee. 1988. Introduction to the Theory and Practice of Econometrics. New York: Wiley.

322

References

Judson, Katherine A., and Anne L. Owen. 1999. “Estimating Dynamic Panel Models: A Guide for Macroeconomists.” Economic Letters 65: 9–15. Justman, Moshe, and Mark Gradstein. 1999. “The Industrial Revolution, Political Transition, and the Decline in Inequality in 19th Century Britain.” Explorations in Economic History 36(2): 109–127. Kant, Immanuel. [1795] 1970. “Perpetual Peace: A Philosophical Sketch.” In Kant’s Political Writings, edited by Hans Reiss. Cambridge, UK: Cambridge University Press. Kant, Immanuel. [1795] 1983. Perpetual Peace and Other Essays on Politics, History, and Morals. Indianapolis, IN: Hackett. Kasza, Gregory J. 1996. “War and Comparative Politics.” Comparative Politics 35: 355– 373. Keck, Margaret E., and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy Networks in International Politics. Ithaca, NY: Cornell University Press. Kennedy, Paul. 1987. The Rise and Fall of the Great Powers: Economic Change and Military Conflict from 1500 to 2000. New York: Random House. Kennedy, Peter. 2005. A Guide to Econometrics. Cambridge, MA: MIT Press. Keshk, Omar M. G. 2002. Stata Program for Estimating Simultaneous Equations Model Using the Madalla Procedure. CDSIMEQ. Columbus, OH: Ohio State University. Keshk, O. M. G., B. M. Pollins and R. Reuveny (2004), “Trade Still Follows the Flag: The Primacy of Politics in a Simultaneous Model of Interdependence and Armed Conflict,” Journal of Politics, 66:1155–79. Keylor, William. 2001. The Twentieth Century World: An International History. Oxford: Oxford University Press. King, Dwight Y. 1981.“Regime Type and Performance.” Comparative Political Studies 13: 477–504. King, Gary. 2001. “Proper Nouns and Methodological Propriety: Pooling Dyads in International Relations Data.” International Organization 55(2): 497–507. King, Gary, and Langche Zeng. 2001. “Explaining Rare Events in International Relations.” International Organization 55: 693–715. Kiviet, Jan F. 1995. “On Bias, Inconsistency, and Efficiency of Various Estimators in Dynamic Panel Models.” Journal of Econometrics 68: 53–78. Kiviet, Jan F. 1999. “Expectation of Expansions for Estimators in a Dynamic Panel Data Model; Some Results for Weakly Exogenous Regressors.” In Analysis of Panel Data and Limited Dependent Variables, edited by Cheng Hsiao, M. Hashem Pesaran, Kajal Lahiri, and Lung Fei Lee, 199–295. Cambridge, UK: Cambridge University Press. Klein, Michael, Carl Aaron, and Bita Hadjimichael. 2001. “Foreign Direct Investment and Poverty Reduction.” Fiscal Policy Papers 2613. Washington, DC: World Bank. Kotov, Vladimir, and Elena Nikitina. 1995. “Russia and International Environmental Cooperation.” In Green Globe Yearbook of International Cooperation on Environment and Development, edited by H. O. Bergesen and G. Parmann, 17–27. Oxford: Oxford University Press. Kummell, Gerhard. 1998. “Democratization in Eastern Europe. The Interaction of Internal and External Factors: An Attempt at Systematization.” East European Quarterly 32: 243–267. Kurzer, Paulette. 1993. Business and Banking: Political Change and Economic Integration in Western Europe. Ithaca, NY: Cornell University Press.

References

323

Kurzman, Charles, Regina Werum, and Ross Burkhart. 2002. “Democracy Effect on Economic Growth: A Pooled Time-Series Analysis, 1951–1980.” Studies in Comparative International Development 37: 3–33. Kuznets, Simon. 1955. “Economic Growth and Income Inequality.” American Economic Review 45(1): 1–28. Lai, Brian, and Dan Slater. 2006. “Institutions of the Offensive: Domestic Sources of Dispute Initiation in Authoritarian Regimes, 1950–1992,” American Journal of Political Science 50: 113–126. Lal, Deepak. 1983. The Poverty of Development Economics. London: Institute of Economic Affairs. Lall, Sanjaya. 1974. “Less Developed Countries and Private Foreign Direct Investment: A Review Article.” World Development 2(1): 41–48. Lall, Sanjaya. 1985. Multinationals, Technology and Exports: Selected Papers. New York: St. Martin’s. Lappe, Frances, Joseph Collins, Peter Rosset, and Luis Esparza. 1998. World Hunger: Twelve Myths. 2nd Edition. New York: Grove Press. Lasswell, Harold D. 1977. On Political Sociology. Chicago: Chicago University Press. Lawrence, Robert. 1994. “Multinationals, Trade and Labor.” NBER Paper No. 4836, National Bureau of Economic Research, Cambridge, MA. Layne, Christopher. 1994. “Kant or Can’t: The Myth of the Democratic Peace.” International Security 19: 5–49. Leitz, Christian. 2004. Nazi Foreign Policy, 1933–1941: The Road to Global War. Oxford: Routledge. Lemke, Douglas, and William Reed. 1996. “Regime Types and Status Quo Evaluations: Power Transition Theory and The Democratic Peace.” International Interactions 22: 143–164. Lenin, Vladimir I. [1911] 1993. “Writings on the Commune.” In The Civil War in France: The Paris Commune, 91–132, New York: International Publishers. Lenin, Vladimir I. 1916. “Imperialism: A Special Stage of Capitalism.” In International Political Economy: Perspective on Global Power and Wealth, edited by Jeffry Frieden and David A. Lake, 110–119. New York: St. Martin’s. Lenski, Gerhard. 1966. Power in Privilege: A Theory of Social Stratification. New York: McGraw-Hill. Levy, Jack S. 1989. “Domestic Politics and War.” In The Origin and Prevention of Major Wars, edited by Robert I. Rotberg and Theodore K. Rabb, 88–108. Cambridge, UK: Cambridge University Press. Lewis, Peter. M. 1994. “Economic Statism, Private Capital and the Dilemmas of Accumulation in Nigeria.” World Development 22(3): 437–451. Li, Quan, and Adam Resnick. 2003. “Reversal of Fortunes: Democracy, Property Rights and Foreign Direct Investment Inflows in Developing Countries.” International Organization 57(1): 1–37. Li, Quan, and Rafael Reuveny. 2003. “Economic Globalization and Democracy: An Empirical Analysis.” British Journal of Political Science 33: 29–54. Li, Quan, and David Sacko. 2002. “The (IR)Relevance of Interstate Militarized Disputes to International Trade.” International Studies Quarterly 46: 11–44. Liang, Kung-Yee, and Scott L. Zeger. 1986. “Longitudinal Data Analysis Using Generalized Linear Models.” Biometrika 73: 13–22.

324

References

Lijphart, Arend. 1977. Democracy in Plural Societies: A Comparative Exploration. New Haven, CT: Yale University Press. Lin, Justin Y., and Jeffrey B. Nugent. 1995. “Institutions and Economic Development.” In Handbook of Development Economics, edited by Jere Behrman and T. N. Srinivasan, 2301–2370. Amsterdam: Elsevier. Lindblom, Charles E. 1977. Politics and Markets: The World’s Political-Economic Systems. New York: Basic Books. Lindert, Peter H. 1989. “Modern Fiscal Redistribution: A Preliminary Essay.” Agricultural History Paper Series, No. 55. University of California, Davis, CA. Lindert, Peter H. 1994. “The Rise of Social Spending.” Explorations in Economic History 31(1): 1–37. Linz, Juan J. 1978. The Breakdown of Democratic Regimes: Crisis, Breakdown, and Reequilibrium. Baltimore/London: Johns Hopkins University Press. Linz, Juan J. 1994.“Presidential or Parliamentary Democracy: Does It Make a Difference?” In The Failure of Presidential Democracy, edited by Juan J. Linz and Arturo Valenzuela, 3–87. Baltimore: Johns Hopkins University Press. Lipset, Seymour Martin. 1959. “Some Social Requisites of Democracy: Economic Development and Political Legitimacy.” American Political Science Review 53: 69–105. Lipset, Seymour Martin. 1981. Political Man: The Social Bases of Politics. Baltimore: Johns Hopkins University Press. Lipset, Seymour Martin. 1994. “The Social Requisites of Democracy Revisited.” American Sociological Review 59: 2–13. Lipset, Seymour Martin, Kyoung-Ryung Seong, and John Charles Torres. 1993. “A Comparative Analysis of the Social Requisites of Democracy.” International Social Science Journal 16: 155–175. Londregan, John B., and Keith T. Poole. 1996. “Does High Income Promote Democracy?” World Politics 49: 1–30. Longworth, Richard C. 1998. Global Squeeze: The Coming Crisis for First World Nations. Chicago: Contemporary Books. Lucas, Robert E. B., David Wheeler, and Hermamala Hettige. 1992. “Economic Development, Environmental Regulation and the International Migration of Toxic Industrial Pollution: 1960–1988.” In International Trade and the Environment, edited by Patrick Low, 67–86. Washington, DC: World Bank. MacDonald, Martha. 1991. “Post-Fordism and the Flexibility Debate.” Studies in Political Economy 36: 177–201. Maddala, G. S. 1983. Limited Dependent and Qualitative Variables in Econometrics. Cambridge, UK: Cambridge University Press. Mainwaring, Scott. 1993. “Presidential, Multipartisan, and Democracy: The Difficult Combination.” Comparative Political Studies 26: 198–228. Mankiw, N. Gregory, David Romer, and David D. Weil. 1992. “A Contribution to the Empirical of Economic Growth.” Quarterly Journal of Economics 107: 69–102. Mansfield, Edward D., and Rachel Bronson. 1997. “Alliances, Preferential Trading Agreements, and International Trade Patterns.” American Political Science Review 87: 624– 638. Mansfield, Edward D., and Jon C. Pevehouse. 2000. “Trade Blocs, Trade Flows, and International Conflict.” International Organization 54(4): 775–808. Mansfield, Edward D., and Brian M. Pollins. 2003. “Interdependence and Conflict: An Introduction.” In Economic Interdependence and International Conflict: New

