Political Survival Politicians and Public Policy in Latin America
Barry Ames
Political Survival Politicians and Public Policy in Latin America
California 5f;ries on Social Choi« and I'oli\;ul Economy Ediud by Brian Barry, Ro/nrt H. Baus, and Sam uel L. I'opkin 1. Markell and Statts in Tropical Africa: The Political Basis of .4g';. cultural Policies RonUT Ii. BATES
2. Political 1-:Conomits
JAMES E. AI.T AN D K. AUC C HRYITAL
3. Abortion and the Politics 01 Motherhood
KRISTIN LUKER
4. Haul Choices: How Women Decide about Work, ClrerT_ and Motherhood KATHLEEN GERSON
5. Regulatory I'o/icy and the Social Sciences
ROGER N OLL.
t,,,,,
6. Reactiut Rid: and Rational Actio", ManaginB Moral Ha:tard Insurl3nu Contracls
jn
C AIIOI. A. IU.IMEII
7. Post· RnJOlutionllrY Nicaragul3'- Slate, Cll3n, and Ihr. Dilemmas 01 Agrarian Po/icy
fORRfSf D. COUURN
a. f..»gys Uti Ihe l'utilkat &UllQmy
vI Rural
Africa
ROlleR... II . BA Il'.s
9. Peasants and King in Burgundy: Agrariall roulldatimls of AbsolUlism IIll.TON I.. 1l00T 10. The Clusal Theory of Jll stice I I. Principles of Group Solidarity
':,.."d,
KAR Ol. ~OHAN MI CH AEl. ~1F.C IlTEk
12. Political SUfl/h'al: Politicl;ms and Public Policy in I.allll Amr.,icl3 !lAIlRY
A M~:S
Political Survival Politicians and Public Policy in Latin America
Barry Ames
University of California PIn$ Bcrktley I 1..0$ Angela I London
Uni~rsity of Californ ia Pr~S5 lI.rkd~y and lo~ Ang.lr!, Californ ia
Universit y of California Press, luI. London, Englan d Q 1987 by The R.gcnu oflh. Uni,·c rsity ofC.liforni.
Fim
P~l'crb~ck
Priming 1990
l ibrary of Congress Ca taloging-in -Publi cation
D.'.
Ames, BMry. Poli,;,al ~u rvival.
ICalifo rn ia ~ ri .s on soci. land poli tiu l economy) Bibliography: p_ Jnc1ud.~indcx.
I. La,in Am .. ic.~ I'olitics and &""""IIl,nl ~ 1948 2. Pol iTks,Pr'ctica l. 3. G""ernlll.nt~pendingpolicy~ utin Americ.. l. TItle. II. Scrics. JL976. M4 198 7 320.9S 86 -24921 ISBN 0-510-06947- 1 (. lk _paper )
Pril\led in ,h e Un;l.-d S,.,•• of Am ••;,.
2 ]
4
j
6 7 8 9
To my mother and late father
rh ""
liliel
•
Contents
T ables and Figures
"
xiii
Acknowledgments Introd ucrjon PART I : THE THEORY OF SURVIVA L
7
COA l ITIONS
,
C ha pler
1;
The Po li tics of Public Spending
C hapter
1:
Survival Strategies and Expenditure T rade-affs
Chapler 3: With T ime
to
Brralhe: Policy in
74
Postcrisjs Adm inistrations P&RT
fI-
TWO Apr ) I CA T IONS OF T HE T HEOR Y
10 1
Chapter 4: The Congress Connection :
IO}
Politics and Expenditu res in Brazil's Competitive: Period
Chapter 5: When Soldiers Need Friends: The Sea rch for Influence by an Autho rita ria n Regime in Ret reat Conclusion: Po litical Surv ival and Comparati ve
& litics Appendix A: Technica l Data for Chap te r
I
Appendix B: T t"chnical D:lt3 for Cha pler
l.
Append ix C: T cchnica! Data fo r Chapter 4
Contents
viii
References
257
Index
'.
Tables and Figures
I.
Summary of HYPOIheses
Multiple-Regression Estimates: Public Spending in l.a tin Ame rica 3. Administ ra tion Residuals 4 . T he Milita ry Pacification Strategy During Political 2.
24 25
30 46
cru~
s . The Bureaucr:lIic Slr3tCgy: Estimated Share of AlIoca li ons ))rvOIed to Salaries in C risis and NODcrjSjS Sit.
48
uatioos 6. Gains and Losses to Pu blic Wo rks Programs When Burca ucralic Recru itment Strategy Is O ptimal , . T ests o f the Local Strategy During Political C rises 8. Tests of the T ransfer Strategy
49
9. Tem of Social Class Strategies 10. Increases in Adversary Programs During Po litica l
61 69
52 j6
~
l!. Effects on Key T fade-offs of Rising Budgets Du rin g PoIjlj ca l C rise·s
70
u . Effects on the Salary Share of Rising Inflation During Political C rises 13 . Military Coups and Prior Changes in Budget Shares 14. Successful Elections and Prior C hanges in Budget
];!.
7S 80
Sha= 15. Midterm Elections a nd Changes in Budget Shares 16. Adminis tration Type and Prior Upsurges in Illegiti mate Political Activity
81. 86
ix
Tablr5 and Figurrs
17. Prior Ups urges, Administration Type, and Currrnt Ra nking of Mllitary's Budget Sha re 111 . Administration Type and Prior Budget CutS 19. Administration Type, Prior Budget Cuts, and Current Ranking of Military Share ~o . Mi litary Pacification, Administration Type, and Expcnditure Response ~ I. Budgetar y Austerity and Administ ration Type u. Budgetary Auste rity, Administ ration Type, and Expendi tu re Response 21 . Cam pa ign Concentration and Key Committee .q . Campaign Concentr:ltion and Committee Efforts 25. One-Term Deputies and Amending 26. Average Permeability Ratings and Cn mmi tte~ Memo ht· rship ~7 . Permeability and Amendment Siz~ 28. Average Spending by Ccmral Governmcm on Slalcs in Rela tio n to Popu lation: 1947 196 4 29. Loans Granted to PrndlK'ers and Cooperatives by the Rural Credi t System : tnO 1981 \0. Final Expenditures of Central Governmem Ministries in Medici, Geisel, and Figueiredo Administrations 3 I . Regional Distributinn of Cent ral Ministrv Expenditures by Administration p. Projects Approved by Industrial Development Council: Fixed Investment as a Percentage of Value of National
86
8, 87
118 8') 90 120 121 IU
114
1 16
I
~8
174 177 18,
Iillal I,. Projects Approved by Industrial Development Council: Fixed I nvestm~nt as a Percentagt.' of Number of Na lional Total \4' Interstate Distribution of Loans of Ihe Bank of Brazi l: '970 19 80 }5. Interstate Distribution of Approved Loans of Ihe National Economic Development Bank : 1970-1 98 2
184
186 190
,
Tables and figure5
,i
Figures I.
unlral Government Public Spending in Three SelcCled
10
La ri n Ame rican N ario ns
Changes in M ilitary Spending in Brazil: 1946- 197' 3. Share of the Budget Devoted to Wages and Salaries in Costa Rica : 19 50-1 97 6 4. Public Works liS. the Military and Foreign Relations in Ecuador: 1947- 1 976 5. Budget Shares in Brazil: 194 5- 197 4 6 . Military vs. Health Spending in Bolivia: 1950-19 75 7 . Program Shares: 194 5- 1965 8 . untral Government Spending on Northeastern States: 1.
47
51 )4
65 67 11 0 1 11
194 8 - 196 6
9. Explained Variance of the Baseline Regression of Total Central Government Spending on States 10. Interstate Distribution of Loans to producers and Cooperatives 1 1 . Index of Loans Approved by Social Operations POrtfolio : 1964 1981 1 2. . Dimibution of COHAB Housing per State and per Urban Dweller: 1967 1982 I} . Distribution of BNH Investments: 1970- 1981 14. Percentage Division of Actual Expenditu res in Ministry of Edllcation and ClIltllre r5 . Evolution of Real Salaries by R:mges : November 197 9 to November 1983
128 16 1 169 171 17 2 180 2.03
Acknowledgments
Noc surprisingly, I ha ve accumulated a great many debts in the course of this projCI;"I. Rohert Bates and Peter McDonough read more than one version of the entire manuscript. In additio n to important insighls, they supplied the kind of encouragement that mmivates constant efforts 10 sharpen one's thinking and writing. Barbara Geddes read the manuscript for the press and made extremely thoughtful and helpful critiques. Samuel Popkin , an editor
of the California Series on Social Choice and Political Economy, pushed me 10 improve the organization and readability of the [ex!. Barbara Nunberg gave the whole manuscript an exceptionally careful reading, tightening it intellectually and edi lOrially. My colleagues at Washington Unh'ersity supported me in this work for a very long time. Their faith that it might finally amount to something was crucial. John Sprague told me so many times that my work constitllled a research program that I finally believed him and began to sec how it all tied together. He, John Kautsky, and Victor i.e Vine read the manuscript in various drafts and made mallY useful suggestions. David Felix and Peter Schwartz helped with particular chapters. John Woolley read numerous versions of the conclusion and aided me to see more clearly the connections between my work and that of others. Many Latin Amcricanists were generous with theIr time and interest. Gaston Fernandez commented on the whole manuscript, and John Bailey and Charl!."s Gillespie read major portio ns. In Bra· zil the University Research Institute of Rio de Janeiro let me use their facilities, and their journal Dados published an early version of Chapt!."r 4. I also had many useful conversations with Jos~ Mu· rilo de Carvalho. Walder de Goes, Glaucio Soares, Edson Nunes, and David and Gloria Vener. Witholll David Fleischer and his assistants the survey of Brazilian deplllies would nO( han' been possible. I I!."arned a great deal about politics in the Brazilian
xiii
•
xiv
Acinowlcdgm~nlj
NOrlheast from Teresa Haguette, Jardelino de Lucena Filho. Yves Chalouh, Rejane de Vasconcelos Carvalho, and Maria AnlOnio Alonso de Andrade. The Research and Documemation Center of the Getulio Vargas Foundation allowed me to use their oral his· tories, and Roberl Paekenham let me examine his imervicw$ wilh Br3Zilian deputies of the 1960s. M:'lnY former members and staffers of the Brazilian Congress patiently explained 10 me how things reall y worked. A project of this Iype is impossible without considerable finan · cial assistance. From 1976/0 1978 I was supporled by the Tinker Foundation, and during this period 1 received office space and computer support from the Lati n American Cemer at Stanford University. Ed Goff, then of Washington University, wrote com· puter routines for archiving and accessing Latin American budget data. My fieldwo rk in Brazil was supported by the American Philosophical Society and by the National Science Foundation, Gram SES-1I209454. In the era of word processors I typed the whole manuscript myself. but Jean D'Wolf ably did the tables.
Introduction
This is a hook ahout the coalition-building strategies of political leaders in Latin America. Strategic coalition theory begins with the assumption that leaders want to hold OntO their jobs. Why they want power is not 31 issue. In their own minds, leaders may seek power in order to assist certain soei:]l or ethnic groups, to improve the well-being of all citizens, to enjoy the trappings of office, or 10 get rich. None of these goals is attainable unless executives can maintain a grip on their offices. Mindful of the high frequency of unscheduled leadership changes in the region, Latin America's executives seek bases of support that will sustain them in office. The central focus of this book is the use by political leaders of public policr, especially public expenditures, as a weapon for survival. The centrality of public spending in this analysis of the coalition-building dfons of executives does nO( mean that other kinds of policy have no importance. Indeed, Chapter 5 analyzes four other policy areas along with spending. But because political actors think of expenditures as rewarding or penalizing different groups and regions, and because they care intensely about allocations, the budgetary process is inevitably a key arena of political conflict. Between groups and over time. budgetary politics is inherently a politics of winners and losers. The first part of this book builds a theory that focuses on no particular country or group of countries. It does not explain, for o:ample. why Argentina spent more in the 19505 on education than Brazil. or why Mexico spent less in the 19605 on the military than Colombia. Instead, the theory explains variations in public spending patterns among "administrations." Why did Ecuador's Velasco Ibarra spend more on public works in his second term than in his first term? Why did he spend more than Kubitschek in Brazil or Arias in Panama~
111 h
,
In ,roduction
To pUi this another way, the units of analysis-the "cases"are the terms of office (or certain yea rs within terms of office) o f individual presidcms. If a series of adminismuions has a common set of institutional relationships. these administrations constitute a "regime," Brazil bctwren 1964 and 1984 waS :l rnilitary regime with fi ve different administrations. Post- 19J 4 Mexico has been a civilian regime with ninc adminiSlrillions, and poSl-1 945 Argen. tina has had a number of regimes and various administrations. Everyday political discourse I::abels administrations with the names of their chief executives: the Thatcher government, the Rtagan administration, the Geisel years, and so on. In such usage the term "administration" includes the chief execurive plus his or her inner circle of advisers, and it excludes legislatures and political parties. In this book I use the names of executives in iust this everyday sense. Each executive has some trusted confidants. Whether every decision is made personally by the executive is irrelevant. In essence, r ignore the inevitable differences and disagreements th:n occur between presidents and their dosest advisers. What governments do, the policies of governments themselves. are outputs. What these outpms lead 10, their consequences for .society, are OUlComes. Government expenditures on education are an outpm; declining illiteracy is all oUlCome. In this book I am concerned onl y with outputs. Whether programs in public works are actually worth their cost is beyond the scope of my analysis. Whether social expenditures really improve the physical quality of life more under one administration than unde r another is an important question. but it will not be answe red here. To dis regard the ultimate consequences of spending is not 10 de ny their importa nce. but public expenditures are significant in and of themselves. Changes in outcomes usually presuppose changes in spending. Expenditures, in other words, arc generall y a necessary bUI not sufficient condition for other kinds of change. Spending allocations arc also important just because political forces fight over them. Political actors U Te abou t allocations, so it is profitable for leaders to manipulate programs as a way of rewarding or punishing friends or enemies. Because administrations afe generally short-lived. and because I strcss executivcs' efforts to secure their jobs. I devote most of my attention to determinants of strategy that fluctuate in the short
Introduction
J
run. In turn, factors that changt gradually-economic dependtnct on advanctd countries, fOT example-do not figure explicitly in the analysis. This is not to argue that dependence has no effect on political actors, for surely n:uional economic vulnerability affectS the choices open to executives. But such phenomena aTe virtually immmable from one year to the next or from one administration to its successor. They get overwhelmed in the decisionmaking calculus of political actors faced with immediatt threats of military revolt, major strikes, or rising inflation.
Explaining Public Expenditures What art the basic elemenls of a model of expenditure policy? In this book the distribution of public expenditures is viewed as a consequence of three clusters of causes: the constraints other political actors place upon the leader; the executive's own preferences; and the limitations of scarce financial resources. The constraints other political actors place upon a leader constitute a structure of political influence. A structure is a set of linked institutions that create claims, eithtr formal or informal, to participate in policy fonnation. Political structures regulate access to the executive. Access equals influence. Certain actors, such as a legislature with the right to approve a national budget, have influence rooted in law. Some actors affect policy through personal ties to the executive. Others trade information for influence. GroU?S such as investors affect policy because their cooperation is crucial to the attainment of leaders' economic goals (Lindblom 1977). Latin Americanists have thought a great deal about the char· acteristics of influence structures that affect policy outputs. One comprehensive list of such structures includes the nature and in· tensity of societal cleavages. the nature of the electoral system, the level of citizen participation in politics, the potential for coercion, the competence of administrators, the method of selection of the executive, the military presence in the executive, and the strength of the legislature and political panics (Remmer 1978). Much more progress, however, has been made in identifying these basic factors than in evaluating their joint impact on policy. Though statistical estimations are common, the dimensions evaluated are generally
""
l~ .
Ie
Introduction
resuicted to a few factors: competitive versus noncompetitive party systems, military versus civilian executives, and so o n. Failure to assess the joint effects of structural characteristics on policy has particularly serious consequences in view of the inherent difficulty of modeling the second cluster of causes. the preferences of executives. Preferences arc of twO kinds: survival and subSlandve. Survival behavior is designed to ensure holding onto the office itself. Substantive preferences-those the executive can implement if he or she maintains office-include everything else. Substantive preferences describe bencficbries from government programs, favored regions, the future of the society, etc. It is not easy to measure substantive preferen(:('\;. The public statements leaders make, fo r examplc, mix substantive with survi val preferences. Before making such statements, leaders calculate how others will react, what is possible. and who can be swayed. "T rue" substantive preferences become, in effect, unknowable. I avoid the problem of identifying substantive preferences by assuming that at certain times the predominant interest of leaders is the maintenance of office. Driven by surv ival motives, executives examine the competing claims made upon them and estimate the cost of attracting new supporters. This calculus leads to a set of policies executives hope will ensure their tenu re. If, as observers, we can make the same calculation of coalitional possibilities executives make, we can evaluale thei r actual stralegic behavior. The third element affecting OUlputs is the constraint of scarce economic resources. If execmivt"s h:ld perfect unders tandi ng of the future. thcy would incorporate resource constraints into their pref· erences; that is, the rank they assign to each objective would be higher or lower depending on its cost. Executives obviously lack sllch timely and complete knowledge of resource flows-indeed, even their information abou t near-term resources is uncertain. Moreover, executives wi th fixed terms of office can choose to postpone paying their bills. T hey can act, in effect. as if they were much less constrained by resource scarcity. As a result, the way executives incorporate resourcc constraints into preferences is nOI always the same, and it makes sense to treat economic resources as a separate intluence.
In,rotilldion
l
The Plan of the Book Pan I of this book analyzes cross-national time stries of ex· pendirures. This multinational statistical approach offers impor· tant virtues. Since the number of cases in Pan I is very largemore than a hundred separate administrluions and almost six hundred annual obstrvations-we can evaluate the efleas on spending of any potential influence while holding other factors constant. And since the range of relevant political characteristics is much wider across seventeen countries than within one, we can assess factors appearing infrequently in any single nation. Pan II takes the theory developed in Part I and applies it to a pair of Brazilian cases. Without denying the strengths of the crossnational analysis, thest single-country applications have vinues of their own. In the analysis of public expenditures, the focus on a single country allows us to examine the local discribution of policy outputs, especially the distribution of public works. Moreover, we can .analyze political actors below the level of the chief executive more systematically, and we can consider nonbudgetary policies that may be especially imponant in a particular country. The Conclusion summarizes the main points of the analysis and discusses the place of strategic coalition approaches in the study of compa!ative politics. Problems of coalition formation, panicularly during leadership transitions, rum out to be central in a number of diverse: studies. Scholars utilizing more traditional modes of analysis, especially modes thaI stress the primacy of social class, need to pay more anention to short·run survival motivations, becaUst the policy manipu l ~uions executives undertake during crises have effects reaching far beyond the crises themselves.
Part I: The Theory of Survival Coalitions
Part I develops and tests a set of propositions that constitute a theory of survival coalitions. Fundamental to this theory is the notion that political leaders cannot attai n their substantive goals unless they hold onto their jobs. In the turbulent politics of developing nations, leaders can never take tenure for granted. Political survival must be actively pursued by manipulating public policy 10 construct suppo rting w alitions. Public ex penditurn are central to survival coalitions. No arena of policy involves so many actors so imensely as public expenditures. Claims on the budget come fro m job-sctkcu, ~ono m ic groups, social classes, and regional interests, and the annual changes in programs unambiguo usly record winners and losers. Chapter 1 explo res budgetary politics in sevent«n Latin American coumries between 1947 a nd 1982. It does 50 in the simplest way possible, by adding together all the va rious programs a nd categories making up the annual central government budget in each country. Latin American exeCIJtives use the budget to rewa rd old followers and recruit new o nes, 10 keep bureaucra ts happy and the military at bay. Because their resources are limited. execulives spend most when their vulnerabililY is greatest, notably ncar elections and JUSt after military coups. The chapter demonstrates that a small set of causal conditions adequately ex plains fiuctu:uions in t01a1 public spending. These conditions include the occurrence of elections o r military coups. Ihe coming to power of an opposition party, suppon by a political party with a workingclass base. and the rise o r fall of potenti:d budget3ry resources. Leadership transitions lead 10 two expenditure cycles, one fo r elected gove rnmems and a no ther for milita ry gov('rnmems. Incumbent executives respond to 3ppro:lChing elections by increasing real 7
,
The Theory of Survival Coalitions
outlays, whereas newly elected leaders reward their followers just after laking office. Military governments coming to power via coups behave like newly elected civilian governments, raising (J(penditures in their own search for popularity.
Chapter 2, "Survival Strategies and Expenditure Trade-offs," broadens the concept of political transitions developed in Chapter I. During certain periods-crisis points-in the tenure of each administration. the executive's hold on office: is weakest. At these crisis points we expect survival preferences to dominate substantive preferences completely. All policies-certainly expcndilllre policies-are devoted 10 survival. The executive tries to determine the mix of budgetary trade-offs thar will maximize suppon. Will raising public sector salaries ensure the suppon of the bureaucrats? Will expanding health care increase working-class suppon? Will construction projects in a cenain town guarantee the support of the local boss? Because expenditure programs such as education, heahh, and public works affect individual social groups and regions differently, the executive builds a survival coalition through budgetary trade-offs: boosting some programs, cutti ng o thers, leaving still others unchanged. Chapter 3, "With Time to Breathe," examines the policies of administrations surviving their immediate crises. It begins by considering the effects of expenditures on long-term survival. Did expenditures help stave off military overthrow? Were electoral challenges overeome? The chapter then considers one of the oldest problems students of comparative politics face: How do military a nd civilian gove rnments differ in their spending on the armed forces? Finally, Chapter 3 extends the central argument of the book to begin an assessment of overall patterns of spending in administrations managing to overcome their survival crises. Five programs-education, health, public works, agriculture, and the military-are ranked according 10 their gains or losses from one administration to the next. This multiprogram comparison enables us to determine which programs lose when a selected program gains, and it is used to evaluate the explanatory power of such regime characteristics as civilian versus military executive and competitive versus noncompetitive party system.
""
1~",1
••
•
I.
The Politics of Public Spending
Public expenditures in Latin America rarely move up or down as smoothly as they do in advanced industrial nations. Consider Figure 1, which traces total central government spending in Argentina, Bolivia, and Peru. In Argenrina spending is remarkably volatile, with sudden leaps succeeded by gradual declines. Bolivia reflects the turmoil of a revolution, the consolidation of a new regime, and even a huge infusion of aid from the United Stales. Expenditures in Peru follow a much smoother course, hut then in four years tOla\ spending triples. These three nations are representative of Latin America's diversity. Indeed, the seventeen countries analyzed in this chapter illustrate just about every pattern of spending imaginable. Despite the apparent differences among these panerns, ate the processes that generated them roughly similar? Arc these processes in some sense "political"? This chapter establishes the political quality of public spending-political in the sense that expenditures are instruments that Latin American leaders utilize in their quest for security. The chapter focuses on total central governmeru spending rather than individual programs precisely to demonstrate that the whole budget, not just a few key programs, responds to the political needs of leaders. The chapter is divided into rwo main sections. The first, devoted to creating a thoory of expenditure fluctuations, emphasizes executives' motivations to build political support through government expenditures and the constraints within which such efforts operate. Expenditures respond to an electoral cycle, to the bargaining power of constituents, and to a military cycle. Moreover, economic, bureaucratic, and ideological considerations all constrain spending. The chapter's second section evaluates this theory with a regression model pooling overall central government spend-
,
Th~
TIII'my of Survival Coati/iOIl5
He
'"
,~
(
'" s- '" ~
J
I
'"
I
'"
~ "" a " '"
".:
~
•
/'
\ .... I
-" ~
">
""
c
";;
:/
! I
"' '" " " " ", ArgEln~na
Bolivia
Poru
Figure
I.
Central Government Public Spo:nding in Three Selecfed Lalin American Nations
ing for seventeen countries between 1947 and 1982.' The results of this statistical test are then illustrated in more depth through a brief discussion of individual countries. Appendix A :11 Ihe back I. The scvcnt«n arc Argent;na, Boliv;a, Bruil, Ch,k, Colombia , Costa Riu . Ecuador, EI Salvador, Guatemala, Honduras, Mexico, Nicaragua. Panama. Para guay, Peru, Uruguay, and Vennueb . The ab~nce of ."f/icient data led to the ex d usion of Haiti and the Domink'an Republic, and Cuba was excluded becau~c its budget after the Cub.n Revolution i. no t comparable 10 the OIhe...
l~""
.
Ie
a
The Politics of Public Spending
"
of the book discusses problems of data quality and sources, operationalization, left-out variables, time-series regression, and interpretation of results.
The Functions of Public Spending Sometimes we think of government as a referee or a broker responding to different kinds of pressures. In all societies, however, and partiallarly in developing nations, il is appropriate to view the government as an institution with its own preferences, usually in harmony with certain private interests and in conflict with others. This does not mean government always seeks the "public interest"; indeed, a government's policies may coincide with no interests but those of its leaders. Executives are almost always interested in maintaining office, and the power of the purse has long been crucial in their search for security. In the pluralist regimes of advanced industrial societies, political elites maximize their chances of reelection by coordinating economic policy decisions with the election calendarthat is, by reducing unemployment and maximizing per capita income near the date of the vote. l Latin American governments, burdened with volatile and open economies and lacking adequate technical advice, utilize spending somewhat differently. They spend 10 recruit and retain followers. Historically, Latin America has been a region of bloated and politicized bureaucracies, with substantial proportions of the work force dependent on government largesse. In Brazil, for example, the proportion of the working population in bureaucratic posts increased steadily during much of the twentieth century. According to Daland (1972:4) changes of regime rllreiy led to "throwing the U5(:3\5 OUI. H Instead, they were moved to innocuous positions, their places taken by new shifts of rascals. In Chile, Parrish (1970:18) found that the increase in bureaucratic employment between 1940 and 1968 was more than ten times the increase in population. New adminisuations regarded nearly all middle and upper bureaucratic pOStS as spoils, but displaced bureaucrats, as in Brazil, maintained their 2. Th~ notion of rh~ ~poliliC3 1 bu.in~ .. cyclc" wu made popular by NordhaU5 (I?7S ) and Tuh~ (l? 78). For an cxulknr review of rhis literature, Stt All and Chrystal (1983).
• •
..
T~
Theory of Survival Coalirio"s
jobs and salaries. The consequence, of course, was a cyclically expanding government payroll. Though leaders may believe the economic objectiYes they are pursuing will benefit the nation 35 a whole. specific policies can rarely be explained withom considering politiCllI effecls. Kahil (1973:330) demonstrates that in post-World War II Brazil the goal of winning the allegiam:e of the urban masses-while still serving other key groups-was as imponant as the goal of in· dustrialization. For Kahil. the suppon-maximizing aspects of Brazilian poli')' help illuminati: th(' incoherent and often inll:Hionary m!turc of economic choices, and they explain the crealion of hundreds of thousands of public sector jobs through the launching of grandiose projects. The political implications of expenditures cxtcild beyond ensuring that one's followers halle gO\'ernment jobs. Because endemic inflation so often plagues Latin Ame rican economics, 5alary adjustments for dllil servants and mi litary personnel become critical economic issues-nOt on ly for government employees themselves but also for economies dependent upon the purchasing power of these employees. The transfcr of funds to other levds of government may also further the political imercsts of central government executives. In Mexico, for example, the authoritarian regime responds 10 election results by manipulating its allocations 10 ind ividual stateS, sometimes rewarding supportive states and some' times bolste ring states where the opposition is threatening (Coleman 2nd Wanat 1973 :18). Job crearion, maintenance of real wages, intergovemmenral transfcrs-these goals 311 make sense in competitive governments. Latin America, however, is hardly a plun.list pan.disc. Is this an argument about a nonexistent world? It is not. Recent authoritarian experiences in the region notwithstanding, elections in the years since World W3r II have been frequent and significant. Be{Wetn 1945 3nd 1982, administrations in Latin Americ3 ended their terms with elections in eighry-{Wo instanceS and wi th military coups in fihy-one cases. Moreover, even in politics like Mexico, where the opposition only dre3ms of winning, elections playa vital role for lower-level politicians trying to demonst r3te their political skill. Last-and this is crucial-military and civilian regimes often feel the same insect.lrities. utin Americ.a's Pinochets and Stroess·
The Politics of Public Sptnding
.J
ners, with their substitution of massive repression for any sort of popular legitimacy, are really quite exceptional. The Electoral Cycle fkcau~ budgetary resources are limited, exeollives must allocate spending when and where it is most needed. The concern of leaders for their own survival is greatest just before and just after elections_ New jobs must be created to recruit additional followers; salaries must be improved to keep old supporters satisfied_ In Chile, this process tu rned the final years of most presidencies into "adminisuative fiascos, with programs completely subiect«l 10 electoral considerations" (Cleaves 1974:24). But the election might not relieve the pressure, for once in office an administration has to satisfy the demands of its new supporters fOf a piece of the public pie_ As a result, we can expect adminiSlr3tions seeking reelection and admi nistrations newly elected to i nCfea~ total expenditures in the year o f the election and in the year immediately after.
Constitumcy Pressures The claims faced by executives go beyond the cyclical demands imposed by elections. One such claim siems from the preferences of the vOlers to whom a party appeals and from whom it reeeives support. It matters little whether leaders and supporters agree on many issues. although such agreement is likely. Leaders respond to party constituencies because they desi re reelection. Parties with constituencies among the poor are less likely to be fiscally conservative.l Support from lower-middle and lower-class voters usually requires commitment to expanded social welfare and infrastructure projectS, and demands for such programs tend to increaK with the entrance of new panicipanLS into the political process. Latin American political systems (like most political systems) generally accommodate the claims of new entrants without injuring the intereSIS of older "power contenden n (Anderson J. FiK~1 ex>nservariVe5 may believe that prosp.-riry and tl«tor~llucc:ess ~ult from bab~ budgets. or thti r linirt for limited "P'"ndinS and low [nBanon mly OU fWt;gh the dnill: for (I«toral v;aory . At ,he nationallcnl in the Unil«i Stues, [)emocrlrs have mldirionally been less ex>nxrvat;ve /isglly than Rrpublic.ans.. Eu· ropan socialist parties Ire obviously less conservltive than Clmll:f Of rishrilt parries (Kir..:hcn and others 1964).
The Theory of Surviv13! CQalilions
'4
1967), so slXnding for welfare and infrastructure ra rely leads to compensatory decn:ases in other areas. Fiscal conservatism !>t-
comes impossible. Obviously the n:trure of political patties varies gready across Latin America. A leftist patty in Bolivia behaves politically q uite unlike a leftist parry in Argentina. The issue, however, is not just the absolute degree of "leftism" of any party but also its representation of groups previously excluded from political participation . In Latin America, political panies mobilizing forme rl y ex' cluded groups should be more likely to inc rease expenditures. No formula exists 10 determine which adminisuations have enjoyed the support of such parties. Alu:r 3 thorough search of the monographic littr::ature on L::atin Americ::an politics, I concl uded that incumbent administrations in twel ve n::alions enjoyed the backing of pa rties with bases in the wo rking class or lower middle class: Arge ntina, 1948- 1955 Bolivia, 1952- 1964 Brazil, 1952- 1963 Chile, 1965- 1973 Colombia, 1959- 1962, 1967- 1970, 1975-1982 Costa Rica, 1963- 1966, 197 1- 1977
EI Salvado r, 1950- 1956 G uatemala, 195 1- 1953 Nicaragua, 1980- 19112 Pa nama, 1969- 1975 Peru, 1964- 1975 Venezue1::a, 1959- 1968 , 1974-1978
A variable called " Ieftb ase" indicated the presence or absence o f a su pportive w9rking.c1ass party. When an administration has th e backing of such a party, expend itures should increase. Insecurity: Nelli ExeClitives and Electoral Margitls The security of executives is usually proportional to the length o f time they hold office. Newly deCled executives are the mOSt insecure. As upset demo nstra tcd for Sa sbtchewan's socialiSI gov· emment (1968:296-298), insecurity encourages budgetary exp an· sion. Administrations s«k 10 reward favored gro ups, bUi they avoid penalizing opponents. Unfortunately for new leaders, their lack of control over leg· islalUres a nd bureaucracies often thwarts their desire to spend and,
The PolitiC! of Public SpendinR
'5
as Ihe ends of Iheir lerms near, personal aUlhority declines. Cleaves, for example, nOies Ihal Chilean presidems begin Iheir lerms wilh considerable legilimacy and SlalUS bUi lillie underslanding of program implementalion. Wilh lime comes experience, bUi once lesser political leaders begin jockeying for positions in Ihe administralion of the next presidem, execUlives lose the ability to manipulate programs (Cleaves 1974:22). In sum, inexperience increases incentives to spend, bUi it denies the skills necessary for implementation. The net effect of these tendencies depends on the executive's ability to acquire necessary information and the speed at which authority declines. The vilriable "nonincumbent" measures the spending of executives serving their initial te!rms in office, but We! can make no explicil pfe!diction of its consequences for overall spending. Insecuriry also resulls from close elections. A term in office pro· vides an opportunity to use the public treas ury to widen the electoral gap between followers and opponents. In a case like Mexico, by contrast, executives lack serious opposition, so they are free to play the role of statesman, deferring the claims of job-seekers. Survivors of close e1eclions have no such luxury. The closer the previous election, the greater should be the incentive to spend. Just as the conslraint of ignorance coumerbalances the incentive of inexperience, the absence of political sup port checks the expansionary motivation of clost' elections. Executives surviving dose elections often confront hostile legislatures. Even worse, dose elections mosl oflcn occur in countries like posl-1958 Venezuela, where Ihe legislalU re wields meaningfu l aut hority over fiscal matlers. Why should the opposition support an executi ve's efforts at political machine building? In effect, the closer the electoral margin, the more limited is Ihe executi ve's ability to spend. Given this ambiguity in the margin-spending relationship, I made a number of attempls to assess the impaci of electoral doseness on expenditures. Initially, the difference between the percentages of the vote received by the twO leading presidential contenders defined the varia ble "'margin." Preliminary tests of this variable in the complete model produced insignifica nt coefficients-Ihal is, the size of the eiecloral margin had no impact on spending. I then examined the possibility thai eXlremely large and extremely small margins might both depress spending. That hypothesis, too, yielded insignificant results. Finally, close elections
,6
The Throry of Survillal Coalitions
(those with differences of less than ten percentage points) were identified with a dummy variable, but it is impossible to make a definitive prediction.
The Military Coup Cycle Military takeovers in Latin America arc normal events. Coups occur most often in the midst of economic difficulties. Prio r to a coup, real government spending has usually declined, because in periods of stagnation and inflation tax collections shrink, new foreign investments diminish, and foreign loans become difficult 10 secure. Declining expenditures may themselves contribute 10 coups, because spending cuts hurt the economic interesu of the armed forces (Villanueva 1969). Once in power, the victorious military faction justifies its coup by nming the economic diStress, corruption, and subversion it in· herited. Attempts will be made to rectify the various ills and in· justices, with the junta taking credit for any pros~rity that fol· lows. Sometimes the new government is temporaril y successful, but in reality the mili ta ry'S success is as likely to be mere luck, because coups lend to occur iust at the bottom of economic slides (Merkx 1969 ; N«dler 1974). Upward readjustments in military sai.aries n«d no t be accom· panied by compensatory CUtS in o ther programs. In facr, the limited evidence available, especiall y my study of the post·1964 Brazilian military (Ames 1973), indicates tha t newly installed officers acr like newly installed civilians-that is, they increase overall spending to win friends and reward followers. If the economy turns upwa rd, the military benefits from rising revenues, and its greater vigor in the collection of taxes further increases available reSOurces. After the initial boost, military regimes allow spending to dedi ne. The military rtlies more on coercion than on fr«ly given support, 50 its sense of secu rit y gradually increases without the rapid loss of authority suffered by civilian regimes. The military is also more autonomous from social forces relying on government programs, and its preferences aTe generally conservative. Finally, military regimes seldo m face truly open electoral challenges. Sometimes military regimes allow controlled elections, and sometimes they accept elections when the economy proves unmanageable or when the prestige of the armed forces falls sharply (as in Argentina
The Politics of Public Spending
'7
after the Malvinas conflict}. But such regimes really worry about countercoups by dissident military factions. Military governments try to prevent their own ouster by co-opting or repressing potential troublemakers. Transfers of recalcitrant officers to distant regions, bribes of cash or appointments abroad, forced retirements or assassinations-such efforts minimize the probability of countercoups. This logic suggests a cycle of spending in military administrations: The year in which the coup occurs should be associated with a decline in spending; the first year after the coup should witness a surge; subsequent years should be associated with declines. At the same time, students of recent Latin American politics know that austerity is nOt always the dominant motif of military regimes. In certain cases-Panama after 1968, Brazil between 1968 and 1973, Peru from 1968 until 1975 , and Argentina between 1969 and 1973-the military expanded the scope of government activities rather sharply. What caused these: expansions? Quite different sta rting points lead to fiscal expansionism. Leftist or reformist military regimes inevitably expand the role of the government. They rejcct the orthodoxy of ~gelling the prices right" as a developmental strategy, and they usc welfare: and job creation programs as carrots to CO-Opt restive working-class and middle-class supporters. In Brazil, by contrast, overall economic policy during the first years of military rule was a gradualist version of orthodox austerity and restraint. Perhaps these: orthodox policies were necessary to reduce wages and increase: investor confid ence, but they failed to lift Brazil from the stagnation precipitating the coup in the first place. Between 1968 and 1973, however, the government became much more aggressive and the growth rate soared. After 1973 economic growth slowed, but the regime continued its expanSionist t('ndeneies, because it feared the disruptive effects of a recession on its program of gradual democratization. (Chapter 5 treats this topic at more length.) In Argentina between 1966 and 1970, President O ngania attempted to control expenditures and "rationalize" government activities. These policies were partly successful in restoring GNP growth. but their social costs led in 196910 the violent outburst known as the "Cordobaz.o." In its aftermath Ongania fired the mininer of the economy, and in 1970 O ngania's military colleagues removed the president himself. As the military regime be-
"
The Thtory of Survival Coalirimu
gan a gradual withdrawal, it relUrned 10 budgetary expansion 3S a way of relieving social tensions.' Is there a general rule in these stories of righi-wing regimes b«orning big spenders?' Cenainly the cases have many common characttrislics. Both Argentina and Brazil had relatively sophisticated economies in which the government had long played a highprofile economic role. Technocrats in both military regimes were ideologically less purist than Chile's neolibcral " Chicago Boys." Both regimes switched policies when austerity programs failed. With so few examples, it is easy to inflate a couple of anCl:dotes into a principle, but the key appears to be sequence. After early surges in spending, right-wing military regimes face various options. In relatively simple economies, the mililary is fiscally conservative_ In more complex economies (whe re [he government i~ inevitably a maior economic force), the military remains anlistatist and conservative as long as the economy performs acceptably. If [he economy fa lrers, the milirary seem~ willing to repl:ace :Ii f:ailed economic tcam with one committed to expansionist policies. The only exception is Pinochet's Chile. where neoliberal economists were so ..... ell entrenched in policy-making positions that failures led to more rigorous application of the same policies. (See the conclusion to this book.) Long-lasting military regimes in each of these categories are actually rather scarce, so the entire scenario cannOl be included in a statistical model. Instead, the model evaluates the firs! four years of military government-four years in which the initial jump should be followed by a declining nend. The fina l section of the chapter examines individual cases for dues about the behavior of the long-term regimes.
The Constraint of Domestic Resources Comparative quantitative research on public expenditures was really born with studies of the policy outputs of the fifty United
•. n..- bal KIOI'''' on At~nlin~ in rhi! po:.i,,.! is William C. Smi.h. -C.isi! of .M Srale and Mil iu.,.·AuIMri larian Rule ;n Argcmina- (19801. A .i",ih r .1<>.,. mi",l be 101d. w;lh lhe ideologic:.l dirtction. reversed. in l'eru w;lh the changrovu from Vcbsco Aluraoo .0 MoralH s"r",ud ••. S. C1urly ~ K~I upan.ion is nOI merdy I eharactc,illk 01 ,he :IO'C~ II.d bu. ruu~r.lfic.au lhotilari3n regime •. Spo:nding in Brazil did ris.c durin!: the heyday of burClUCUlic IUlhoriurianism, bu, in Arg~nnn~ cxpcndi'U'''' shun k during ,h~ inmllanon ph~..e QI the t 966-1969 SA ~ta\e and txp.nckd only ~Ilcr i.. coll.p..e. Chi le under Pinochc, , tcldily redu ced public .pend;ng, .nd Plnama and Peru fail to qu.lify a~ bU!'C'lucuti c-a u\hnr;urian regimes. Sec O'Donnell {l97J1.
TIlt Politics of Public Spending
.,
States. This literature emphasized the primacy of the social and economic environment as a determinant of demands for governmental services. Urbanization and industrialization were thought to have created both the need for an expanding social welfare sector and the wealth to pay for that expansion. This argument, which assumes that political systems process the demands generated by the socioeconomic environment into outputs, makes less sense in Latin America. True., over the long run social changes increase the demands on political leaders, but the adoption of policies in areas such as social security can be explained much better as a result of a pattern of diffusion from wealthy to poor countries than as a response to demands made by populations at cenain economic levels (Collier and Messick 1973). Besides, in Latin America the public sector has always been large in relation to levels of national wealth, because government has provided the employment opportunilies lacking in perennially weak economies. Latin America lacks the industrializalion-demand creation mechanism that in the United States translates resources into expenditures, but it does nor follow that the empirical linkage will disappear as well. Economic growth should still lead to increased public spending, not necessarily because growth increases societal needs-most leaders alre::ady believe needs are enonnous-but simply because growth increases resources, giving the government the ability to respond to ongoing problems, ::and because the sheer level of economic devdopmenl creates expectations about the kinds of services government ought to provide. The economies and tax structures of Latin America are much more heterogeneous than those of the fifty Uniled States. Latin American nations earn subslantial income from import and export taxation, and they remain vulnerable to sudden shifts in commodilY prices. Rather than measuring Ihe revenue base of these sevenleen nations with a single indicator, the analysis separates the export-dependent countries from the more diversified. The former tend to have tax StrUCtures geared to exports, while the latter colleer more revenue from taxes on perso n::al income, property, and sales. Colombia, for ex::ample, colleered more th::an 50 perCent of its taxes in the 19605 from income ::and property and about 25 percent from impons and exports. Honduras collected less than 20 perCent of its l::axes from income and property and almost 50 percent from imports and exports (Instituto Americano de Estalistica 1971). In the model of public spending used in this
'0
Tile Th rory
of S"'lJilNll COillitions
chapter, the gross domestic product (GDP) served as the me:lSure of domestic rC:SOl,lrct:S for the $Cyen cou ntries in which th e per-
c(mage of revenue raised from direct taxes was 3t I,:ast 45 percent of IOlal taxa tion. For the ten counlfies in which diucr taxes ..... ere less than 45 percent of the 10131, income received from exports measured domestic resources.· When resources expand, onc c:m predict that public expenditures should rise.
The Constraint of International Resources Because Latin America depends so heavily on export earnings. balance of paymems positions are critical. Levels of imern;uional reserves. including gold holdings, International Monetary Fund (IMFJ drawings, and foreign exchange reserves, funclion 3$ impotunl signals of the heallh of the impon-cxport balance and as indicators of future restraints on growth. When reserves drop (or when debtJreserve ratios grow tOO brge), count ries may redutt cxpcndilU rc~. Sueh eUl ~ e:m be p:nl of :lIIemplS 10 righl tile fi · nancial ship without resoning to the lMF, or the IMF may impose them as a "conditionaliIY- fo r the extension of its assistam.:e. Information on the health of the reserve position is nOl available to Latin American execUlivcs quickly enough 10 :lllow inSlant ad· apt:ltion in spending. Instead, changes in the currell! reserve po. sition generally affect economic plans one year latcr.' In prelim. inary tests of the model, measures of reserves h:ld no significant effecls on spending. It seemed plausible, however, to wonder whether executives might respond only to large changes in re' serves, ThIS possibility was operationalized by crealing IWO dummy variables. one for increases in resc:rvcs grc:ater than 50 percent and the other for decreaS(.'s greater than 50 percent. We can expect spending to mo\'c in the same directions as these brge flu ctuations in rCSl'rvcs, When a country borrows from the IM F. it proceeds through a series (Jf credit lines. or Mtranches.- each with more stringcllI con· ditions. In tlJ l' higher tr:lIlches. the IMF may require the borrowl'r n:ltion to agret: to a formal program of economic reform. Since 6, The CDr'r"c~'ur~d M~xi co, ~",l r~nJ"'A, Thr
""!lnlli .. ~tt A r~( tllin" , 1I,~zil, Chilr, Cnlombi~, P.m, rxpo,Hnu\urtd wunuits indude IIoli,ia. Riu. Eru~Jo,. 1:1 S~I.ador. G~'lt.m"I". H otldu,~" NicJraguJ. I'Jragu3y, UruGu ay, and VrnuudJ, Th. d,la com. from Iht IM~'s /n"""",lo",,/ I'm,m ,;"/ St"r'II!n. 1, 1k"u~ '<Srr""" ur • funn;on of the "gj:r.g~l • .<11.< "I Ihe econo my. Ih~ ,<:tull IndICa I'" " Ih. ,alio "I '."''' •• [0 Ih. Gnp,
Co."
The Politics of Public Spending
such programs almost always call for fiscal austerity, the existence of an IMF program should be associated with declines in public spending.' At times, grants from the United States have constituted an important source of revenue in Latin America. Aher the CIA helped overthrow Guatemala's President Arbenz in 1954, the U.S. government sought to use aid to make Guatemala a showcase of democracy and capitalism. In Bolivia in the late 1950s. the United States supplied up to 40 percent of the central government's budget (Wilkie 1971). And in Chile between 1964 and 1970, Washington's interest in the Frei administration led to another sharp increase in economic aid. The potential impact of foreign aid depends upon its size relative to the country's economy. Each nation's gross domestic product was therefore used to deflate annual U.S. grants. So adjusted, we can expect grants to be positively and significantly related to spending. Incrementalism i" Public Spending
In the short run, decision makers have only a limited ability to affect the level of overall spending. Once started, programs develop their own momentum. Earmarked taxes often fund government-owned enterprises and autonomous agencies. Retirement pensions may be untouchable. Dubious projects begun in the distant past continue receiving public funds. As a tesult, levels of spending change in small increments and the potential influence of political factors is therefore reduced. In the United States, "incrementalist" theories explain why there are limits on the short-run flexibility of expenditures (Wildavsky 1964).' These theories argue that spending in prior years is the beSt predictor of current spending. Previous spending is not, however, likely to dominate the statistical explanation of Latin American expenditure movements as it does in the United States: 8. For analyses by lMF economists of the effectS of IMF programs on coumries' behavior,"" Rciclu:nann and Still..,n (1978) and Kelly (1982), The programs ana' lyud by the IMF economists were used here as well. 9. fatly apmditute analysts argued that changes in expenditures are incremenIal becausc decision makcn have limiled cognitive ability and high motivalion 10 avoid conflict. But Gist [19n) and othen discovered thaI the mosl controllable ptogra .... in the U.S. federal budget are those thai conform least to the incmnental
modd.
"
"
Tht! Tht!Ory of Surflival CO
Latin American budgt'ts arc legally much more manipubble in the shon run than those of the United States, political struCturt's arc less institutionalized, and resource fluctuation s are greater. Still, although incrementalist influences will not overwhelm political and economic efft'Cts, we should see a signifil;am positive e{{t'Ct. We can t'XpeCt spending in any year to be positively rel:u ed fO spending in the previous year. 1'he Diffusion of Deve/opmentalism M
M
In an aut'mpt to promote t'conomic devdopmt'nt and prevent Tt'pe titions of the Cuba n Revolution, a host of intemational agencies t'xpanded their activities in Latin Arm-rica in the 19605. For the United Nations, the period became known as the "Development Dt'cade." With the help of the UN, the Alliance for Progress, and the international lending instilUtions, a new generation of technically trained :administrators encouraged their governments to takt' a more active rote in economic management. More resources became available, and (partly through international efforts to improve tax systems) mo re revenue was squeezed out of existing resources. Demands for government allocations with;'1 Latin American na· tions also grew. As Latin American societies became more competitive politically, and as more groups pressed their demands, political dialogue moved slighlly to the left (Schmitter 1971b; Anderson 1967). Everyone became. in Albert Hirschman's words, a Mcard-carrying devcloper. If this change in the political cli mate led to increases in public spending, how was the spending financed? A substantial share of the expansion of the late 19605 came from international and domestic borrowing, as Latin America spent roughly 15 percent more than it took in. A few countries, p:l.rtil;ul:lrly Argentina, Chile, and Peru, at limes ran up deficits equal to 25 percent of tax receipts. The consequence of continual deficits was a heavy burden of dcbt servil;e. In 1965, service p:aymentS on external debts ave raged 9 percent of the value of exports for all of Latin America . By the [ate 19605. Argentina, Chile, and Peru we re spending one· sixth of their export receipts to service external debts, and in 197 1 service payments requi red 13.5 pereent of exports for Latin America as a whole (Ruddle and Barrows 1974:520-52 1). Such levels of debt service pale in comparison, of course, with the burden of debts M
The Polities of Public Spending
'3
accumulated after the 1973 and 1979 oil shocks. Were pre-1973 levels of debt service seen as a drag on spending? Apparently not. Debts could simply be pushed forward, postponed to be p:lid by future administrations. After 1973, of course, the cumulative burden of debt made its cost unbearable even to incumbent leaders. In the more advanced economies, increased government intervention was linked to policies of import substitution. Although import substitution led to a brief period of rapid expansion, growth eventually slackened. Politicians, unfortunately, were late in recognizing the boom's end. They continued through the sixties to spend heavily on projects of little merit, finally contributing to inflation and, in the end, a decline in the rate of growth (Hirschman 1968). This expected jump in spending was a sudden (or discontinuous) shift in the middle sixties rather than a gradual increase over the whole post-World War II perio(LAfter some experimentation, the period from 1965 to 1975 was selected as hcst reprC5enting the Development Decade. In this period we expect spending to n~.
Analysis The first section of this chapter linked, one by one, a series of political and economic factors to movements in central government public expenditures. Table 1 summarizes these hypothetical linkages. To measure the independent effect of each factor, the individual linkages must now be evaluated together. The appropriate technique for such statistical evaluation is multiple regression. So that data from seventec:n countries can be evaluated jointly, we must convert the monetary variables to constant local currency (using the price deflators of the IMF) and then to 1963 U.S. dollars. Next we index the constant dollar figures so that 1963 equals 100. An index is also used for the annual ratio of reserves to GOP, and U.S. aid grants are measured as percentages of GDP. All other variables are dichotomous "dummies," taking the value of 1 if a condition exists and 0 if it does not. The dependent variable-total central government spending for each year-aggregates all types of expenditures: ministerial budgets, decentralized agencies, debt payments, bond purchases, and so on. Table 2 presents the results of the multiple regression. In the
"
"
Tht' Theory
f'f Surl/il/al
Coalitions
Table I. Summary of Hypothests Independent
V",i~bl
• Po/itic,,/" V~n"bles El
Marginl.ow, Nonincumbcnt, Coup, Po,;troup, I'onroup ... I, POSl("OUp t 2, POSiroup t J,
Pred,cud Erf,",,1 of Indepc ndent Variable On T"I~ 1 Spt' nding
In cr.a..;e In=3se Increase Unclo .. Und.~r
DNCr.ase Iner •• "" < 1'0. I("Oul',
<
PO. 1COUp
<
Po«roup
·Setting" Variablfs Spcnd;"s $, , Ruout« . S, Re,ervn S, _, l~tF Ag,e.menl, Grants, Devolopmom,
• I, •"
Incrc~>c InCt~.,~
(ncrca,< tk"caso: In~...,aso:
Inc...,...
case of the monetary variables, the coefficients should be interpreted as the number of percentage points the index of spending changes when the indexed variable changes o ne percentage point. In the case of the dichotomous dummy variables, the coefficients measure the perctntage change in spending when the selected condition is present. The resulls fit the theory wdl. The modd explains more than 90 percent of the variance of tOlal spending in the seventeen countries, and the explained variance is not less than 70 percent in any country. In general, the coefficients are substantial and carry the appropriate signs. When a hypothesis forecast the order of a series of coefficients, the results are supportive.
Domestic and In/emational Resources, the Effect of Prior Spending, and the Development Decade Consider the context in which support-maximizing budget makers operate. Increases or decreases in resources, wherher domestic
The Politus of Public Sptnding
"
Table 2. Multipk· Regrcssion Estirnatts: Public Sptnding in LAtin Ammca Indepmdenl Variable
Parameter
Inlnapt El«rion, Postekction, Lntba~,
M...rginl.ow, NoninculllbcTll t Coup, POSlOOUPt Postcoup + I , PO'itcoup + 2, POSlcoup + J, Spmding St~ ' RcooW"CeS S, R~rvtS Up, . , RO$tfVC< I)own, . , IMft\~"
Gums, Development,
Estim~et
-8.975 6.267 - 7.569 8.082 - 3.275 16.482 -4.270 2.I43 -1.619 - 3.769 - 0.029 0.922 0.201 10.338 -IS62 - 0.253 !S.7B 1.669
Slandud Error
T StalisbC
2.1S 2.21 1.8 1 2.29 3.35 3.82 2.8 1 3.10 3.U 4.0S -4.82 0.02 0.03
-4.174 2.830 - 2.690 3.SH -.976
J.52 2.7 1 2.07 120.61 2.67
43"
- 1.S I8 0.72.) - 0.4n - 0.932 - 0.006 43.783 6.913 2.944 - O.94S
- O.lll 0.213 0.625
NOk: II' _ 0.93: DF _ $95.
or intcrnational. were cxp«lcd to lead 10 parallel movemenls in total spending. In the case of domestic resources, Ihe hYPOlhesis proved correct: When domestic r~urces rose or fell 1 percent, spending rose or fell an addilional one-fifth of 1 percenl . Changes in internalional reserves, however, had a more complex effect. Pre· liminary teslS revealed that only large mnve menrs in reserves had any effect on spending, but we eXpC:cted rymmetriCIJI effects- thai is, large movemenlS up or down should slimulate parallel changes up or down in cxpenditures. In fact, big improvements in reserves did encourage executives 10 spend more, bUI big declines elicited only weak and statistically insignificant cuts. IO Leaders prefer to increase: spending. SO good news mmivales them more [han bad news discourages them. Contrary 10 our expectation, recourse 10 the Inlernational Mon10. 8cfW..,n 1947 and 1912, dlanges from one yur fO fh~ ..."ff of SO pornnt Or mOre occurred forry-fWO fimes in our .. venf..,n counfries.
The Theory
,6
of SUrviI.lD/ Coalitions
clary Fund did not affect spending. The really draconian phase of IMF intervention began, of course, just after the end of this analysis. Ignoring the IMF may no longer be casy, but the major dcblOrs still seem willing to ny. Through 1986. neither Brazil nor Peru met any of the: IMPs fiscal deficit targets. Mexico made progress in 1983 and 1984, but the fiscal floodgates rcopened for the 1985 congressional election (Bagley 1986). Aid from the United Stales was expected to ha ve a positive impact on spending, bUi the size of aid granls is extremel y variable between countries and over time in the S3me country. so sliuisrical verification is difficult. In facl, the regressio n revea led no consistent link between gtaniS of aid and spending. In cerlain casesBolivia in the I:n (' fifties, G u:u emala after the o venhrow of Arben2., EI Salvado r durin g its civil war, Chile under Frt'i-U.S . .aid undoubtt'dly added to overall spending. The sma ll amounts rt'· ceived by other co untries produced no mt'asurablt' d fects. The weight of past spending had, as ex pected, a smoothing effect on currem spending. Increases o r decreases of 1 percent in spending in the previous yea r led to increases o r d«rt'asts of 0.92 perccm in the current year. The statistical reliability of this result is especially notewonhy. Past fluctu ations in spending have a con· stam impact on current spending regardless of time o r coumry. Indecd, the weight of the paS! is the most consistent va riable in the model. The years from 1965 \0 1975 wt're expected to be a time of higher spending. They were nOl. 'n a n estirn:uion made when data were available o nly through 197 1 (Ames 1977), post· 1964 ex· penditures rose about 9 percent. After the energy crisis, however, spending increases in the o il·importing nations shrank so much that the overall 1965-1975 results are insignificantY The Devel· opment Decade, in the end, was something less.
The Electoral Cycle, Constituency Pressures, /nsu IITity, and the Military Cycle The political as pect of the budget was strongly and unequivoc.ally confirmed. Executives respond to political incentives and constraints with substa ntial cha nges in overall spending. 11. Ika use 0 / other pt,/«tly a>tnpa,able.
im pl'OYcm~nn
in the rnotkl, th_
carlin .esuln are nOt
"
The Politics of Public Spending
'7
Is there an electoral cycle? Because governments running for reelection sought to buy support, spending before electoral tests rose more than 6 percent. 12 For postelection spending. however, the prediction of increased spending proved quite wrong. Spending actually declined by the same amount it had risen the previous year. NOIe, however, that nonincumbent administrations increased spending more than 16 percent. Since most incumbents lost or could not run, newly elected administrations were typically nonincumbents as well. The huge nonincumbent effect swamped the postelection effect-that is, the few incumbent leaders or parties who were reelected immediately began lowering their expenditures, perhaps to deter inflation. Newly elected leaders postponed that readjustment while they constructed their own bases of suppon. Did the size of the electoral victory affect spending? No unequivocal prediction could be made earlier in the chapter, because the stimulative effect of a close election might be countered by the depressive effect of the executive's political weakness. The results reflect the evenness of these opposing forces: Actual spending hore no relation to the size of the electoral victory. This model attributes no special significance to the nature of the election itself, that is, to its importance in legitimizing and constituting national authority. Still, Latin American elections vary considerably. Mexico, for example, has regular but hardly competitive presidential elections. In EI Salvador, elections have at times appeared open, but clearly cenain competiwrs are never allowed to win. In Colombia. elections between 1958 and 1974 simply ratified an agreement between the major panies mandating alternation in office. In other countries, including Costa Rica, pre-1973 Chile, post-1958 Venezuela, pre-1972 Uruguay. and pre1964 Brazil, elections were true determinants of leadership. Is the electoral-budgetary cycle in such cases heightened? In other words, does the openness of the election stimulate spending? In order to evaluate this possibility, I examined competitive elections separately. Their electoral cycle coefficients, however, differed little from the aggregated elections whose coefficients are reported in Table 2. 12. If w~ npcriment with the dcaoral cycle by 3dding dummy ... ariables fnr j,trond year before an rleaion, it becomes even more evidenr thaI the jump in ,pending occur> typically just one year before Ih~ ronlesl. th~
"
"
The l1uory of SurVIlld/
"
Co.JI' lio~s
In addi tio n 10 the cyclical effects o f eleCiions, pressu re from wo rking-class conSlilUcnci~ was ex pcCled to boost 0\-er311 spending. In adminisuations supported by pa n ics wi th worki ng-class backing, IOlal spending inc~a sed mo re tha n 8 percent . In essence, even modest class polarization o f the party system raised the COSt of rCl:lining o ffice. Natura lly such polarization a ffects some government p rograms mo re th an OIhers. In the nex t cha pler WI.' Sh311 examine the consequences o f class polarizalion on trade-offs between progra ms; here it is imporlant to note merel y th aI demands fo r cerlain prog rams a re not matched by compensato ry cuts in others. Wo rk ing-class demands are ac ro mmodaled al the cost o f overa ll budgeta ry expansio n. Is there a mili tary coup cycle? Thc results arc not extremely solid statisticall y, hUi they d e,Hl y move in the right directio ns. Coups occur d uring economic recessions, at times when overall spending is shrinki ng. U Ikeause military regimes initiall y bring or· der, bt: nc r tax collection,
~IIlJ
higher public seeto r salaries (espe-
cia lly fo r the armed forces), and because military regimes seek 10 buy po pularity, ex penditures rise more than 6 percent fro m th.· year in which a coup occurs to the first subsequent year. Then declines set in, lasting until the fo unh year o f milita ry gO\'t~rnm ent. Military gove rnmems, in other ..... o rds, behave for o ne yea r like civilians. Then their sense o f secur ity increases, and they begin 10 em phasize austerity and o rthodoxy. What about the bureaucratic-autho ritarian cases and the few left-leaning milita ry regimes? If these regimes truly follo wed d is· tinctive co urses. courses not captured by o ur theory, the model sho uld systematica lly err in its predictio n o f their ex penditures. The spending of ex pansio nist ad ministrations should be underpredicted. and the spending o f o rthodox regimes sho uld be o ver· pre
T~
'.
Politics of Public Spmding
Chile and Velasco Alvarado in Peru, ideologically farthest apart of all the military leaders of the 19705, were both within I percent of the model's prediction. In these cases the factors included in the model explain spendi ng quite welt, so we can conclude that bureaucratic authoritarianism is nO[ characterized by a distinctive ap proach to public spending. Indeed, the only consistent errors in predicting the spending of military regimes carne with regard to the populist caudillos of the late 1950s: Odria (Peru), Rojas Pinil1a (Colombia), and Perez Jimenez (Venezuela). Each spent considerably more than the model's prediction. Populist authoritarianism, in other words, was truly the missing factor in the model.
Some Deviant Cases Any statistical model fits some cases better than others. An ex· amination of poorly explained cases can help us identify important faCtofli left out of the original theory. Table 3 includes the 40 administrations (out of 102) deviating most (either underspending or overspending) from the model's prediction. 14 A number of cases were extreme for quite idiosyncratic reasons. Paz Estenssoro's firstterm underspending in Bolivia resulted from the general chaos following the 1952 revolution, and the overspending of Paz's second term and the subsequent administration of Siles Zuazo were made possible largely by huge infu sions of U.s. aid. Colombia 's Ospina Perez confronted the beginning of the violencia that wracked his country and nearly became a civil war. In Guatemala. foreign grants substantially in Oated the budget of President Castillo Ar· mas. And in Nicaragua, the Sandinista government faced an enor· mous rebuilding task after the depanure o f Somoza. These exceptional C
Earn of !hac
adminJmations Mid
pcI¥o'i:f
ar leur fWO years.
The Theory of Survival Coalitions
3' Table 3. Administration ResidUiJls - Maximum
'-
Ovt!t$pntd"$~
n.
Al,m an (Mexico)
".
2. Pn, 1961- 1964 (Bolivia ) 3. Sil.. Zuuo (Bolivia )
U. 14. 15. 16. 17.
<. Ruiz Cortin.. (Mex ico)
,.
Ospina Pan (Colombia) 6. l<:mu ~ (EI Salvador) 7. Perez Jimtncz (Vcnczuda) 8. FSlN (Nic ragu. ) Lu is Somou (Nicaragua ) 10. ea"ilIo Arm • • (G u.!c ma l~ )
>S.
,.
".
t.op.,~ Portillo (M.,,;co) Galvez (Hondura.) ArtvaJo (Guatemala ) Figueres, 1953-195& (Cosra Rica ) Pach«o (Uruguay) Uuas RCSlrq><> (Colombia ) Odlibe, (COS!, Rica) Iklaundc, 1963-1968 (Peru)
RaIdc'>. (Ecua dor)
20. B.nur (Bolivi a )
:Uro O.vianc. from Prediction
2'-
3'-
Ba rrientos (Boli via) I'rodo (Peru ) 23. Owri o (EI Salvador ) H. Rivera Carballo (EI Salvador) Robl ... (Panama ) 26. VJlencia (Colombia)
".
".
F. Gesrido (Uruguay) lB . Mendez MOnfenegro (Guar(m~ l a) 19. Echandi (Cosr~ Rica) 30. Dial Ordaz (Mexico)
IkI.rI
32. Son.hn Hcrnin
".".
"Maximum U"dnspmdt,,· Not" Praid...", .... h"«l in 0'.1., of ,h.i, .I..i.,i"" lrom ,I>< modd', p,«lic1icn, Iron! tl>< S'=,.., o. .... pm ... r jAI....:ln ) ,I>< un
,0
y."'"
Comparing overspenders with underspenders also confirms the argumenl Ihat the competitiveness of eleclions tarely determines budgetary pressure. Mexican elections are hardly examples of effective competition, but Mexico placed three ad ministrations among the top overspenders. Costa Rica, with very compelilive elections, and Colombia, where elections were much less competitive, each placed presidents al both extremes of the spending spectrum. Competition may increase incenlives to spend, but it denies executives the power to do so. When compared to the other executives from his country, an
"
"
,.