References

325

Perspectives on an Enduring Debate, edited by E. D. Mansfield and B. M. Pollins, 1–30. Ann Arbor: University of Michigan Press. Mansfield, Edward, and Jack Snyder. 1995. “Democratization and the Danger of War.” International Security 20: 5–38. Mansfield, Edward, and Jack Snyder. 1996. “The Effects of Democratization on War.” International Security 21: 196–207. Maoz, Zeev. 2001a. “Democratic Networks: Connecting National Dyadic and Systemic Levels of Analysis in the Study of Democracy and War.” In War in a Changing World, edited by Zeev Maoz and Azar Gat, 143–182. Ann Arbor: University of Michigan Press. Maoz, Zeev. 2001b. Data Set DYMID Version 1.1. http://spirit.tau.ac.il/∼zeevmoaz. Maoz, Zeev, and Bruce Russett. 1993. “Normative and Structural Causes of Democratic Peace: 1946–1986.” American Political Science Review 87: 624–638. Markusen, James, and Anthony Venables. 1999. “Foreign Direct Investment as a Catalyst for Industrial Development.” European Economic Review 43(2): 335–356. Marquand, David. 1997. “The Great Reckoning.” Prospect, July. Marshall, Monty G., and Keith Jaggers. 2000. Polity IV Project: Political Regime Characteristics and Transitions, 1800–1999. Center for International Development and Conflict Management, University of Maryland. http://www.cidcm.umd.edu/ inscr/polity/index.htm#data. Martin, Hans-Peter, and Harold Schumann. 1997. The Global Trap: Globalization and the Assault on Prosperity and Democracy. London: Zed Books. Martin, Michael L. 1994. “National Security and Democracy: The Dilemma from a French Perspective.” Armed Forces and Society 20: 395–421. Marwick, Arthur. 1988. “Introduction.” In Total War and Social Change, edited by Arthur Marwick, x–xxi. London: Macmillan. Marx, Karl. [1867] 1990. Capital Volume 1: A Critique of Political Economy. London: Penguin Classics. Marx, Karl. [1871] 1993. “The Civil War in France.” In The Civil War in France: The Paris Commune, 23–90, New York: International Publishers. Matanduno, Michael. 1992. Economic Containment of CoCom and the Politics of East-West Trade. Ithaca, NY: Cornell University Press. Mazumdar, Dipak, and Ata Mazaheri. 2002. Wages and Employment in Africa. Aldershot, Hampshire, England: Ashgate. McMillan, S. M. 1997. “Interdependence and Conflict.” Mershon International Studies Review 41: 33–58. McNeill, John Robert. 2001. Something New Under the Sun: An Environmental History of the Twenty-Century World. London: W. W. Norton. Midlarsky, Manus I. 1995. “Environmental Influences on Democracy: Aridity, Warfare and a Reversal of the Causal Arrow.” Journal of Conflict Resolution 39(2): 224– 262. Midlarsky, Manus I. 1998. “Democracy and the Environment: An Empirical Assessment.” Journal of Peace Research 35: 341–361. Midlarsky, Manus I. 2002. “Review of Democracy and Development: Political Institutions and Well-Being in the World, 1950–1990, by Adam Przeworski et al., 2000.” Journal of Politics 64: 670–672. Mill, Stuart J. [1860] 1951. “Considerations on Representative Government.” In Utilitarianism, Liberty, and Representative Government, edited by H. B. Acton. London: Dent.

326

References

Milner, Helen V., and Robert O. Keohane. 1996. “Internationalization and Domestic Politics.” In Internationalization and Domestic Politics, edited by Robert O. Keohane and Helen V. Milner, 3–24. Cambridge, UK: Cambridge University Press. Minford, P., J. Riley, and E. Nowell. 1997. “Trade, Technology and Labor Markets in the World Economy.” Journal of Development Studies 34(2): 1–34. Mintz, Alex. 1985. “The Military Industrial Complex: American Concepts and Israeli Realities.” Journal of Conflict Resolution 29(4): 623–639. Mitchell, Sara McLaughlin, Scott Gates, and Havard Hegre. 1999. “Evolution in Democracy-War Dynamics.” Journal of Conflict Resolution 43(6): 440–461. Modelski, George, and Gardner Perry III. 1991. “Democratization in Long Perspective.” Technological Forecasting and Social Change 39: 23–34. Moomaw, William R., and Gregory C. Unruh. 1997. “Are Environmental Kuznets Curve Misleading Us? The Case of CO2 Emissions.” Environment and Development Economics 2: 451–463. Moon, Bruce E. 1991. The Political Economy of Basic Human Needs. Ithaca, NY: Cornell University Press. Moran, Jonathan. 1996. “Contradictions Between Economic Liberalization and Democratization: The Case of South Korea.” Democratization 3: 459–490. Morrow, James D. 1997. “When Do Relative Gains Impede Trade?” Journal of Conflict Resolution 41: 12–37. Morrow, James D. 1999. “How Could Trade Affect Conflict?” Journal of Peace Research 36(4): 481–489. Morrow, James, Randolph Siverson, and Tressa Tabares. 1998. “The Political Determinants of International Trade: The Major Powers 1907–1990.” American Political Science Review 92(3): 649–661. Mousseau, Michael, and Yuhang Shi. 1999. “A Test for Reverse Causality in the Democratic Peace Relationship.” Journal of Peace Research 36(6): 639–663. Muller, Edward N. 1988. “Democracy, Economic Development and Income Inequality.” American Sociological Review 53: 50–68. Muller, Edward N. 1995. “Economic Determinants of Democracy.” American Sociological Review 60: 966–982. Muller, Edward N., and Mitchell A. Seligson. 1994. “Civic Culture and Democracy: The Question of Causal Relationship.” American Political Science Review 88(3): 635–652. Muller, Ronald. 1979. “The Multinational Corporation and the Underdevelopment of the Third World.” In The Political Economy of Development and Underdevelopment, edited by Charles K. Wilber, 151–178. New York: Random House. Nafziger, Wayne. 1997. The Economics of Developing Countries. Upper Saddle River, NJ: Prentice Hall. Nafziger, Wayne E. 2006. Economic Development. New York: Cambridge University Press. Neumayer, Eric. 2002. “Do Democracies Exhibit Stronger International Environmental Commitment? A Cross Sectional Analysis.” Journal of Peace Research 39: 139–164. New York Times. 2004. “The Sky is Falling! Say Hollywood and, Yes, the Pentagon.” New York Times, February 29. Nicholson, Walter. 2005. Microeconomic Theory: Basic Theory and Extensions. Fort Worth, TX: The Dryden Press. Nickell, Stephen. 1981. “Biases in Dynamic Models with Fixed Effects.” Econometrica 49: 1417–1426.

References

327

Nordlinger, Eric. 1970. “Soldiers in Mufti: The Political Impact of Military Rule upon Economic and Social Change in Non-Western States.” American Political Science Review 64: 1131–1148. Normand, Charles. 1996. “Health Insurance: A Solution to the Financing Gap?” In Marketing Education and Health in Developing Countries, edited by C. Colclough., 205–221. Oxford: Clarendon Press. North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge, UK: Cambridge University Press. North, Douglass C., and Barry Weingast. 1989. “Constitutuions and Commitment: Evolution of Institutions Governing Collective Choice in 17th-Century England.” Journal of Economic History 49(4): 803–832. O’Donnell, Guillermo. 1973. Modernization and Bureaucratic-Authoritarianism: Studies in South American Politics. Berkeley: Institute of International Affairs, University of California. O’Donnel, Guillermo. 1979. Modernization and Bureaucratic-Authoritarianism. Berkeley: University of California Press. O’Donnell, Guillermo. 1993. “The Browning of Latin America.” New Perspectives Quarterly 10: 50–53. O’Donnell, Guillermo. 1994. “Delegative Democracy.” Journal of Democracy 5: 55–69. O’Donnell, Guillermo, Philippe Schmitter, and Laurence Whitehead. 1986. Transitions from Authoritarian Rule. Baltimore: Johns Hopkins University Press. OECD (Organization for Economic Cooperation and Development). 1994a. The Performance of Foreign Affiliates in OECD Countries. Paris: Organization for Economic Cooperation and Development. OECD. 1994b. The Environmental Effects of Trade. Paris: Organization for Economic Development and Cooperation. Olson, Mancur. 1982. The Rise and Decline of Nations. New Haven, CT: Yale University Press. Olson, Mancur. 1991. “Autocracy, Democracy and Prosperity.” In Strategy and Choice, edited by Richard J. Zeckhauser, 131–157. Cambridge: MIT Press. Olson, Mancur. 1993.“Dictatorship, Democracy, and Development.” American Political Science Review 87: 567–576. Oneal, John R., and Bruce Russett. 1997. “The Classical Liberals Were Right: Democracy, Interdependence and Conflict, 1950–1985.” International Studies Quarterly 41: 267– 294. Oneal, John R., and Bruce Russett. 1999a. “Assessing the Liberal Peace with Alternative Specifications: Trade Still Reduces Conflict.” Journal of Peace Research 36: 423–42. Oneal, John R., and Bruce Russett. 1999b. “Is the Liberal Peace Just an Artifact of Cold War Interests? Assessing Recent Critiques.” International Interactions 25: 213–241. Oneal, John R., and Bruce Russett. 1999c. “The Kantian Peace: The Pacific Benefits of Democracy, Interdependence, and International Organizations, 1885–1992.” World Politics 52: 1–37. Oneal, John R., and Bruce Russett. 2000. “Comment: Why ‘An Identified Systemic Model of The Democratic-Peace Nexus’ Does Not Persuade.” Defense and Peace Economics 11: 197–214. O’Rourke, Kevin. 2001. “Globalization and Inequality: Historical Trends.” NBER Paper No. 8339, National Bureau of Economic Research, Cambridge, MA.

328

References

Paehlke, Robert. 1996. “Environmental Challenges to Democratic Practice.” In Democracy and the Environment: Problems and Prospects, edited by William Lafferty and James Meadowcroft, 18–38. Cheltenham, England: Edward Elgar. Panayotou, Theodore. 2000a. “Globalization and Environment.” CID Paper 53, Center for International Development and Harvard University, Cambridge, MA. Panayotou, Theodore. 2000b. “Economic Growth and the Environment.” CID Paper 56, Center for International Development and Harvard University, Cambridge, MA. Park, Tong, Farid Abolfathi, and Michael Ward. 1976. “Resource Nationalism in Foreign Policy Behavior of Oil Exporting Countries.” International Interactions 2: 247– 262. Parson, Edward. 2002. Protecting the Ozone Layer: Science, Strategy, and Negotiation in the Shaping of a Global Environmental Regime. Oxford: Oxford University Press. Payne, Rodger A. 1995. “Freedom and the Environment.” Journal of Democracy 6: 41– 55. Penn World Tables 5.6a. Center for International Economic Comparisons, University of Pennsylvania, Philadelphia, PA. http://pwt.econ.upenn.edu/ (April 5, 2009). Penn World Tables 6.1. Center for International Economic Comparisons, University of Pennsylvania, Philadelphia, PA. http://pwt.econ.upenn.edu/ (April 5, 2009). Pindyck, Robert, and Daniel Rubenfeld. 1991. Econometric Models and Economic Forecasts. New York: McGraw-Hill. Platner, Marc. 1993. “The Democratic Moment.” In The Global Resurgence of Democracy, edited by Larry Diamond and Marc Plattner, 31–49. Baltimore: Johns Hopkins University Press. Polachek, Solomon W. 1980. “Conflict and Trade.” Journal of Conflict Resolution 24: 55–78. Polachek, Solomon W. 1992. “Conflict and Trade: An Economic Approach to Political International Interactions.” In Economics of Arms Reduction and the Peace Process, edited by Walter Isard and Charles Anderton, 89–120. Amsterdam: Elsevier. Polachek, Solomon W. 1997. “Why Democracies Cooperate More and Fight Less: The Relationship Between International Trade and Cooperation.” Review of International Economics 5: 295–309. Polachek, Solomon W. 2000. “Trade-Based Interactions: An Interdisciplinary Perspective.” Presidential Address, Peace Science Society. Presented at Yale University, New Haven, CT, October 28. Polachek, Solomon W., and Judith A. McDonald. 1992. “Strategic Trade and the Incentive for Cooperation.” In Disarmament, Economic Conversion, and Management of Peace, edited by Manas Chatterji and Linda Forcey, 273–284. Westport: Praeger. Polachek, Solomon W., and John Robst. 1998. “Cooperation and Conflict Among Democracies: Why Do Democracies Cooperate More and Fight Less?” In The Political Economy of War and Peace, edited by Murray Wolfson, 127–154. Boston: Kluwer Academic. Pollins, Brain M. 1989. “Does Trade Still Follow the Flag?” American Political Science Review 83(2): 465–480. Porter, Bruce D. 1994. War and the Rise of the State: The Military Foundations of Modern Politics. New York: Free Press. Potter, David, David Goldblat, Margaret Kiloh, and Paul Lewis, eds. 1997. Democratization. Cambridge, UK: Polity Press.