The Politics of Public Spending
administration like that of Chile's Carlos Ibanez seems quite anomalous. Total spending for all other Chilean presidents-AIessandri, Frei, Allende, and Pinochet-fell within 3 percent of the model's prediction. Why did Carlos Ibanez underspend by almost 22 percent? In a nation whose political panies were the most classspecific in Latin America, Carlos Ibanez was elected "above party." Chosen because of his great personal popularity and owing election to no single social group, Ibanez envisioned a kind of Peronist future for Chile. He found it easy 10 avoid spending, but Peronism was inappropriate in a polity where the party system accurately reflected social cleavages. Ibanez's Peronist strategy failed, and his administration became unpopular in practically all quaners (Bray 1967). The two widely separated terms of Peru's Fernando Belaunde Terry illustrate the complexity of the relationships among incentives, constraints, and ideological preferences. In his fim administration (1964-1968), 8elaunde confronted a strong labor movement and a hostile legislature dominated by the opposition apristas. Labor's power-particularly its potential for violencecould compel favorable wage settlements and bring down governments (Payne 1965). In 1967, for example, Belaunde's budget included a wage increase for public employees of only 10 10 15 percenr. The unions reacted strongly, and the apristas-political allies of labor- backed them. Though APRA had claimed to be a "sound money" party, it would not support Sclaunde in im· posing new taxes and it refused 10 oppose the unions' wage demands. As the deficit rose, 8elaunde overspent the model's prediction by nearly 8 percenr. His term was abruptly ended in 1968, however, by military ouster. In 1980 the military withdrew and 8elaunde was reelected to the presidency, but the economy was in disarray as inflation and the foreign debt soared. After an initial boost to the budget, Belaunde opted for orthodox economic policies. Overall spending in his second term fell below the model's prediction by nearly 4 percent. Why spending policy was so different remains to be explained: The structure of politics had changed; the «onomy had changed; perhaps even 8elaunde himself had changed. Finally, these deviant cases suggest that sequence matters. In 1958 Colombia's rwo parties signed a formal power-sharing agreement. This agreement specified alternation in the presidency,
"
"
The Theory of Survival Coalitiol1S
J' ~quality
in the legislature and bureaucracy, and heavy majorities pass legislation. Such an agreement could work only if the two panies were rdatively close on substantive issues. In tenns of our model. however, President Guillermo Leon Valencia underspent by more than 12 percent, while his successor, Carlos L1eras Restn-po, overspent by nearly 9 percent. The key 10 Lleras's reversal of Valencia's policies lies precisely in the requirement of party alternation. Lleras sought to undo the destabilizing effecls of Valencia's conservative economic policies. Realizing that his party could not succeed itself, L1efa.'> spent heavily in order (0 wmmit the next presidem 10 his programs (Latin America 1967: 111152). Since the spending of Misael Pastrana, Lleras's su(:(:cssor, matched almost exactly the model's prediction (neither under- nor overspending), Lleras must surely have felt vindicated. This cursory look at the model's under- and overpredictio ns provides confirmation that the model works. It is also an exercise in speculafion- an exercise that leaves plenty of puzzles and clues for furth er explo ration.
10
Conclusion This chapter has sought a plausible, general explanation of lO\al central government expenditures in Latin America. The model developed here sees the government neither as a mere referee for the conflicts of social groups nor simply as a representative of some domina nt economic interest. Instead, the model emphasizes the survival motivations of political leaders. Public expenditures, as an inSTrument for the realization of leaders' goals, become political expenditures. Most Latin American leaders would like 10 satisfy every dema nd fo r public money, bUi they cannot: In poor and volatile economies, resource constraints are ever present. Fluctuations in the gross domestic product and in exports exert a powerful influence on spending. But the pressure 10 spend must he intense, because leaders ignore wheneve r possible such economic warnings as falling international reserves and IMF agreements. Spending cannot, in this difficult economic environment, rise constantly, so ext'cutives husband their resources for use against the strongest threats. Some leaders must accommodate the demands of working-class conSlituencies for job creation and social
"
"
Tht Politics of Public Sptlfdi1lg
JJ
programs. Othtrs face tht cyclical threat of d ections. Bc:fort elections, spending rises as incumbents try to ensure electoral success, and after d ections spending rises again if a new leader has a«ained the presidency. The importance of d ections in constituting national authority was not a factor in the size of spending increases. Why do elections of varying significa nce dicit similar expenditure responses? O ne answer is that the more important the election, the more likely it is to be close and the tougher will be opposition to the executive's attempts to buy victory. Another answer is that achieving the widU t possible margin of victory may be important even when an election obviously will not be close. Mexico's PRI cannot lose, but the campaign is ta ken seriously by a host of lower-level politicians on thei r way up. Finally, what about Latin America's many authoritarian gO\'ernments? In light of the repressiveness of a Stroessner in Paraguay or a Pinochet in Chile, is the political function of public spending irrelevant? It is not. Classifying governments as authoritarian and nonauthoritarian ovtrsimplifits: In rtalilY few govtrnmenls are 50 securt that the budgt t is not a tool fo r incrtasi nl;! support. Military governments may opt for austerity once: they fed safe, but in thti r first ytar in power thty Covtr thtir political bascs. Aggregale spending remains, however, a blunt political instrumenl. Guidi ng expenditures to serve political needs requires precist strategic caiculuions on tht part of txttuti vcs. Tht ntxt chapter tlaboratts this model of u tin Amt rican political survival by disaggregating lotal spending into the specific programs that bind togetht r the executive's survival coalition.
2.
Survival Strategies and Expenditure Trade-offs
This cbapler analyzes the strategies chosen by Latin American executives to guide the allocation of public expenditures into individual programs. The central question is whether executives ado pt allocation policies increasing their chances for sun'ival in office. Since the choice of an appropriate policy depends on each country's political and soci31 environment, the firSI seerion of the chap. [t' r ('xplores the conditions t'xecutives consider when they frame their strategic approa(bcs. The second section evaluates five strategies available to Lliin AmeriGln political leaders. Each str.:llcgy stresses specific programs of public expenditures, and each strategy is directed at a specific political target-the military, bureaucrats, local interests, receivers of direct transfer paymentS, or specific social classes. The third section treats the economic context of allocations, focusing on the relatio nship between expenditure stralegies and changes in rhl' rates of economic growth and inflation.
What CO lllHS in the Calcu lus of Survival? The executive's problem is one of coalition formation. ' In building survival coalitions, the key cboices of Latin American leaders revolve around two Sl·ts of facturs: the sites, currencies, and sizes of winning coa litions and the conditions governing the prices paid to secure allies. 1. Of th~ vaSt lit.ratur~ in coal ition the"ry, [ have found particularly IIS.,(,,\ the "'~Y' by ui .. r",n. Kenw"rthy, and De Swan n in Grocnnins, and "thers (1970).
Survival Strategies and Expenditure Trade-offs Site, Currency, and She: The Dimensions of Winning Coalitions
"
Since most research on the formation of political coalitions has concentrated on the United States and Western Europe, we know most about coalition processes in legislatures and cabinets. In these sites votes are the major currency, and the influence wielded by political actors depends on the number of votes they command. In Latin America, by conlrast, multiple sites and currencies are the rule, and political forces losing in one arena may not accept the decision. Legislatures, if they exist, are only one of many arenas of decision making. Other sites include international finandal agencies, transnational corporations, military headquarters, bureaucracies, and even the streets. Currencies of influence include violence (or the potential for violence), the ability to unseat a government by force, Ihe market power of inveslors, and the potenlial 10 block or advance the implementation of programs. Scholars have often noted the failure of Latin American nations 10 agree on the dominant currency of authority. In an attempt to apply coalition analysis to Latin America, Eldon Kenwonhy (1970) offers a "dual currency" model in which coercion and popularity are the dominant power r"sources. COt'rcion, mostly in the hands of the military, is a resource of the upper and upper middle classes. Popularity, defined 3S the ability to elicit support from large numbers of people across occupational strata, is a currency more available to middle-class and working-class groups. Kenwonhy's dual currency model is a good beginning, but its restriction to coercion and popularity is too narrow. The cenlrality of entrepreneurs, for example, comes from their roles as investors and employers.l The mere threat of organized disruptive behavior by entrepreneurial elites is often sufficient to derail government effons at reform, and even where economic compelition is too fierce for entrepreneurs to collude, government policies that frighten investors lead 10 economic decline. In the same way, Ihe bargaining power of organized labor extends beyond getting out the vote. Labor's ability to disrupt economic activity can bring down governments by encouraging military intervention. J In es2. Stt Lindblom (1977), l liberal rest3t(mem of this long-argued Mane;st positio n. J. This ugument i. ma de by Payne (1965).
• •
,6
Tht Th([Jry of Su,"illal Coalitio"$
senee, coalition fonnalion in Latin America is more than a IWOarena game. l atin American coalitions involve muldple sites and currencies, and banles fought in onc site remain indecisive as long
as a major political fo rce finds better odds elsewhere. Wh:n determines coalition size? Suppose we distinguish, alol1S with Kenworthy, between " reigning~ and Mruling." If leaders
merely want to preserve the status quo-that is, 10 reign-they require smaller coalitions that need not include every currency. But when leaders want to effect change-to rule-they must develop larger coalitions including more of the major currencies. The mo rt: a program demands active support-rather than mere acquiescence-the larger the coalition necessary to ensure its success. Whether leaders choose to rule or reign is surely no accident. Reformist leaders representing middle-class o r working-class groups need the power to rule. Leaders representing higher-statuS groups are usually content to hold the line, to reign. Beginning in the middle 19605, however, military regimes representing industrial and agro-cxport clites ca nlC to dominate Brazil, Argentina, Uruguay, and Chile. Not content merely to reign, these regimes attempted to transform the political and economic structures of their societies. The nature of each regime's program determined the minimum size of its suppOrt c031ition. In Chile, Pinochet adopted a program of shrinking the st3le, a program thai commanded so linle domestic support the regime had to rely on heavy doses of coercion. In Brazil, on the other hand, the Geisel administration tried hard after the 1973 o il , risis to auran the suppOrt of the domestic bourgeoisie, because without its support the regime's ambitious plans for promoting capital goods production could not succeed.
The Costs of Coalition Building Putting wgether a walition requires determining which political actors a re potential participants . A society's cleavages define its political actors. Totally latent cleavages create no political demands. Cleavages become manifest through organization. Organized interests, represented by political acwrs, constitute the bases of conflict, the building blocks of the executive's political world. Once having identified the society's cleavages and relevant political aCtors, Ihe executive divides them into allies and adversaries. Allies 3re those who contributed support 3t the executive's acces-
,
Survival Strategies and Expenditure Trade.offs
17
sion to power (or at some other critical test); adversaries are those who were opposed. The executive can convert old adversaries into new allies, but conversions may come at the COSt of losing some previous allies. Old supporters defect because resources-in the form of control over programs-are transferred to the new recruits: The executive therefore seeks to minimize expected losses and maximize expected gains. The crucial elements in this tactical equation include the bargaining resources and unity of potential coalition members, the nature of intermediate organizations like political parties, and the motivating issues themselves. The resources with which political actors bargain have no fixed or absolute weight. Instead, the perceptions of other actors determine the weight of each actor's resources. The military, for example, can always threaten to overthrow the government, but such a threat will be taken much more seriously in Argentina than in Mexico, simply because coups occur much more frequently in Argentina and because olher political forces might support a takeover. Similarly, the pre-1973 Chilean legislature paid more attention to working-class groups than did the Colombian legislature, because in Chile the parties divided along class lines and represented every major social group, while in Colombia the elite-dominated National Front largely excluded labor. The fundamental point is simple: Rules and structures translate interests into bargaining resources at particular sites. In industrial countries, legislatures and bureaucracies are the main sites. In Latin America, legislatures are sometimes important and bureaucracies always so, but claims are also pressed through personal and family connections, demonstrations, elections, strikes, and insurrections. When executives are aware of these bargaining sites, they can estimate the value of any potential coalition participant 10 their chances of surviving and implementing a program. Executives often find it useful to divide or disaggregate political forces. If the executive can splil a once-united group, and if one of Ihe new subgroups dominates the other, cheaper coalitions hecome feasible. Thus an executive seeking to cut wages may try to deal separalely with labor's leaders. If the leaders will accept 4. Ncr distributable po!itical-«onomic spoil. may illerca.., a. a (Oll"'quon"" of a new coalition tht inerea..,. producti~ity. gross outpUt, or foreign aid, but sucr. incru..,s arc un""win. To induc. a political aeror ro accept furu~ for present ~waro.
requires a premium.
""
l~
-
. Ie
,8
Tht Theory of SurufllO/ Coalitions
bribes to counsel workers against striking, it will be much easier to ignore the interests of the workers themselves.
The Importance of Intermediate OrganiUltion.
The notion of or-
ganized interests implies intermediate authority between the execurive and the population itself. The slrength of such intermediate organizations varies rremendously, from structures thoroughly dominating followers 10 those safely ignored by executives. In the former case, organizational leaders may have a direct "charismatic" authority over followers, an 3U1hority based on personal dependence. At the other extreme, an organization such as a political party may exist only at elections and there may be little agreement between party leaders and followers even on basic issues. The effects of intermediate organizations on coalition formation depend on two factors: the nature of group loyalties and the ease of entry of new competitors. The more intense the links between allies and the executive, Ihc easier it is to woo the opposition without losing the allies' support. Com'ersely, the more intense a group's loyalties to an adversary, the more difficult it is for the executive to pursue them. Desertion of old allies usually pays only if the executive can keep a significant part of their support. [n a case like Mexico, for example, since competition with the ruling Institutionalized Revolutionary Party (PRI) is very difficult in the first place, defection of allies from the ruling coalition is nearly impossible. As a result, Mexican presidents can adopt policies that are contrary 10 Ihe interests of important members of the coalition. 5 Ideological Distance. Coalition theorists recognize the concept of ideological distance. Coalition partners must not be ideological opposites, and coalitions form more easily among political actors with shared interests. Still, such simple rules obscure the richness of political life. Ideological distances change with the passage of time. When import substitution in Chile, Argentina, and Brazil had reached a point where an emphasis on economic stabilization be· 5. The ,«en! eleaora1 gains by The opposition PAN in Mexico only reinfor", This argument. Though by 1985 the f'RI had 10st a .;gni~cant percen t3ge of its electoral .hare, it w •• not d.,... 10 losing control of the mech.n;sms of policy making •• nd it did not suffer the dffecrion of important members of its coalition.
"
u.
Survival Strategies and Expenditure Trade-offs
"
came imperative, populist coalitions of industrialists and urban laborers were no longer possible, because each side came to see its interests as quite opposed to those of the other (O'Donnell 1973). Moreover, if issues are perceived as zero sum-if IMF help is contingent upon a slowing of land {efonn or if the army hinges its support on repression of workers-then certain alliances are ruled out. Divisibility and Symbolism. Executives seek to avoid zero-sum situations, so they resort to divisibility and symbolism. Perhaps they can divide a peasant movement by conceding bureaucratic jobs to its leaders o r providing a few symbolic gifts of land for the masses. Perhaps the executive can announce grandiose plans for education and health care or the arrival of Peace Corps volunteers. Symbolic policies afe promises. Even though real benefits are postponed, executives expect prompt rewards from the grateful recipients of symbols. Promises made to one group imply costs for another. Even though costs, like real benefits, are postponed, why do those on the cost-paying side of promises not see their long-term implications and penalize the executive accordingly? Perhaps they realize promises are subject to future renegotiation that will lower their cost or even void the deal entirely. Beneficiaries of promises should come to the same conclusion and hold back their rewards as well, but they often do not, particularly if they are poor. Such groups may be victims of an "ideological hegemony" preventing them from comprehending the consequences of political action.6 Symbolic outputs satisfy those with the d out to compel a better Outcome but insufficient understanding to demand more than promises. Suppose, however, that political actors refuse to accept symbols in place of material allocations. If a demand must he met with an immediate material response, then its benefits ought to he as spe6. In an interesting p~per on public opiniOl1 in authoritarian Bruil, Geddes and Zaller (1985 ) found that people of lower education tended to increase thei r .uppon of [he regime'. policies u their information in"",ased. Be<;o.use the media in this period (1972_1973) we~ controlled so miClly [har only government·spon· sored infonnati on penetrated, the situarion was truly one of ideologio;;l l hegemony. Rapid economi c growth s.ati.~ed mosr people, and compariwns with the (mo.dy verbal ) exccssa of [he old pluralist regime favored the anned forces.
,0
The Theory of Survival
Coalition~
(ifi, and its cOSts as diffuse 3S possible! If costs are spread out so much that each penon's share is very sma ll, individuals may not he able to perceive them at all. When this happens, thc execurive maimains ;I level of perceived rewards higher than Ihe level of pen;eived penalties. As Lowj's work suggests (1964), executives prefer distributive "rivers and harbors" policies, which carry spe-
cific benefits and diffuse
COstS,
to redistributive policies like pro-
gressive income taxes, which have clear and specific as benefits.
Information g.:uhering is
;1
!:;Isk essential
10
cost S 305
well
the formation of a
survival c031ition. Some questions- the n.:nure of a society's duv ages or the bargaining r~utces of differem political fOTcrs-are so obvious they hardly nerd asking. Othrr kinds of information, such as the specific demands a political actor makes as the price of adherence, need to be repeatedly and explicitly sought. Querirs conccrning the intensity of loyalties or the willingnrss to accept symbolic rewards may be unanswerable umil the executive com· mits real political resources. Coalition formation is always uncertain and experimemal. If, as Kenworthy suggests, reigning requi res a smaller coalition than ruling, it i ~ no surprise: thai ambitio us programs often give way to mere continuance in office. o
Applying Coalition Theory to Budget Allocations The range of slrategies available to executives depends upon the elementS crucial [0 coalition formuion. Individually, these elements rened: the great diversity of L:uin American politics. Barriers against entry 10 political competition, for example, vary fro m practically impregnable to nearly no nexistent. People may harbor intense loyalties [0 political parties, to individual leaders, o r to neither. Effective budgetary control by legislalUres va ries from substantial to practically none. And, finally , the techniques used by executives to manipulatc followers unge from direct personal con· Irol to payoffs to powerful intermediaries. Still, shared historical, cultural, and political rootS combine to produce a limited numhcr of coalition patterns. Each pattern corresponds to one of flve strategies of budgetary allocatio ns: paci fy [he military, recruit bureaucratS, target local interests, increase di · 1, On 'hi, $Ubi«! '" Echol, and Rundqllin 1l919),
Surviual
Strategit~
.'
lind Expenditurt Trllde-o{{s
rect transfer payments, and reward social strata _ This correspondence-coalition patterns to budgetary allocations- exists even though economiSIS find it difficult 10 attribute the costs and benefits of public expenditures to individual groups or social classes_ (See Selowsky 1979 and Webb 1977.) Public expenditures are central to survival behavior because politicians and other political actors struggle over allocations as if their incidence were clear. The remainder of this chapter is devoted to explaining these strategies and determining whether political leaders actually implememed them.' Before we can begin that task, however, the strategies must be ~operationaliled"-that is, they must be defined precisely. Three steps arc critical: choosing the appropriate su rvival-maximizing strategy for each country in each year, defining the concept of political crisis, and measuring expenditure tradeoffs. (Additional details of a methodological nature will be fo und in Appendix B at the back of the book.)
Choosing the Right Strategy. For each year within each country in this study, it was necessary to decide which strategies would be appropriate (or "optimal ") for an executive trying to survive. Choosing strategies requires judging the bargaining resources of various political groups. How can bargaining resources be measured? On occasion voting results or survey responses might serve as indicators, but such data are quite scarce. At the same time, Larin Americanists have little trouble agreeing that political parties have reflected class cleavages more in Argentina than in Uruguay, more in Chile than in Colombia. They agree as well that military coups were more likely in pre-1964 Brazil than in post-1948 Costa Rica. Indeed, Latin Americanists would probably agree on the recent status of practically any country on any of these dimensions. Because such a consensus exists, it was possible 10 use a "sophisticated coder" methodology-a technique relying on the Latin Americanist literature and frequent consultation with country expens. In essence, the analysis used the monographic literature for 8. Norc, however, th~t leaders are likely to pursue multiple $tr~tcgin simul· tancouoly. The military, for example, may be pacified while tran,fen are inc.uoW. BUfUUCtat:s may be ~ctuited while 10<:<>1 interestt ate targeted. But it would be difficult ro recruit butUUCf~ts, urget local intcn:.a. ~nd incruse tr:I.nsfcf!I all at the same time, bccau~ thesc expenditure categories alone would exhaust the wbole budget. Moreover, many theoretically possible strategies, as we shaJJ 5«, prove un.ulistic in practice.
rho
l~
h
The Theory of Surviv
each counlry as a basis for judgments about changing social and political conditions. These judgmenls wefe translated into dummy va riables; thai is. Ihey look Ihe value 1 if a strategy was optimal and 0 if it was not. Consultants for each country (an average of two or three and at least one per country) verified and corrected the prelimin:uy codings. These c13ssifications are still open to debate. of course, so Appendix B contains a complete listing.
The Concept of the Political Crisis. Political leaders have many goals they would like to pursue, but budgetary rt'So urces are always in shon supply. Because survival is not the sole objeCtive of political leaders. the devotion of budgetary resources to survival strategies is likely to be greatest when leaders arc most vulnerable-that is, during political crises. Two kinds of events, elCClions and military coups, define crises. If an incumbent president or party is compering in an election, then we can expect the year just prior to the event TO reflect Ihe executive's allac:>!iona1 n.-sponse. Aher an elCClion, the winners seck to secure their positions with their own modillcations of spending.' In the case of coups, a different pattern of shifts is predicted. Since the incumbent cxCQItive does not plan for a coup, we expect no prior budgetary movemenl, bUl a newly installed military regime (unless it envisions itsclf as merely a caretaker) faces the same coalition-building problem as civilians, even to the point th:lI it may need to increase the military budget to prevent a cou ntercoup by disgruntled officers. The first year in which a new regime has budgetary control is thus defined as a year of political crisis.tO Measuring Expmditure Trade-offs. Budgetary programs (public works, education, foreign relations, and the like) arc competitors for scarce resources. A program's com~titi ve success is mtasufed 9. Only p.esidenrial elections are included he.e. In aUple. J I consider midIcrm dectillnl. 11 1M OUlCOmc oIlh~ tlcnion iI quile a nain, II i, not ( amide.ed 10 provok~ a aisi. unks. 1M elcnion ~ ..... o_u in Mexic;o-.rcruirmml and !nObiliry funcrion •. If 1M
Sllrviual Strategies
ll1ld
Expenditllre Trade-offs
by iu share of toul final expenditures-that is, the percentage of lotal sPf:nding allocaled 10 it. Hence the military share: of total expenditures is the ratio Army
+
Air Fo rce + Navy + Other Military Allocations Total b:penditure for All Programs
and public works l1 would be the ratio Communications + Transponation + Public Works + ... TOIaI b:penditure for All Programs Since the arguments of this chapter are framed in terms of expected movements in shares, the dependent variables (what we are trying to explain) art typically the percentage changes betwccn twO consecutive years in some program's share of the central government budget. If, for example, executives try to keep the military happy during a year of political crisis, the change in military allocations between that year and the previous year should be significamly greater tha n the change in other years. These strategic hypotheses predict more than just a high growth rate in the share of a given program. Indeed, growth in one specific program comes at the expense of other s~cific programs. To express this trade-off, annual fates of change were first calculated fOf each of two programs and then the difference between these rates was taken. If A, is the military share of the budget at time t, for example, and B, is the publie works share at lime t, then the military-works Ifade-off is
M"I l -W ork5,"" A, - A. _ t A' _I Five Strategies of Budget Allocation Pacify the Military
For most contemporary Larin American executives the threat of military ovenhrow
CondjtjolfS fullOring M ilitary Pac:i{iClJtio n.
I t . In 101M cownries minisrries such as communications art "high Icch· comIndirional po.k ba.rd pmgn.m....ch as public workJ. ~.all, howev.,!, no consr.lml diotincDon of Ihi. Type an ~ mack in 1..o.rin Amtric:l. As a ,""h, Ihil mnw •., il DOl ali sJ,a'l' ali OM would lik~. ~.rd fO
Tht! Tht!ory of Surv;t'ai Coali/;on5
is very real. Whether the chief executive is a civilian or a soldier, a wary eye and ready checkbook are indispensable. In all pOSt1945 Latin America, only Mexico and Nicaragua have been entirely free of overt military intervention. The probability of coups varies across countries and across time. In a pioneering study of intervention, John Fitch argues that the military acts as a "monilOr of the political scene, intervening when, in its judgment. a 'national crisis' exists." Officers' perceptions of the level of crisis depend on the military's ratings of the constitutionality of governmental actions. the officers' personal ties or antagonisms toward the government, public opinion toward the government, government attentiveness to the institutional needs of the armed forces, government policy toward any perceived McommuniSl threat,~ the level of public disorders, and the need for socioeconomic reforms (1977: 160). These "triggers" of crises have changed over the years. As the level of military professionalism has increased, the social backgrounds of the officers have become less important than their institutional interests. Professionalization has also raised the military's confidence in its own ability to govern while lowering its opinion of the abilities of civilians. The threat of communism be-came much more important as a Irigger after the Cuban Revolution, because military officers generally believed thai a coup against Batista would have prevented the rise of Fidel Castro (Fitch 1977: 160-164).
These observations refer to Ihe perceptions of individual officers, but coups are not made by individuals. Coups occur when a minimum number of officers agrees on their desirability_ The formation of that minimum involves more than just a hard core of conspirators expanding until enough moderates join to guarantee victory. Successful conspiracies form in different ways and act wilh varying sm:ngths. Fact'd with such uncertainty, the executivt' assesses the probability of a coup. A leader knows coups are more likely if (1) the military has a tradition of intervention, (2) the economy is declining, (3) mass disorder is increasing, (4) the military's financial requests have nm been met, and (5) therc is ideological conflict between the executive and the officer corps. Direct budgelary rewards are a logical response to the military
Survival Strategies and Expenditure Trade-offs
threat. If the armed forces' primary motivation is simply their own instinnional enrichment, the executive can spread the COStS of meeting their claim over various nonmilitary programs. If, on the other hand, the military not only wants more for itself but less for its enemies, then its budgetary gain should be accompanied by specific cuts in programs the officers dislikeY
Evaluating the Strategy. Table 4 shows what happens during periods of political crisis. When the military pacification strategy is appropriate, the military's share of the budget rises. The big losers are programs in health, welfare, police, and foreign relations (although foreign relations is too small to contribute much to the military). Education yields lesser amounts, and public works and pensions defend themselves quite well. The success of public works is probably due to the substantial number of cases, especially in Central American coun!ries like EI Salvador and Honduras, where military leaders combined construction projects with payoffs to the armed forces. Pension programs in Latin America are often !reated as entitlements, so they are less subject to short-term manipulation. u Table 4 applies only to conditions of political crisis. In noncrisis years the same regressions produce no significant trade-offs. Did these executives, all of whom were vulnerable to coups, turn against the military when the crisis abated? They did not. Note that military budget shares gradually decline in about three-fourths of our seventeen countries, and remember that this decline is sharpest in administrations with linle to fear from the military. For the vulnerable govemments in Table 4, on the other hand, the absence of shifts to or from the military budget in non crisis years means that the anned forces still have enough strength 10 neutralize the trend against them. Figure 2 illustrates the response of military spending to political 12. An ~mphasi. on budgr:ury .. ward is nOf munt to imply that no Ofh.. srr.otcgit$ ~rt availabl~. A luder may fry 10 maintain a baSl' of support in on~ faction of the umed forc.-s u a way of minimiring overall weakness vis·l·vi. 1M military. Such an app....,..ch would ckarly nOf prcx:lude a si multaneous budgtlary reward. 1.1. The only budg~ alqory with the wrong sign;s uansfer pa.ymml$. Official cI ....i6arion. sqnrating traruf~ .... ftom other kinds of spending happa! to be quile weak, perhaps the wakest of any budgtt e,negory. M=remenr ~rror is thus a distinct pouibiliry.
• •
"Table 4. Shift
10
The Theory of Survival Coalitions The Military Pacification Sfrategy Du ring Political Crises
Military From
Huhh Welfue
Police Educal;on Public Works Agriculture Foreign Rdalion.
Pen.ion. T.. nden
b 0.1105 0.0757 0.0689 0.03 05 0.0358
0.0445 0.0888 0.0452 - 0. 198 1
ITI
R'
N
0.0002 0.02.57 0.001 4 0. 1143 0 .33 1 0. \ 9 0.0001 0.210 0.08
28 15
80 73
2J 17 OJ
'" '"97 '" "
p <
123
28 15 73
17
15
Note, Each row repr ....'110 • ><'POl.'e «juation in "'hich 1M ";>de ... ff i. r.V"""'"" on
. 1r.tcgy dum"';« for yu~ of en,i •. TIm. /). Mil - t:. Hnlth _ A ~ b (mili,.ry dummy) + < + d (nW0n.1 dum my) +
( 1)u~.u.cr.rH:
,t..
dummy)
Th, dcpn>&nt •• ri.bl.. ;n ,h... "'Iu otion. or. the u, .. of chang< of tl>t mil".,), ,h. " minu, .h. rAt.. of ,h'RB<' "I , .... p'''P.m in ,I.. -..It .ol.. mn. II". b . nd p ",I., to II.< <<>
d~rity , .. "&n' hayt bftn ...... n.«I •• i ~ log< of numb.,.. I.. , ,h.n I ... "'P,i'.' Th. coc-ffici.n" a,. un,,.,,,j,,diuJ. 11' . nd N ,dor '0 .. .,i"K" lor , .. rompl ..r "11"'';011 rqu",ion<; ,hal i••• 11 "'''rgy dommi .. "~,rK.iuJod .
inchid..! in the "'Iu.rion. For
crises in Brazil." The increment in the military budget share in 195 1, the firs! year of the elected government of Getulio Vargas, was strongl y positive. The military had removed Vargas in 1945, so upon his rerum to office Vargas was attentive to his military support. In 1956, the first year of Juscelino Kubitschek's rather insecure administration. the military share shrank, but this decline was reversed by 1957. In 1962 and 1963. Joao Goulart bucked the negative trend in the increments the armed forces had received after 1957, but their shares continued to shrink, and that shrinkage may have contributed to the collapse of his administration. In 1965, 1967, and 1970, military presidents began terms." In the first two cases the military share jumped immediately, and in the last-Medici's government- it gaint~d in the second year. In all, 14. Except in Figore 5, the horizontal dotted lines On each plot represen t 3 aero rate of change. Po inu aboy~ and b.low the line rq>lnC1lt posit;yt and n~ti"e percentage in~ments, respectivdy. Since the data are differenca of loW..! .hares, the actual Kale on the vertiCIl Ui5 has been omitted. 15. Th. coup initiating Brazil', ",·. nty·y'3r prriod of authoritari an rule oc· curred March J t , 1964, but tl>e 6r., full yea r in ,,·hieh the general. h.d bu..!g~tary comrol was 1965.
"
"
SurvivlJl
Strattgie~
and Expnlditllre Trade-offs
r! j
r ..J
I ~9~,C5""--;,C9'5"O"--;,C9'5"5"--+'~9'6;0 -""'9"6'5+-""--'9;7CO"~ Figure 1. Changes in Military
S~nding
in Brazil : 1946- 197J
of six administrations for which military pacification was crucial, the military share rose immediatdy in th ree cases, rose in the second year in two, and rose not at all (though the !rend was up) in the soon-to-be-ousted Goulan administration. Recruit Bureaucrats Conditions Favoring Recruitment of Bureaucrats. Bloated bu reaucr:tcic$ in Luin Ameria date from colonial times, and public employees have long constituted a larger fraction of the politically active population than in industrial countries. The imponance o f government employees becomes even greater when ethnic or linguistic divisions cut the size of the politically rdevant population. In the early 19605, for example, Peru's teachers (plus their immediate rdati ves) received about 3 percent of the nation's personal disposable income and consliluled 6 10 8 percent of the tOlal numbet of voters (Kuczynski 19n). When bureaucrats constitute a large fracti on of Ihe politically significant population, their pay becomes a pivotal public issue.
,8 Table S.
The HurtaUuali, Str/Jltgy' Est;maltd Sbilfe of Al/oeD/iom Devoted 10 Salaries in Crisi$ and Noncrisis Situations l>olitic~1
S lr~trgy P~ci fy
"
Military
R«ruit
Rur.IUCf~n
[ncru$C
Trtn $f~"
Rew.rd Local R(w~rd
I mc r.>1~
Cri sis' I'
<
ITI
No 1'0Iot;,.. 1 Cri,i, ' b
p <
ITI
O.OS I 0. 131 0.OJ7
0.38 0.02
0.015 -0.0 12
0.75
O.OJ
- 0.0 19
0.78
-0. 150
0.02
0.007
0."
- 0.Q10
0.'#2 0.'" 0.58 0.8S
- O,O\S 0.012 0.04 1 0.019
0.115 0.86 0. 64
0.8 1
Socill Slrul
Domin.m CluS('>; Intumrdiatt CI'~'tl SubordinOf~ d. ~ Muhi d~ ls
- 0.005 - 0.097 -0.0 12
0.82
.'" * 9.1, R' .. )$ ,,, .. 1.' 1: 1',01>:> F .. 0 . 11 . ' N .. ]96, II ' .. O.OIi,p " I.J6: Pro!> > f" .. 0.1 1. ' /In< /, .,14"
,,,f,,,
' u , ..... c... fn..:i.... " 'nJ
"5",fi<.nc~
.....1.
Q{
,10,- ,ftdo •• lu.1 ,I~ n 'm ....
In > "n~~ r.g ... " .....