References

329

Pridham, Goeffery. 1994. “Encouraging Democracy.” In The International Dimension of Democratization in Eastern Europe, edited by Goeffrey Pridham, Eric Herring, and George Sanford. New York: St. Martin’s. Przeworski, Adam. 1985. Capitalism and Social Democracy. Cambridge, UK: Cambridge University Press. Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Reforms in Eastern Europe and Latin America. Cambridge, UK: Cambridge University Press. Przeworski, Adam, and Fernando Limongi. 1997. “Modernization: Theories and Facts.” World Politics 49: 155–183. Przeworski, Adam, Michael E. Alvarez, Jose A. Cheibub, and Fernando Limongi. 1996. “What Makes Democracies Endure?” Journal of Democracy 7: 39–55. Przeworski, Adam, Michael E. Alvarez, Jose A. Cheibub, and Fernando Limongi. 2000. Democracy and Development: Political Institutions and Well-Being in the World, 1950– 1990. New York: Cambridge University Press. Przeworski, Adam, and Fernando Limongi. 2003. “Political Regimes and Economic Growth.” In Development and Underdevelopment: The Political Economy of Global Inequality, edited by Mitchel A.. Seligson and John T. Passe-Smith, 435–448. Boulder, CO: Lynne Rienner. Pugel, Thomas. 2003. International Economics. New York: McGraw-Hill/Irwin. Putnam, Robert D., Robert Leonardi, and Raffaella Y. Nanetti. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Pye, Lucian. 1966. Aspects of Political Development. Boston: Little Brown. Quinn, Dennis. 1997. “The Correlates of Change in International Financial Regulation.” American Political Science Review 91(3): 531–551. Raleigh, Clionadh, and Nils Petter Gleditsch. 2005. War Sites 1946–2003. Centre for the Study of Civil War, PRIO. Rao, Vaman. 1985. “Democracy and Economic Development.” Studies in Comparative and International Development 19: 67–81. Rasler, Karen, and William R. Thompson. 2000. “Democratic Peace and the Reversed Causal Arrow Hypothesis: Sequentially Relating Threat, Democratic Dyads, and Reduced Conflict Propensities.” Presented at Midwest Political Science Association, Chicago, April. Ray, James Lee. 1995. Democracy and International Conflict: An Evaluation of the Democratic Peace Proposition. Columbia: University of South Carolina Press. Ray, James Lee. 2003. “Explaining Interstate Conflict and War: What Should be Controlled For?” Conflict Management and Peace Science 20: 1–32. Reiter, Dan, and Allan C. Stam. 2003. “Identifying the Culprit: Democracy, Dictatorship, and Dispute Initiation.” American Political Science Review 97: 333–337. Remmer, Karen. 1998. “Does Democracy Promote Interstate Cooperation? Lessons from the Mercosur Region.” International Studies Quarterly 42: 25–51. Reuveny, Rafael. 1999a. “The Political Economy of Israeli-Palestinian Interdependence.” Policy Studies Journal 27: 643–664. Reuveny, Rafael. 1999b. “Israeli-Palestinian Economic Interdependence Reconsidered.” Policy Studies Journal 27: 668–672. Reuveny, Rafael. 2000. “The Trade and Conflict Debate: A Survey of Theory, Evidence and Future Research.” Peace Economics, Peace Science and Public Policy 6: 23–9.

330

References

Reuveny, Rafael. 2001a. “Bilateral Import, Export, and Conflict/Cooperation Simultaneity.” International Studies Quarterly 45: 131–158. Reuveny, Rafael. 2001b.“Disaggregated Trade and Conflict: Exploring Propositions.” International Politics 38: 401–428. Reuveny, Rafael. 2002. “Economic Growth, Environmental Scarcity and Conflict.” Global Environmental Politics 2: 83–110. Reuveny, Rafael, and Heejoon Kang. 1998. “Bilateral Trade and Political Conflict/ Cooperation: Do Goods Matter?” Journal of Peace Research 35: 581–602. Reuveny, Rafael, and Quan Li. 2003. “Economic Openness, Democracy and Income Inequality: An Empirical Analysis.” Comparative Political Studies 36: 575–601. Reuveny, Rafael, and Aseem Prakash. 1999. “The Afghanistan War and the Breakdown of the Soviet Union.” Review of International Studies 25: 693–708. Reuveny, Rafael, and William R. Thompson. 2004. Growth, Trade, and Systemic Leadership. Ann Arbor: University of Michigan Press. Ripsman, Norrin M., and Jewan-Marc F. Blanchard. 1997. “Commercial Liberalism Under Fire: Evidence from 1914 and 1936.” Security Studies 6: 4–50. Risse, Thomas, and Kathryn Sikkink. 1999. “The Socialization of International Human Rights Norms into Domestic Practices: Introduction.” In The Power of Human Rights: International Norms and Domestic Change, edited by Thomas Risse, Stephen C. Ropp, and Kathryn Sikkink, 1–38. Cambridge, UK: Cambridge University Press. Robbins, Donald. 1996. “Evidence of Trade and Wages in the Developing World.” OECD Development Center Technical Papers, No. 119. Paris: Organization for Economic Cooperation and Development. Roberts, Bryan. 1996. “The Social Context of Citizenship in Latin America.” International Journal of Urban and Regional Research 20(1): 38–65. Robertson, Roland. 1992. Globalization: Social Theory and Global Culture. London: Sage. Rodrik, Dani. 1997. Has Globalization Gone Too Far? Washington, DC: Institute for International Economics. Rodrik, Dani. 1998. “Symposium on Globalization in Perspective: An Introduction.” Journal of Economic Perspectives 12(4): 3–8. Rosato, Sebastian. 2003. “The Flawed Logic of Democratic Peace Theory.” American Political Science Review 97(November): 585–602. Rosecrance, Richard. 1986. The Rise of the Trading State: Commerce and Conquest in the Modern World. New York: Basic Books. Rudra, Nita. 2005. “Globalization and the Strengthening of Democracy in the Developing World.” American Journal of Political Science 49(4): 704–730. Rueschemeyer, Dietrich. 1991. “Different Methods – Contradictory Results? Research on Development and Democracy.” International Journal of Comparative Sociology 32: 9–38. Rueschemeyer, Dietrich, and Peter Evans. 1985. “The State and Economic Transformation: Towards an Analysis of the Conditions Underlying Effective Intervention.” In Bringing the State Back In, edited by Peter Evans, Dietrich Rueschemeyer, and Theda Skocpol, 69–92. Cambridge, UK: Cambridge University Press. Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John Stephens. 1992. Capitalist Development and Democracy. Chicago: Chicago University Press. Rummel, Rudolph J. 1997. Power Kills: Democracy as a Method of Nonviolence. London: Transaction Publishers.

References

331

Russett, Bruce. 1993. Grasping the Democratic Peace. Princeton: Princeton University Press. Russett, Bruce, and John R. Oneal. 2001. Triangulating Peace: Democracy, Interdependence, and International Organizations. New York: Norton. Russett, Bruce, John R. Oneal, and David R. Davis. 1998. “The Third Leg of the Kantian Tripod for Peace: International Organizations and Militarized Disputes, 1950–1985.” International Organization 52(3): 441–467. Salvatore, Dominick. 1998. International Economics. Upper Saddle River, NJ: Prentice Hall. Salvatore, Dominick. 2004. International Economics. New York: John Wiley. Sassen, Saskia. 1996. Losing Control: Sovereignty in an Age of Globalization. New York: Columbia University Press. Sayrs, Lois. 1989. “Trade and Conflict Revisited: Do Politics Matter?” International Interaction 15: 155–175. Sayrs, Lois. 1990. “Expected Utility and Peace Science: An Assessment of Trade and Conflict.” Conflict Management and Peace Science 11: 17–44. Schamis, Hector E. 1991. “Reconceptualizing Latin American Authoritarianism in the 1970s: From Bureaucratic-Authoritarianism to Neoconservatism.” Comparative Politics 23(2): 201–220. Scharpf, Fritz. 1991. Crisis and Choice in European Social Democracy. New York: Cornell University Press. Schelling, Thomas C. 1958. International Economics. Boston: Allyn and Bacon. Schmitter, Philippe C. 1996. “The Influence of the International Context upon the Choice of National Institutions and Policies in Neo-Democracies.” In The International Dimensions of Democratization: Europe and the Americas, edited by Laurence Whitehead, 26–54. Oxford: Oxford University Press. Schneider, Gerald, Katherine Barbieri, and Nils Petter Gleditsch. 2003. “Does Globalization Contribute to Peace? A Critical Survey of the Literature.” In Globalization and Armed Conflict, edited by Gerald Schneider, Katherine Barbieri, and Nils Petter Gleditsch, 3–30. Lanham, MD: Rowman & Littlefield Publishers. Schultz, Cynthia B., and Tamara R. Crockett. 1990. “Economic Development, Democratization, and Environmental Protection in Eastern Europe.” Boston College Environmental Affairs Law Review 18: 53–84. Schultz, Kenneth A. 1998. “Domestic Opposition and Signaling in International Crises.” American Political Science Review 92(December): 829–840. Schumpeter, Joseph A. 1942. Capitalism, Socialism and Democracy. New York: Harper & Row. Schumpeter, Joseph A. 1950. Capitalism, Socialism and Democracy. New York: Harper & Row. Schwartz, P., and D. Randall. 2003. An Abrupt Climate Change Scenario and Its Implications for United States National Security. Report Commissioned by the United State Government Department of Defense. http://www.gbn.com/ArticleDisplayServlet.srv? aid=26231. Scruggs, Lyle A. 1998. “Political and Economic Inequality and the Environment.” Ecological Economics 26: 259–275. Segal, David R. 1994. “National Security and Democracy in the United States.” Armed Forces and Society 20: 375–393.

332

References

Self, Peter. 1993. Government by the Market?: The Politics of Public Choice. London: Macmillan. Seligson, Mitchell A., and Edward N. Muller. 1987. “Democratic Stability and Economic Crisis: Costa Rica, 1978–1983.” International Studies Quarterly 321: 301–326. Sen, Amartya. 1994. “Liberty and Poverty: Political Rights and Economics.” New Republic 210(10): 31–37. Sen, Gautam. 1984. The Military Origins of Industrialization and International Trade Rivalry. New York: St. Martin’s. Shafik, Nemat. 1994. “Economic Development and the Environmental Quality: An Econometric Analysis.” Oxford Economic Papers 46: 757–773. Shafik, Nemat, and Sushenjit Bandyopadhyay. 1992. “Economic Growth and Environmental Quality: Time Series and Cross Country Evidence.” Background Paper for the World Development Report, WPS0904. Washington, DC: World Bank. Sheth, D. L. 1995. “Democracy and Globalization in India: Post-Cold War Discourse.” Annals of the American Academy of Political and Social Science 540: 24–39. Siegle, Joseph T., Michael M. Weinstein, and Morton H. Halperin. 2004. “Why Democracies Excel?” Foreign Affairs 83(September/October): 57–71. Simpson, Miles. 1990. “Political Rights and Income Inequality: A Cross-National Test.” American Sociological Review 55(5): 682–693. Singer, David, and John Keating. 1999. “Military Preparedness, Weapon Systems and the Biosphere: A Preliminary Impact Statement.” New Political Science 21: 325–343. Sirowy, Larry, and Alex Inkeles. 1991. “The Effects of Democracy on Economic Growth and Inequality: A Review.” In On Measuring Democracy, edited by Alex Inkeles, 234. New Brunswick, NJ: Transaction. Skocpol, Theda, Ziad Munson, Andrew Karch, and Bayliss Camp. 2001. “Patriotic Partnerships: Why Great Wars Nourished American Civic Voluntarism.” In Shaped by War and Trade: International Influences on American Political Development, edited by Ira Katznelson and Martin Shefter, 134–180. Princeton: Princeton University Press. Slaughter, Matthew J., and Phillip Swagel. 1997. Does Globalization Lower Wages and Export Jobs? Washington, DC: International Monetary Fund. Smith, Adam. 1776. The Wealth of Nations. New York: The Modern Library. Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics 70: 65–94. Soros, George. 1997. “The Capitalist Threat.” Atlantic Monthly 279(2): 45–58. Souva, Mark, Dale Smith, and Shaun Rowan. 2008. “Promoting International Trade: The Role of Market Protecting Institutions.” Journal of Politics 70(2): 383–392. Stark, Jeffrey. 1998. “Globalization and Democracy in Latin America.” In Fault Lines of Democracy in Post Transition Latin America, edited by Felipe Aguero and Jeffrey Stark, 67–98. Miami: North-South Centre. Starr, Harvey. 1991. “Democratic Dominoes: Diffusion Approaches to the Spread of Democracy in the International System.” Journal of Conflict Resolution 35(2): 356– 381. State of the World’s Forest Report. 2001. Food and Agriculture Organization, Rome, Italy. Stern, David I. 1998. “Progress on the Environmental Kuznets Curve?” Environment and Development Economics 3: 173–196.

References

333

Stiglitz, Joseph. 1998. “Towards a New Paradigm for Development: Strategies, Policies, and Processes.” Prebisch Lecture at UNCTAD, Geneva. Strand, H˚avard, Joachim Carlsen, Nils Petter Gleditsch, H˚avard Hegre, Christin Ormhaug, and Lars Wilhelmsen. 2005. Armed Conflict Dataset Codebook: Version 3–2005. Oslo (PRIO): International Peace Research Institute. Strange, Susan. 1996. The Retreat of the State: The Diffusion of Power in the World Economy. Cambridge, UK: Cambridge University Press. Stolper, Wolfgang F., and Paul Samuelson. 1941. “Protection and Real Wages.” Review of Economic Studies 9(1): 58–73. Sullivan, Gerard. 1983. “Uneven Development and National Income Inequality in Third World Countries: A Cross-National Study of the Effects of External Economic Dependence.” Sociological Perspectives 26(2): 201–231. Suri, Vivek, and Duane Chapman. 1998. “Economic Growth, Trade and Energy: Implications for the Environmental Kuznets Curve.” Ecological Economics 25: 195–208. Sylwester, Kevin. 2005. “Foreign Direct Investment, Growth and Income Inequality in Less Developed Countries.” International Review of Applied Economics 19(3): 289–300. Tarkowski, Jacek. 1989. “Old and New Patterns of Corruption in Poland and USSR.” Telos 80: 51–62. Templeton, Scott R., and Sara J. Scherr. 1999. “Effects of Demographic and Related Microeconomic Change on Land Quality in Hills and Mountains of Developing Countries.” World Development 27(6): 903–918. Thompson, William R. 1996. “Democracy and Peace: Putting the Cart Before the Horse?” International Organization 50(1): 141–174. Tilly, Charles. 1992. Coercion, Capital and European States, AD, 990–1990. Cambridge, UK: Blackwell. Torras, Mariano, and James K. Boyce. 1998. “Income, Inequality, and Pollution: A Reassessment of the Environmental Kuznets Curve.” Ecological Economics 25: 147– 160. Trent, John E. 1994. “Democracy in Danger.” Presented at the 46th World Congress of International Political Science Association, Berlin. Tsai, Pan-Long. 1995. “Foreign Direct Investment and Income Inequality: Further Evidence.” World Development 23(3): 469–483. Turner, B. L. 1990. The Earth as Transformed by Human Action. New York: Cambridge University Press. UNEP (United Nations Environment Programme). 2003. Global Environmental Outlook: Past, Present and Future Perspectives. Nairobi and London: Earthscan Publications. UNEP. 2006. GEO Data Portal. http: //geodata.grid.unep.ch/ (May 3, 2006). U.S. Department of State. 2007. How Political Democracy and Economic Development Are Mutually Reinforcing, Paula J. Dobriansky, Under Secretary for Democracy and Global Affairs, Remarks at a Freedom House Symposium on Democracy and Development, Washington, DC, January 19. Van Hanen, Tatu. 1990. The Process of Democratization: A Comparative Study of 147 States, 1980–1988. New York: Russak. Verdier, Daniel. 1998. “Democratic Convergence and Free Trade.” International Studies Quarterly 42: 1–24.

334

References

Vernon, Raymond. 1971. Sovereignty at Bay. New York: Basic Books. Vogel, Steven. 1992. “The Power Behind Spin-Ons: The Military Implications of Japan’s Commercial Technology.” In The Highest Stakes: The Economic Foundations of the Next Security System, edited by Wayne Sandholtz, Michael Borrus, John Zysman, Ken Conca, Jay Stowsky, Steven Vogel, and Steve Weber, 55–80. New York: Oxford University Press. Wade, Robert. 1996.“Globalization and Its Limits.” In National Diversity and Global Capitalism, edited by Susanne Berger and Ronald Core, 60–88. New York: Cornell University Press. Wallerstein, Immanuel. 1974. The Modern World System. New York: Academic Press. Waltz, Kenneth N. 1970. “The Myth of Interdependence.” In The International Corporation, edited by Charles P. Kindleberger, 205–223. Cambridge, MA: MIT Press. Waltz, Kenneth N. 1979. Theory of International Politics. Reading, MA: Addison-Wesley. Weart, Spencer R. 1998. Never at War: Why Democracies Will Not Fight One Another. New Haven, CT: Yale University Press. Weiss, Edith Brown, and Harold K. Jacobsen. 1999. “Getting Countries to Comply with International Agreements.” Environment 41: 16–23. Weitzman, Martin L. 1993. “Capitalism and Democracy: A Summing Up of the Arguments.” In Markets and Democracy: Participation, Accountability and Efficiency, edited by Samuel Bowles, Herbert Gintis, and Bo Gustafsson, 314–335. Cambridge, UK: Cambridge University Press. Westing, Arthur P. 1990. Environmental Hazards of War: Releasing Dangerous Forces in an Industrialized World. Newbury Park, CA: Sage. White, Halbert. 1980. “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity.” Econometrica 48: 817–838. Whitehead, Laurence. 1986. “International Aspects of Democratization.” In Transitions from Authoritarian Rule, edited by Guillermo O’Donnell, Philippe Schmitter, and Laurence Whitehead, 3–46. Baltimore: John Hopkins University Press. Whitehead, Laurence. 1993. “Introduction: Some Insights from Western Social Theory.” World Development 21: 1253–1260. Whitehead, Laurence. 1996. “Three International Dimensions of Democratization.” In The International Dimensions of Democratization, edited by Laurence Whitehead, 3–24. Oxford: Oxford University Press. Wiggins, Vince. 1999. Comparing XTGLS with Regress Cluster ( ). STATA Corporation. http://www.stata.com/support/faqs/stat/xtgls rob.html. Williamson, Jeffrey G. 1997. “Globalization and Inequality, Past and Present.” World Bank Research Observer 12(2): 117–135. Wolf, Martin. 2000. “The Big Lie of Global Inequality.” Financial Times, February 9, p. 25. Wolfson, Murray, Patrick James, and Eric Solberg. 1998. “In a World of Cannibals Everyone Votes for War: Democracy and Peace Reconsidered.” In The Political Economy of War and Peace, edited by Murray Wolfson, 154–176. Boston: Kluwer Academic. Wood, Adrian. 1994. North-South Trade, Employment, and Income Inequality: Changing Futures in a Skill-Driven World. New York: Oxford University Press. Wooddall, Pam. 1995. “The World Economy: Financial Markets Survey.” The Economist, October 17, pp. 1–37.

References

335

Wooldridge, Jeffrey. 2002. Introductory Econometrics – A Modern Approach. Cincinnati, OH: South-Western College Publishing. World Bank. 2002. World Development Indicators. Washington, DC: World Bank. World Bank Poverty Net. 2000. World Bank, Washington, DC [Online]. http:// www.worldbank.org/poverty/inequal. World Resources Institute. 1999. World Resources. New York: Oxford University Press. World Resources Institute. 2001. World Resources. New York: Oxford University Press. Zacher, Mark W., and Richard A. Matthew. 1995. “Liberal International Theory: Common Threads, Divergence Strands.” In Controversies in International Relation Theory: Realism and the Neoliberal Challenge, edited by Charles W. Kegley, Jr., 107–150. New York: St. Martin’s.

Author Index

Aaron, C., 68 Abolfathi, F., 162, 179 Achen, C., 197 Ahluwalia, M. S., 72 Aitken, B., 68n12 Alderson, A. S., 68n13 Alesina, A., 65n5 Allan, N. J., 269 Allen, L., 280 Almond, G. A., 131n9 Alvarez, J. A., 52 Alvarez, M. E., 28, 91, 93, 97n3, 99n5, 109n6 Alvarez, R. M., 91, 127, 153, 157 Amemiya, T., 114, 152, 197 Amin, S., 31, 33n17 Anderson, T. W., 115 Angell, N., 13, 158n1 Antweiler, W., 248 Arad, R. W., 161n7, 181 Arellano, M., 116 Armijo, L. E., 34, 34n20 Ashley, R., 161 Aten, B., 231, 259, 283 Baldwin, D., 161n8, 162, 162n10 Bandyopadhyay, S., 237, 288 Baran, P. A., 67 Barbier, E. B., 237, 250, 288 Barbieri, K., 160, 161 Barnet, R. J., 67 Barrett, S., 212 Barro, R. J., 56, 89, 94, 95, 110n8, 111 Bates, J. M., 237, 287 Batra, G., 68n11 Beck, N., 50, 59, 82, 87n28, 115, 115n15, 116, 197, 234