Bureaucratic recruitment is d irected row.:Ird political :u:ti visls. It works best when 13yers of intcrmediilte le3ders control the m3SS b3se supporting the cxecutlvc. JIISI as Lyndo n Johnson's War on Poverty benefited well-placed activists morc than the poor, so a bureaucr3tic recruitment strategy rewards labo r leaders mor(" than fa ctory workers, It'achers mOTC tha n pupils. By comrast, bUrt'aucratic recruitmem is less useful when intermediate organizations arc weak or absem- th:1I is, when the relationship belWet"n It"aders and followers is dirc-cl 3nd pt"rsonal. A bureaucratic strategy ma y be implemented either by enlarging the size of the bureaucracy or by increasing public sector salaries. The executive's problem is to determine when such tactics would be effective. Su rely one import3nt d3tum is the size of the bureaucracy relative 10 the politically relevant population. If the numbe r of government employees is large, perhaps compared to the actual or pOlenljal e1ecror3le, then bureauef3tic r«ruil mem makes scnsc." The feasibility of emphasizing the bureaucracy de16. An inte~rin, .tn iR,e from Cou~ R,,:a i~ pro"idni by Osat Anas Sinck, (1976). The pcrcnuagc Qf fe
.,
Su rvival Strattgits and E.:cpenditure Trade-offs
Table 6. Gai" j lind Louts to Public Works Programs When Bureaucratic Recruitment Strat f'gy Is Optimal Program &ndiring Trad~-olf
!M.rw..,n Programs
Works-DdclIK Works-EduclIion Works- Hu hh Works- Foreign Rd~!;ons WOfk$_Welf~r~
Works- Poli", Works- Agricu hure Wor kl_ubor Workr.- I'cnsions
NO,,";J;J
Crisi,
Works Work> Ntilhtr Work s Works Works Wurks lob<" Works
DdenK EduClIfion Works Foreign Relations Wtllut Polin AlViculru.., lobo, Pcnsion~
pends also on the absence of alternative power bascs. If there arc strong regional cleavages-that is, if executives must appease local or regional bossu-then a bureaucratic emphasis would be coun' terproductive, beca u ~ bureaucratic rewards cannot easily be limited 10 a specific geographic area.
Evaluating the Strategy. Since bureaucratic strategies affect wages and salaries, the most straightforward ttst of the app roach is the change in the percentage of the whole budget devot('d 10 personnel. Table 5 shows that when the bur('aueratic strat('gy is optimal during crises, the sabry share of ('xpcnditures riS<'S.' 7 When, on the other hand, local or regional interests must be satisfied, s:alaries fa ll. Public works, with its rdativtly low wage and salary conlpDnent, is antithetical to bureaucratic reeruimlent. If an administration seeks 10 maximize bureaucratic support during a crisis, public works progra ms ought to suffer. Table 6 shows that public works does indeed declinc vis-.l-vis all ma jor programs except labo r and health. U The individual coefficients :tssoci,u ed with :til categories 17. NOlc Ihl! Tabk S prnenl$ fn.ul~ from a >ing!e r""ts~ in which all 1M orr:u..gy dummies If~ includn!. Ill. Labor and he;ohh programl U~ 1}'piCllily among 'M . malln! of .;mllli g"'·~rnmmf minimico, so Ihu~ is link payQff in J(jutnint IMm.
,0
The TbnJry of Survival Coaiitio.u
ar~ quirt low, howtver, and only the consislency and direction of Ihe results lend credence 10 the argumenl." Figure 3, which chans percenlage increments in Ihe share of
of Table 6
total expenditures devoted to governmem wages and salaries in Costa Rica, illuslrates the bureaucratic recruitrncllI suattgy. Of the thinecn cases in which the salary share dedint'd, only four1962, 1965, 1966, :md 1970-occurred in preelection or postelection years, and the declines of 1966 and 1970 were quite small. Moreover, salaries rose before every election exeC'pl the election
of 1965.10 Thus in Costa Rica salaries were more likely near elections and more likely
10
10
risc
fall between them.
Target Local Interests Co1lditions Favoring a Local Strategy. t:undamental cleavages often develop along geographic lines, In Ecuado r, the landow ners of the highlands (sierra ) tradi tionally oppose the planters and bankers of the coast (costa ), In Colombia tWO highly organi7,ed parties recruit and maintain intensely loyal fo llowings down to the village levc:1. When local cleavages manifest the msel ves in such a renas as elections a nd legislatures, executives need a suitable response. I ' Such a response is found in pork barrel spending, NOIhing distributes government largesse like public works. An executi ve can place a school, hospiu l, or da m just where il visibly rewa rds loyalists. The whole coumry Ix-ars the CO SIS of the project, while carefully selected localities enjoy the benefils (including construction jobs). If a society's major clea vages all fa ll a lo ng geographic lines, this strategy predicts an expa nsion of public wo rks cxpcndilU res a t the expense of every o ther pro gram, h ut executives need not abando n 19. II _ co n'pu~ ,"" 1'I1,1
<>n(nu. olt~ p"'t:r.'m, .Juring C1'.1n ~nd noneri,"- no .Jllk .... n~s ap~u In ,he: comlXl""'C s'rrn~,h of publIC ,,·o. ks pto. gr~m ., Thi, ~n ding ~U PPOrlS 'M hypotlles., thu uch mlntSHY', w~gc b.1I shou ld m~ i main in rd ~,i"e position. BUI on« ~8~in I.... I<stll" ~rc C<>n, .. ,~m III d i.ccritln '~lh,. ,han imprC:SI' v, In sia, 20. In 1950, I"" begin ning of ' he ~rsl ""np'(WI!,on~1 g.,..trnmml ail, . J..... Flgu eres 's uprisi ng. no prm«IIon d (e" IS nr«,~d. 21 . l."c~1 , " aleg;" ar e "'" II ,"e,su ll)' ~PP"'JI "~ IC (0. ~II polllicalllluhlnC'>. iu" for Ihose rompcllllve ~I 'h e loc31..,. ' '1:ion.II.,·.1. ,\ l3ch, n... II .. , ..e domIna", n31ionally, such os M.,,;co ', I'ltl , m. y ...... u" . vcry dl ffncnt appro1
Survival
i
StTattg~j
and Expenditurt Trade-offs
+
~
~
~
j £
•
0
f
u
•a •
~
. • ~ c
<
I I I I I , I I I I I I I I I I I I I I I I I I I I I I I
1945
o
1950
1955
1960
1965
1970
1975
PreeleClion Years in Wh ich Elecloral Effect Is Expected NOTE; EleClions were held in early February in 1958. 1962. 1966. 1970. and 1974. In these years the crucial budget year was one year earlier. In 1953. the election was held July 27. so the budget efieci should have occurred the same year.
Figure 3. Share of the Budget Devoted to Wages and Sabries in Costa Rica: 1950-1976
public works strategies even if geographic cleavages combine with a horizontal cleavage such as social class. Government programs favoring cenain social classes also include physical construction components. Both primary schools and universities arc made from bricks and mortar, but the class backgrounds of the students attending these schools vary dram:nically, so a political leader can
1~",1
Ie
a
The Theory of Survival Coalitions
"
Table 7. Tests
of the
Local Strategy During Political Crist's
Shift 10 Public Works From
b'
p >
ITI'
R"
N'
0.1 \0 0. 106 0.107 0.113
0.007 0.010 0.006 0.001
IJ
lIb
15 11
TT3nsfers
0.422
70
Wdfore Poli"" Agr;';'ulture
0.106 0.128 0.10l. 0.206 0.1 76
0.001 O.OU
71 1" 21
Military
Education Hcahh For~ign
Rdarions
L>h", Pensions
• '" .,
15
O.OC}!
12
0.044
25
101
0.003
""
J5
0.007
'" "
' H< "",Ifiri"".. . nd u lI"ific.",,< 1...1, 01 the local,n""",' dummy "'I>m .11 ".her "ri.bln ... md..d
i,,""
use the school construction budget
10
appeal to class and local
interests at the same time. Similarly, an airport or road project may serve military purposes as much as local interests.
Evaluating the Strategy. Table 7 provides unequivocal evidence for the local strategy. When loca l approaches are optimal during crisis periods, all other programs lose to the public works minist riL'S. When local approaches are optimal during tlOncrisis periods, no program los",s to public works. Executives reward local or regiona l interests on ly for the brief interva ls when such rewards are mos t critical. Figure 4 provides a concrete illustration of the local strategy . It relates the public works budget share in Ecuador to the shares of two competitors: the traditionally weak fordgn relations ministry and the powerful military ministries. Politics in twentiethcentury Ecuador has been dominated by Jose Maria Velasco Ibarra. Five times president but only once able to complete a term as scheduled, Velasco was truly a charismatic figure. His volatile temperament and his penchant for translating popular mandates into quests for dictato rial powers made his political survival con· tinuously problematic. Only once, from 1953 to 1956, did Velasco complete an administration. In that term he tempered his enthusi:Jsm for big construction projects with the realization that a mil-
,.,.,.
l~
-
Ie
Survival Strategies and Expenditure Trade-offs
"
itary that had already removed him twice might repeat the favor. As Figure 4 demonstrates, Velasco increased public works in 1953 at the expense of foreign relations, but he favored the military over public works. Many factors conuibuted to his survival, of course, but his early pacification of the armed forces could not have hurt. When Velasco returned to power in 1961, he ignored the military and, as Fitch (1977:198) :and Pyne (1973) make clear, turned away from his Conservative party supporlers altogether. Public works were favored over both the military and foreign relations. Velasco's administration collapsed almost immediately, and he was replaced by Carlos Julio Arosemena, the vice president. Arosemena reduced public works vis-:\-vis the military, but the new president's overall budgetary strategy was puzzling. Why should a traditional ministry like foreign relations gain more than a politically important program like public works? The answer is that Ecuador's relations with Cuba had become the chief issue of his administralion. Arosemena's efforls to sleer a palh of neutrality against U.S. efforts to isolate Cuba infuriated Ecuadorean eonsc::rvatives, particularly in the military. The legislature, the bureaucracy. and Ihe armed forces combined to block the president's initiatives, and the government drifted until his ouster after only eighteen months in office (Filch 1977:55--64). In this case budgelary allocations reflect an uncaring and perhaps uncomprehending attitude toward survival. In July 1963 a military junta wilh certain developmental pretensions replaced Arosemena . Public works grew more rapidly than foreign relations throughout the junta'S rule, and works even gained initially against the military. After the armed forces withdrew in 1968, Velasco managed 10 win a fifth term in the presidency. He repeated his earlier strategy. with public works growing faster than both foreign relations and the military. The result was repeated as well-Velasco was ovenhrown again.u 22. V~bloCO', La$! ousrer may not ha~ b«n 50 darly a rau lf of hi, cxpo-ndiNfC $(f.rqy. ln 1972 he fCCIMd almosllO be inyitinl a coup ill order 10 preYenl a pofMIlu o:andidarc from winnin. the: _ prnidrnrial n«flOn. Thm, whm rhe WI.I p bcamc ;DeY;labl~, h¢ rried 10 make common cause with I~ candidacc. Too
•••
Th~
H
Tbtory of Survival Coalitions
Works
1 o
J
Milltarv L.._ _ _ _ _ _ __ _ _ _ _ _ _ _ _ _ _ _ __
•
Works
1 o
!
i""".I
13
Aelations 11145
6, FI~I
"" posHlarthquake yelr
o First year 01 First year 01 'V First year of O FItsI year 01 FiBl year of
o
o ---
1960
1965
1970
1975
!of Galo Plaza
Velasco', ' 9~ 1956 rElgime Velasco's aborted 1961 regime C&r1oe JuliO Arosemena junla ( 1964-1 966) Velasco', aborted 1969-1971 regime
Oillefen0Et8 In rat' 01 change 01 public WO!1oIs and militaly DilfenltlO!S in r.'e 01 change 01 public WOItts and foreign relalion$
Figure
In"~Q.Se
~I'ld
f orc-ign Relations
Transfers
Conditions FallOring a Transfer Strategy. When no intermediate organiz.uions claim fewards and no stratum of aaiv;slS demands jobs, instant cash becomes the logical strategy and transfers become" the app ropriate medium. A transfer strategy succeeds when
weak political panics c()(:xist with fie rcely competitive leaders. Old
Survival Strategies and Expenditurt Tradt-offs
"
parties disappear; new parties form; platforms hardly maner. The personal qualities of leaders, their charisma, antact followers. Miners, urban workers, and middle-class students may be barely organized but still intensely loyal to some political leader. Executives can buy their support with such programs as pensions, social security benefits, and university scholarships. This strategy combines well with other approaches. An executive seeking working-class support can concentrate transfers in programs that benefit workers, such as health, rarner than programs that benefit the middle class, such as education. Thus transfer payments retain class selectivity, but the recipient populations are much broader than those affected by specific programs. In general, then, a transfer strategy succeeds when intermediate organizations are weak. When leaders seek to recruit masses rather than reward machines, when they desire immediate responses from broad populations, direct payments are an appropriate tactic. Evaluating the Strategy. The transfer strategy is the most difficult to evaluate, because the measures of transfer payments arc quite weak. Classifications of transfers across countries are less consistent than classifications of specific programs, and in many cases data are missing. In Table 8a two measures of transfers (pensions and social security payments) reveal no signs of a transfer strat· egy.u In Table 8b, however, where the expenditures that governments themselves label as transfers are evaluated, the results are better. Here all the signs are in the predicted direction, and most of the coefficients are substantial. The most encouraging results come from Table 8e, which examines shifts from salaries to transfers in periods of crisis and noncrisis. Here the coefficients are substantial and appropriately signed. In general, however, the data are inadequate to confirm the existence of a transfer approach. Whether these provisional results would hold up with less ambiguous data is uncertain.
Reward Social Strata Conditions Favoring a Reward Strategy. Few Latin American societies qualify as "advanced industrial," but the social class deav23. Social S«\Iriry programl doubtless have imponant politicol aspects (Mesa ugo 1975), but paymenl5 have Sene.... lly followed a smooth upward path.
1~",1
Ie
a
,6
Tbe Theory of Survival Coa/ilium
Table 8. (a)
Shift
Tests of the Transfer St'lltegy
GQ"",,,m~"1
[0
Co"triUu,;an, to Soc;3i SUll,jty Fund. 0"';"8 Polili".1 C,;,~,
Goy~rnm~n1
Soci~l ~cur;ty
b'
Contributions From
p
>
111'
R"
N'
37
2J
Wolf3r~
- 0.176
Military
0. 107
0.07 0 .1 2
Police
- 0.OJ4
0.96
Foreign Relation. Public Works Ag.riculrure
-0.082 - 0.100 -0.015
0.32 0. 28 0.87
j
Education
- 0.050
0.55
II
40
R"
N'
" '"
" " " " ""
(1)) T.:ln.{r. PlZ)'menu Du.ing Polith;al
Shifts
10
Transfers From
b'
Polic..
Educuion
Public Worh Agriculrure Furago Relations Military
Ie) Tu",s/",.
Shift from 5aJaric. 10 Transfen
Political Cri,;.
ITI'
0.01
o.on
0.49
0.103 0.121
0.011
VS.
II
." J5
40
..
er;"S
0.123 0.053 0.285
0.06
0.5\
0.09
25 J5
""
Sot..rics
b' 0.270 -0.105
No Political Cri.;s
p >
"",
P>
lT1'
0.12 0.08
R"
N'
40
"
"
53
til ... b and p ..f
.m.
ages that characterize Western Europe and North America are often critical in Latin America as welJ.H In such diverse countries as Nicaragua, Brazil, Chile, and Haiti, recent political conflicts have involved interclass struggles over the distribution of national 24. Some ddinirions: "Dominant dasKS~ means up",,' "nd upper. middle groups- the bourgeoisie, big landowne •• , high-level profcuionals and m"nagers,
,
Survival StTalegies and Expenditure Trade.offs
"
wealth. How can we link the rise and fall of newly organized social forces to the evolution of public spending?2J Effective challenges to the power of traditional oligarchies in Latin America have occurred roughly in step with each nation's economic evolution. In Argentina and Uruguay, middle-class groups challenged the political power of the old oligarchy in the first quarter of this century. In Brazil the challenge started around 1930; in Bolivia it began in the forties. Working·class groups achieved substantial power later-in Argentina after 1945 and in Brazil around 1960. Beginning in the mid-1960s, however, the flow of political influence to lower-status groups proved quite reo versible. In Brazil, Argentina. Uruguay, and somewhat later in Chile. coalitions of domestic and foreign bourgeoisies joined the armed forces to reassert their dominance. The political expression of this coalition, the "bureaucratic-authoritarian" regime, was modernizing in the sense of promoting greater economic integration with advam;:ed industrial economies, but it was reactionary in the sense of promoting greater inequalities in the distribution of income. In budgetary terms this strategy implied, on the one hand, CutS in those social programs helping the middle and working classes and, on the other, boosts in military spending.
Dominant-Class Strategies. What policy choices funher the interests of dominant classes? If a politician is simply the latest in a long line of leaders representing dominant groups, we expect linle change. But when a leader represents the reaction of donli· nant classes against the challenge of subordinate groups, then the latter may lose some previously won benefits. Defenders of dominant classes will claim such retractions are in the public interest, part of the restoration of entrepreneurial optimism so necessary for economic stabilization and growth. Even in the days before IMF conditionalities, a Banzer, Castello Branco, or Pinochet would and the lih. ·lntennwi~te classn ~ includes middle bur~ucrats and professionals, shopkeepers, teachers, whi~·collar employees, and so on. "Subordinate dasses ~ mean. th~ urban working cia"", muginal group. such a. "luallen, and the rural ,",,'25 . It might be objectw that the attribution of the COSts and benefits of public expenditures to individual social cia..... has provw to be quite a difficult problem for economists. This is undoubtedly true, but it is beside the point: Political acton fight ovn allocations .u if their incidcnCl: w(~ dear. Sec Selowsky (1'79) and Wdlb (1977).
• •
,a
Tbt Theory of SUrl/nuli Co
assume that entrepreneurial confidence requi red a reduction in "unproductive" public cxptnditures, notably those on education, health, labor, and land reform. Because reductions in social spending cannot be enforero without in crea~ in the reprtsSive capacity of the government, dominam-class trade-offs benefit military and
police forces. Dominam-dass rea c;:tio ns differ fundamentall y from the pauician, liberal-democratic style: of a leader like Eruador's Galo Plaza 3nd from anachronisms like Somol.3 or Stroessner. Typical ex:amples of aggressive defenders of upper.class interests include: Ongania and Vidda in Argentina, Banzer in Boli via, the post-1964 Br:lziiian junt3, Pinochct in Chile, and Castillo Armas, Peralta, and Arana in G uatemala. In these cases dominant groups rea!;;ted after subord inate dasses had already mobilized and a!;;hieved vinories. In other cases of dominant-dass reanions, notably those of Odna in Peru, Perez Jimenez in Venezuda, and Ospina Perez in Colombia, popular mobiliza tio n was subst:lnria lly lower. The administrations of Odria and Perez Jimene-l began as rea!;;tions against popular mobilizatio n, but, not surprisingly, they soon turned to a Mpopulist" style, emphasizing public works and social welfare spending. In Colombia the level o f social mobilization was so low that a class strategy seems incongruous, but the feuds between Liberals and Conservatives tempor::nily polarized elites, and Ospina Perez saw himself as preserving civilization against the mindless hordes. The ease with which Colombian elites accepted the National From after the vio/enda and the repressive administration of Rojas Pinilb demonslrates that Ospina's strategy missed Colombian reality.
/ntermediate·CLtss Strategies. Conditions bvorable to targeting the intermediate classes have exiSted in both Cennal and South America. In Chile, for example, political connin has long revolved around class, with distinct political panies representing the Idl, center, and right in a welt-defined ideological spectrum. From 1958 until 1964, Conservative Jorge Alessandri held Chile's presidency. He was succeeded by Christian Democrat Eduardo Frei. In CLtss Conflict and Economic D~/opment in Chile (1978), Barbara Stallings a rgues that Alessandri defended the interests of the big bourgeoisie while Frei favo red the smaller bourgeoisie. In terms of the over.lll economic consequences of Alessandri's and
SUrvillO( StraugitJ and Exptnditurt Tradt-off$
"
Frei's policic:s her analysis i5 convincing. In lerms of expendirures, howc:ver. their incentives were c:qually center-dirc:ctc:d. Alessandri had little reason 10 fear a righrward ddc:aion of his consc:rvative supponers, and he n«ded the backing of the cemrist Radical party in the legislarure. When his party lost support in the midterm legislative election of 1961, he was forced to move toward the clectorally crucial cenler. The lerm of anOlher Chilean presidem, Gabrid Gonula Vidda (1947-1953), contrasts sharply wilh thai of Alessandri. Elected with Ihe support of Ihe left, Gonzakt VideJa's first cabinet ineluded Ihree Communists. But personal belids and international financial pressure soon made Gonzalez Vidda a cold warrior. He removed the Communists and began a cr3ckdown on labor. Moran's (1974) description gives the impression of a complele turn toward the dominant el3Sses, but Gonula Vidda should have adopted an inlermediate strategy. In Chile such a str31egy would necessarily be more repressive than Ihe same approach in a coun"1' like Peru, simply because the mobilization of the left was so much greater in Chile (Bowers 1958). Fernando Bdaunde Terry in Peru and Arturo Frondizi in Argentina also followc:d intermediale str3legies. 8claunde combined an intermediate strategy with a regional approach. Frondizi began his term pursuing economic policies favorable to the Peronist support that had put him into office, but overall the policies of his administration reOected the interests of the intermediate classes supporting his branch of Ihe Radical Civic Union (Zuvekas 1968). In a number of Genlral American countries the overthrow of old Somoza-style dictators lc:d 10 mililary-sponsored anempts at reform. Under such executives as Osorio and Rivera in EI Salvador and Villeda Morales and L6pez Arellano in Honduras, Ihe scope of governmental activity grew. ThC$(: admjnj~tratjons beg::an infraslructural projca.s and attempled minimal social reform. U Until the early seventies political conflict in these societics was not organized along class lines; Ihat is. the working class and peasantry had no legitimate mechanisms 10 articulate their intercsts. As a result, the resistance of the dominant classes could easily stymie military effons to encourage a larger role for governmenl. In EI Salvador, where Ihe replacement of subsistence farming wilh cap26. The lin, souru on Cmlnll Ameria in this pmod is Anderwn 0"7).
00
The lh"ory of Srm,ival Coalitio"5
italiSI agri<.:uhufC sharp ly increased social tensions, the political center became increasingly isolated as class conflict grtw. CcnWTdominated reform proved illusory, giving way finally to civil war. SlIbortiina!e and MlfltidaH Coalitions. few aJrninj S lr~ltiOI1S ernb':l rk on a strategic defense of the interests of the subordinate classes alone. A generous definition of administrations pursuing subordinate-class strategies might include the prt:-1951 government of Juan Peron in Argl.'ntina, the Popular Unity administration of Salv3dor Allende in Chile (1970- 1973), the incompetent govcnllllcnt of T eodoro Picado in Costa Rica, and the administrarions of Juan Jose Arevalo and Jacobo Arbcnz in Guatemala. All camc to unhappy endings at the hands of milit:!ry uprisings Y Multicbss coalitions arc more common . In Argcntin:l, the poSt1950 phase of Per6n's gOVl'rnment reprcsented a hro:ldening of its carlier targeting of subordinate cbsses, but transitions in this di· rection (s ubordinate to multiclass) h:lve been rarc. More often, multicbss strategies have immedi:nciy followed one of two con· ditions. One condition is simply the existenn' of plentiful reo sources. In Venezuela, the Democnl1ic Action (AD) governments of Romulo Betancourt and Raul Leoni (1958-1968) took good care of the military, but th:mks to oil resources AD was :!ble to forge a multiclass alli:lnce at the samc time. In 1968 AD's splil allowed its centrist opponents to take the presidency. The victory of the Christian Democr:lts might have led to an intermedi:!te· cemered slr:!tegy, bUl the central government budget could easily expand to accommodate new claimams, so expenditure policy re· m:lined substami:llly unch:lnged. Multicbss coalitions h:lve also follow ed a second TOUle. If classbased mobilization is too weak to permit a strategy b:lsed on eithe.>r the interme.>diate or subordinate.> cbsses :llone, and if there has not yet occurred a successful ch:lllenge to upper·cbss domination, then such a challenge could come from an alli:lnce of intermediate and subordinate classes. This condition was met in Bolivia, where the 27. Picado, it should k nQCed, w3S '>USCN not by his own milic~ry hut by a p<>pular rebdhon led by Jose FiguorC'S . Poron was overthrown aftor (but not k· couse) hi, regime h.d .hiftN 10 , mulrida...tr.t~gy. ArbenL d;ffers from OTher J~~dcrs who loll OWN working..:la .. woregif!", including hi. pred«cs
.... '
l~
u•
6.
first government of Victor Paz Estcnssoro's National Revolutionary Movement (1952- 1956) rapidly expanded benefits to its middle-class :tnd working-dass supporters while cuning spending on the milit3ry, a military closely associated with the previous regime,J' A setting quite different fro m Bo livia existed in Brazil in the c3rly sixties. ~cause the government appar:uus was already in the hands of challengers 10 traditionally do minant dements, the feasibility of coalitions depended upon the experience of that earlier SCI of challengers. The government of Juscelino Kubitschek (19561961) followed no dass strategy, but it increased gove rnment intervention in the economy and it supported intermediate-class interests. Formnatdy for Kubitschek, Brazil was in the heyday of im port 5uhstilUtion, 50 he had the luxury of dividing an expanding pic. By the earl y sixties the mobilization of the subordina te dasses was sufficiently ad vanced to make a multidass alliance unavoidable but very costly. J030 Goula rt tried to preside over such an all iance, but stagnation had replaced rapid growth and trade-offs were made over a shri nking pic. Economic and political chaos followed, a nd in 1964 the military ousted the democratic regime.
fllaluating Dominant·Class Strategies. For the dominan t-dass strategy the key trade-oUs involve expenditures o n the military and police versus those on education, heallh, agricu hu re, public works, and fo reign relations. Table 9 shows thai during erises military spending grows at the ex pense of education and health hOI not at the expense of public wo rks and foreign relatio ns.lf Education and 28. Similu ~ .. enfS probably also o«urrnl in lhe !ale 19So. in Ecuador. S« FilCh (1977). 29 . Conlnry In nJ>K'~rion •• politt .nd milir~ry
"Rtt we..,
8ci •oci
~, • ~
~, •~•
,
,
~,
!,
~,
~,
- , :, "~' +
~
•
•
~
•
ci
+
~, ~,
• ~+
~, •"+
~,
~
, ~, ~,
"'
"' ~ ,
~,
1e
o
~ ~ •+ ~, •+
•
~ •+
~
, , ci, ci,
ci
~
•,
:•, :, :-,
~, §,
~, •~, ci, ~, ci, ~, •~,
•
~, :, !, ~, ~,
•"'ci •
0-
•
•~ ,
"o
~
~,
"•
o
•
N
• •
N
iiici ~ ~ ~ ~ • , , , ,
~,
, ~
o
•
!• •
~ o
~,
,
,, J
-,
o o ci
~
,
~
o
ci
• •
§o ~ ~ , , o
~, ~, ~ •
~,
,
~,
o
~ ;; ,
~
~ 8ci, ci , ,
",
~
• •
•
o
o
~
o
:. N
q
,
~,
>
~
•
~ ,
o
"
h~31 th
The Thfi)ry of SUf"ifllll
~fjtio"$
arc programs identified with intermediate and subordinate
classes, while agriculture, public works, and foreign relations lack dear class beneficiaries. JO
Figure 5, which traces 1945- 1974 budget shares for four key programs in Brazil, illustrates a vigorous dominanl-dass reaclio n, J' In 1965, the fim full ycar in which the Castello Branco
government controlled the budget. military spending reverSfii its long post-World War II slide (a slide common to most Latin American countries) and quickly moved to higher levels. Spending on police rose:u an enormous r:lle in 1965 and 1966, maintained a vc ry high level until 1969 (by which time an dficient and re-
pressive' poliu apparatus was in place), and never returned to precoup lC'vds. Educatio n and health pro grams, o n the o ther ha nd, behaved quite differently. Education jumped in the first postcoup budget, when the military sought populari ty thro ugh spending o n primary schools, but then declined until 1970. Subsequent rebellions of middle-class youth stimulated an ex pansion of unive rsities that led to an increase in the tot:al education budgct.lZ Heahh progra ms increased their budget share in 1966 but then went into a ta ilspin (rom which they did not recover until aft er 1974.
Eva/llating intermediate-Class StrQtegies. What happens when executi ves pursue such intermedi:lfe classes as teachers, white-colla r employees, shopkeepers. or middle-level bu reaucrats? According to Table 9. the key program is education. In crises, education gai ns against every program except police a nd agriculturc. H Health progra ms, which arc comparatively mo re important to the poor. lose to education. public wo rks. police, and foreign Telations. Public works programs ge ner:all y suffe r as wcll. o nly holding their o wn vis-a-vis the military and losing against everything else. Many administrations pursuing an intcrmedi:ate srrategy fa ce :a military threat as well as a class cleavage. As a result, military .ID. foreign ",b rions conTinues hert as an example of a ruditionll program. The diplom~lic KfYice has always bftn an outlet for the cafttr a. pinrion. of upp¢r-da$l gcntkmcn. J L NC>le Ihal these number. do nOi (n thq dod in .,ulin fifl,urcs) rtprocnt ntcs of chany oyer lime. 32. FOI' an eXlendni dHcussion 01 educationa l polirin, s« Arne!; (197)). JJ. Ndlhc-t u«l"ion i. 'mporunl . IkIfh poI,« and afl,ricu lru", atc much sm~lkr , h~n niucalion, .., .m~lI ,""I'tionalc dfea. on Iht ovcralltradc-off. The loss 10 af!,riculruf'C wu largely a ,."...1, of the long·lerm upwud lrend in ~fl,ricu lru", spl'nding in Chile.
6,
....: 1945
1950
1955
1960
1965
1970
Military
'''''enor (police) Education H..M
FigLire
j.
Budget SharC$ in Bruil : 1945- 1974
pacification may accompany an intermediate-class strategy, and the total effect on military spending rould he positive. The rcal winner, however. is education: Middle-class Icachen and middleclass pupils ind«d creatc a powerful consliluency. Evaluating Subordinate-Class Strategif's. T CSIS of Ihe subordin:u cclass strategies produced leSS consistent cstim:ucs, bllf they did reveal the priorities of administrations involved in such approaches. In crises, ai ucalion and hcahh lose some: of Ihe edge they maintain over the military in nonctisis situations, and bOlh education 2nd hc:ahh gain SUbSI3ntiaily against agricult ure. Public
,"
The Theory of Survival Coalitions
66
works is favored over the military in hOlh crisis and noncrisis silUations. Heahh docs well against education in noncrisis periods hut loses during crises. Overall, then, subordinate-class strategies include some dfons 10 satisfy the military and social programs gct a lower priority, hut the key programs arc those generating
jobs. Evaluating Multiclass Strategies. Leaders typically adopt multiclass approaches when they realize a subordinate-class strategy is no longer viable. Since these administrations usually face severe economic problems, and since economic instability can lead to military takeovers, they ought 10 seek immediate military support. Instead. adminimations pursuing multicJass strategies generally make no such efforts or make them only after their initial survival crises. In Argentina and Bolivia, for example, Peron and Si les ZII~ azo lowered military spending upon beginning their terms bur raised it each year thereafter. j 4 )030 Goulart steadily lowt'red military spending. The fact that the military overthrew all three administrations helps us unde rstand why multiclass strategies have as little success as strategies focllsing on the subordinate classes alone. H Figure 6, which charts the movemenls of military and health programs in Bolivia, presenlS an example of a muhiclass strategy. The first governmenl of Paz Estenssoro's National Revolurionary Movemenl (MNR) held office from 1952 until 1956. In four of the five years health programs gained more than the military. But the MNR could not sustain a multiclass approach- with its redistributive changes in policy OUtputs-when inflation was accelerating and the GNP was plummeting. The regime then turned to the social groups that had provided its original leadership. Paz's successor, Siles Zuazo, rebuilt the military while cutting social programs such as health. When Paz returned to office in 1960, he continued the new pattern with steady increments for the ar med forces. Unfortunately for the MNR, paz was unable to appease the generals. Colonel Barrientos overthrew the MNR in 1964, and the colonel's own effort at military pacification soon produced another health y increment in the armed fo rces' budget (Mitchell 1977). 34. Note that in Pu6n's ca~ this is tne term beginning in [951. 3S . Oil-rich V~nczllela i. an IIndn.tandable n cept;on.