Beck, U., 31, 32n12 Bennett, D. S., 195–196 Bennett, M. R., 162, 179 Berge, E., 209 Bhagwati, J., 28, 28n4, 29n5, 247n4 Bhattarai, M., 250, 288 Bienen, H., 96 Birdsall, N., 65n4, 67, 72n15 Blanchard, J. F., 66, 161 Bliss, H., 251 Blomstrom, M., 68, 68n11 Boehmer, C., 161, 182, 198 Boix, C., 64, 97n3, 99n5, 109n6 Boli, J., 28, 29n7, 30n9, 56 Bollen, K. A., 27n3, 31, 37n21, 49n25, 58n27, 64n3, 65, 82, 153 Bond, M., 188n16 Bond, S. R., 116 Borenszstein, E., 68 Borjas, G. J., 70n14 Borner, S., 97 Bornschier, V., 68n13 Boron, A., 31, 33n16 Borrus, M., 161, 161n9 Boswell, T., 68n13 Boyce, J. K., 211, 212 Brack, D., 243 Bremer, S. A., 128n4 Bronson, R., 251 Brunetti, A., 97 Brunn, S. D., 28, 29n7 Bruno, G. S., 115 Bryan, L., 31 Bueno de Mesquita, B., 89, 93, 129, 131, 164n12 Bulte, E. H., 250

337

338

Author Index

Bun, M., 116 Burkhart, R. E., 26n2, 27n3, 37n21, 49, 49n25, 56, 58, 61, 82, 91, 97, 97n2, 112, 112n11, 116 Bush, G. W., 24 Bussmann, M., 68 Buzan, B., 161 Cameron, R., 24n1 Cammack, P., 31, 31n11, 32n13 Camp, B. See Carlsen, J., 283 Carter, N. G., 72 Cavanagh, J., 67 Chan, S., 63–65, 65n6, 127, 128n3 Chan, Sylvia, 90, 90n1 Chapman, D., 248 Chase-Dunn, C., 68n13 Cheibub, J. A., 28, 91, 93, 97n3, 99n5, 109n6 Chinaware, H. B., 72 Chirot, D., 95 Choucri, N., 161 Clauge, C., 95 Clinton, W. J., 24 Coe, D. T., 68n11 Cohen, Y., 94, 95 Colaresi, M., 97n2 Cole, M. A., 237, 287 Collins, J., 64n3 Congleton, R. D., 209, 211, 212 Copeland, B. R., 248 Cox, R. W., 31, 31n11, 32n12, 33n16, 33n17 Crescenzi, M. J., 133, 161 Crockett, T. R., 208 Cross-National Time Series Data Archive, 111, 111n9 Crutsinger, M., 243 Dahl, R. A., 26n2, 31, 32n14, 92 Dailami, M., 29n5, 30n10, 32n12 Dasgupta, P., 73, 268 Davidson, R., 117 Davis, D. R., 135, 136, 143 de Bruyn, S. M., 237 de Soysa, I., 68, 248 de Tocqueville, A., 92 Deacon, R. T., 269 Dean, J., 250 DeGregorio, J., 68 Deininger, K., 56, 65, 71, 72, 73n16, 80, 81, 81n19, 83n23, 85

Diamond, J., 205, 206, 226 Diamond, L., 24n1, 28, 29n6, 29n7, 31, 31n11, 32n13, 49n24, 56, 58, 63, 92, 109n6, 247n4 Diaz-Alejandro, C., 246 DiNardo, J., 116 Dinda, S., 231n3, 245n3, 250, 283 Dixon, W. J., 68n13, 126, 128n4, 129, 135, 230, 251 Dollar, D., 68 Domke, W. K., 161n7 Dorussen, H., 159, 162, 179, 180, 195, 198 dos Santos, T., 161 Downs, G. W., 89, 93 Doyle, M., 129n6 Drake, P. W., 26, 28, 29n6, 48, 63n2 Drucker, P. F., 31, 32n12 Dryzek, J. S., 210, 210n1 Easterly, W., 71, 80, 81, 81n19, 85 Enterline, A. J., 133 Environmental Performance Measurement Project, 222, 238 Eriksson, M., 232, 283 Esparza, L., 64n3 Esty, D., 243 Evans, P., 28, 29n6 Farber, H. S., 130 Farrell, D., 31 Fearon, J. D., 129, 230 Feenstra, R. C., 68n12, 194, 195 Feng, Y., 18, 26n2, 37n21, 56, 60, 91, 95, 97, 109n6, 110, 111, 111n10, 119 Ferreira, F., 63 Findlay, R., 95 Flandreau, M., 24 Food and Agriculture Organization, 230 Forest Watch Indonesia, 206 Forland, T. E., 161n8, 162 Frankel, J. A., 243 Freedom House, 53–59, 61, 212 Frieden, J. A., 34, 34n19, 34n20 Friedman, M., 89 Fruhling, 28, 30n9 Fukuyama, F., 89 Garrett, G., 34, 34n19 Gartzke, E., 130, 144, 151, 155, 161, 182, 198

Author Index Gasiorowski, M. J., 26, 27n3, 37, 37n21, 48, 49n24, 49n26, 53, 65, 65n5, 65n6, 97n2, 109n6, 111n10, 112, 113, 119, 134, 154, 161n7, 161n8, 162, 179 Gates, S., 131, 131n9 Germain, R., 69 Gill, S., 31, 31n11, 33n15, 33n17 Gilpin, R., 161, 161n9 Glasgow, G., 91, 127 Gleditsch, K. S., 127 Gleditsch, N. P., 128n3, 160n6, 209–212, 232, 283 Global Forest Watch, 206 Goldblat, D., 1, 2, 17, 23, 24n1, 62, 67, 69–70 Gonick, L. S., 97n2 Goodell, G., 95 Goodin, R. E., 95, 96 Gordon, H., 68n12 Gore, A., 205, 226 Gowa, J., 130, 247, 251 Graddy, K., 212 Gradstein, M., 65 Graham, E., 68n12 Gray, J., 31, 31n11 Greene, W. H., 59, 115n17 Grieco, J. M., 247 Griffiths, W. E., 49 Grossman, G., 248, 250 Guilkey, D. K., 153, 157 Gujarati, D. N., 197 Gurr, T. R., 47, 81 Haddad, M., 68n11 Hadjimichael, B., 68 Haggard, S., 26, 31, 32n13, 34, 34n20, 48, 93, 94, 113 Halperin, M. H., 11, 90 Hammig, M., 250, 288 Harbaugh, W. T., 250n5 Hardin, G., 210 Harris, J. M., 232n5, 241n1, 243, 250n5, 273 Harris, S. L., 280 Harrison, A., 68n11, 68n12 Hatzius, J., 67 Hayek, F., 89 Heckman, J. J., 140, 152 Heerink, N., 73 Hegre, H., 131, 195, 196, 283 Heilbronner, R. L., 210 Held, D., 1, 2, 17, 23, 24n1, 28, 28n4, 31, 31n11, 62, 63n2, 67, 69–70, 247n4

339

Helliwell, J. F., 26, 37n21, 91, 97 Helpman, E., 68n11 Heston, A., 231, 259, 283 Hettige, H., 248 Hewlett, S. A., 94, 95 Higgins, M., 71, 72, 80 Higgs, R., 131 Hill, C., 49 Hillmann, H. C., 180n15 Hirsch, S., 161, 181 Hirschman, A., 13, 158n2, 161n8, 162n10, 162n11, 181, 246 Hirst, P., 34, 34n18 Hoffmaister, A. W., 68n11 Howell, C. H., 280 Hsiao, C., 115, 116 Huber, E., 92 Hufbauer, G., 243, 246 Hughes, B., 62 Human Development Report, 64n3, 67, 69 Huntington, S., 24n1, 26, 26n2, 28, 30n10, 48, 89, 93 Ichino, A., 70n14 Im, H. B., 28, 28n4, 31, 32n13, 32n14, 33n16, 33n17, 63n2, 93 Inkeles, A., 64n3, 65n7 Inter-American Development Bank, 66 Intergovernmental Panel on Climate Change, 229, 266, 273, 280 International Financial Statistics CD-Rom, 149 International Herald Tribune, 89, 90 Islam, N., 110, 111, 118 Italianer, A., 193, 194 Jackman, R. W., 64n3, 65, 94 Jacobsen, H. K., 209 Jaggers, K., 47, 81, 109, 109n7, 283 James, P., 127, 127n1, 134–135, 144, 150 Jenkins, R., 67 Johnston, J., 116 Jones, R. J. B., 31, 31n11, 34, 34n18 Judge, G. G., 49 Judson, K. A., 115 Justman, M., 65 Kang, H., 162n10, 162n11 Kant, I., 12, 13, 28, 29n5, 30n10, 129, 158n1 Karch, A., 131

340

Author Index

Kasza, G. J., 131 Katz, J. N., 50, 59, 82, 87n28, 115, 115n15, 116, 197, 234 Kaufman, R. R., 26, 31, 32n13, 34, 34n20, 48, 93, 113 Keating, J., 269 Keck, M. E., 28, 29n7, 30n9, 60 Keefer, P., 95 Kennedy, P., 114, 181, 197 Keohane, R. O., 34, 34n20 Keshk, O. M., 157, 161, 195, 197 Keylor, W., 181 Kiloh, M., 24n1 King, D. Y., 96 King, G., 59, 85n27 Kiviet, J. F., 115–116 Klein, M., 68 Knack, S., 95 Knutsen, T. L., 131n9 Kokko, A., 68n11 Kostyal, K. M., 280 Kotov, V., 208 Kraay, A., 68 Krueger, A. B., 248, 250 Kummell, G., 28, 29n7, 31, 33n17 Kurzer, P., 34, 68 Kurzman, C., 97 Kuznets, S., 68, 72, 84 Lai, B., 196 Laitin, D. D., 230 Lal, D., 94 Lall, S., 67 Lappe, F., 64n3 Lasswell, H. D., 130 Lawrence, R., 68 Layne, C., 130, 131 Lee, J., 56, 68, 111 Lee, T., 49 Leinback, T. R., 28, 29n7 Leitz, C., 180n15 Lemke, D., 130 Lenin, V. I., 13, 89, 158n2 Lenski, G., 65 Leonardi, R., 92 Levinson, A., 250n5 Levy, J. S., 128n3 Lewis, P., 24n1, 73 Lewis-Beck, M. S., 26n2, 27n3, 37n21, 49, 49n25, 56, 58, 61, 82, 91, 97n2, 112, 112n11, 116

Li, Q., 50, 63n2, 95, 97n2, 109n6, n7, 111n10, 112, 112n11, 112n13, 113, 114n14, 119, 161, 165n13, 182, 189n17, 198, 233, 235, 247, 283 Liang, K., 59 Lijphart, A., 26 Limongi, F., 27, 28, 49n24, 52, 90, 91, 91n1, 93, 97n3, 99n5, 109n6 Lin, J. Y., 24 Lindblom, C. E., 31, 31n11 Lindert, P. H., 65, 65n5 Linz, J. J., 26, 92, 93 Lipset, S. M., 26n2, 28, 28n4, 37n21, 65n5, 92, 93, 96, 97n2, 102, 237 Lipsey, R. E., 68, 68n12 Londregan, J. B., 26n2, 47, 47n22, 51, 82, 97n2, 109n7, 111n10 Longworth, R. C., 31, 32n12, 34, 34n20 Lucas, R., 248 Lutkepohl, H., 49 MacDonald, M., 31, 32n13 MacKinnon, J. G., 117 Maddala, G. S., 140, 145, 152, 153, 157, 197 Mainwaring, S., 26 Mankiw, N. G., 110–111 Mansfield, E., 47, 47n22, 48, 82, 130n8, 160n6, 195, 247, 251 Maoz, Z., 128n4, 129, 134, 136, 150 Markusen, J., 68n11 Marquand, D., 31, 32n12 Marshall, M. G., 109, 109n7, 283 Martin, H., 31, 32n12, 33n15, 247n4 Martin, M. L., 131n9 Marwick, A., 131n10 Marx, K., 89 Matthew, R. A., 63, 63n1 Mazaheri, A., 68n12 Mazumdar, D., 68n12 McDonald, J. A., 162 McGrew, A., 1, 2, 17, 23, 62, 67, 69–70 McMillan, S. M., 159n4, 160n6 McNeill, J. R., 269, 270 Midlarsky, M. I., 90, 91n1, 131n9, 211, 212 Mill, J. S., 89 Milner, H. V., 34, 34n20 Minford, P., 66n9 Mintz, A., 131n9 Mitchell, S. M., 131 Modelski, G., 131 Montesquieu, Baron de, 13, 158n1

Author Index Moomaw, W. R., 237, 287 Moon, B., 65, 251 Moon, B. E., 65n5 Moran, J., 31, 32n12 Morrow, J., 50, 129, 161, 165n13, 182, 189n17, 247, 251 Moses, J., 131n9 Mousseau, M., 130n7, 133, 133n12 Mroz, T. A., 153, 157 Muller, E. N., 26, 26n2, 28, 28n4, 31, 32n12, 37n21, 49n25, 50, 65, 65n5, 82, 85, 93, 112n11, 114n14 Muller, R., 67 Munson, Z., 131 Nafziger, W., 67, 69, 80n18, 96 Nanetti, R. Y., 92 Neumayer, E., 211, 248 Nicholson, W., 188n16 Nickell, S., 115 Nielson, F., 68n13 Nikitina, E., 208 Nordlinger, E., 96 Normand, C., 69 North, D., 89, 95 North, R., 161 Nowell, E., 66n9 Nugent, J. B., 24 O’Donnell, G., 26, 31, 94, 32n13, 33n16 Olson, M., 38, 89, 94–95, 247 Oneal, J. R., 18, 28, 29n5, 47, 47n22, 48, 50, 68, 126, 127, 127n1, 128n3, 128n4, 129n6, 134–136, 143–145, 150, 151, 155, 157, 159n4, 161, 195, 198, 233, 235, 247, 251 Opschoor, J. B., 237 Orejas, D., 243 Organization of Economic Cooperation and Development, 68n11, 232n5, 241n1 Ormhaug, C., 283 O’Rourke, K., 70 Owen, A. L., 115 Paehlke, R., 210 Panayoto, T., 231n3, 232n6, 241n1, 245n3, 250, 259, 283 Park, T., 162, 179 Parson, E., 243 Payne, R. A., 208 Penn World Tables, 109–111, 149, 151

341

Perraton, J., 1, 2, 17, 23, 62, 67, 69–70 Perry III, Gardner, 131 Pindyck, R., 81n22 Plattner, M., 28, 28n4, 92 Polachek, S. W., 130, 158n3, 159, 161, 161n7, 162–164, 164n12, 169, 170, 179, 186, 192, 193n18 Pollins, B. M., 160n6, 161, 165n13, 189n17, 195, 197, 247 Poole, K. T., 26n2, 47, 47n22, 51, 82, 97n2, 109n7, 111n10 Porter, B. D., 131, 131n10 Potter, D., 24n1 Powelson, J. P., 95 Prakash, A., 131n10 Pridham, G., 26 Przeworski, A., 27, 27n3, 28, 30n10, 31, 33n17, 49n24, 52, 90, 91, 91n1, 93, 95, 97n3, 99n5, 109n6 Pugel, T., 232n5, 241n1, 241n2, 250n5 Putnam, R. D., 92 Pye, L., 95 Quinn, D., 69 Raleigh, C., 283 Randall, D., 280 Rao, V., 94 Rasler, K., 133 Ray, J. L., 128n3, 195 Rayner, A. J., 237, 287 Reed, W., 130 Reiter, D., 196 Remmer, K., 251 Resnick, A., 95, 247 Reuveny, R., 18, 50, 63n2, 97n2, 109n6, 109n7, 111n10, 112, 112n11, 112n13, 113, 114n14, 119, 131n10, 159n4, 160n6, 161, 161n9, 162n10, 165n13, 170, 189n17, 193, 194n19, 195, 197, 197n20, 232, 233, 235, 283 Riley, J., 66n9 Ripsman, N. M., 161 Risse, T., 28, 29n7, 30n9, 60 Robbins, D., 67 Roberts, B., 28, 30n8, 30n9 Robertson, R., 31, 32n14 Robst, J., 130 Rodrik, D., 31, 32n12, 62, 65, 65n5, 65n6, 66n9, 67 Rogowski, R., 34, 34n19, 34n20

342

Author Index

Romer, D., 110–111 Rosato, S., 130 Rosecrance, R., 161 Rosenblatt, A., 162, 179 Rosh, R. M., 97n2 Rosset, P., 64n3 Rowan, S., 247, 251 Rubenfeld, D., 81n22 Rubio, L., 243 Rudra, N., 29n6 Rueschemeyer, D., 27, 28, 29n6, 49n24, 92 Rummel, R. J., 128n5 Russett, B., 18, 28, 29n5, 47, 47n22, 48, 50, 126, 127, 127n1, 128n3, 128n4, 129, 129n6, 134–136, 143–145, 150–151, 155, 157, 159n4, 161, 195, 198, 233, 235, 247, 251 Sacko, D., 165n13, 189n17, 247 Salvatore, D., 67, 241n2 Sammon, R., 280 Samuelson, P., 66 Sassen, S., 31, 33n16 Sayrs, L., 159n4, 160n6, 161n8 Schamis, H. E., 65n4 Scharpf, F., 34, 34n18 Schelling, T. C., 162 Schmitter, P., 26, 28, 29n5, 29n7, 31, 31n11, 247n4 Schneider, G., 160n6 Schott, J., 243 Schultz, C. B., 208 Schultz, K., 129 Schumann, H., 31, 32n12, 33n15, 247n4 Schumpeter, J. A., 28, 28n4, 89 Schwartz, P., 280 Scruggs, L. A., 212 Segal, D. R., 131n9 Self, P., 28, 30n8 Seligson, M. A., 37n21, 49n25, 50, 82, 93, 112n11, 114n14 Sen, A., 209 Sen, G., 161n9 Seong, K., 92 Shafik, N., 211, 237, 288 Sheth, D. L., 28, 30n8 Shi, Y., 130n7, 133, 133n12 Siegle, J. T., 11, 90 Sikkink, K., 28, 29n7, 30n9, 60 Simpson, M., 65 Singer, D., 269

Sirowy, L., 64n3, 65n7 Siverson, R., 50, 129, 165n13, 189n17, 247, 251 Skocpol, T., 131 Slater, D., 196 Slaughter, M. J., 66n9 Smith, A., 129, 158n1 Smith, D., 247, 251 Snyder, J., 47, 47n22, 48, 82, 130n8 Solberg, E., 127, 127n1, 134–135, 144, 150 Sollenberg, M., 232, 283 Solow, R. M., 110, 110n8, 118 Soros, G., 62 Souva, M., 247, 251 Squire, L., 56, 65, 71, 72, 73n16, 80, 81, 81n19, 83n23, 85 Stam, A. C., 195–196 Stark, J., 28, 30n9 Starr, H., 27n3, 28, 30n10, 48, 112, 112n13 Stephens, E. H., 27 Stephens, J., 27, 92 Stern, D. I., 250n5 Stiglitz, J., 68 Stokes, S. C., 97n3, 99n5, 109n6 Stolper, W. F., 66, 66n8 Strand, H., 232, 283 Strange, S., 69 Sullivan, G., 68n13 Summers, R., 231, 259, 283 Suri, V., 248 Sverdlop, B. O., 209–212 Swagel, P., 66n9 Sylwester, K., 68 Tabares, T., 50, 165n13, 189n17, 247, 251 Tan, H., 68n11 Tarkowski, J., 31, 33n17 Taylor, D., 153, 157 Taylor, M. S., 248 Templeton, S. R., 284 The Economist, 89, 269 Thomas, G. M., 28, 29n7, 30n9, 56 Thompson, G., 34, 34n18 Thompson, W. R., 18, 97n2, 131, 133 Tilly, C., 131 Torras, M., 211, 212 Torres, J. C., 92 Tovias, A., 161, 181 Trent, J. E., 31, 32n13

Author Index Tsai, P., 68n13 Tucker, R., 197 Turner, B. L., 270 United Nations Environment Programme, 218 Unruh, G. C., 237, 287 US Department of State, 90 van den Bergh, J. C. J. M., 237 Van Hanen, T., 28, 29n7 van Soest, D. P., 250 Venables, A., 68n11 Verdier, D., 251 Vernon, R., 34, 68 Vogel, S., 161n9 Wada, E., 68n12 Wade, R., 34, 34n18 Wallensteen, P., 232, 283 Wallerstein, I., 24n1, 31, 33n17 Waltz, K. N., 158n2, 161 Wang, J., 162, 179 Ward, M., 127, 162, 179 Weart, S. R., 129 Weder, B., 97 Weil, D. D., 110–111

343

Weinstein, M. M., 11, 90 Weiss, E. B., 209 Weitzman, M. L., 28, 28n4 Werum, R., 97 Westing, A. P., 268 Wheeler, D., 248 White, H., 50, 116, 153, 234, 260 Whitehead, L., 26, 28, 30n10, 31, 33n16, 63n2 Wiggins, V., 50, 82, 116, 153 Wilhelmsen, L., 283 Williamson, J. G., 70n14, 71, 72, 80 Wilson, D. M., 250n5 Wolf, M., 62 Wolfson, M., 127, 127n1, 134–135, 144, 150 Wood, A., 66 Wooddall, P., 69 Wooldridge, J., 117, 233 World Bank, 62, 64, 195 World Resources Institute, 229, 282 Zacher, M. W., 63, 63n1 Zak, P. J., 26n2, 56, 60 Zeger, S. L., 59 Zejan, M., 68 Zysman, J., 161, 161n9

Subject Index

2SLS. See two-stage least squares 2SLS Kiviet. See two-stage least squares Kiviet estimator 3SLS. See three-stage least squares absence of conflict, 132, 180 Afghanistan, 269 Africa, 56, 112, 253 AH estimator, 115 Asia, 48, 60, 112, 181, 218, 253, 260–262 autocracy, 1, 11, 12, 16, 25, 29, 36, 47, 82, 90, 93–96, 98, 103, 104, 109, 129–130, 132, 134, 195, 206, 208–210, 212, 214, 219, 222–227, 230, 238, 240, 244–245, 253, 255–258, 260, 273, 298, 300, 303, 306 bilateral trade, 14, 158, 160, 161, 163, 165, 168–170, 178, 181–184, 189, 193, 243, 247, 251, 297, 299, 304 California, 23 capital, 10, 28–30, 33, 36, 37, 42, 44, 45, 47, 62–70, 72, 76, 77, 79, 93, 95, 96, 100, 110, 249, 301, 306, 307 ceteris paribus, 42, 52, 163, 164, 169, 170, 180, 182, 192, 193, 198, 227, 261, 292 China, 25, 46, 93, 96, 98, 106, 127, 134, 140, 196, 206, 218, 247, 284, 303, 306 classical economics, 10 climate change, 16, 229, 239, 259, 266–268, 272, 280–281, 303, 304 clustering, 49, 59, 82, 116, 145, 153, 197