"
"
6,
Survival Srrategies and Expenditure Trade-offs
~
Mililary
11 ~
t
o
tj
10._
I
Heallh
1945
1950
1955
1960
1965
1970
1975
Figure 6. MilIlary vs. Health Spending in Bolivia: 1950- 1975
The Economic Context of Allocations Thus fa r the discussion has emphasized the share of the budget allocated 10 cenain programs. In principle, shares of spending can rise or fall regardless of movements in the gross domestic product (GOP), exports, or price levels. Still, the absence of a necessary connection does not deny the possibility of an empirical relationship. This section develops and tests some simple linkages between the health of the overall economy and trade-offs in budget shares. Suppose we discover that whenever the GOP rises, the share of the budget going to the military declines. Why might that happen? Surely an increase in resources does not inevitably cause a decline in the military'S budget share. When the GOP rises, however, the total budget is likely to rise, and a rising total budget makes it easier to cut the military's relative share without the political danger of an absolute CUt.
• •
68
Tht Thtory of Survival Coalitions
ExKUllvts make three decisio ns rd:lIcd to budgets. They select
a Strategy. deu:rminc the 100ai size of the budget. and decide program shares. Their first step is the adoption of a strategy; then they determine the sile of the 10 lal budget and the sile of each program. Decisions to exp:md or conlTat l the total budget can only be understood in the context of the resource base. A political milieu in which the 101:1.1 budget grows while resources shrink is very different fro m a milieu in which the total grows while resourcu gro w. In the former case alloca tion policy is aggressive and risks fiscal a nd political 51T3in ; in the latter it proceeds from a secure and ex panding base.
Strategies and Global Increases When :I. strategy calls for expanding a cerlain program, do other programs gain absolutely even if their relative sizes shrink ? If, fo r example, an administration follows the military pacification approach, do nonmilita ry ex~nditures rise absolutely in order to cushion the competitive struggle? Table 10 focuses on programs receiving increments during political crises_ Column A represents increases in the total of no nstrategic programs. In fo ur cases-military pacific.1 tion, bureaucratic recruitment, local inlerests, and intermediate classes-losers are cushioned by enl:uging the total budget. But when elCecutives adopt do minant-class approaches and boost the military budget, their fi scal conservatism forestalls increases in other progra ms. Column B of Table ) 0 examines the movements of a n indicator of fiscal stress-that is, the ratio of all nonstrategic (or nOnla rget) programs to the G DP."6 Except fo r adm inistrations choosing mul· ticlass strategies, total no ntargeled spending ex pands no fa ster than resources. Administra tio ns adopting muhiclass approaches, ho wever, seem 10 be under great pressure, so when they sP<'nd on their target programs (education and public wo rks), they augment nonfavored programs eno ugh to prevent absolute losses. The military, fo r example, may sec: its share of the budget shrink, but aCtual expendilU rcs will go up. In other words! Multiclass gov· J6. GOP was 1a&&N orw: yu. b«2UIo<' upend"u.n (lIem,ely.. affect (h~ GOP a/ld ~"'UK ;/lfOflTUrion abotillhe GOP IS no\: always tim~lr. EXC'C\lliy~. mus! uS(: some sim ple decision tul~_£or tumpl., 'M usumpl'ion ,lI a, ,he .ale of , hng. of Ih. GOP will he ,II. o..ImC' at ;n .al. of (h~ngc ;n Ih. p~;O\l1 yu •. Thu, an ;nc'US(: ;n the budgt"1 ;n yu . I tqU;v.aicnl 10 Ihc ;OCN:llo<' 1/1 the G DP in yu • • - t would ~ "strnoJns.-
h
I
Ma
6,
Survival Strategies and Expenditure Trade-offs Table 10_ IncrtllSts in Adllerwry Programs, During Political Crises Mun lncreaM' in
(8)
Progf,m
NrmlllTgll P.ogrl2ms (%)
Rl2tio of All Nrmtorgtl Programs 10 lAgged ChIInge in GDP
MiliLuy Salaries Public Works Military Military Education Public Works
0_091 (O.Ol)' 0.047 (0.04) 0.084 (O.D03) 0.03 (0.33) O.lU (0.08) 0.031 (0.16) 0.031 (0.30)
-5.46 (0.44) 1.18 (O.SS) -0.589 (0.86) 5.79 (0 .11 ) -1 1.44 (0.31 ) 1.90 (0.08 ) 2.60 (0. 10)
Tug~1
Mililary Pacification Rcauit Bu~aucrats Local Interests Dominant ClnM" lnttrmN1iate Cbsses Multidass
Expendituru
1M All
Strategy
Absolut~
'The numbc:, in p.«n,het.es is ,h. prob.bili u,o.
ernments cannot cut budgets even when faced with economic crises. No wonder they have (rouble surviving. When executives cushion relative losses, which programs do they favor? Table 11 compares me effects of various str:uegies on key programs in two contexts: while total budgets are rising and while they are stable or declining. 1? When the military pacification strategy is optimal and budget totals are rising, education and public works programs cushion their relative loss completely. Because a rising total could al10w an even greater relative shift to the military, this cushioning is quite unexpected. In a sense, the finding underscores the curious competitive position of the armed forces: Spending on the military may he politically crucial, but economically il is pure waste. If the total budget cannol grow, military pacification requires relative shifts. When executives can augment the whole budget, popular job-creating programs like education and public works hold their own against the armed forces. The bureaucratic and local-interest strategies reveal the same tendency for trade-offs to shrink when executives increase the whole budget. Leaders know that losers punish more Ihan winners reward. Hence they reason that il is better to maintain shares and 37. Stnltegies with insuffiriem castS have !>fen excluded.
"
"
70
The Theory of Survival Caalitions
Table 11.
Effects on Key Trade-offs of Rising Budgets During Poliric.Ji Crises (a) Military Padfication Estim~ted
Ex~ndiru...,
Tr~de-off
Ris;ng BlUIge/
Categories
Parameter
Stable or Dedi";"g BlUIget
Military (+ ) Public Work. (-)
-0.034 (0.40),
+0. 104 (0.0 1)
Military (+ )
+0.022 (0.36)
+0.055 (0.02)
+0.050 (0.13)
+0.138 (0.0001 )
+0.048 (0.21)
+0. 1 IS (0.01)
+0.074 (0.003)
+0.060 (0.02)
+ 0.055 (0.03)
+0.127 (0.0001 )
£ducation (-) Military (+) Health (-) Military ( +) Wclbre ( - ) Military (+ ) Police ( - ) Military (+ ) Foreign Relllions ( - )
(b) Ex~ndirurt
Bu...,aucrat;~
Recru;tmmt Estimat~d
CaTegory Salaries ( +)
Parameter Stable 0' Dedi";"g Budgn
+0.045 (0.47)
+0.226 (0.0007)
(e) LfX~1 l"tefc,1S Ex~ndirure
Tradc..,ff Catcgori.. Public Work< ( + ) Mililary (- ) Public Work< ( +) £ducalion (- I Public Work< ( +) Health (- I Public Work< ( +) Foreign RcI'lion$ (-) Public Work. ( + I Welfare ( - ) Public Works ( + ) Poli"" (-)
Estimated Paramefer Rising Budget
Stable
Or
Duli"i"g Budget
- 0.007 (0.89)
+0. 159 (0.0003 )
+0.060 (0.22)
+ 0.126 (0.008 )
+0.0]4 (0.73)
+ 0.191 (0.0001)
+0.052 (0.20)
+ O.HI (0.0005)
+ 0.085 (0.08 )
+0.119 (0.05)
+ 0.050 (0.08 )
+0.179 (0.0001)
'Th< .ipJ.d entf)' i. tM .. timat«l codfirin with Thr entf)' in p.>M'lt~ i. tho .W>ci.t<'d .igni!iea"'% lovol.
~II
dommi .. indud«l.
"
u.
Survival Strategies and Expenditure Trade-ofts
7'
boost everyone, particularly since some budget participants may be more concerned with absolute than relative gains_ Strategies and Inflation
Are budgetary allocations affected by the endemic innation characteristic of so many Latin American countries? If prices change at a constant rate, executives learn to adjust, and spending should be unaffected. When prices jump suddenly, however, the real incomes of govemment employees decline. Executives seeking to recruit bureaucrats are forced to increase salaries just to compensate for the unexpected innation. When the rate of innation jumps, the share of the budget going to wages and salaries drops.JI Since (alling salaries can hardly cause increasing inflation, executives must fail to adjust salaries quickly enough when inflation accelerates. The salary share declines because executives maintain other components of the budget at the expense of wages. If executives really care about bureaucrats-that is, if the bureaucratic strategy is optimal-the salary share ought to rise, not fall, when inHation accelerates. Table 12 confirms the hypothesis: The coefficient of the bureaucratic recruitment variable roughly doubles when inflation is rising. Once again, losers punish more than winners reward. When bureaucrats are the political target. executives prevent losses in wages. I t 38. Consumer price increaoes are calculated Irom the JMF's ',,'ema'ional fi"allcwl St4liJlicJ (line 6-4). The equarion takes lhe lorm Salary share - A + B (mili~ dummy) + . . . + f (dous dummies) + Z (inHarion) where: Z is the pu::u""tu of inteRS!. 3~. The condwion that jumps in inHarion cause administrations pursuing bureaucralic recruitment 10 raise salaries is risky, beause s:olary increncs may themsel\'elI cause inHation. Note, howe\'er, that ino;reases in the rate of change of prices were associated with declines in wages when nonburcaucratic approaches WCre in dfeet. Ginn equal overall .pending increases, il would be anomalous il some .trategies led to il\CfC;tSQ in inOalion while others led 10 dcauses. It i5 mo« rc.uonable to conclude thai gains in the salary share arc consequences of burHuaatic strategies. If it i. assumed that sudden changes occurring in the first two quarters will be refl.cted in Ihe salary budget thai year, while ch.nges oa:urring in the 1....1 two quarters will affect uJaries Ihe nexl year, lhen estimales using quarterly price dat.. b«ome even stronger. If the bureaucratic recruinnent paramCttr Wett nOI a re.ponse to inflalion, the estimates would become weaker.
"
"
71
The Theory of Survival Coalition5
Table 12. Effects on the Salary Share of Rising [n(/alion During Political Crises Burcaue.a.ic R(eruitmcnr (Esrimatw P.r.m~er) Exptnditurc Calcgo'l'
RiSIng In(lation
Sr4blr o. Declining In(liJ/;on
Salarie< (+)
+ 0. 153 (Om)"
+0.08 1 (O.20)
"'Tho ,igM
11>< entry in
Conclusion We learned in Chapter t that ex.ecutives use expenditures as a weapon in their struggle to hold onto their positions. To use the budget efficiently , executives must break it down into specific compom:ms [argetablc for maximum political dfcet, and they must conccnlrate changes in these components in periods when they ace most vulnerable-Ihat is, in periods of polilical crisis. The political use of Ihe budgel requires a strategy. Executives gather information ahout the cleavages of their societies, the resources of each political force, and the slate of their economics. They filter that info rmation through their own ideo· logical beliefs and they choose, in the end, one or more of five coalition strategies. Because social and political groups in Lalin America consistently demand specific programs, the five strategies can be linked to trade-offs in Ihe components of public expendilUre. We found that survival politics indeed have allocational consequences. Four of the five strategies- military pacification, bureaucratic recruitment, local imcrests, and social class-yielded essentially [he results predicted. One-the transfer strategy-did not, bUt severe measurement problems hindered evaluation of this approach. If military pacification is an optimal strategy during political crises, executives boost the armed forces' budget at the expense of programs in health, welfare, and police. Education defends itself fairly well, and public works programs prove invulnerable. When the crises pass, these same leaders aller their military strategy, allowing the armed forces neither gains nor losses.
,.,.,-
l~
-
Ie
Survival StrllugieJ and Expenditure Trade-orrs
7)
Salary and public works approaches turn out to be strategic opposites. Bureaucratic recruitment leads to gains in wage and salary shares at the expense of public works (with its substantial physical construction component), while the targeting of local interests reverses that pattern. The social class strategies aTe really a group of sharply variant substr:uegies. Leaders adopting dominant-class approaches stress the military at the expense of education and health. Agriculture, public works, and traditional ministries like foreign relations survive unscathed. Intermediate-class strategies favor education against almost everything, including programs that benefit the poor. When leaders target the subordinate classes. they try to keep the military happy while boosting job-creating programs like public works. Welfare programs enjoy some support, but the central target is the creation of jobs. ExecUlivcs seeking a multiclass (.ualilion try to do it all: military increases for pacification, public works for job creation, welfare programs for social needs. Leaders soon discover that losers punish more than winners reward. They respond by attempting to cushion programs whose budget shates decline. But cushioning losers is expensive, because it creates a constant impetus toward budgetary expansion. Multiclass and subordinate-class coalitions fail in part because they cannot reconcile a "no losers" allocation srrategy with sufficient economic growth to support that strategy. For politicians, survival is just one goal among many. Here we asked what governments do to survive. The next step is to determine whether they do anything else.
• •
3. With Time to Breathe: Policy Postcrisis Administrations
In
This book has been exploring the struggles of political leaders (0 hold OntO their offices. Though many exttUlives losl their positions, a majority survived to enjoy a time of grealer security and autonomy, a time in which they could pursue longe r-range goals. This chapter begins the analysis of public spending in such postcrisis periods. It focuses on four questions; Were the allocations of deposed executives different from those managing to compiete their terms? Did the allocations of elected leaders succeeded by an administration of the same political party differ from those succeeded by the oppositio n? Were the budgetary priorities of military governments different from those of civilians? And, finally, were there broadly similar patterns of overall postcrisis spending? These four questions are logical extensions of the analysis of survival strategies. Since leaders rearrange spending priorities because they believe such rearrangements will help them, expenditures ought to increase success in avoiding coups and winning elections. The possibility that military and civilian executives allocate expenditures differenlly has stimulated a substantial body of comparative research, but the issue remains unresolved. Finally, overall postcrisis spending is important as an illustration of the limits of survival-centered models and as a direction for future investigation.
Victims of Coups: Bad Strategists ? Chapter 2 linked expenditures with the need to pacify the military. In the end, administrations either complete their terms or
With Time to Breathe Table 13.
Military wups and
Prior
Pnuntase
Challges ill Budget Shares
a.an~
"
from PuviOllI
Adminimation in P"'&""nI', Sha.e of TOfal Expcndirurcs
p...... Miliury Eduction Agriculrun: Public Works Hulcl! To tal Budg"
Su,viwd
O...ud
-0.071 0.068 0.)00 -0.034 0.085 0.086
0.122 0.092 0.171 0.170 0.051
O.OH
....... >0.05 > 0.26 > 0.«< > 0.0) >0.3 4 > 0.10
'The prol>.bililJ ..tim.tN i. . .. ociatN WIth the pncentose of rima. difference 0C>tn of that m"IJU'ud. could """,. by ,h._. II p,obo.bility of 0.05, 10. ~.mpk , muM th •• 5 tim .. OUt of 100 • dilfffcnct in militory .pending I>nwkn ,un'i ...... and .;mml II 1>,.. .. 'M difttr=« reported he.e could oceu. by
they are ovenhrown. 1 Will !he survivors be the administrations that most increase the military share of the budget, or will the military inevitably remove some administrations-lavish rewards to the officers notwithstanding? For five important programs, Table 13 compares adminis!rations that finished their terms with .administrations that failed. Compared to ousted executives, the survivors CUt military spending, decreased public works, and expanded total budgets. This test is too simple, however, and, as a result. tOO ambiguous. Survivi ng governmenlS mighl CUI the military budge! for a variety of reasons: They mighl be more secure, or they mighl be· lieve the military will not respond strongly to budge!ary losses. Perhaps ousted administrations lasted as long .as Ihey did because they paid o ff the milit:u y-that is, the mililary might have over· thrown even more administrations if executives had not granled them substantial budgetary resourc6. More precise hypotheses and a morc r~fined measure of spending are required. Table 13, which estim:u es spending changes be· tween administrations for each of five programs, faits to capture the concept of spending priorities. If one ~xecutive pays for a mit· 1. Among the COUntries in OIIf I nalysi., II,., only uapOOn$alC Boli~ia in 1952 and Nicarar;ul in 1979, wM.e popular revolutionary plMIps OVcnhKW the es.abli.h.ed military.
"
Tht: Theory of Surviva l COl/litions
jl;ary Increase at the expense of public wo rks while ano ther fi nances the same increase at the expense of education, tht, two ad ministrat io ns 3rc not equivalent. The concept o f prio rities requi res idemifying losing as well as winning programs. Suppose we wish to measu re the I; hanges in priori ty a n administration gives to thl!Se five programs. First, calculate the percentage cha nge in cach program between a gi\'cn adminisu a tio n and ilS prcdCCC'Ssor. Next, rank these percentage changes fro m the program gaining the most to Ihe progra m losing the most. Each rank ordering describes a single ad mi nisu a t:on.: An adminisU3 tio n favo ring the mili tary over ed uca tio n wi ll be di.slinci fro m one whose priorities 3re reversed, even if the aClUal (h ang~s in th ~ Inili ul")' ,ha r~ a r~ th~ sa m~ . Likewise, an adminiSlr3tio n giving gr~'lI es t priority to health, with miliury seco nd and educatio n laSt, will be distinct from one thai ran ks heahh first , ed ucation second, and Ihe military laSt. Now consider agai n thc diffe rences between Ihe allocations of surviv ing and failing admi nist rations. Sincc the re is lillie rcason to l!Xpt'ct expend iture shifts from executives with no real possi· bility of being overrbrown, only those admin istratio ns vulnerable to coups (that is, those for whom mili ta ry pacific:uio n was an optimal slrategy) will be included. Among such admin istrations, twenty·two ra nked military spending last o r next to last among the five budgeta ry programs. The armed fo rces ovcrrh rew twelve of t hOSt' twemy- two, or 55 percent. Sixteen ad lllin islr:ltions fa · vo red miliury programs, ra nking them fi rst o r second, :mel thc 3rm~d forces ovenh rcw six of these six teen, o r 37.5 percent.' Vul· 2. This JC~]ing pt=rw" only tlrdiru l infoftn:;l tioll. S~pp<~ olle ati minillf~I;On lhr« l" ogr:J.m. + 10. O. ami - 10 perCent anti 3norhe r changes lhe ... me p'ogNm, +5,0, :;I nd - 5 percenl. ~or rankings ...·""'141 be the nn .... Si nce the p,otnbility of . .. m lu ge diffe,~ ,n In. 8"in or los~ of ,mlivi,Jual progr am. dcpc1'fal u pend., ....... arlcoliun In me I, " er .hould minimize .!iilomon, "" .."'.! by the f... ,mer. For ~hern3nY~ tru t· men1$ of th( mu s.. rcment problmt in b.. dgn:lry analysis. su Gist (1, 79 ) ~mI P. ruff ~mI Pooohk ·W..,co ( I ' ' '). 3. Among admi nistrations ranking In. mih .. ry fo .. n h ",1' ~fth and for ........... milita ry pa cific'llion w~. opIim>ll, the "",ncd ind..detl L.:m ui anti S-inchn Her· n"ndn (EI Salvador), Torr .. (Roii"'l ), Cutillo A .m~, (G.utcmab ), L6pc~ ArdL.no (Hontlu,a s, 1972- 1975 ), Allentle (Chile-), Vups (by sIIki ...) .mI eo.. ll" (Bn»I ). PerOn (A,gemina), B.. st~ m .nt. anti B.t. un.Je (Peru ). ~".J Veincn Ib •• n (F.cuado. ). Survivor. included Anna and Lucu Ga,ci. (G .. atemala ), Lo.ano Diu, Melga, Cas"o, and 4\p.z Ard lano ( 1963_ 1'.171) (Hondu r:IIsl, Cuzm:ln (Dominin n Re· public), tI. Ia Gua,di. and Ch ia,i (panam. ), Kubitschek (B.au l), :;I nJ O,J,i. (Peru ). ch~ng<s
With Tim/[ to Brtatht
m: rable administrations that responded by according the military a high priority were thus more likely to suo'ive, but the differences are not terribly impressive. Consider, however, juSt those (xecutives who came 10 office after a competitive election. Since they rarely enjoyed close ties to the officer corps, their administrations should be especially vulnerable. Twelve such executives failed to payoff the military (according them fourth or fifth budget prio r' ity); eight were victims of coups. Six elected executives did au end to their military suppon (according them first or second priority); all but one survived. Thus highly vulnerable administrations found that anention to the military's budgetary nceds yielded substantial payoffs. The military is hardly the only interest group making claims on executives, so leaders threalened by military revolt arc seldom able to devote unlimited resources to the armed forces. Twenty·two executives with a military pacification problem stressed spending on education and health programs, ranking neither lower than third .~ The armed forces removed thirteen, or 59 percent, from office. Of thirty·dght administrations not ::according the same high priority to education and health, only eleven, or 29 percent, were given the boot. J Thus ('xecutivC$ who placed a heavy emphasis on For !h~ who gavr 1he ulmd loras r:lnks of lirst or s«ond, OUStW ;ulm;n;srra· lions induded Chn~ ( P~nsuayl . Pu (Bolivia), Onpnia and Vidda (Argmlina). Roia~ Pintlll (Colombia]. and /'ira Jimenez (Venezuelal. Survivors induokd Rom· ero (EI Salvador), l.:Iugtnlf,l and Prralla (Gualcmalal. Roman y Reyes {Niaragw.l. TorliiO'l and Robin ( Panam~), Ban ter (Bolivia), Pinochn (Chilt), Cruz (Hnndurasl, and VcI:I.KO Iblrra (Ecudor, 1952-1956). 4. The .;ri!erion of ranking In !ht lop rhrtt is adminwl)' arbiITll')'. h wa, chmcon 10 yi(ld mough U'ln for analy,is, 5. Gus«'guay). Survivon indud«) GlUm;! n (Domlniun Republic), Rodriguez Lara and Vcluco (blrr~ (1956-1960) (Ecuador l, lucas Garda, Ydigoras Fucn!es. A.... na, and lauge.uJ {Guu(malal, CI'II1, Gj lvn, and Mclpr Curro (Hondura.). Roma n y Reyn and Tacho {Nicaragua l. Arias Espino ..., Alcibiadel Al'05Cmma, Chiari, and Roble. (Panama ), SI~ntr (Paraguay ), Ptron (Argrnrina, 1951- 1955), SilC5 Zuno and
,.
Tb~
Theory of Survival Coalitions
social programs were almost twice as likely 10 be overthrown 35 executives who did not. Without question, such "reformist" administrations were likely to favor working-dass interests in OIhtt aspects of economic policy as well, thereby augmenting their vulnerability. According high priority to the military budget wa~ an overwhelmingly 53fer strategy.
Elt:ctoral Success: Can Spending Buy Votes? Executives oftcn pursue policies thar bring immedi:uc rewards but have costs they can shift 10 successors. In Latin America the nearly universal prohibition on immediate reelection of the exec-
utive reinforces this myopic attitude. If you can't be around when the bills corne due, why worry about them? Why even Ix: concerned with the next elt:ction? In fac;t, exec;udves do c;arc about the election of the ir succesro rs .
Though constitutions may p rohibit immediate reelection, many leaders harbor dreams of assuming the presidency again after a term on the sidelines. Even without such hopes, loyalty might stimulate leaders 10 help their panics' candidates. Viaory by the incumbent pany in the election of a successor puts the stamp of approval o n an administration. If, on the other hand, the election is lost, the president may be a bme duck for many months as his lenn nears its end. If executives do care about the election of their successors, particular panerns of spending might increase their chances of victory. Success, unfonun,uely, occurs pretty rarely, because incumbents nearly always lose. In the entire World War 11 period, incumbent panies in Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, and Venezuela won only six times in thiny ekctions. If elections are generally lost, prior expenditure shifts may be irrelevant or even counterproductive. Perhaps the armed forces still need pacification, but group, class, and loca l strategies become measures of intent and hope rather than predictors of success. In an imponant study of Latin American elections, Roben Dix (1984) anributes high uccul ive turnover 10 both political and ecoBanur (Bolivia), KubitJd>ck alld Castello 8n.nro (lkuil). l'illochfl (Chi It). Odria (PmI). Molina and Romero (El Salvador), and Pba Jimenn (Vrnaufla).
With Time to Breathe
79
nomic factors. Constitutional prohibitions on immediate reelection weaken presidential control. Presidents cannot tfansfer their authoriry to the candidates of their parties, and these candidates may even choose to :lttack the president in order to mobilize their own followers. Even without this inevitable weakening of incumbents. executives have a hard time coping with steadily rising levels of social mobilization. As Latin American populations have become more urban, they have become more restive and better organized politically. The result, in Dix's view, is an ever-increasing gap between popular demands and the limited resources of executives. How can the linkages between expenditures and electoral out· comes be discovered ? In some cases the executive's party has no chance at or interest in reelection. When a party exists only to support a certain leader, the victory of an opposing party in the next presidential test is not an instance of electoral turnover. Ecuador's Jose Maria Velasco Ibarra, a charismatic figure par excelience, was constitutionally prohibited from running in 1956. Velasco supported Camilo Ponce, the candidate of another party. Ponce's victory did not make Velasco a winner, but surely he should not be classified as a loser. In the election of 1952, outgoing Ecuadorean President Galo Plaza chose to back no candidate (Marn 1972). Like Velasco, Plaza neither won nor lost. Personalist parties represent only one of the difficulties in classifying outgoing executives as winners or losers. Democratic Ac· tion (AD), the dominant parry of Venezuela. split over il5 attempt to elect a successor to Raul Leoni in 1968. Together AD's two candidates took a majority of the popular vote, but singly neither could overcome the candidate of the opposition. Since dissatisfaction with Leoni did not cause the split in AD, it is hard to call the electoral result a negative judgment on his adminisrration. Eliminating such ambiguous cases leaves twenty incumbents clearly interested in the election of their successors. Eight were victorious; that is, the incumbent president or his party captured the presidency for another term.' 6. Tht twtlvc Iostn, Frti and A1QUndri (Q.ik), And." Pf rn and Oddcra {Vcnuueia), Edu.ndi, Ulart, Odilbc., Tujos, and Cuaw (CosQ Ilia ), Ibl~gucr {Dominican Republic), Kubiachck (Bealil). and M~ MonraltgrO {G",rtmala). The tight winners, Pcr6n {ArJentina. ' 9S II, Apra {Costa Rica, 197-4 ), Guzm.in {Dominican Rcpubti<;), do la Guardia and Chiari {Panama), I..6pn: Michelson (CoIombia), BnancoUrt (VclICWtla), and Osorio (EI Sa .... ado.).
80
The Theory of Survival Coalition$
Table 14.
Suc:cenful Elec:tiOlu and Pn'Qr Change5 in Budget Share5 Pe,centage Change from l'rC\'io\ls Administration in Program'. Share of TOlal ExpendiTures
Program Military Educalion Agriculture Publi c Works Health Toul Budgel
Won Succuding Lost Succuding Elution Elu/ion - 0.035 0. 190 0.364 0.102 0.245 0.046
-0.132 0.013 O.llJ - 0.103 - 0.058 0.089
I'rob
> 0.23 > 0.10 > 0.44 > 0.32. > 0.06 >0.20
In part, elections are referenda on the policies of outgoing administrations. Table 14 presents the records of the incumbents
in terms of the priorities [hey gave to each program. Executives who were successful in the subsequent election did not necessarily increase total budgets, nor did they increase the budget share of the armed forces. Only two programs, education and health, grew significantly faster in victorious administrations than in defeated ones. Consider the interprogram priorities of winning and losing administrations. Of twelve losing executives, seven ranked education and health spending among their three most advantaged programs. Winning executives were only slightly better, with six of eight favoring education and health. None of the eight winners gave military spending a first or second prioriry, bur only [\vo of twelve losers favored the armed forces either. If executives try 10 structure program changes 10 maximize political payoffs, they might cut programs that have done well in the recent past on the grounds that payoffs from such programs are diminishing. In fact, winners do appear to try that strategy. If we compare the absolute size of the budgetary shares these winners inherited, we find that executives cut or granted smaller increases to programs beginning at higher levels. Among losing administrations, there was no relationship between the inherited size of education and health programs and subsequent boosts or cuts. Winners, in other words, acted more strategically. If education and
,
With Time to BTeilthe
8.
health were already at high levels, additional money would produce smaller payoffs, so leaders gave these programs lesser increments. Losers fai led to make such calculations-their increments were unrelated to the budgetary base. These differences, no netheless, arc small. If changes in spending explain only a small part of electoral success, what factors do contribute to electoral victory or defeat? Remember Dix's social mobilization argument: " lncrelSing demand [social and political} outruns avai lable resources" (1984:442). If this argument implies that economic downturns lead to electoral defeat, it is incorrect, because losing executives enjoyed an increase in GDP tlllice that of winners. If, on the other hand, Oix is suggesting that voters are more likely to reject incumbents when social mobilization reaches a ccnain absoh.lle level, then the results are more promising. The eight winning executives generally ranked below the losers on such social mobilization indicators as literacy and newspaper circulation. Of the twelve dear losers, nine were from Costa Rica, Chile, and Venezuckl. Of the eight winners, fou r were Caribbean or Central American (excluding Costa Rica). Thus polities with high levels of social mobilization and politicization make life harder for incumbents. A further due comes from examining the countries in Latin America where presidential and legislative elections were nOi always simultaneous-that is, where presidenrs f:tced midterm legislative elections.' Table 15 exposes the consequences of midterm elections by comparing changes in the expenditure shares of executives who fa ced midterms with those who did not. Clearly midterms affected expenditu res. The growth fate of overall spending almost doubled when adminiSlr3tions faced midterm pressures. Spending on education- a program with immediate polilical payoffs- was 40 perccll! higll<:r. Publ k wurks. which grew more than 3 percent in the absence of midterm elections, failed to increase: at all in administralions facing midterms. The biggest loser was the military budget, which absorbed a 78 percent larger cut when executives could not avoid midterms. 7. Such d~rnom occurred in Br:l1il. Chir~. Colombi a. Ecuador . and EI Saluoor. ExJl<'ndit\lr~ dat~ W<:", anilablt from lilt following midlmn rltcrion., Bra_ u l, 1!154. 1!l58, 1962; Chile. I!JSJ, 1957, 1961 . U6S. t969. 197); Colombia. 1%0, 196-4. 1968: Ecuador, 1950. t954. 1958. t962: EI Salvador. 1!164, 1966. 1968, 1970. Data on Ecuador. ho,.,tvtr, a", Incomplt lt.
"
,
The Theory of Survival Coalitions
Table IS. Midterm Eltttio"s a"J Changes in Budget Shares P(r~nta't Ch:tnge from l'I(.ioo5 Adminislruion in Shift' of TOlal Exp<:ndituf",
M jdu .... ugiJl.triw
Program M'litary
Educuion A&ri nllru ....
Public Works I-kallh
Tof:tl Budgn
F.ltctioll
- 0.118 0. 141 0. 11 0 0.002 0.09 1 0. 102
No
Midr~'m
ugisl.triIJt Ekctwll
-0.072 0. 101 O.J60 O.OJ ~
0. 10) 0.068
In terms of interprogram prioritits, the results are si milar. Ten of the twelve executives fa cing a midterm election ranked military spt'nding (ourlh or fi fth. Only half the no-midterm civilian Ic:adcrs tanked the military equally low. Health and eduC2tion programs finished in the top th ree in ni ne of twdve midterm-facin g administrations hut onl y 40 pt'rcent of those fret" fro m midte rms. Although the num ber of casts is small, these rl'$ults establish the high cost of midterm elections. Executi ves react to midterms by spt'nding more overall, cutting the nlilitary budget, and favoring programs with immediate payoffs. The risks arc obvious: Sudden incrtascs in the government budget may stimulate inflation, the military might k come dissa tisfied and restive, and economic instability could prove damaging 10 the incumbent party in the next gener:lI election. In the long run, mo reover, the budgeta ry shift toward consumption (a constq uence of the rel:lIive decline of infrastructu re programs) diminishes the government 's contri bUTion 10 economic growth. For executives, unfort unately, Ihe pressure of midtl'rm elcclions is not a pressure they can avoid. Military Govern ment :"Ind Military Spending Though the link between military govc.-rnment and military spending has become part of the st3ple fare of specialists in comp;lTative politics, schobrs ha ve bl·cn able to cstablish no consistent
8,
With Time to Breathl!
relationship between the two. Most recent work has argued that military governments spend neither more nor less on the armed forces man civilians, but research on this topic remains curiously tentative and unsatisfying. There are still disputes about the appropriate standards for measuring military spending, and analysts continue to ask whether military budgets grow under military rule rather than asking what factors cause a government to spend more on its armed forces. In a rttent anempt to settle the question, Gary Zuk and William Thompson (1982) measure spending on the armed forces in two ways: military spending as a percentage of the total budget and military expenditures per capita. They find that the military's budget share grows more under military governments than under civilians, but military expenditures per capita are tied mostly to GNP growth, not to the nature of a country's rulers.8 The authors conclude (rom these contradictory results that military rulers do not increase the size of military budgets "either in general or in comparison with their civilian counterparts" (p. 71). The analyses presenu,:d earlier in this chapter suggest that Zuk and Thompson's negative conclusion is unjustified. Unless we compare per capita military expenditures to per capita expenditures on education, public works, and other programs, per capita spending cannot measure the priority given to the armed forces . When the GNP rises, the whole budget will rise as well. Military expenditures per capila share in that expansion, but so do expenditures for education, health, and public works. The question, then, is whether the military gets more than other programs. The military spending question, in other words, is one of relative priorities. When Zuk and Thompson use an indicator of priorities (the military budgcr sharc), thcy find that military administrations do spend more and that military coups trigger additional increments. Still, little anention has becn given to rhe determinants of spending on the military. Within the framework of interprogram priorities presented here, how can we think about these factors? Philippe Schmitter (197 1a) has suggested a useful way to categorize administrations. For each year between 1950 and 1967, Schmitter classified every Latin American country as civilian comB. Zuk and Thompsun employ ~ -pooled crms·sc,:t;onal and rime ... ri"" an~l · y.is for .ixll'·six ooumrie •• induding all Larin America excepl. oddly enough, Chil. and Venezuda. M
1~",1
Ie
a
Tht Throry of SlIrviWli
8,
~/jtiOtfJ
pclilivc:, civilian noncompetitive, or military no ncompetitive. Latin Americanists have adopted this typology widely. and it can easily
be extcnded backw:ud to 1945 and forward
to
1980. Levels of
competitiveness a nd militarizatio n rarel y change during the tenufe o f ind ivid ua l leaders, so we: can a ggregall~ ,he ra nkings to whole administrations. The nature of the regime leads to sharp differences in the gains or lossn of budget sh:atcs. Sixteen of fo rry military-noncompetitive
administrations (40 percent) gave the armed forces the first or second largest budgetary increase among the five basic progr3ffiS (military, health, education, public works. and agriculture), Only six of fifty-four civilian-compctitivc admini.m atio ns (1 1 percent) and only five of Iwcmy-seven civilian-noncompetitive administrations (185 percenl) ranked Ihe mililary fi rsl o r second_ Overall, mililary-Ied adminisualions we re Ihree limes more likely Ihan civilian admi nisn alions 10 give a firsl or second budgel:lry prioril y to Ihe armed forces_· Wh y did mililary governmenlS spend mo re? In lighl of the theoretical argu meOls elaborated earlier in this book, fou r sources of ch:m ges in spending on the military merit invt'Stigalion: in crea~ in " illegitimalc" polilical panicipalion before Ihe military too k power, a low priority for Ihe milita ry'S budget in Ihe preceding administration, Ihe threat of a mililary coup, and a tight overall budget _ Once agai n, administralions will be labeled military no ncompelili\-c, civilian noneompeliti\-e, and civilian competitive. Expenditure data exist for only 109 administrations, so the analysis will be rt.'S u icted to simple bivariate relationships. "llIegitimate ~ I'articipatiofl
Certain forms of politics, panieularl y Slrikcs, demonstrations, and :lrmed ;maeks, question Ihe government's monopoly on Ihe use IJf force_ These actions th rea ten govern ments. Because elites dominate the language of political life. such flJrms of participation arc usuall y ci;lSsed as "i llegilimate." 9, MortO"tr, Ih CO'"P<'l ;l i~c"~~ o f IWO of II\( ",;,'i li.n-comP<',ilivc . dmin;._ I" UO"~
Ihal "lid f,,,'ur 1M .""cd fOfc""" "","cwh.1 douhtful. Robles on " anorn. by Ihe NaUQ/lal A~:.tmbly an d hp. in power by II\( N.n"n.' f ue n'" in GU~I. ",o ta w••
"' J~ imP<'Jeh.d Gu~,d. Ydlgo. u
vn ,r.
8,
With Time to BrCJJthe
How does illegitimate participation translate into budgetary rewards for the armed forces? Overall levels of military force are a function of past domestic insurgencies and international conflicts. Given the virtual absence of interstate disputes in Latin America in this period, the change in prioriry accorded to the military ought to be a function of change in the level of insurgent threat. Such changes, or "upsurges," of illegitimate activity were defined as a 300 percent increase in any of four indicators: protest demonstrations, riots, armed attacks, and deaths from domestic violence. lo Tables 16 and 17 reveal that military administrations arc most likely to follow periods of rising illegitimate participation. Civilian-competitive administrations, on the other hand, are least likely to appear after such an upsurge. The expenditure consequences are even more striking: Military administrations are much more likely to respond to upsurges in illegitimate participation by strengthening the armed forces. Civilian-noncompetitive leaders are slightly more likely to spend on the military than their civiliancompetitive counterparts, but both are quite distinct from military administrations. If illegitimate participation has been stable or declining, military leaders are still about three times as likel)' as civilians to favor the military budget, but at such times even military executives seldom respond with expenditures. The central finding, then. is the extraordinary fiscal response of military-led administrations to upsurges in illegitimate participation. Civilian regimes buy carrots; military regimes buy guns_ 10. The standard source for counts of semilegal and ill egal political activity is the World HIl"dbook of Po/i(iCil/llnd Socjll/lndicllro,s, authored by Bruce Ru.sen ~nd ochers (1964) and by Charle, Taylor ~l1d Michad Hudson (\983). Co.ering 1948-1\177, its volumes provide annua l country-specific in/ormation. The data were aggregated to whole administrations. The staodard of 300 pc:=nt for defining u~urgcs is obviously arbittary. Th.,.., administrations were all coded as (01l0w;"8 such upsurges: Peron (Algentina. 1\151-1954, 1\155); Paz Estenssoro (1%6-1960. 1960-1\164) and 8=ientOli (Bolivia); Castello and Colla e Si lva (Brnil); Allende and Pinodtet (Chile); Lau",ano GOmez. Rojas Pinill., Albtrto Uel'3l Camargo, Ca riOli Ueras Restrepo, and Turbay Ayala (Colombia); Cosra Rica, nooe; Balaguer (o.,minic.tn Republic, 1966-1970, 1971-1977); Velasco Ibma (Ecuador, 1961); Molina and Romero (E1 S:1I"3OOr); Castillo Armas, Peralta. and Mendez Monrenegro (Guatemala); Rlliz Cortines, Echeverria, and L6pc:z Portilln (Mexico); Tacho (Nicoragua); Akibiades Arosemena, Chiui, Robles, and Torriios (Panama); S[roess~r (Paraguay); Odria, Bcbundc, and Moral ... Bcrmudcs (Peru); P~rez Jiminez. 8cuncourt, and lroni (Venezuela).
111 I
•
86
The Theory of Survival Coalitiolls
Table 16. Admini$!ratiOIl Type and /'rior Upsurges ill Illegitimate Po/iti,al Activity Activity Rose in Adm;ni.I •• r;on of
No Chang~ in Activit}, in Adminimation of
I"edecasot
I'redecaso.
"
17
Administration Tyl'" Military Noncompetitive Civilian Non
10
" --"--
-"'"
"
Table 17. Prior Upsurges, Adminis/ration Type, and Currelll Ranking of Military 's Budget Share Activity Rost AtlminiSfr.niun Type
Military ( I ' lnd
1st
Mi li tary Noncoml"'titivt Civilian Noncom!",titiv. Civilian Co mp<'tilivt
9 1
- '-
"
Miltw)' J'd-5th 6
•
-"17
Activity S,able u. Declining Militar y J .d-50,
Military
1st o r lnd
J
"
"
)
-',
--"--
"
Prior Budgetary Cuts Leaders ignore the corporate illlercsts of the armed forces al their peril. How do different types of administrations respond 10 cutS made in the military's budget share during prev/otls administrations? Table 18 compares administrations following a period of low priority for the armed forces (budgetary rank 4-5) to those following a period of high priority (rank 1-3). Military administrations arc actually slightly more likely 10 succeed an administration that treated the armed forces well, hut this is mostly a consequence of the greater likelihood of one military administration following another. AI first glance, Table 19 indicates that military executives react strongly to prior budgetary cuts. Military-led governmentS are fi ve limes more likely than civilians to reverse budgetary neglect. But military governments respond to periods of high rewards for the armed forces almost as strongly; thai is, military administrations
,"
•
8,
With Time to Breathe
Table 18. Admini$!ration Type and Prior Budget Cuts Administr~!ion
Previous
R ~nkcd Mili!~ry Sh~,e
Administration Type
4th Or 5th
Mil;!ary Noncompe!itive Civilian Noncompetitive Civilian Competifive
Previous Adminimation Rankro Military Sh ..c ist, 2nd, 3rt!
""
--1L
" Table 19. Administration Type. Prior Budget CuIS. and Cu rrent Ranking of Military Share Previous Admini.lrarion Rankro Miliury Shut 41h or 5fh
Administ",rion Type Military NonCQmpetitive Civilian Noncompetitive Civilian Competitive
Previous Adminisrrafion Ranhd Military Sha", is!, 2nd, 3rd
C""1!tI1
C""1!tI1
Cu" e"t
Shart
Shart
Shart
'-2 5
-'8
3-5
, 5
--1L 59
C""l!tIf Shart
'-2
,,
3-5
~3_
-1L
""
35
H
are four times as likely as civilian-competitive administrations to favor the anned forces even if they did well under the prior administration. In other words, prior budgetary neglect is a positive but weak determinant of expenditure response. Military Pacification
In Chapter 2 we saw that administrations faced with the threat of military ovenhrow respond by increasing the military's budget share. Here the question is narrower: How do different types of administrations respond to that threat? When the military must be pacified, military-noncompetitive administrations are five times more likely than civilian-competitive administrations to respond by favoring the military budget (see
"
"
.
Th~
Table 20.
Throry of Survival Coalitiolls
Military PacifjUltion, Administration Type, Ilnd Expetldill/ rt Rt$ponu~
AdminiJInrion
Ty~
M i1il~ ry Noncom~lilivt
Civilian Nonrompericivc Ci.ili~n COnlpt'lilive
Mili/Ilry
Military
Mililll'Y
Military
SharI
Sha re J-J
5/"""
/-,
ShllTt
12
S
6
/
IS
--"-
-'9
--'L
/-,
,
•
- '"
"
'-5
•
"
·Admin.",.,;"n, ..... cI"li~n1 In "'ml ul ,II< ,mJlO'''''''' 01 p..i",in5 tho miliury .
Table 20)." In the a bsence of a p:lcificalion problem, military. nonco mpetiti ve administ ra tions :I re still 3.2 limes mo rc likel y to favor the :lrnu:d forces. Table 20 also shows that civiliancompetitive and noncompctilivc adminismuions behave very diffe rently. Civilian-nonco mpetiti ve administrations rcspond dra maticall y to Ihe need 10 pacify t he armcd fo rces, whik civilianco mpetilive adm inistrations fa il 10 respond a l a ll. Pe rh aps non· compelitive :'!dminislration§ are frec to res hape budgetary allocalio ns because Ihey do nOI b ce h051ilc legislatures or honeS( d CClions. The real split in dealing wit h potenl;:'!l coups is Iherefo rc nOI between mi lilary and civilian administralions; it is between competitive and nonco mpetitive administ ralions. Competitive ad ministratio ns have trou ble find ing the resources to buy sticks. because the demand for ca rro ls is ve ry hi gh.
The Com traint of Austerity We saw ea rlier th at per capila levels of mi lita ry spending are a poo r indicator o f the: military 'S budge: tary prio rity, because per capita indicators miss Ihe co ntexi of budgetary change. M ililary expansion whi le o lher components of the b udgel a re: growing implies somelhi ng qu ite d ifferent from mil itary ex pansio n whil e: ,he: base is stead y o r shrinkin g. II . The mil itary budgn is f~Yo.(d by ~,n of ni~u'(11 mililuY 'noncompo:l'liy( adtnilliwalions bul only two of tw'nry.,i~ civilian-mmpo:t,ri., admilliural'onl.
8,
With Time to Breathe
Table 21. Budgetary Austerity and Administration Type Ch~nge
Administration TylX' Military Noncompetitive Civilian Noncompetitive Civilian Competitive
UJJ
in Total
Than PtMUenor
S~nding
More Than P,edecenor
16
15
,
IJ
~
--1L
so
Table 21 shows that administration eype is unrelated to changes in the volume of overall expenditures. But in Table 22 we see once again the sharply different proclivities of military administrations. When total spending is rising fasler than it rose under the previous administration, military administrations are much more likely to expand the military's budget share. Civilian-competitive administrations rarely use budgetary largesse to benefit the armed forces, and civilian-noncompetitive administrations are even less likely to spend overall expenditure increases on the military. !1 These findings are far from obvious. One might expect the military to reward its own whatever the overall budgetary constraint. Actually, the possibility of a larger budget liberates military administrations more than civilians. Military regimes do feel the constraint of a tight budget. Conversely, a rising total budget under a military regime trebles the chances of a high priority for the budget share of the armed forces. Recapitulation
Two problems-insufficient attention to the measurement of military spending and insufficient attention to the factors that motivate it- led us to question the traditional conclusion that military government and military spending are unrelated. When we measured spending on the armed forces with a multiprogram ranking, it became clear that military-led administrations do favor the armed forces. Military and civilian administrations respond quite differently to upsurges in illegitimate panicipation, prior budget12. It might ~ obj.tcted rhal the inclusion of military spending itself OOnt~m inales Ih( roral budg(r. Wh(n the analysis was run u~ing !h( total minus milifary upcnditura, the result. wer( very dose to !hose ginn above.
1~",1
Ie
a
~o
Tht The()ry ()f Survival CO/llit;ons
Table 22.
Budgetllry Austerity. Administrlltion Type. Il"d Expemliturc ResponJe Changr in lOlal SpcnJing Lnl Til"" Pudurssor
Mllu~ry Non~(Impcllt;.c Ci,,;I;~n Ci.;I;~n
N(ln.c:ompct;t;W Compcririvc
Til,... Prrrirussor
Milil... .,
M,/il,"y
S},,,. e
s}'"ft
Sh",~
1- 2
J- J
1- 2
.I
IJ
Mi/il"., Administranon Typ4'
Mor~
2
,
_1-
,
-"-
"
,
2 _J_
.
Mdilu.y Sh" ,c J -J
, _ 1"_'-
"
ary neglect, threa ts of coups, and fi scal austerity. These concepts b)' no means exhaust the causes of military spending, but they offer imporlant clues abou t the sources of military behavior.
Overall Patterns of Postcrisis Spending Our inqui ries in this chaple r have thus far been related quile directly to the theoretical center of the book-the survi\'al beh;)vior of executi\·es. Leaders. we have ass umed, are concerned primarily with keeping their positions. This limiting ass umption, that executive preferences can be equated to mere retention of office, is defensible during politic;)1 crises, but 10 assume tha t we can explain p O$U:r;$;$ policy with such an extrellll: simplification would be wrong. Implicit in the survival model is ;) broader view of the determinants of policy. In this broader view, public policies are consequcnces of the interactions of politically relevant actors, each with a ~t of preferred outcomes and each with a cenain degree of power or influence to promote these outcomes. Both dimensions, power and preferences, ;)ffect policy. When we speak of the "characteristics of administrations," we mean the institutional arrangements-the structures of po .....er-that magnify or shrink p0.litical influence. The nurnber of char;)cteristics with the potemial to affect policy, even ir policy is limited 10 government spending, is very large:. Simple typologies like ~military versus civilian control of the ex-
With Timt to Buatht
••
ecutive" actually subsume a multiplicity of widely va rying regime characteristics, including cohesiveness. degree of control over mass media. coerciveness. nature of societal cleavages. elector;)l system. level of citizen participation in politics, bure3ucr3tic expertise, and so on. Research based on cross-nation31 statistiC31 estimations typo icall y focuses on a much more restricted set of administration characteristics. Whether constructing new typologies or linking established typologies to outputs, cross-national studies avoid concepts not easily quantified and concentrate instead on three kinds of regime characteristics. In terms of the organization of civil society, one theme is the ba1:Jncr of vertical versus horizontal organization. the strength of challengers to the existing distribution of resources (especially unions), and the degree of ~ il1egitimate~ mobilization (especially the level of civil violence). A S('cond theme in cross-national research is the nature of the executive in terms of military involvement or the competitiveness of executive selection. A third theme is the balance of state versus society, especiall)' the dominance of one or the other or the functioning of inter· mediate institutions such ;)s legislatures or political parties (Rem· mer 1978 ). Implicit ly or explicitly, cross-national st:llistical analyses view regime char;)cteristics as constraints on leaders or, more precisely, as constraints on leaders' preferences. Me;)suring preferences, unfortunately, is very difficu lt. Even if we could c.:n egori ze a regime's public expressions of preferences as either survival or substa ntive, the collection of comparable information across many countries and periods is ;) daunting lask. Cross·n:ltional statis.jcal research usually strives to include large numbers of nations, but for many countries the monographic literature is quitc weak. As a result, models analyzing regime charactcristics ignore the question of leaders' preferences. Most social science models leave out significant aspects of reality, so the absence of explicit measures of prefcrencrs in policy models is hardly a unique failure. Regrettably, cross-nation:ll policy studies al most never offer enough information about specific cases to allow readers to tcst their own hypo theses. Me:lsures of preferences-surely a major determinant of outputs-are left OUl, but scholars interested in elaborating and testing such measures get little help. An extended discussion of preferences is a research project in itself, :l project beyond the scope of this chapter. Instead, in this
Til .. Th ..vry v( Survival Coalitions
section I identify clusters of administrations exhibiting similar ex· penditure patterns. Such dusters can guide future investigation by offering clues about executive preferences. The first cluster is based on the traditional notion of Latin American reformism. Our image of reformist adminisuations usu· ally includes an expansion of social services coupled with are· duction of military influence (Anderson 1967). Refornlist administrations S!Tess education and health spending at the expense of the socially wasteful armed forces. If these "social liberals" are defined as adminisuations giving highest priority to increases in education and health (in either order) and ranking military spending last or next to last, only eight executives meet the standard: de Ia Guardia (1956-1960), Panama Dutra (1946-1951), Brazil Lleras Camargo (1958-1962), Colombia Lopez Mateos (1958-1964), Mexico Luis Somoza (1956-1963), Nicaragua Sanchez Hernandez (1967-1972), Guatemala Torres (1970-1971 ), Bolivia Trejos (1966-1970), Costa Rica AdminiSlrations of this type are about as likely to be competitive as noncompetitive. Social liberal administrations are almost never military: Bolivia's Juan Jose Torres is the only military leader in the group, and he lasted less than one year. The social liberals also tend to come either from countries at low levels of social mobilization (Panama, Guatemala, Bolivia, Nicaragua) or from more advanced countries in an earlier period (Brazil in the late 1940s, Mexico and Colombia around 1960). II is striking, indeed, that many come from the early years of the Alliance for Progress. The scarcity of competitive governmentS in Ihis group results from the high cost of social liberalism. For noncompetitive socbl liberals the "economic Slrain" of government expenditures (measured by the ratio of the change in total spending 10 the change in the gross domestic product) actually declined-that is, their increase in expenditures was less than the increase in GOP. for competitive administrations, however, fiscal strain was part of the
"
,
,
With Ti_
10
Breathe
"
price of social liberalism- their total expenditures increased 2.4 times faster than the GDP. These governments felt obliged to cush· ion, by expanding the [Otal budget, the relative loss suffered by the armed forces. Social liberal governments rarely followed particularly liberal executives. None, for example, succeeded administrations in which education and health were ranked fim and second, and in five of the eight cases they followed military leaders who had expanded the military budget and whose nonmilitary expenditures emphasized large-scale public works. As Anderson (1967:289) notes: "The presence of one regime in some ways makes the achievements of its successor possible.'" Why are social liberals not found in more advanced countries? Note that these profilcs classify administrations according to thc changes they implemented in spending, not the levels of their budgetary shares. The: fastest growth in demand for social S<'rvices, and thus the fastest growth in spending, should occur during the emergence onto the political scene of middle-class and workingclass groups. This period of emergence has occurred more recently in countries likt" Guatemala and Bolivia, but it is long completed in Chile, Argentina, and Uruguay. In the latter countries other demands. especially thoS<' for job creation. have come [0 the fort" . At the same time, the post-World War II decline in military spending that is characteristic of advanced countries like Argentina, Chilt, Brazil, and Uruguay was dramatically re\'erscd with the coming to power of authoritarian regimes in the 1960$ and 1970s. Completely reversing the expenditure pattern of the social liberals, one group of administrations gave military spending first or second priority and accorded lowest rank to health and education. Total expenditures increased barely more than the increase in the GDI). These repressive leaders were very confident, and they felt no need to cushion losers by expanding the total budget. Their ranks include some of the most infamous regimes in recent Latin American history:u Sanzer (1971-1976). Bolivia Chaves (1949- 1954), Paraguay Laugerud (1974-1978), Guatemala I) . In .he C20f of AfIW.ina 's Onpnla, howevn, mili~ry ~ins w:os not dir«:t«l 50 milch al ........cuiun al ar .implc mridl~1 of rhe olfM:u corps.
The Theory of SurviV(J1 Coalitions
Medici (1969-1973), Brazil Ongania (1966-1969), Argentina Paz Estenssoro (1960-1964), Bolivia Perez Jimene1. (1952-1958), Venezuela Pinochet (1973-), Chile Ro jas l)inilla (1953-1957), Colombia Romero (1977-1 979), EI Salvador Torrijos (1968- 1978), Panama Videla (1976- 1981 I, Argentina Military administrations were somewhat mo re likely to fit [his pallern than civilian-noncompetitive administrations, but among fifly -two civilian-competitive administrations, none fell into this category. Only twO of the presidents were not actually military offi cers. Federico Chaves look power in I);uaguay during a period of intraparty feuding following the overthrow of dictator Higenio Morinigo. Chaves was never elected on a contested ballot, and Alfredo Slrot.""Ssnc:r finally pushed him Olll. Bolivia's Victor I'az ESIc:nssoro came into office for the first time in 1952 after an insurrection backed by a populist coalition of middle-class and working·class groups. When he was voted back into power in 1960, the c:conomy had collapsed and the U.S. government was supplying 30 to 40 percent of the Bolivian central gO\'ernment budgel (Malloy 1970). The price of U.S. aid was a rcoric:ntation of Paz's coalition away from the poor. That reorientation could only be consolidated by rebuilding the armed forces to carry Out thc:ir functi()n as protectors of ordcr (Mitchell 1977). Leaders in this group often justified their policies as necessary to comrol armed insurrections, but in faClthc:y were not especially likely to face violcm opposition. Laugcrud, Romero, Vidda, and Rojas Pinilla did come to power in the midst of guerrilla or ter' rorist activities, but T orrijos and Samer began their terms ..... ith litt le or no armed opposition. and Dngania hardly fa ced any at 31\. Even Medici .....3S very much in cont rol of Ihe weak armed opposirion th31 existed in Br3zil . Perh3ps this group's most suiking qualilY is ils diversilY of «0' nomic and social levels. Medici and Videla fall into the more ad· vanc('d bureaucratiC-3U1horilarian (SA) camp and Dngania represents a BA regime unable 10 esta blish ilself, but clearly none of
With Time to Breathe
the others are at that level. For this group personal substantive preferences (rather than economic or social factors) led to their particular combination of reaction and repression, but those preferences remain unexplained. A third group of administrations stressed health and education while also according the military a high priority. The group included: Arbenz (1951-1954), Guatemala Arias (1949-1951), Panama Balaguer (1966-1978), Dominican Republic Betancourt (1958-1963), Venezuela Diaz Ordaz (1964-1970), Mexico 1111a (1963-1966), Argentina Leoni (1963-1968), Venezuela Mendez Montenegro (1966- t 970), Guatemala Osorio (1950-1956), El Salvador Peralta (1963-1966), Guatemala Remon (1952-1955 ), Panama Roldos (1979-1981), Ecuador Valencia (1962-1966), Colombia Villeda Morales (1957-1963 ), Honduras These "defensive liberals" were mostly competitive or at least semicompetitive. Only one military ruler rearranged expendilUres in rhis pattern, and competitive administrations were about three times as likely to appear as noncompetitive ont!S.!· With the exceptions of l11ia of Argentina and Betancourt and Leoni of Venezuela, defensive liberals came from less-developed countries. Betancourt and JIIia were bOlh successors to military governments, and both faced serious military pacification problems. Leoni's army faced the task of coping with an armed rural guerrilla movement. A striking difference berween the social liberals (education and health first, defense last) and these defensive liberals lies in the fiscal strain each accepted. Social liberals spent at a rate three times 14. Thi, group waf defin(d by its ranking of military, tdUCI.!ion, and hulth in Ihe top Ihree, in any order.
1~",1
Ie
a
,6
The Tbw ry of Survival Coalitions
higher than their GOP gai n. Defensive liberals, on the other hand, increased spending little more than the incuase in the GDP. Was the support the defensive liberals gave to the military budget sim-
ply a way of ensuring that the military rn:limained its absolute level of expenditure receipts evcn when economic gro wth was
100
low to support an increase in the total budget ? Actually, just the opposite is true: During their administrations the defensive liberals enjoyed 30 aVl!rage increase in GOP mo re th an double the increase experienced by the social li berals_ In o ther words: Social liberals accepted a much greater inflationary risk when th ey increased so-
cial spending
al
the expense of the military.
OUf final expenditure cohon might be referred 10 as "builders." These administrations ranked public wo rks spending flrst or sc:c· ond and milit:uy spending fou nh or fifth . They included: Aleman (1 946- 1952), Mexico BalW.er (1975- 1980), Boliv ia Barrientos (1966- 1969), Bo livi:a Belaunde (1963- 1968 ), Peru Bllstamente (1945 -1 948 ), Peru C:astillo Arm:as (1954- 1957), Gu:atcm:al:a Figueres (1970- 1974), COSt:a Ri c:a Frei (1964- 1970), Chile Kubitschek (1956- 1960), Br:azil Lanusse (1971-1973), Argentina Gomez (1950-195 1), Colombia Ller:as Restrepo (1966- 1970), Colom b;:a Luc:as Garcia ( 1978- 1982), Guatem:ala Odria (195 0- 1956 ), Peru Orlich (1962- 1966), Costa Ric:a Qu:adros (1961 ), Brazil Ulau~ (1949- 1953), Costa Rica Vargas (1951- 1954), Brazil Velasco Ibarra (1968- 19 71), Ecuador Builders were equally li kely 10 be competitive or noncompetitive. civilian o r military. O nl y four of ninetee n rep resented countries
Wnh Time to Breathe
97
as poor as Bolivia or Guatemala. Levels of fiscal strain were moderate, with expenditure increases averaging a little less than twice GOP increases, but the average level of GDP growth was a very high 5.6 percent. These administrations were as likely to face a military pacifi· cation problem as social or defensive liberals, but they adopted policies penalizing the armed forces and favoring projects with payoffs further in the future. For some, the rewards from additional social spending might have begun to diminish. For others, accelerated economic growth required better infrastructure. The fact that GDP was increasing may well have been crucial. Executives expecting substantial resource inc teases can afford a longer time perspective. B«ause nOI every project has to produce an immediate payoff, public works and agriculture can enjoy budgetary largesse. Most imponam, total expenditures can grow, so losses in the shares of individual programs ate nOI inevitably transformed into losses in absolute spending. Conclusion The main argument of this book concerns a set of choices made by executives confronted with a specific problem. Among the many policy areas leaders affect, the book focuses on only one: public expenditures. Among the many problems leaders face, il considers just one: survival. That this perspective opens up imponant, even critical strategic dilemmas should be evident. This chapter went beyond the narrowest limitations of the survival perspective to show the usefulness of the approach in a wider but still cross-national framework. The inquiry centered on the consequences of allocations for avoiding coups and winning elections, on the differences in allocations between military and nonmilitary administrations, and on the existence of overall allocation patterns. The answers to these questions have no bearing on the rightness or wrongness of the survival theory itself. Instead, they are guides to using strategic approaches to answer some of the broader questions posed by Larin Americanists and political scientists in general. The contribution of spending to avoiding coups and winning elections is a logical step in the theoretical argument. Granted Ihat leaders spent the way they were expected to spend, did it help?
1~",1
Ie
a
,8
The Theory of Srlrl/i,'a/
OJ~ljlio" s
Spending on (he military did help ket:p the armed forces:1t bay, but leaders making it 10 the next election found the resources they could devote to social and econo mic progr3ffiS insufficient to improve their elecloral chances by very much. Mo((.'Over, the electoral process itsdf proved costly: AdminisH31ions facing midterm
dcctions were forced to increase expenditures iust to maimain popularity and legislative strength. The old questio n of the policy outputS of military governments was broached precisely because no one has resolved it adequately, although mOSt researchers deny the existence of a military go\'· crnmenllmilita ry spending rd:uionship. In this chapter. I used a muldp rogram ranking of expenditure changes to eval uate leaders' intentions. Using this ranking, the search for the factors determining civilian-milil3ry differences revealed Ih3t military 3dministrations do spend more on their armed forces. M ilitary governments spend maTt:, it turns out, because they respond differently to domestic violence, hudgetary neglect. threats of Coups. and fi scal austerity. Finally, we made a brief search for overall patterns of postcrisis spending. Explanation of these patterns proceeds no further here, because: the assembly and codification of systematic information abotll le3ders' policy preferences is a research project in itself. What we have, instead, is a restatement of the puzzle in the shape of a simple typology. Thc 3dminiSlTations grouped in this typology are import3nr as much for what they fil ii to explain ;l.S for thei r sucttsses. The repressive-reactionary group cannot be explained either by certain levels of socioeconomic development o r by the intensity of popubr mobiliZ3tion, a nd "builders" could be competitive, noncompetitive. military, or civilian. Our inability to explain these administradons takes us back to the importance of substantive preferences. How did these leaders come to adopt the patterns of spending revealed herd Th;u question is left 10 futu re rese3rchers. Does a dominant thellie emerge from this brief exrursion inlo postsurvival policy? If so, it is the fragility of pol itical power. latin American excrutives do survive their political crises, bill they rarel y attain the security and 3ulonomy of Ih~ir counterpans in industrialized nations. If an exerutive represents a ci\'ili3n-competilive regime, the chances of his party electing its successor are slim, and the possibility of im plementing a policy package signif-
With
Tim~
to
Br~atht
"
icandy improving those chances is equally poor. When competi· tively elected leaders face midterm elections, the cost of the p0litical process itself distorts budgelS and adds to inflationary pressures. If leaders wan! to increase sJX'nding on social progra ms, they must boost military spending as well. The only escape from this tup is budgetary expansion, but expansion merely pushes the trade-off into the future. Democratic political leadership in latin America faces nearl y unbeatable odds. Military regimes differ little fro m civilians du ring their initial surviva l crises, but once crises abate the policies of milita ry·led governments di verge radically. Overall budgets grow more slowly under military leadership, but the increase in sJX'nding on the armed forces is marked. Numerous factors trigger spending by the military on the military, including civil violence, the threat of a coup, and growth in resources. Military regimes seem to replace the politics of compromise and coalition with austeri ty, corporate scM·interest, and repression.
"
u
Part II: Two Applications of the Theory
Part II applies strategic coalition theory [0 Brazil. The rwo chapters do not test the theory developed in Part I. Rather, Part II extends the theory: examining aspects of expenditure policy that cannot be studied cross-nationally, analyzing the strategies of lower-level political actors, and evaluating policy areas unrelated to public spending. The ability of survival coalition theory to guide the detailed studies of Part II is a demonstration of the broad utility of the approach. Chapter 4, "The Congress Conm:clion," examines the role of the Brazilian Congress in budget making between 1947 and 1964. This period was the high point of pluralism in Brazil, a period in which the Congress exercised real authority over central government expenditures. The chapter focuses on the Congress rather than the president because the Congress was the center of an in" fluence structure involving state and national elections, local and regional elites, and linkages to the presidency and the ministers of the central government. In effect, a particular political structure, a congress full of politicians with their own survival strategies, affected the strategic options of the president. Though elected on a statewide basis, depUTies sought to attract funds to their local electoral bailiwicks. To do so they manipulated committees and arnendmellls in the style of classic pork barrd politiCS. The overall success of states (rather than individual deputies) in attracting central government largesse turns out to depend in part on the efforts of these individual legislators and in part on the organization of the legislative body, on intrastate political conflict, and on the socioeconomic bases of sute politics. Chapter 5, "When Soldiers Need Friends," treats the policy consequences of the transition to democracy of Brazil's post-19M military regime, a transition that took place during the 1974-1984 administrations of Generals Geisel and Figueiredo. The chapter
'"'
'"
Two Applications of the Theory
(xtends the notion of a "survival strategy" 10 consider a strategic option not contingent on maintaining office. In this case, the main political 3etor, the military, sought to build suppon so that it could remain influential after the end of formal military rule. It was not the survival of incumbency, but of influ(m;c, thai the generals sought, and they wanted to construct a base of support th:n would make them indispensable to the survival of fu ture incumbents. After suggesting a series of strategies the regime could have followed, I 3ttempt to determine what the Geisel and Figueiredo :administrations :tcru:llly did. Because the expendilUres of the ceowII ministries arc only a small pan of Slate activity, I consider four policy a reas in addition to expenditure trade-offs: agricultural programs in the Nonheast. low-cost housing, industrial dewncemr:Hion, and wage and sabry setting. In the end. the regime fared poorly. Two factors proved fundamental in impeding its effo rls at building support : the accc:1er:Ilion of Ihe econonlic decline and the fragmentation of the structure of policy making caused by the su rvival strategies of lesser political actors.