CO2 , 15, 206–207, 212, 214–228, 232–238, 259, 267, 271–282, 284–290, 293, 297–298, 303 communication, 1, 48, 92, 112, 150 comparative advantage, 161, 232, 242, 245 composite environmental indicator, 219, 222, 238 composition effect, 242 control variable, 7, 36, 39, 43, 46–48, 52, 53, 56, 74, 80, 87, 104, 108, 112, 114, 116, 118, 126, 128, 141, 146–147, 153–154, 173, 174, 178, 182, 195, 198, 202, 214–216, 219, 231, 236, 238, 252, 253, 258, 260, 261, 273–275, 279, 287–291 country fixed effects, 59, 77, 87, 101, 113, 114, 117, 272, 286 country fixed effects estimator, 117, 285 covariance matrix, 50, 113, 116, 153 COW composite index, 196 data cross section, 25, 49, 53, 110, 115, 140, 153, 214–215, 274, 284, 285, 288 panel, 250 time series, 49, 53, 56, 101, 105, 115, 140, 152, 214, 215, 233, 260, 274, 284 time series cross section, 101, 115 DCs. See developed countries deforestation, 15–17, 206, 207, 212, 214–222, 224–229, 233, 234, 236–237, 239, 240, 242, 250, 252–259, 261, 263, 267–269, 271–274, 277–282, 284–285, 287–290, 293, 298, 303–305 demand, 14, 28, 30, 64, 92, 163–166, 169–170, 182, 184, 187–193, 195, 229, 242

344

Subject Index democracy, 1–6, 8–16, 18–20, 23–65, 70–81, 89–119, 125–157, 174, 175, 178, 186, 195, 197, 199, 205–238, 243–266, 270, 273, 283, 292–300 democratization, 29, 35, 37, 39, 53, 79, 93, 99, 104–107, 126–128, 131–133, 138, 145, 147, 225, 263, 272, 302, 303, 306 dependent variable, 6, 7, 12, 26, 35, 47, 49, 50, 53, 56, 71, 80–82, 85–87, 90, 97, 99, 100, 108, 115, 126, 145, 146, 173–175, 179, 196, 197, 206, 213, 214, 228, 233–235, 251, 252, 260, 272, 282, 284, 285, 291, 293 developed countries, 16, 32–34, 38, 41, 44, 64, 66–70, 73–76, 78, 79, 83–85, 96, 98, 101, 240, 241, 253, 256–258, 261, 263, 271, 274, 280, 306 display of force, 134, 138, 173, 179, 180, 185, 202 domestic conflict, 129 dummy variable, 56, 59, 60, 85, 133, 150, 196, 215, 260, 261, 263 dyadic, 3, 12–14, 18, 125–128, 132–141, 143–148, 151, 153, 155, 157, 159, 165, 172–175, 178, 193, 196, 202, 297 economic growth effect, 245 economic openness, 1–6, 8–14, 18, 19, 23, 24, 30–34, 36, 44–46, 60, 62–66, 70–74, 77–82, 85, 87, 100, 107, 125, 158, 186, 205, 227, 247, 266, 281, 292, 293, 299, 301, 302, 305–306 education, 1, 11, 28, 43, 56, 60, 66, 72, 73, 92, 100, 110, 111, 118, 208, 215, 233 EKC. See environmental Kuznets curve emission, 15, 16, 206, 207, 211–212, 214–219, 224–229, 232–234, 236, 239, 248, 259, 267, 268, 270–291, 293, 297, 298, 303 endogeneity, 102, 127, 215, 233 endogenous variable, 97, 99, 100, 109, 111, 113, 115, 126, 135–138, 140, 145, 151–153, 157, 194, 197 environmental degradation, 2, 9, 15, 16, 205–225, 228, 231–235, 238–242, 244–246, 248–250, 252, 259, 266–268, 272–274, 279–281, 283, 285, 288, 292, 293, 297, 298, 303–305 environmental economics, 6, 18, 231

345

environmental Kuznets curve, 19, 68, 77, 212–214, 225, 227, 231, 236, 237, 241, 245–246, 248, 249, 253, 273, 283 Environmental Performance Measurement Project, 222, 238 Environmental Sustainability Index, 222, 238 Environmental Systems Quality composite, 222, 238 EPMP. See Environmental Performance Measurement Project equality test, 88, 200 error, 51, 60, 68 error term, 8, 38, 49, 59, 73, 81, 87, 102, 113, 114, 140, 175, 197, 215, 233, 253, 274, 285, 286 ESI. See Environmental Sustainability Index estimator, 43, 46, 49, 53, 59–60, 77, 79, 85–87, 91, 101, 113–117, 153, 157, 197, 234, 237, 285 Europe, 3, 44, 48, 60, 65, 112, 206, 253, 260–262, 270, 284 exogenous variable, 97, 99, 100, 110, 111, 113, 135, 137, 138, 152, 233 export, 14, 29, 47, 66, 111, 147, 149, 158–160, 162–173, 175–181, 183–194, 197, 198, 200, 232, 233, 245, 248, 250, 251, 256, 259, 283, 297, 305 externalities, 162, 258 factors of production, 30, 45, 66, 95, 245 famines and human life argument, 244 fixed effects, 101, 110, 111, 114–115, 213, 235, 286 fixed effects estimator, 59, 77, 85–87, 114, 117, 234 Florida, 23 foreign direct investment, 10, 11, 19, 25, 28, 36, 37, 40, 44, 47, 62, 66, 72, 79, 81, 111, 228, 249, 299 foreign financial capital, 66, 72, 79 formal model, 6, 7, 17, 147, 159, 165, 182 France, 24, 46, 127, 134, 140, 155, 196 Freedom House, 53, 56–59, 61, 212 freedom of information channel, 244 GEE. See general estimating equation General Agreement on Trade and Tariffs, 228 general estimating equation, 59, 60

346

Subject Index

Gini coefficient, 10, 56, 64, 65, 71, 72, 80, 81, 85 global warming, 226, 239, 304 global-commons thesis, 243 globalization, 1, 4, 8, 18–20, 23–46, 50–53, 62, 76, 79, 228, 240, 266, 299 Greece, 77, 96, 269 gross domestic product, 37, 47–49, 56, 68, 72, 81, 82, 85, 87, 96, 97, 100, 102–104, 108, 110–112, 118, 134, 138, 149–151, 154, 178, 195, 199, 202, 214, 231, 232, 237, 248, 252, 259–261, 273, 276, 279, 283, 286, 288–290 Gulf War, 266, 269

international organization, 18, 32, 78, 79, 146, 258 international political economy, 6, 18, 147, 158, 185 international trade, 13, 15, 66, 78, 228, 232, 239–242, 245–248, 254, 257, 258 inverted U, 231, 245, 246, 250, 283, 296, 298, 301, 302, 304 Iraq, 269, 270 joint democracy, 12, 13, 125–127, 130, 132–135, 141–147, 151, 251, 297, 304, 305 Kuznets curve, 72, 74, 83, 84, 87, 245

heteroscedasticity, 49, 59, 60, 82, 116, 153, 197, 285 homoscedasticity, 59 Huber-White robust standard errors, 59, 82, 234, 260, 285 hypothesis, 7, 14, 60, 70, 72, 74–76, 82, 92, 117, 132, 134, 150, 153, 160, 163, 168, 170–175, 193, 194, 197, 198, 253, 259, 261, 297 null, 50, 59, 117, 235 hypothesis testing, 60 IMF. See International Monetary Fund Im-Pesaran-Shin panel unit root test, 286 import, 14, 47, 66, 111, 147, 149, 158–160, 163–173, 175–194, 197, 198, 200, 232, 245, 248, 259, 283, 297, 305 income effect, 231, 245, 246 income equality, 63, 78, 80, 295, 306 income inequality, 2, 6, 10, 18, 28, 32, 40, 43, 46, 56, 60, 62–87, 95, 125, 245, 247, 281, 292–296, 299–303, 307 independent variable, 7, 12, 35, 48, 49, 71, 87, 90, 99, 135, 152, 173, 197, 213, 214, 230, 235, 251, 272, 273, 275, 282 inertia, 37, 49, 52, 84, 110, 112, 118, 119, 154, 215, 233, 236, 285, 287, 296, 298 inflation, 18, 26, 37, 48, 52, 53, 100, 111, 113, 134, 138, 149, 150, 154 international dollars, 72, 82, 102, 231, 236, 259, 283, 287 international economics, 18 international labor mobility, 66 International Monetary Fund, 51, 78

lagged, 233, 235, 236, 274, 276, 283, 284 lagged dependent variable, 49–51, 53, 82, 87, 101, 110, 114–116, 150, 216, 218, 234, 235, 274, 276, 279, 284, 287, 290, 291 land degradation, 15, 16, 206, 207, 211, 214–216, 219–228, 230, 232–234, 236–237, 239, 240, 250, 252, 253, 255–259, 261, 263, 273, 293, 298 LDCs. See less developed countries LDV. See lagged dependent variable least squares dummy variable, 114, 115 less developed countries, 16, 32, 33, 38, 39, 41–42, 44–46, 53, 58, 59, 64, 66–70, 73–76, 78–79, 83–85, 95, 96, 101, 102, 105, 106, 111, 118–119, 239–241, 253, 256–258, 261, 263, 265, 270–271, 274–280, 287–291, 298, 302–304, 306 Levin-Linard, 286 long run change in democracy, 103, 104, 117, 118, 235 Lorenz curve, 80 LSDV. See least squares dummy variable Maddala, 152 correction, 145, 152, 153, 157 estimator, 152 method, 140, 145, 153 major powers, 133, 136, 146, 150, 196 market argument, 243 mean, 7, 75, 82, 143, 177, 178, 218, 219, 254, 260, 263, 278 Middle East, 112

Subject Index military conflict, 2, 4, 6, 9, 12–15, 18, 19, 107, 125, 126, 128, 130–131, 135, 138, 145, 147, 158–164, 172, 175, 182–185, 205, 212, 232, 258, 293, 295, 297, 299, 304, 305 military interstate dispute, 12, 125–128, 133, 135, 138, 139, 141–148, 150, 152–157, 160, 172–182, 185, 195–202 model variables AFFINITY, 138, 139, 144, 151, 155, 157 AFRICA, 56, 253 AGRICULTURE EXPORT, 173, 175, 194, 199 AGRICULTURE IMPORT, 173, 175, 194, 199 ALLIANCE, 138, 139, 143, 151, 155, 173, 174, 196, 199 ALLY, 134 ASIA, 56, 253 AUTOCRACY DUMMY, 214, 230, 233 CAPABILITY RATIO, 138, 139, 151, 155, 173, 174 CHEMICAL-MINERAL EXPORT, 173, 194 CHEMICAL-MINERAL IMPORT, 173, 177, 194, 199 CONFLICT, 163, 168, 173, 174 CONTIGUITY, 138, 139, 143, 150, 155, 173, 174, 195, 199 COUNTRY FIXED EFFECTS, 99, 100, 111, 214, 272, 274, 284, 285, 288 DEMH , 136, 138–141, 143–145, 147–150, 152, 154, 155 DEML , 136–145, 147–150, 152–157 DEMOCRACY, 36, 40, 41, 47, 71, 72, 74–77, 81–88, 99, 100, 104, 107–110, 112–113, 117–119, 137, 173, 252, 254–256, 260–265, 272, 273, 283, 287, 288 DEMOCRACY DUMMY, 214, 230, 233 DEPENDENCE, 139, 151, 155 DEVELOPMENT, 37, 48, 52 DIFFUSION, 25, 27, 30, 36, 41–43, 48, 53, 56, 58, 60, 92, 99, 100, 112, 119, 137, 139, 150, 154 DISTANCE, 173, 195, 199 ECONOMIC DEVELOPMENT, 52, 53, 56, 60 ECONOMIC GROWTH, 52, 53, 99, 100, 119 EDUCATION, 56, 60, 99, 100, 118