4. The Congress Connection: Politics and Expenditures in Brazil's Competitive Period
Elires are nOI the only political aelOTS who care about survi\'al. Wherever parliaments have influence independent of executives, IcgislalOrs pursue their own survival strategies. When legislators represent distinct 5ubn3tional entities such as states or local communilies, they strive to increase pork barrel programs, because such programs allow them 10 claim credit for the benefits they bring to theiT constituents. Executive a nd legislative survival strategies operate in different ways. Though executives sometimes adopt loealist stralegies, they use such approaches to rew:ard some localities and penalize OIhcrs.
Legislative localism, by comrast, stresses mutual cooper,uion and the sharing of rewa rds among as many legislators as possible. It is no surprise, then, that presidents a nd legislatures are often :I t odds. The decline of Chile's Congress in the 1960s, for example, has been indirectly linked to the overthrow of President Sa lvador Allende (Valenzuela and Wilde 1979), and Ecuador's Congress played a central role in the overthrow of a democratically elected president in that nation (Pyne 1973 ). Legislatures are also a frequent object o( criticism launched by technocraticalJ y oriented civilian planners. Their allacks frequently center on the very stuff of legislative politics: bargaining, trading, logrolling- in sum, the politics of pork. But for legislatures, they argue, budgets could be implemented in acco rdance with national priorities and resources rather than local needs and the reelection strategies of incumbent legislato rs. Roberto Campos. a civilian architect of Brazil's mil itary regime, made the point succinctly: Because of ils hypu-sensi riviry to regional pressures capable of destroying th e coherence and balance of plans and pragums, Ihe Congress
'0,
Two Applications of tht Thtory
had lu:en tran§formed into an Uinflation machine" multiplying the budget and inlO a "distonion factor" vis-a-vis investments (Mendes 1975:36). Is Campos right? Do legislatures inevitably increase and redistribute spending to suit electoral needs? Many schola rs agree. James SCOtt, for example, suggests that elections in Southeast Asia shift the balance of exchange in ,he direl.'tion of clients. Regimes under pressure to engage in distribution "resort 10 budgel deficils, especially in election years, to finance their networks of adherents" (Scott 1972:1 13). Raymond Hopkins (1979) reports that in Kenya and Tanzania congressmen used central government funding to boost their electoral chances. On the other side, Arturo Valenzuela and Alexander Wilde (1979) agree that Chilean legislators logrolled on behalf (If local interesls, hut Ihey argue that legislative politics had the beneficial effect of reducing conflict and Ihat only a small percentage of the budget was diverted to truly wasteful projects. For which legislators is the pursuit of pork-of geographically separablt:· public spending-a survival slr:lTegy? Unde r what conditions?' In developing countries we still have very few ;answers to these questions. The biggest obSl.1de to understanding legislative survival strategies is the absence of dala. Rarely is in{ormllion about both legislative budgets and final expenditures avai lable over a period of years, and even more rarely can final expenditures be linked to specific constituencies. This chapter explores legislative strategies affecting the distri· bution of central government expenditures 10 Brazilian states between 1947 and 1964. In these years the Brazilian Congress was an active participant in the budgelary proccss. ~ The chapler treats t. A promi,ing SUr! in an 'w~ring these questions h~s ~n m~d~ in rese~rch on South Kor~a. Kim and Woo (1 975 ) found that con.titu~ncy scrvicr in the Ko· rean a~sembly was ,dated to a variety nf f3ctors. eenai" panies 'Hr. more ( On· 51itu~ncy·ori~l1Ied than olhe .. ; rural assemblymen were more active than their ur· ban rounterpar!$; Ye!er3n~ we.. les. active thon junior mrmbf. ..; and memb.. .. from single-m~mber district. wete mOre aClive than thosc eleCied at large. Although Kim and Woo\ concept of (on~titurncy service ududed budgotary appropriation. (which the legi.larurc could nOt modify ), thei, findings could extend !O a broader notion of logi.latiH pork barrel. 2. Brazilianim haye nOt liy~t.mati ea ll y treated thi s top'c, but th~re ha~ been considerable writing about otMr a.p<"C1. of the COngr ..... For .,say' on ~,,,h topics a. p3rty alignments, voting me<:hani.m •• so<;iai background. of deputies, budgetary incrementalism, recruitment, and styl.. of rep"",ntation, $tt Santo, (1979 ), Car.
,
The Congreu Connection
the Congress as the focal point-though in no sense the sole detenninant-of an influence process involving state and national elections, local and regional elites, and linkages 10 the presidency and the ministries of the central government. The strategic activities of congressmen tum out to be consequences of the decisionmaking structure of the legislative body, intrastate political conflict, the socioeconomic bases of state politics. and the individual motivations of legislators to seek geographically targetable public goods. The discussion is divided into four parts. The first section begins with a brief introduction to the formal and infonnal structure of pre-1964 Brazilian politics. I then examine movements of selected expenditures in order to demonstrate that considerable change occurred over the course of the pluralist epoch. The next section develops a model of legislative survival strategies affecting the expenditure process. The following section evaluates the hypotheses generated by the model and explores a number of deviant cascs. The chapter concludes with a brief comparison of the pluralist years to the authoritarian period that began in 1964. Appendix B at the end of the book discusses some problems in evaluating the model developed here.
The Development of Competitive Politics Political Background
Getulio Vargas assumed the presidency of Brazil in 1930, victor in a revolution that marked the beginning of the modem Brazilian stale (Fausto 1970). That revolution had ended the Old Republic, a regime dominated by coffee planters and the governors of economically powerful Sao Paulo. Minas Gerais, and Rio Grande do SuI. The origins of the revolution of 1930 remain historically controversial, but the struggle clearly led to a more centralized and interventionist government, a government that could break the hold of the planters and ad mit the military and middle class into political life. v~ lho (1973), F1~j scher ~nd Packenham (1971).
( 1976, 19n. 1981).
ljm~
Junior (l9n). Nunn (1978),
1~",1
Ie
a
'06
Two Applicat;o"5 of the Theory
Vargas gradually consolidated presidential authority until, in 1937, he created the Estado Novo, the New State: ,enualized, amiparty, repressive, and nationalistic. Free from the restraints of liberal constitutionalism, the government began to innovate, expanding welfare policies, foslcring and comrolling unions thro ugh its new Ministry of Labor, and investing di rectly in steel, iron ore, and riveT valley development. Although the tcnt3tive industrialization that followed was partly a consequence of the collapse of Brazil's import capacity in the world depression and World War II, the government's self-transformation had enabled it 10 become an active participant in economic managemem. Centralization during the Estado Novo also affected the political process (Souza 1976)_ Vargas dosed the Congress and replaced most governors with powerful federal appoi ntees. The ne w Administrative Department of the Public Service (DASP) acted as a supenninistry at the federal level, and little DASPs served the same function in the states. To circumvent the traditional bureaucracy, Vargas and his advisers created new auta rchic agencies a nd public enterprises. Increases in the power of tht' armed forces (at the expense of the once largely a utonomous state mil itias) expanded the military's role in national po licy ma king to include such issues as industrial dt'velopment. At the conclusion of World War II, pressure to liberalize the regime became irresistible. In a bloodless coup, the military removed Vargas. A constituent assembl y soon produced a ne w con· stitution, a document that establ ished the fo rmal structure governing Brazilian political competition until the military coup of 1964. The Constitution of 1946 created a presidential system with a bicameral legislature. Three senators, elected to stagge red eightyear terms in majority elections, represented each state. In the Chamber of Deputies each state was entitled to seats in proportion to its population except for two deviations: No state had fewer tha n seven seats, and above twenty seats t'ach new seal required 250,000 electors rather than 150,000. The rules thus penalized large industrial states, especially Sao Paulo, in favor of small, rural, and poor states. J Deputies were elected to concurrent fourJ. Con.. rvative rcprcsmt:nive. to the ron,rituent ~ .oe mbly [rom induotrial >fates accepted thi s arrangement, b«:.uoe li miting the influence of !h( urb an work·
,
The COffgTeSS Connection
"7
yur terms th rough a system of proporlional represcn!ation. Whole states constituted districts, and electors (literate adults) selected one candidate from a party list. After Ihe balloting, votes for all candidates on each party's list were added together, and the frac· tion of the total vote represented by each party's share determined the number of seats allocated to that party. The number of votes each candidale had received determined Ihe actual holders of the sealS. Gellllio Vargas's influence over Brazilian politics did not disappear with his ouster in 1945. At the closc of the Estado Novo, Vargas and his followers organized two political parties. The Social Democratic Party (PSD) rooted itself in the state political machines Vargas had fostered. Though influential everywhere, the PSD was particularly Slrong in rural areas and poorer Slates. The Brazilian Labor Party (PTB), by contrast, was organized by Varguistas in the Labor Ministry and Ihe unions. The PTB's strength was concentrated in big cities (though not in Sao Paulo) and in Vargas's home statc of Rio Grande do SuI. As time passed. the PTB also became home to a diverse collection of dissidenls from OIher panies. The major opposition to Ihe Varguisras came from Ihe Nalional Democratic Union (UDN). The UDN combined losers in the intra· oligarchic power snuggles of Ihe Estado Novo, constitutionalists who opposed the dictatorship on principle, and upper-middle-class city dwellers repelled as much by the populism of the PTB as the conservalism of the PSD (Benevides 198J ). Though laissez-faire constiluted the mosl important currenl of UON economic Ihink· ing, its raison d'ctre was opposition to Vargas. The UDN's hopes thai the diclator's fall would lead to its own ascension to power were dashed at the end of 1945, when the join! candidate of Ihe PSD and PTB, General Eurico Dutra, was elected presidenl. The PSO received 42 perctnt of tht I;ongressional vote and took 52. per«nl of the scats in the Chamber of Deputies. Tht UON came in sccond. with 2.9 perunt of the seats, and Ihe ITB fini shed a distant Ihird with less than 8 ptrl;ent. The Duna administration was cautious in its economk policy. By 1947 Brazil's hard OIrrency reserves from World War II had in, cluJ w;u "'OIt j"'pOIunr rhan rhe ;n rUtSB of rheir own sutn. (!itt ScNu ]976, Soares ]\173, and Kinm 19110.1
'0'
Two Applications of tfft Tht!ory
disappeared. and huge deficits appeared on the cu rrent accoum . Rather than devalue the cu rrency. the government adopted rigid import controls. Import substitution in consumer durables began, but the government's share of nalional income stagnated and blockages in infrastructure became evident. The eltttion of 1950 returned Cerulio Varg~lS 10 office. The
PSD had refused
10
go along with the PTB's nomination of Varg3s,
but many PSD supporlers backtd the former dictator anyway, and Vargas won an absolute m3jority of the popul3r vOle. In the Chamber of Deputies the PSD lost its majority, never ag3in surpassing 37 percelll. The PTB doubled its scat share (10 abou t 17 percent), and sma ll parties became a major force. Vargas's administriuion was both activist and limited. Public investment in energy and transpo n .uion increased, the National uonom ic Development Bank (BN DE) became a major source of investment capital, and the government controlle-d exchange r:lles to help rat ion imp""i. The government was Imahle. however, to increase taxes, and it met infb.tionary pressures in 195 4 and 1955 with another try at contraction. BUI comraClion and austerit), were difficult to accept in the middle 1950s, because the vigorous in· dustrial sector was quite import-dependent and Brazilian nationalism (encouraged by the goverllment and fueled by Vargas's suicide in 1954) had grown enormously." In 1955 the PSD and PTB rebuilt its fo rmer alliancc, electing Juscelino Kubitschek to the presidency. Kubitschek's ~ TargetI'ro gram" was Brazil's first thoroughly developmenralist strategy (Lessa 1975). Emph:lsizing tl'1lnsport, energy, basic industry, :lnd the construction of the city of Brasili:l, Kubitschek paid less attention to price stability, the b:d.mce of p:lymentS, :lgricultuu, and income distribution. Though the resources fo r the program had to come from the government and foreign capital, domestic industrialists wert supportive. because the government was cre:lting new areas of growth and leavi ng imaCI the prerogati ves of domestic capital. Initially Kubitschek's strategy wo rked. 8etwctn 1957 and \96 1 domestic production climbed 7.0 percent ann ua1Jy, and key sectors 4. At his dc~th, V~rps kit a ktttr anad,onIJ. ror~i", nlt~'l'riH$ for m~kon, exorbitant profits ~nd fM blockUl, me plYemm~nt 's dforrs '0 ~,d workers and takr a biWr rolt in the ecooomy. The len~' unlushnl a way~ nf pub!i, symp.rhy
for Va r&-,s and
~trrngthr~
narioR3lisl fttling. Stt Skidmorr
(I~67].
Tht ColIgrtSJ
COlIlIUriOIl
"9
like transporl3tion advanced even faster. Brazil began the sixties with a more mature economy, but the financing of the developmem program inevitably encouraged a hyperinflation. Agricultural production per person had barely increased, regional disparities had worsened, ;;and employment had grown less than the increase in the work force. A political crisis paralleled the developing economic crisis. Rapid industrialization and urbanization had broken the dominance of the conservative parties, especially the PSD, but malapportionment and electoral instability prevented the emergence of a stable legislative realignment. Increasingly, deputies were elected by multiparty coalitions whose composition va ried from state to state. These coalitions, moreover, were rarely based on enduring ideological differences, so they changed from election to election. The electoral rules also encouraged small parties to ally with larger ones as a way of guaranteeing themsdves at least one legislative seat. Along with the overrep r~enta tion of traditional regions like the Northeast, this combination of electoral instability and the diminished weight of the thrtt major parties made the Congress consistemly more conservative than the president and produced a legislative deadlock.' While President J030 Goulart (first elK ted vice president in 1960) was drihing lehward in the face of pressure from populist groups and challenges from conservative economic and military interests, Congress was in a state of disorder and paralysis. The legislative process came to a halt as both the number of bills introduced and the number passed shrank. The rcsult was political and ad ministrative chaos-a chaos that contributed, in no small measure, to the overthrow of the competitive regime at the end of March 1964. elmlral Government E.xfnnditures
The Brazilian Congress had constitutional authority over thr national budget, but legal control over a budget docs not translate automatically into effective control. Depurirs might not sec budgets as part of their survival stratcgits, and informal pattems of influence outside the Congress could negate their efforts. But public expenditures were strategically important for Brazilian congressmen, and though some important political actors bypassed
s. 5« ~ru (19'-4, 1971, 191)), Sdlwamnu.n (1970), and Souu (1976).
of rhe Theory
Two ApplicationJ
.. 0
0 .35
0 .30
f•
0.25
i,
0 .20
J ~
D. I S
~
,;
~
0 .10
,,
.
\/\
'/
I 'I I"
, \.
f
,.
'-_'
_ ._ '
,
"
, \\ ,, ,
"
-
\ r
"",
/
'
-""" \ . / ,.,."\
~
0 .05
,,
, ',',,
,I',
,
\ .\ .I
... ................. .. ~...~...••.
..~..
v
i
,_ .1.
..........................~ ... ~
M i litary
Public Works
EducaHon and Hoallh AlI'icullure
Figure 7. Program Shares: 1945- 1965
the legislative branch to deal directly with the executive, most budget-related political activity did involve the Congress. We begin our examination of congressional behavior with some simple expenditure patterns. Figure 7 shows the shares of final expenditures devoted 10 key programs between 1945 and 1965. The military was the chief loser, fallin g from JS percent of expenditures in 1947 to just 14 percent in 1964." Public works pro6. Th~ military .hur i .....fincd a. tho .urn of all t"'" military rnin lS{n(S and other mililary progurns divickd by I"'" IOI~' 01 all «nlnl govrmmcnl Jpmding.
The Cnngfl'5S Connl'ction
'" _t.4
"\
""
"
"
0.50
0.75 ~aui
Ma ranhao
Alagoaa
-
/, 1.10
"
1. 6
"
"-0.90
1.25 Rio Grande
SerQipe
"
Ceara
do Nor1e
NOle: Numerical entries represent the ratios of the percentage of all spending received by the sta te 10 its percenlage of Brazil's POputation.
Figure 8. Central Government Spending on Nonheastern Stales: 1948_1966
grams stagnated during Dutra's administration but prospered during the terms of Vargas and Kubitschek. Education and health programs, which aided increasingly powerful middle-class and working-class voters, grew steadily until the military takeover. The programs enjoying the sirongesl growlh are those often
1~",1
Ie
a
Tu.v AppUallions of lbe Theory
called "pork barrel"-that is, programs targetable to specific regions or localities. Since the severe malapportionment of the Chamber of Deputies fav ored less-populated regions, small Northeastern states should have received more than their share of such programs_ Overall, the small Stales of the Northeast did indeed improve their position. But if we examine these Slates separately, as in Figure 8, their budgeta ry reco rd seems quite mixed. Four states clearly gained, twO showed no trend, 3nd one aClu3l1y declined. Figure 8 presents a pU7.1.le. What explains the diverse movements of expenditures into states and progra m s~ We know, of course, that pork barrel politics is legislative politics par excelience? Did the legislative process produce the patterns in Figure 8? And if so, how? Congressional Influence in the Expenditure Process Why Legislators Wallt Pork
Legislators care about reelection . In the hope that voters will reward them, deputies suppon projects aiding their electoral bases. To ensure that their pet projeers will pass the legislalUTC, deputies seek aUies who wi!J join coali(ions backing their pro jectS. This search for aIlies leads to the time-honored practice of logrolling, each member supponing the projects of others in relUrn for their suppon on his. The division of the total budget shifts to fa\'or programs with more pork potential. Funds for hospitals, schools, :md roads expand. Programs that cannot easily be subdivided or th;1t barely affect individ ual districts (such as foreign relations or military wages) suffer.' Unless legislators can claim credit for new government projects placed in their districts, they have no incentive to seck such projects. As a result, legislative survival strategies are affected by 1_ St1: Lowi (1964). The seognpki... lly s.eparabl~ compon ~nt of miliruy 'IKn.!. inS docs not .s«m to be as import.>nt for tke miliury ;n lati n Ameri ... as ;1 i. in Ikc United State$, probably ~ u.., tken: i. much leu local procurement in utin Ameri .... 8. The belr theorerial statement of this position is found in She~le and W~jn Sur (198\).
n, c1tttoral laws. If ~ach legislative distria electS one representative, crroil is unambiguous, but representation in Brazil was proportional, with whole sr:nes serving as multimember electoral distrieu. Did incenrives for the pursuit of pork exist nonetheless? They did. for two reasons. Voters cast ballolS for individual candidates, not for party lislS. so voters could reward individual legislators. Moreover, rn3ny states developed a system of informal districlS akin to the "bailiwicks" of Ireland.' Candidates coneentrau:d their campaign efforts in these: districts and dn!w most of their VOltS from them. So ( ven though Brazil used a system of proportional representation, legislalOrs seeking reelection should have attempted to influence the programmatic a nd geographic distribution of public expenditures. Deputies seeking more cennal government spending in their states had a variety of options. to They could ame nd budget proposals with projects benefiting their bailiwicks, they could seek membership on committees dealing with budgets, and they could try to join the leadership of the Chambers or the party directorates. If all deputies' survival motivations were equally intense, no matter which Slates they represented, the malapponionment of the Brazilian Congress would benefit the poor, rural states of the Northeast. This was not the case. NO[ all Northeastern states were winners, and some SlateS gained much more than others. What else counted, then: What conditions m3de some members of Congress more eager and mo re successful in the pursuit of targetable budgetary goods? Inlecurity. If deputies cxp«e to seek future terms in office, the insecurity of their current $COltS stimulates their sc3rch (or district· specific benefits. In Brazil's system of proponion31 representation. deputies' c1eaor31 fortunes were determined by their p3rty list ra nking after the balloting. Deputies with low ranks or bare mar9. On Imand _ Sack. (1976). for evidentt of ba,liwick$ in Srnd _ Fkischn (1976. 19m. 10. Aftu proposed budgeTs were .ubmi"~ by [he praidrnt ,n 1M ~d quarrn- of uch yel<. hurinp ";en: Mid b,. propam·sp«ili, iuix.-omminttS of the Budget Comm,n= of the Chamber and xnate. Presidenti al budgetS coul d be amended in rommitt« and on The floors of bOIl. cham""". Budge" were origin3liy unde. the control of tlK ~'n .na Commin«. bUT in the u.ly 19505 a Budget Commi= wu auted. I ihali uwally n:fer to the Budt;et Committtt. but In budt;et5 m.de befon: its c.urion [his is [he f in.nce Co mmtn«.
".
Two Application$ of tht Theory
gins of victo ry wert: uncertain of seats in {ulUre legislative sessions. Visible a nlTal government projects could help them rise in thc party list 3t the next declion. As deputies accumulated morc seniority, however, their security increased, 50 the longer legislators held office, thc lower were their motivations to seek pork.
Party Ethos. The National Democratic Union (VDN) was a partial exception 10 the undisciplined, clientd islic. and d ecentralized nature of Brazilian po litical partics. As Benevides (1911 1: 172) points OU! , the r.1isons d 'ctTe of the VON illd uded fi ghting President Vargas, denou ncing adminiSlr31ive corruption, and o pposing go ve rnment intcrvcmion in the economy. UdeniSllS emphasized ordrr, austerity, and the importance of a national rather than a local vision. Vdenista-held ministries tended to be in Foreign Re· lations, Jusri(e, and the Treasury-the classic "collective goods~ ministries. M any VDN deputies ignored the pieties of party leaders, of course. but to the extent that VDN legislators sh ared the party ethos, they would decrease their usc of budget·inereasing acti vities as survival strategies and increase their visibility as "Guardians of the Treas ury" (Fenno 1966). Bailiwicks. Though deputies in each state weTe eJected at large, they often limited thei r campa iglls to clearly delimited zones of politic;)l influence and elector;)l support (Fleischer \976). The effect s of such redutos ("strongholds") on legislative behavior de· pended on two (onsiderations: the number of candidat es from each deputy's own party who emen'd his electoral zone and thc number of candidates from other parties who entered. If no opponents at all entered, the depmy obviously had a safe scat and no SUNiv3l problems. If a few candidates from the dep uty's own party entered, their VOtes would push the deput)· down in Ihe pa rty lisl. reducing his chances of d cction, bill these vOles would add ro the party's total, thereby increasing the deputy's chances. As the number of competitors from the depu ty's own parry grew, they would take VOles from each other, and the likelihood of changes in rank would dc<:reasc, The deputy would benefit. and his budgetary incentives would dc<:rease. Candidates from other parties posed a greater threat. Votes going to these candidates pushed the I.:andid:u e do ..... n in his own party lisl, so the deputy wo uld be less likely to be 3 winne r. BUI
T~
CoIIgrtu Connection
.. s
candidates from OIher panics also cut into the votes going to any candidate of the depUly's own party, 50 fewer total Rats would be allotted to members of his party. We can expect competition from other parties, therefore, to srimul:ue deputies to ~ek projects for which they can claim credit. Why States Get Mort Por.
Insecurity, party ethos, and bailiwicks all affected the budgetary strategies of individual deputies. The consequences of these individual survival strategies for allocations to states in toto arc the subject of our second model. This model includes five delegationwide characteristics: comminee representation, stability, leadership, ministeri31 linkages, 3nd occupational skills. Committee Representation. How was comminee membership dctennined? Seats on com minces were distributed proportionately, first by party and then by State. There were too few members to enable each pany and each state to have just the right number of members on each committee. Since party proponionality was the dominant criterion, states might be over- or underrepresented by as many as nine or ten members on 3 large committee. Indt.'Cd, the median misrepresentation per state OYer the whole competitive period was such that a typical state might have 15 percent too many or 15 percent 100 few members on a comminee. "Extra" seats were importlnt 10 the deputies, because seats increased stale innuence over l comminee's subsll mive business. In both the Senate lnd the Chamber. the Budget Commiuees were the key commiurts affecting expenditure legislation. Aft!;:r the executive scm a budget proposal to the Congress, the !eadership of the Budget Comminee divided the proposal inlo programs and picked individual committee melllbers fO be relators, that is, "na rr:uors,~ for clch program. Relators conduCted hearings, gave opinions on amendments. wrote substitutes subsuming the individual amendments, and defended Ihe comminee bill on the floor. A negltive committee opinion usually killed an amendment. and l fl\'orab le opinion almost always guaranteed plSsage. 11 Sinct' extra members on a kt'y committee would help states
II.
II . Vonl oppmiuon HI the domin~ nct of the II-uJj;tt Commutft' was common, "p«Ially dunng WGuardian of ,he Tru.ury " phaK in 1M Dulra adm;ni.lrarion. Sec.. fQt tnmpl~. ,he 'JX'C'('hH and monOM o( lR"...ry J".andor Pirn, o;Jrio do
.,
Two AppliClltions of th~ Theory
in th~ business of that comminec, we c:m expect ovcrrcprcsemation on the Budget Comminet to increase spending in a state.
Delegation Turnover. Though the legislature began in 1947 as a totally new body, seniority soon b«ame a prerequisite for influencc. States with unstable internal politics suffered higher turnover and thus had fewer senior members. Such delegations were less likely to produce leaders of the Chambers or party
dir~orates
and less likely to hold SUIS on key comminees. AI the s:ame lime, deltgalio ns with high flIrnover were more likdy to include eleetotall y insecure members. and insecurity, as we have secn, incu ascs the motivations of individual deputies to pursue survival strategies. Thus delq;;uions with high turnover were simuhancously more eager and less able 10 ali raCl funds 10 Iheir sl:ues. As ~ resull, we cannot prcdia the dfeCls of dd egation SlabililY·
Each branch o f Brazil's le:gi51:lI ure: had IWO kinds o f le~de rs: party leaders ::and Ch::ambtr leade rs ("'('$a ).!! Party leaders organized positions on issues, arranged quorums on lIotes, and so on. B«ause the party direclorates Iried 10 apponion leadership Sl0l5 regionally, r~l hcr junior depulies could as pire fO them . These party direclOrates. however, we:re: somewhal less prestigious Ihan the Chamber leaderships. because the latter conrrolled comminee assignments and Ihe legislarive ca.1endar. Leaders were we-II posilioned 10 garner bigge-r shares of gOIle:rnmenl programs, so we can expcci slate:s whose membe:TS panicipate:d in these twO leadership groups 10 recei\·e more fede:ral largesse. Corrgressio lla{ Leadership.
Vertical Lillkugc$. Though Ihe legislalUre was fhe pivot of hudgel making, il opcr::n ed in : I. rich conrext of miniSlries, stale governors. ::I.nd presidents. Linkages worked in both directions:: Pafties and Slates made claims on ministries, and ,he: president involved himCo ~Ii.tSJO ~
NaClonal. S/:plcmbt-t N , 1948. The !:u .. dj ~1\ .oIf IS
m~<:h.
suonge.
In
'yJlem toke Ih>l .. f lh" Un nnl SIa l.... Sn- F"nno ( 1966) ~nd I'.....,n~n (1980) fo.
p~'Jl)d
Ilea ll"e"" . 12. f'ury lude .. incluJ.cd. for nch. pl ' l)". I ~Iu
Thl! Congrl!$s Connl!ction
'"
self in the selection of Senate and Chamber leaders.l l Ties between governors and presidents increased state bargaining power, and a successful electoral campaign in a state might encourage a president to look with favor upon its requests. Executive survival coalitions often included ministers from politically important states. These states gave suppOrt to the executive and gained influence over programs. A state's lobbying would be facilitated if its residents filled ministerial positions. Skills. In any legislature, state delegations lobby for the interests of their states. Because good lobbyists need to be familiar with the details of programs, Brazilian congressmen usually gravitated to legislative activities related to the skills and expertise they brought into the Chambers. Military officers involved themselves in military programs; engineers worked on big construction projects; doctors dealt with health problems. A congressman should benefit when the fit between his skills and the committee's business was close. Overall, governmental activities encompass many areas of expertise. The more limited the range of occupations in a state's delegation, the less often will its members find a committee matching their own specialties. If a delegation included members with a variety of professions, the overall match between programs and skills would be closer, and the delegation could place an expert near more programs. Thus delegations with more varied professional backgrounds were bettcr positioned to attract governmcnt spending.
Analysis In this section the hypotheses developed above are tested empirically. The analysis begins by considering the budget strategies of individual deputies. Next, I assess the budgetary success of whole states. The analysis then shifts from a cross-sectional to a dynamic mode, comparing congressional to presidential survival strategies over the whole pluralist period. Finally, I consider a select group of states individually, facilitating the examination of such factors as intrastate political conflict. 14 13. ~ Cart Filho (l966), Benevides (1976), and Cintr.> (1979). 14. A brief discussion of problems in operationalizing the hrpoth~1 developed in this section is presented in Appendix C at the end of the book.
""
1~",1
Ie l.Ia
,,'
Two ApplicdlionJ of th e Throry
Individllol Strategies
Budget Co mmiuee Membership. Without exception, deputies agreed which comminces in the Chamber werc the most sought afterY Constitution and Justice was crucial because it had the power 10 obstruct key legislatio n. Finance-later divided into IWO committees (one on finan ce (and a second specifically on the budgct}-was im portam because it handled central government expenditures. An individual deputy's chances of getting a seal o n a choice committee would improve if he had done well in the previous dection. State party lists were often headed by some popular VOIegener who might well garner 50 percent of the IOtal rcreived by all the party's candidates. Because the VO tes of such deputies increa~d the pafty total and facilitated the election of less popular depUlies, they were powerfullegislalOrs. In 1948, for example, the twenty-one members of the Chamber Fimmee Commiu ee included six who had finished first on th r' r lists. Since Ihe average number of members elected per party was abolll nine, the twenty-oneperson commillee had more than double the number of first-place finishers it would have received on a random basis.,e In 1952 the commillee boasted twelve fim-place finishers. once again double the number expectcd. The succe5S of fim-place finishers in gelling on the Finance Commillce does not prove that electoral success always dominated insecurity and need in predicting commiuee membership. In only one session (195 1- 1955) did the budget-making commillce's ollerall composition-rather than merely its tOp vote-geliers-reOect electoral success. In Ihal session twemy-seven of the Ihirty-eight members of the Financc Comminee fi nished in Ihe top half of their
lists. The reason why top vOle-getters-the 513te hcavyweightschose budget-making commiltees in Ihe 195 1- 1955 legislarure lies in thc:: change of presickntial administra tions. During the fiscally consc::rvalive administration of General Dutra, the Finance Commin ce functioned as 3 budget3ry watchdog preventing additions IS. lnlffVi~Wi conducted in 1\183. 16. 11 .~nk in 1i11 Iud no df«t on coffunin« phcemem. 11 j)(K'm1 of 1M co""nin« would have b«n oompoKd oi li"" pl~~ nnishc.., II p~'
The Conguu Connu tion
'"
to the president's budget. During tht! populist.devd opmenralisr administration of Dutra's sucCt!Ssor Gt!tulio Vargas, tht! Finance Commintt 100000n«l its purst! strings. With hudgcury opponunitit!S tOO good to pass up, the dt!ctorally powerful exercised their prerogativt!S and took scats on key commintts. Vargas's suicide in 1954 was succeed«l by twO years of austerity budgets submin ed by provisional prt!Sidents. 17 Opportu nitit!S for pork were paltry, and the Budget Committee's new members came as often from the bottom half of tht!ir pany lists as the tOp half. Tht! Kubitschek and Goulart presidencies brought a renewed budgetary expansion, and after the congressional dection of 1958 new members once again came from the upper half of pany lists. Why did deputit!S want these Budget Comminee scats? For top \·ote-geners electoral anxiety was hardly a strong incentive. In· stead, they treated the Budget Committee as a source of long-term political capital. Passing out budgetary largesse to colleagues was an important way of amassing influence, both in their own states and nationally. Lesser vote-getters, by contrast, nl.'eded [0 assure their own reelection. They sought Budget scats because a position on the committee would increase their political resources and im· prove their immediate chances for survival. Deputies sought or avoided scats on the Budget Committee for three reasons: party ethos, vote concentration, and competitive pressure. As we saw in the first section, the National Democratic Union (UDN) had a streak of antipopulism and fiscal conservatism. This orientation inclined Udenistas away from the budget committttS. When deputies were asked (in the mail survey) which of their committees was most important to them, Udenistas were half as likdy as members of other parties to name the Budget Committee, and they were much more likd y to seck scalS on [he Constitution and Junice Committee. When ask«l to rate the importance of " helping local areas" in committee selcttion, UDN members were slightly more likely to dOWPIgrade local benefits. Nonetheless, Udenistas did serve on the Budget Committtt, because party shares of committee scats correspond«l closely with pany shares of total Chamber scats. Why, with their aversion 10 pork, did UDN members accept Budget Committee spots? When 17. Loal ludm in Rio (Orand.: do Notu: tc$rified m31 Proicknt Cafi Fitho particularly dipppoinltd his follow~. who up«mlloa of money.