347

ENERGY EXPORT, 173, 177, 194, 199 ENERGY IMPORT, 173, 177, 194, 199 ENVIRONMENT, 214, 228, 272, 274, 282, 283 EUROPE, 56, 253 EXPORT, 164, 168, 173, 177, 199 FDI, 25, 33, 36, 37, 40–47, 50, 51, 53, 56, 58, 62, 64, 67–68, 71–79, 83–87, 249, 295, 296, 301–303, 305, 307 GDP INITIATOR, 173, 174, 195 GDP TARGET, 173, 174 GDPPC, 71, 72, 82, 83, 99, 100, 108, 109, 137, 138, 149, 252, 283, 287, 288 GDPPC2 , 71, 72, 83, 252 GINI, 56 GINI coefficient, 46, 60 GROWTH, 37, 112, 134, 137–139, 149 HOSTILITY, 134, 135 IMPORT, 164, 168, 173 INCOME INEQUALITY, 64, 71, 295 INFLATION, 37, 48, 52, 53, 99, 100, 113, 119, 137, 138, 149, 154 INGOs, 29, 30, 56, 60 INITIATOR CAPABILITY RATIO, 196, 199 INITIATOR DEMOCRACY, 174, 195, 199, 202 INITIATOR GDP, 199, 202 INSTABILITY, 99, 100, 111, 119 INVESTMENT, 99, 100, 110, 111, 118 LAGGED ENVIRONMENT, 214, 215, 234, 272 LEVEL OF DEMOCRACY, 214, 230, 233 LONG RUN CHANGE IN DEMOCRACY, 117, 118 LONG RUN PERCENT CHANGE IN GDPPC, 118 MANUFACTURED EXPORT, 173, 177, 194, 199 MANUFACTURED IMPORT, 173, 177, 194, 199 MID, 136, 137, 141, 173 MIDAB , 155 MIDDLE EAST, 56, 253 MINOR POWER, 173, 174, 196, 199 MISCELLANEOUS CONSUMPTION EXPORT, 173, 194 MISCELLANEOUS CONSUMPTION IMPORT, 173, 177, 194, 199 MISCELLANEOUS EXPORT, 173 OCEANIA, 253

348

Subject Index

model variables (cont.) PAST INEQUALITY, 71, 72, 77 PERIPHERY, 56, 61 POLITY, 109 POPULATION, 100, 252 POPULATION DENSITY, 214, 232, 236, 238, 252, 272, 273, 283, 284, 287, 288 POPULATION GROWTH, 99 PORTFOLIO, 10, 25, 36, 37, 40–43, 46–48, 50, 53, 56, 58, 71, 72, 75, 77, 78, 81, 83, 87, 293, 301, 305 POWER BALANCE, 174, 196, 199 PPP, 112 PRIOR DEMOCRACY, 37, 52, 53, 58, 99, 100, 102, 112, 117, 119, 137 PRIOR RGDPPC, 99, 100, 104, 118 PROXIMITY, 134 REAL GDPPC, 214, 231, 233, 236, 272, 273 REAL GDPPC SQUARED, 214, 231, 233, 236, 272, 273 REGIME, 134, 135 REGIME DISSIMILARITY, 174, 195, 199 REGIME TYPE, 214, 230 RGDPPC, 99, 102, 104, 111, 117, 118 SEMIPERIPHERY, 56, 60, 61 SOUTH AMERICA, 253 STABILITY, 134 TARGET DEMOCRACY, 174, 195, 199 TARGET GDP, 199 TERRESTRIAL ENVIRONMENT, 252 THIRD-PARTY MID, 137, 139, 150 TRADE, 36, 39–43, 47, 56, 58, 71–72, 74–77, 99–100, 111, 112, 118, 119, 137–139, 149, 151, 154, 155, 255, 256, 260, 261, 283, 287, 288 TRADE DEPENDENCE, 138 TRADE OPENNESS, 214, 232, 233, 236, 252, 254–256, 263, 265, 272, 273 WAR, 214, 218, 232, 233, 236, 272, 273, 275, 277, 282, 283, 286–288 WAR ABROAD, 273, 276–279, 287 WAR AT HOME, 273, 276–279, 287 WAR AT HOME OR ABROAD, 276–279 YEAR, 37, 42, 52, 53, 99, 100, 112, 119, 137, 138, 150, 154, 272, 274, 284, 285, 287, 288, 291 YEAR FIXED EFFECTS, 214, 215 YEARLY FIXED EFFECTS, 99, 100, 111 monadic, 3, 13, 126, 128, 129, 132, 133, 135, 136, 145, 147, 297

Monte Carlo Simulations, 115 Montesquieu, Baron de, 13 multicollinearity, 114, 197, 200, 234, 237, 286, 288 neoclassical economic theory, 105 neoclassical economics, 10, 63, 111, 118, 159, 296, 306 neoclassical trade theory, 242 North America, 3, 253, 260, 261 NOx , 15, 206, 214–216, 218, 219, 222, 224–225, 227, 229, 232–234, 236–237 occurrence of conflict, 133 Oceania, 253, 260, 261 OECD. See Organization for Economic Development and Cooperation OLS. See ordinary least squares ordinary least squares, 59, 81, 82, 85, 115, 117, 152, 285 Organization for Economic Development and Cooperation, 38, 51–53, 73, 76, 77, 87, 88, 101, 162, 253 panel corrected standard errors, 59–60, 87 panel data, 59, 116, 118, 140 percent change in DEMOCRACY, 42 policy inaction argument, 243 politically relevant dyads, 127, 140, 143, 147, 155 POLITY, 230, 259 POLITY III, 47, 48, 56–59, 61, 81 POLITY IV, 109, 195, 214, 283 polity score, 214 pollution, 1, 15, 206, 209, 211, 212, 214–218, 222–229, 231–237, 246, 248–250, 266, 268–270, 293, 297, 298 population growth, 244 Portugal, 96 post conflict effect, 133 PPP. See purchasing power parity preparation for conflict, 132 profit, 33, 67, 159, 163, 182, 210 purchasing power parity, 48, 72, 82, 100, 102, 109, 112, 231, 259, 283 R2 , 83, 197, 198, 234 random error, 35, 71, 173, 213, 251, 272

Subject Index random-effects estimator, 59, 85, 114 real gross domestic product per capita, 96–100, 102–104, 107–111, 117–118 Reducing Environmental Stresses composite, 222, 238 regime dissimilarity, 126, 136, 138, 141, 145, 178, 199 regime type, 16, 18, 47, 91, 214, 216, 219, 222, 225, 226, 230, 240, 255, 257, 258, 265, 296, 304 regression, 35, 81, 82, 85, 109, 117, 132, 133, 180, 182, 206, 234, 284 regulatory effect, 242 relative risk, 143, 144, 177, 198 responsiveness argument, 244 robust standard errors, 49, 59, 82, 85, 116, 145, 153, 197, 234, 260, 285 root mean square error, 115 rule of law argument, 244 Russia, 25, 46, 58, 94, 96, 106, 127, 140, 196, 269, 306 USSR, 130, 174 sample mean, 42, 102, 177, 198, 199, 218, 235, 238, 255, 276, 287 scale effect, 231, 245, 246 SEM. See simultaneous equations model sensitivity analysis, 43, 62 serial correlation, 49, 60, 82, 116, 153, 197, 234, 285 simultaneity, 50, 91, 97–98, 101, 107, 113, 114, 117, 126–128, 132–135, 140, 143–146, 155, 197, 233, 286 simultaneous equation, 50, 127, 140, 141 simultaneous equations model, 11, 13, 91, 97–99, 101, 102, 107, 113, 116, 117, 126, 128, 134, 135, 140, 144–147, 149, 150, 152, 299 SITC. See Standard International Trade Classification Smith, Adam, 13, 245 South America, 48, 251, 253, 260–262 South Korea, 96 Spain, 77, 96 standard deviation, 42, 43, 75, 83, 143, 144, 177, 178, 198, 218, 235, 238 standard errors, 49, 59, 82, 83, 115, 116, 145, 152, 153, 157, 234, 286 Standard International Trade Classification, 194, 195

349

structural effect, 241 supply, 14, 163–170, 177, 184, 187–193, 195, 303 sustainable development, 15, 249 Taiwan, 96 technology effect, 242 Texas, 23 theory, 96, 107, 110, 111, 149, 159, 173, 178, 180–184, 194, 196, 199, 245, 257, 273, 288 three-stage least squares, 101, 104, 113, 114, 116, 119 trade, 1, 5, 10, 13–16, 18, 19, 24, 25, 28, 32, 33, 36, 38, 39, 44–47, 50, 51, 53, 56, 62, 64, 66, 70, 73, 76–79, 81, 83, 87, 95, 100, 111, 112, 118, 119, 130, 138, 139, 147, 149, 151, 154, 158–195, 197–199, 205, 212, 214, 228, 232, 236, 239–243, 245–259, 263, 270, 273, 283, 293–299, 301–306 trade barriers, 258 trade flows, 10, 14, 147, 169, 178, 179, 181, 182, 185, 193, 197, 198, 228, 246, 247, 256, 304 trade liberalization, 25, 44, 248, 250, 303, 304 trade openness, 10, 16, 36, 46, 47, 53, 77, 78, 83, 138, 149, 213, 232, 240, 248, 251, 252, 254–258, 295–296, 298, 301, 303, 305 TSCS. See data Turkey, 96 turning point, 236, 237, 287, 298, 301, 302, 305 two-stage least squares, 101, 114, 116, 118 two-stage least squares Kiviet estimator, 101, 104, 115, 119 UK, 127, 134, 140, 155, 196 Britain, 25, 269 United Nations, 128 United Nations Conference on Trade and Development, 51 United Nations Environment Programme, 229, 243 United Nations General Assembly, 139, 151 use of force, 134, 138, 162, 173, 179, 180, 185, 202

350

Subject Index

variance, 49, 58, 59, 61, 85, 113, 197, 237, 260, 285, 286, 291 Variance Inflation Factor, 197, 200, 234, 237, 286, 288 VIF. See Variance Inflation Factor war, 12, 16, 61, 62, 92, 128–135, 138, 173, 179, 202, 214, 232, 235, 236, 279–283, 286–291, 298, 305 war channel, 244 waste, 209, 222, 229, 231, 246, 250, 261, 268, 269

World Bank, 47, 62, 64, 78, 195 World Development Indicators, 47, 81, 112, 113, 228, 229, 232, 259, 282, 284 World Resources Institute, 229, 282 World Trade Flows, 194 World Trade Organization, 78, 228, 243 WTO. See World Trade Organization year fixed effects estimator, 59