Two AppliCillions of the Theory
t1.o
Table 23. Campllign Conuntrlltion lind Key Committee Art'a of C2 mpaign
K..y Commin""
Rudget finance Oth~r
No Regionll/
Conctnl"lIion in Contiguous
Conctm,"/ion in G,oup of Sulle,ed
Conunrr3tjon
Alunjc;p~/jrje.
Munjdp;>/jljts
11.1 % 1 L! % 77.8% 1000/. (27)
20.6% 15.6% 63.8% 100% (34)
20.0% 26. 7% 53.3% 100% (30)
Not" Thi, ubi. i, • , .. ult of fWD qu .. rion" "When you ran lor , ...1 in lhe o..mbcr, wnne did 1"" ~., .. p.>"
cone."".,. your
"''''p.i",?" and
"~
..... wh .... you co""'"""..! .h.
of your ",mp.ip ..,.r•. .. ." TIre "umbe,.. in p.,.mh ........ ,he ""mbe, o'
=pondcnts.
asked which committee members were particularly associated with the role of "Oefend .. rs of Order" (a Brazilian euphemism for fiscal conservatism), a long"time Budget Comminee staffer responded with six names: Adauto Cardoso, Pedro Aleixo, Aliomar Baleeiro, Herbert Levy, Gustavo Clpanema, and Wagner Esrdita. All were distinguished deputies, but four of the six represented the VON. Thus a party that averaged less than one"quarter of all Chamber seats from 1947 until 1964 generated two-thirds of the leading budgetary conservatives. VON members got on the Budget Com" mittee in order to resist pressures to spend, not to get a piece of the action. Ideology-in this case the ethos of the party-pre" vented adoption of a strategy that might have broadened the po" litical appeal of some VDN deputies. Deputies representing discrete electoral zones could claim they were responsible for attracting governmt!nt programs to thost! zones. Survey respondents were asked whether they concenlrated their campaigns or spread their eHom all around their states. As Table 23 shows, concentration led members 10 rely on the Budget and Finance Committees. Asked what percentages of their votes came from their areas of greatest suppon, deputies who believed that a single key area delivered more than 60 percent of their votes were twice as likely to regard the Budget Committee as most im" portant as those whose key areas delivered less than 40 percent. Respondents were then asked why certain committees were most important to them. Their responses were coded into fifteen
,
The Congress Connection
Table 24. CompDign ConU1ltrDtion and Committu Efforu Helpin! kse or R~on
Campaigned in Whole SllIe
Most Important 2nd Mosl lmponant 3rd MO$I Imponanl
CamJnigned in Some Areas More Than Orhers
7.0%
30.1% 23.8"-
18.S% ~
6.)"-
27% Nort: Column. do
no<
add '0 100'4 beause respondm .. mlr
ci'~
""
I cril<1ion no rank.
cattgorits. Two arc especially localistic; "because of my region" and "because of my voters." Deputies with concentrated c;ampaigns were twice as likely to give these constituency-oriented reo sponses. Moreover. as Table 24 demonstrates. these deputies were three times as likely as those with no campaign concentration to believe that helping their base or region was the most important criterion in comminee selection. In sum, deputies sought places on the Budgel Committee because they campaigned in and drew support from political subregions in their 513Ies." They undersrood Ihat uwarding these subregions was a marter of political survival.
Amending Bchavior. Deputies unable 10 attain seats on Ihe Budget Committees were far from helpless. They could sponsor and promote amendments 10 budget bills in comminee hearings and on the floor of the Chamber of Deputies or the Stnate. Among those programs subjeci to frequent amendments, the gram·in-aid budget of the Education and Culture Minislry was typical. Its gr:mts, ranging from a few hundred to Ihousands of dollan. went to a variely of social services, including shehers, asylums, and clubs. Though only a small part of the total Education and Culture budget, these funds wert imporlant 10 depulies, because their reo cipients would become loyal supporters. In 1957. some 23 percent 18. Richard W;lIftn has 'Il~f~ in I pc~1 communication mar dtpuries with una.u;1I riecroral prospectS ma y hav~ b«n forced 10 CI)rIUIIInIU. ~t cam· pa;!" activity .nd 'UbwqUftll vof~ ring. Iluk th~ d.a.la 10 p"nuf Ihi. illff,tRillS hyp<MhQ,H, althollsJ! in nullY (lIIQ bailiwicb prnbred 1M dtl'llty; thaI is. tkey Wfre cr~ariON of family Illd hfrif~.
r 2..1.
Table 25.
TUJ()
Olle-Term Deputies alld Amelldillg
PropoKd Amendment.
Onr Tl'rm Only
y~
15% 85 'YG
No
(93 )
Sm."" 1I, ..il. eo"Y""'" N.""".I, DeN. Octubc:r 4.
p.r.nth~
Applications of the Theory
.,. tlr< num b.-r of r.. pornknt ..
All
Otb~
DepulieJ
26% 74 % (2 871 19H. pp. S_107. The nun,b.-.. in
of Ihe 10131 Ch3mber membership made amendments to the 1958 educalion gram-in-3id budgt:l.' · Wh31 factors exp13in 3mending hehavior? P3rty 3ffiliarion malleTt:d: UDN members continued to kt:ep their dist3nce from budgeta!)' politics, making only about two-thi rds as many amendments as members of any other party. The number of terms a deputy had served did not 3ffee! 3mending. But, as T,lhle 25 shows, deputies in their in itial term who did not serve againexiting defeated or choosing not to run- we re considerably less likely to be active un the budget. 1o Perhaps thc number of terms a deputy had served was less importam than the time remaining until the next electoral test. In this regard il is helpful to examine amending in the Senate, because senators served staggered terms. Some of the senators serving in 1957 wo uld be up for reeb.:tion in 1958, while others faced no challenge until 1962. Senators facing a 1962 ele..:toral tes t made about TWice 35 many amendments as those facing an immediate test in 1958 . When an electoral challenge was only months away, it was too late to spend money. Because elections four years distam afforded sufficient lead time, deputies were able to use expenditures as part of their smviv31 strategies.~t 19. Thi s calculation e xclud~. m~mber> uf ln~ Budg~t Commi nee. bocau"" tnei r could be anended without .mendment5. IO. If approp ri.", dau were available. o ne migh, ask whether th ese depuri~ reduced their aCTi~ity tr..c.usc they knew they int~nd.-d to ....·ithdraw Ut whether their inactivity led to ddeat or withdrdw.1. II. ,\-byhew (1974) argues thal,n the Un ited Staf~ cI.iming cred it for future performance (tha t i~. prom;... ) may be more imponam th~n u.king cr.dil {or proj «:fS already completed. In Brazil, howcv~r. political communication is tOO ""eok lur th. cI«:tor. te !(r an(1 boli.'·e .uch plomises. n~d.
Ile.,
The CongreJ1 Connection
The Issue of l'ennea/Jility. How did the existence of electoral Strongholds affect budgetary behavior? The survey data discussed earlier tied deputies' perceptions of electoral zones to their committee choices. Rather than relying on perceptions, can we calculate individual bailiwicks' actual permeability (or penetrability) to the campaigns of other deputies? Such a calcubtion requires county-level denion returns, but these returns randy exist. Suppose, instead, we could identify localities benefiting from the individual amendments each deputy made. Assuming that no deputy would waste political capital seeking benefits from places that contribute no votes, localities getting aid constitute deputies' bail iwicks.u If no other deputy from the same state made amendments for any of the places aided by a particular deputy's amendments, then the latter had a completely impermeable bailiwick. If all the other deputies in a state made amendments for every site aided by the deputy, the bailiwick was completely penetrated-that is, it was no bailiwick at all. The key, then, is the bailiwick's permeability, defined as the ratio of actual to potential entrants in one's lone:
a.cc-p. " ,--",
(n - l)s,
where
p. a. n
'.
permeability of depllly x, ranging from 0 to number of sites mentioned in amendments by deputy x that are also mentioned by each other deputy in the state total number of deputies in the state number of pbces for which deputy x makes amendments
Permeability can also be defined in a second, more limited way, in terms of bailiwick incursions made only by members of other panies. This alternative version, in other words, ignores entries by members of the deputy's own party because they are less threatemng. Focusing on members of three important committees, I used the 1953 grants-in-aid budget of the Ministry of Education and Health 22. In a few srares all rhe: deputies ioined rogerher ro sponsor a single amend· ment. Joint sponsorship indicates cooperation, bur if does nor mean rhal all rhe ckpuries """"ived vOles cycrywheu in the Slate.
""
1~",1
Ie
"'a
Two A.PP/;Clltion$ of the Theory
'"<
Table 26. Averllgt Permeability RlltingJ and Committu Mtmbus!Jip Commin«
s..d,tl eN-
il,
Con.ritution and Jllnia (n ..
Diplomacy (If _
$0..,.,,,, B...,zlI.
$) ~
'"
All-Party Pe'mubiliry
Ofhtr-Plrty Pt.mubiliry
0.J43 0. 1$4 0.187
0.364 0. 142 0.210
Norionol. t>CN, 0.:.... 4, " H, pp. S- I07.
to caleulalc Ihcsc two cxpcrimcnlal vcuiolls of bailiwick permeability. On both meas ures Minas Gerais matched its reputation in Brazilian political folklore 35 the stale wilh Ihe mOSI impermeable districts. Experts o n Sao Paulo and Bahia confirmed that these locality-defined bailiwicks corresponded to Ihe elector:!! dislricts they knew. Finally. county-level electoral data for the state of Rio de Janeiro were available for the year 1962, and although ,he overlap between the experimental bailiwicks and the declio n results was imperfect (due to IUrnover among the deputies), it was quite high. Did the exinence of electo ral bailiwicks affect survi val strategies? Table 26 demonst rates that, with both measures, representatives of more permeable (that is, Icss secure) distriru were mo re likely to hold seats on the Budget Commintt. Suppose the actual monetary size of amendments is affected by bailiwick permeability. Do deputies from more permeable districts make larger amendments? Table 27 answe rs that question in the affirmative. In o ther words: Deputies whose d~Cloral bases were easily invaded sought committees that could reward their districts and made larger amendments to budget proposals.l.I
Determinants of Sliccess at the State Level We now turn to the budgetary success of whole statcs. More precisely, what determined ave rage per capita spending on each
me
23. One olh~r Ioel of amendment.!, 1M.. "l<)
"
"
us
The Congreu Connection
Table 27.
Permeability and Amendment She Correlation Between lliputy'$ Nonparty Enuy Ratio and Site of Avcr~gI' "mrodmmt
St~t'"
Ome4>tion
"'
Minu Gtr:ai. Sio Paulo Bahia
0 •.39
5.48
0.52 0.44
32 23 IS
~ari
0.63
10
5.5 1
"
7.99 3.12
$00<, ..... !o.u;I, eon" ..... Nacional, DeN. Ocrobt, 4.1952, pp. $-101. -n...w ...," -'"" odc-cted bt<:>.... lhoy h ... d.. 1.'"8e 0.05 le>d .uepe of Sahl., whidr 'oI.
,Iu,
state over [he entire 1947-1964 pe r iod?l~ Pursuant 10 the model developed in the second section, the effeas of five factors wefe estim:'IIed with multiple regression techniques.2J The five v:lnables and their indicators arc as follows : Budget Committee Representation: the total of Ihe :mnual over' or unde rrepr~ntation on the Budget Commi[t~ of each sta(t as compared to its percentage of all Chamber scats Party Leadership: the average annual ra tio of a s(;ue's per(en(.age of positions in the pany leaderships to its percentage of tolal Chamber sc.ats Chamber Leadership: the average annual ralio of a Slate's per· 2.4. For each year, the per~ntage of total nntralplvemmc:n1 Jpendin, Coin, into each SCatC ....·a5 divided by the STate', percentage- of Br:azil', population. A SI.Te receivi ng a share of central governmenT spmding equ.1 In ils populu;o., .ha", would receiY~ a ..:ore 011, statel proportiona tely bett~r off would receive a SCOU g.UtU than I, and so on. H. The ambiguity of the Ktutnove," hypothesi, was noted ~bovt. High del egale turnover seemed likely to inCfl'aJe mntivation to s«k po,k. but mJua tM prn;.ibility of IUCcns. In preliminary "tim~rion. of tht multi pit reycssion, ,he uro-orde. correl ation of this nriable wilh spending was low .nd irs partial ' tIVrs · sion .ocflkient was in,igni~unt. Sinct it wu both theoretically and naTistically wuk. il WaS d,opped from Ihe analy,i,. In .ddition. a control ,uch as "economic development" might be expected. Initially, the per.:entage of the work foro: participating in industry was included n such a (UnlTol. It il not reported he,e beau ... iii partial waS elOC11tially zero, with a 0.99 signi~Clln~ loel. Note .1", ,ha t the Federal Dimicr was 'emoved f'om the ana lysis since 10 much federal ,pendin8 t«hnio.lly uricinat" tM",.
1:.6 Table 28.
Two Applicatio11$ of the Theory Avtrage Spending by untral GOllernment on States in Relation to Population: 1947-1964
Vatiablt
Esrim~le"
0.0003
0.041 -0. 126 0.250
3.6] -1.0] 3.32
0.0027 0.320 0.005
O.ot8
2.27
0.039
- 0.008
- 1.48
0.162
1.04
RepTcsent~rion
No,,, Il' • 0.62: F • 4.51: P"ob :> F •
'Th... rorffici
Of<
111
4.7 1
Imetcepl Budgtl Commin..., Party Lud",.hip Ch,m~. Le.dership Verti£~1 Linb&" (Ministerial Positions) Skill, (l.i~r31 Professions in Delegalion )
Prob >
T
o.on.
umun.urdi ..d. Th
to m"'Uf< tho rd"i"" import.n« 01 th<
1,"10".
cemage of positions in the Chamber leadership of tOlal Chamber seats
to
its pen;entage
Vertical Linkages: the lotal number of yea rs in which someone born in a state directed any of the following ministries: Agriculture, Justice, Public Works, Education and Culture, Health, and Labor Skills: the average annual percentage of the Chamber delegation whose principal occupation was medical doctor or lawyer 6
The results of the multiple regression arc sho wn in Table 28. Three factors produced strong effects in the directions expected. States with "excessive" representation on the Budget Committee (in comparison to their share of all Chamber seats) pulled in a bigger share of central government spending. Cea ra, for example, which placed ten extra members on the Budget Committee o\'er the whole 1947-1964 period, received a 41 percent boost in its receipts because of this overrepresema(ion. Sta(t:s active in (he Chamber leadership also did wel1. Minas Ge rais had only half the number of leadership positions its scat share merited, a nd this 26. O<:rupJlions ,..ere coded~ , medicin~, i3w, publk functionary, journ~li.m, mmmcrce or banking. indu",)' or entrepreneur, milil.')·, engineering. I<.~hing, agricullUll', clergy, low·level b~nking, and O1hen. The medica! Jnd leg.1 prof..· .ions ~re rcprded gener~lIy as the Imdn;onal "Iibe.al prof.. ,;on'~ in Bra,ii. Stt also Gr""nficld (1977)_
,"
• •
~ Congress ConnecriOll
'"
factor lowered the state's r«eipts by JO pcn:ent. States that placed many of their residents in ministerial positions also profited. Residents of Bahia, for elCample, oc(:upied ministerial positions rwent)' times, and the slate's gain in spcnding ruched 36 pt"rcent. Two factors (in addition to the "turnover" (actor ~jected ear· lier) proved less important. States with delegations dominated by doctors and lawyers- that is, states with deputies whose skills were 100 na rrow for effective lobbying-re,eived, as elCpected, less spending, but the effect was small. The party di rectorates in the Chamber, expected to be less important than Chamber leadership, in fact manered little. Since most of the Chamber's folk wisdom attri butes more influ ence to the leadership than to the committees, Ihe centr.ll find ing is surely the importance of Ihe Budget Committee. Stalf$ that managed to gel their deputies on the Budget Commin ee were rewarded with significant boosts in cenlral government spending. Congressional vs. Presidential Survival Strategies
Earlier chapte rs ullcovertd an electoral-expenditure cycle in latin Ameri,a. Spending in response to de"ions was not dt·signed-as it might be in the United States-to inOuenee macroeconomic conditions. II was meant , instead, to provide jobs and projects for supporters and allies. In a case such as Brazil, where the legislature obviously plays a central role in the spending process, it is important to seek the source of the elecloral-elCpenditu re cycle. Is it the Congress, the president, or both? The notion of an electoral-elCpenditure cycle implies that expenditures have deviated from some norm- that is, from the distribution that would have occurred without de"inns. In Brazil normal receipts were dete rmined by each state's economic needs, often defin ed in terms of per capiu income, and its clilpacity to abso rb sptnding. frequently measurcd by the size of the lotal population and urba n population. Suppose annual central government spending on a state is a function of these th ree faclOn. The proportion of the total fluctuation in spending elCplained by these three determinants is a ro ugh baseline. an "apolitical" norm, from which to eval uate short-term "political" deviations. In Figure 9 the percentage of the total fluctuation in spending elCplaine
R'
8"'" P
''''''
C
80%
'0%
C C
'''''' 20%
C P
pCP
C
/: Pc
c .. CoIIgresslonai elect.io!'l eHeet Ilq)e(:led P _ PrHidential election IIMeet Ixpec1ed NOTE : The dilia lor income, lotal population. and urban population come from the decennial IigUf8S 01 !he Anu6rio EsIa1istioo do Brasil. They were interpolated to yield I _I estimales. In lhe years COIIsred by Ihesa dala. presidential e!ecli0n5 were held on October 3, 1950: October 3. 1955;
and OcIobef 3. 1960. ElecOons Iof the Chamber of Oepulies were IwlId on 0d0tIeI 3, 1950; Oc\obef 3, 19~ : Odober 3, 1958: and October 7, 1962. Because II careUlkef g0vernment was ., o tfiee in 1955, no preelection elleel is expecled. The year 1962 t\as also bHrI id&nlifild with • 'p- because il was the /irS! year 01 !he admInlSI•• ,iDn of JoAo Goulart, who took oI1iCe atter !he resignation of JaniO Ouadlos.
Figure 9. Expbined Vari:lncl.' of the B:lStlinl.' R~rl.'5Sion of TOlal Central Government Spt"nding on Sillies
n.
The Congress Connection
ralist years. The higher the percentage of total fluctuation ex· plained, the less important are "political" factors. The dramatic decline after 1950 in the explanatory value of the baseline equation reflects the increasing politicization of expen· diture politics that resulted from the end of PSD dominance. Within that decline, elections played a significant short-term role. In 1950, the first election year, the baseline regression actually improved its performance (Rl rising from 52 to 76 percent). The increased weight of income, total population, and urban population in explaining state shares resulted from the apolitical quality of the outgoing Dutra administration and the conservative role taken by the Finance Committee. In 1951, however, President Var· gas and the new Congress had their own ideas about state shares. Political coalition building became much more important, and the percentage of the interstate distribution explained by the equation declined from 72 to 17 percent. The congressional election of 1954 did not affect the distribution, but in the aftermath of the 1955 election-held by a caretaker government after Vargas's deaththe baseline regression fell back to 21 percent. The congressional election of 1958, like its predecessor, left distribution unchanged, but another low in the baseline model was reached in the presidential election of 1960. Two years later, when a new president took power and congressional elections were held, the predictive power of the equation declined again. Deviations from the baseline regression represent an increase in the use of the budget as a survival weapon. Electoral deviations attributable to the executive were much stronger than those at· tributable to the legislature. Indeed, me midterm congressional elections of 1954 and 1958, which did not coincide with presidential elections, affected the distrihution of spending very little. Why did the distribution of expenditures change so little around midterm congressional elections? In Chapter 3 we saw that total spending rose during midterms, because the executive's survival chances were improved by spending more. For individual deputies this one-time largesse would be beneficial, but it would not caust' them to rearrange their own distribution pattern. Each deputy could bring in a litrle more spending without incurring the political cost of pursuing a major budgetary redistribution. In other words, the willingness of the executive to spend more enabled incumbent legislators to engage in their own form of shock-cushioning. The
1~",1
Ie
a
'JO
Two Af'pliwlio"s of tht: Theory
fact that the interstate distribution-of expenditures changed during presidential elections may be a sign of the president's weakness relative to the legislature- for only at presidential elections could the president modify an interstate division basically determined by the Congress.
Some Indiuidual Cases The slUdy of individual cases is a way of enriching our understanding of the underlying processes generating statistical estimates. Six states were selected for intensive analysis on the basis of the cross-sectional regression's residuals (t hat is, the differences between the model's predictions and the :lellla1 federal spending the states received).
Rio Grande do SuI, Parand, and Sama Catarina. These relatively wealthy states ;lre all in Brazil's prosperous SOUlh. In budgetary terms th<'y differ sharply: Rio Cr:mde do Sui received extraordinary amounts of central government expenditures; Santa Catarina did quite well; and Parana received far less than its population merited-and far less than our model predicted. Rio Grande do SuI has long been politically blessed. Because it borders on once-hostile foreign countries, it hosts huge contingents of Brazilian soldiers and receives a substantial chunk of the mil· itary budget. It also receives "excessive" shares of almost every other program. To the student of Brazilian political history, this success is no accident. The revolution of 1930 was made by gau· cho (citizen of Rio Grande do Sui) Getiilio Vargas, and his gov· ernment was always dominated by gauchos. Before Vargas's coup, Rio Grande do Sui was already receiving twice the share of expenditures its population merited, and by 1933 it was collecting three times its per capitJ share. Post-1945 administrations were equally kind to the ga uchos. The sole exception was General Dutra-during his government the state's share of all feder::!1 spending declined 20 percent. With Vargas's return in 1951, Rio Gra nde do Sui resumed its privileged position. Vargas's successor, Juscelino Kubitschek, needed thc strong support of the state's PTB for his election, so he gave the vice president, gaucho Joao Goulan, comrol over patronage in agriculture, labor, and social secu rity. The last administrations of [he competitive period. the ill-fated Quadros and Goulart presi·
,
The Congress Connection
.J'
dencies, maintained Rio Grande do SuI's budgetary preeminence, and even in the supposedly apolitical military regime the state did well. Four of the five military presidents were gauchos, and the state's share of federal largesse never dedinedP Rio Grande do SuI's neighbor, Parana, was a budgetary loser, In the competitive years Parana was predominantly a coffee producer. After 1947, when me United States removed its wartime price ceilings, coffee became extraordinarily lucrative. The coffee elite in Parana wanted coffee's profits, but they also wanted a share in the direction of the Brazilian Coffee Institute, the government autarchy that sct domestic coffee prices and controlled marketing. It was difficult for any state to plead poverty~an important argument for pork~while simultaneously claiming that its economic importance entitled it to a share in policy making.2a Not all political leaders in Parana were linked to coffee. Why were those outside the coffee sector equally poor at attracting pork? The extreme fragmentation of state politics was one major hindrance. In the elections of 1954, 1958, and 1962, five different parties elected deputies. The local VDN split into two antagonistic wings, one centered in the capital (Curitiba) and the other, comprised mostly of natives of the Slate of Minas Gerais, located in Londrioa. The PSD and PTB were also fragmented and contentious, and as a result Parana had the highest turnover of all congressional delegations. With such conflict, relationships between presidents and the state's governors were often strained. Kubitschek actually finished third in Parana's ba[[oting (even behind PHnio Salgado, the candidate of the neobscist Integralist pany), and the president got along quite poorly with the state's governor. Neighboring Santa Catarina, on the other hand. was highly successful in obtaining federal funds. Santa Catarina's internal political struggles were much milder than those of Parana. Only the three main parties ever elected deputies, and the st:ne supponcd the winning ticket in al1 four presidential contests. Whereas Parana's leaders were predominantly from the liheral professions, in27. Joao Figueiredo CounU a. a g;lucho MCaU5e even Ihough he was born in Rio de Janeiro his family was all gaucho and many of his formative experiences were in Rio Grande do SuI. 28. Information based on interviews in July 1983 Wilh Parana deputi .. from lhe UDN and rhe PSD.
•
'J'
Two AppficatiollS of tlH Throry
dumial and commercial groups (groups who could profit from federally supplied infrastructure) dominated the importa nt parties in Santa Catarina. Santa Catarina's UON, for example, was very diHerent from the moralistic and austere UON of Parana. The VDN leadership in Santa Catarina came from [wo dynami, families, the Konders and the Bomhausens. When members of these two bmiliel married. the union joined a bank and a family of uaditional politicians. Though elections in Santa Catarina were vigorously contested, cooperation on budgetary issues never broke down. During the presidential administration of Cerulio Vargas, the VDN governor of Santa Catarina (Irineu Bornhauscn) persuaded Vargas 10 approve fund s for the "Wheat Highway," a road running across the state to connect Sao Miguel d'Oeste with Itajai. As the price of Vargas's cooperation, the first stretch linked the towns of Campos Novos and Lajes. Lajes, it happens, was the political seat of Nereu Ramos, PSD boss and Vargas ally in Sama Catarina. The Ramos and Bornhausen families competed fiercdy, but their competition brought rewards to Santa Catarina. u
Ceara , Maranhiio. and Rio Grande do Norte. These Northeastern states 3re not equally poor or backward, but they are all poorer than Rio Grande do Sui, Parana, or Santa Catarina. In rdation to its population, Ceara was Brazil's champion in anracting central government expenditures. It was also the champion in Budget Committee members and Chamber leaders. Why did Ceara do so well? Located in the center of a region subject to severe droughn, Cea ra is a state whose leaders have lea rned to profit from adversity (Cuni ff 1975). With a vulnerable economy based on cattle and couon, Ceara's dites discovered even before Brazil became a republic tha t federal la rgesse was no threat. In the major drought of 1877- 1880, representatives of Ceara led the lobbying effort in Rio de Janeiro. Overcoming the usual bitterness of local politics, they achieved a scrupulous bipartisanship. The results were heartening: The Liberal government taking over in 1878 spent 29. Inltrvicw1 with San ti Cat ..ina dePllri .. in JlIly ]983. II i. wonh IlOling ThaT aftu 196" 1M 8omh.uKn and Ramos "mi]i~. ~nally did unit~ in Th~ goy. trnmen' put)'. ARENA. For. similar UtllgglC bCf',I,..,.n the Andrada and Bi .. Fon.. fami]i .. in Barbac.:na, Minas Ge,.is, Stt Carvalho (1966).
The Congress Connection
'Jl
almost thirty times what its predecessor had spent jU5t one year earlier. In the early 1950s Ceara persuaded the directors of the new Northeast Development Bank to locate the bank in Fonaleza, their capital, ovtr the opposition of both the Vargas administra· tion and most bankers..)O While other Northeastern delegations split over the issue, every Ceara deputy supported the coming of tht bank. By 1958 one Pernambucan newspaper, comparing Ptr· nambuco with Ceara, put it nicely: "The Cearenses unite; we split. The Cearenses work; we fight. The Cearenses take care of life; we take care of politics."" O ne factor hthind Ceara's success was the unimportance of ideology in Sl3te elections. Party was just a label. Candidates frequently jumped from the UDN and PSD 10 the PSP (Social Progress party) or PTB, and the smaller parties could tip tht electoral balance either way. The price of their adhesion was not principle but patronage. The absence of working·class mobilitation made ideological flexibility easier; indeed, the PTB (nominally a working-class pa rty) was basically con5trvativt and rural in Ceara, gelling less than 20 percent of its vote in the capital (Monttnegro 1960). Deputies from otht r states joint d Budget Comminte staff mt mbers in steing Ceara's deputies as the most pork-oritnted dtit· gation in the Congress.)1 Ceara's lobbying centtred on the Budget Comminte. At one lime the presidtnt of the committee was Paulo Sarasate, a Ceara rcprtSl!ntariv t from the usually conservative UON. No defender of economic order, Sarasatt was widely known as a spender. Whtn a Ctara commiute member left the Chamber, he was immtdiau:ly replaced by anothtr Ceara dtp uty from tht same party.lJ In 1962 the PSO and UON in Ceara united, tht Slate was ablt to achieve its highest degree of overrepresentation on the Budget Committee, and expenditures grew. 30.
!ire RtWla Ba"ed,;" B,~ikira. 00. 228, D«nnbel" 12, t951, p. I. ...... ic .. Rtciu linl, Campin~ Gr~nde 1«On.d••n.d Fort~lua third. 3 1. Sft Di:irio dt Perna",bouo, Dc<:emMr U, 1958, p. 4. S« ~Iso &.Il00 do Nordc.te (1958) a nd Jo ...al do Comh(io. May 13. 195 I, p. 12. 32. Thil observation is based on. interviews ",il" former drputir:5 and t\lrrmt luffen.. CUra depuJ)' Jose Colombo de Sou..., ....I>m alked if he "ad rYer fried r~n.ktd
to limit s~nding. could only re_bel" one inridcnr: He "~d ana offered all. , ""'ndm.... t to prohibit gOVC1lllmnl oflieu from buyin.8 smokin8 wppr ... for dwir cmploy«s. 33. For ex~mplc. Colombo de Souu ....as tq)Iucd by Alvaro Lins, then by Paco de Andnde, ~JJ from thr ISP.
')4
Two Applicati01U of the TMory
Rio Grande do Nortc, which resembled Ceara politically, was another substantial recipient of federal funds. The: leading politicians ill both the UDN and the PSD were 3.ssoci:lIcd with com-
merce: and cotton.l O The panies downpb)'cd ideological differences, and cross-parly transfers were commo n. En~n VDN candidates sometimes ran as populists, and the two main cities, Natal and Mossoro, domina ted political life by virtue of their weight in the 10l'al Slale electorate. Maranhao was the poorest of these three Northeastern states, blll its poveny did 110 \ lead ils congressional delegadon to pursue central government programs. Indeed, per capita federal expenditures in Maranhiio were vastly inferior to almost all other states. Why did its ddegation do so lilll e~ l bc passivity of politicians from Maranhao stands in stark con· trast to the aggressiveness of politicians from Ceara and Rio Grande do None. Free from the periodic shock of drought, the o liga rch y of Ma ranhao merely desired solitude, and the politic:ll machine it developed was well anuned to its goa1. H Throughout the competitive period the last federally appoi nted gove rnor of the E.stado Niwo. Vitorino Freire, conuolled politics in Maranhao. His political machi ne funnionro with extraordinary smoothness, drawing its energy not from proj«'ts but from nominatiolls, particularly judgeships. Freire's extn:mdy conservative PSO was totally dominant. In the federal elections of 1954, 1958, and 1962, the PSO always controlled a majority of Ma ranhao's deputies. and a majority of the electorate always supported PSO presidential candidates. The second largest pa rty, the I>SP, picked up vOles primarily in rura l areas, and it was as conservative as the PSO. The PTB and UON wefe also conservative, but they wefe so weak tha t often they failed e\'en 10 compete in fede ral cI«'t;ons. 3~. In Ih~ I'SD, t~ JK>wers W (rC Gwrgi!>Q Avdhoo, who li,'( d in Rio; Thco· dorico Ikuna, w,th intercsts in COltOn and hotd.; JOlO Cam .... conon; ~nd JCSIoC Pinto Freire, I,nk<"
cl
M
I
Tbt CongrtJJ Conntaion
'll
Where, in the midst of this malaise and apathy, was a new generation to catalyze the process of chiJnge? Potentially, one such leader was Renato Archer, son of a former governor and scion of a traditional PSO family . Archer was the prOtotype of the modem, progressive leader Maranhao lacked. Linked to the left-nationalist wing of the national PSO, his position on the Chamber's Budget Committee was just the place from which to start something in Maranhao. But Renato Archer had distanced himself from the st,lte. Living in Rio and spending little lime in Maranhao, Archer developed a "national vision"; indeed, he was known in Brazil as a nuclear power expert. U On the Budget Committee, Archer took charge of the navy budget (he was a former naval officer), a branch of the military that could hardly uansform Maranhiio. In a sense, the tragedy of Maranhiio was that it skipped the pork stage. Its leaders reflected either the iron bands of tradition or a modernity alien to the needs of theif state.J7
Recapitulation. This inquiry into individual states highlights the importance of political climate, of context, both in determining the intensity with which deputies pursue budgetary allocations as a survival strategy and in determining the success of whole states in attracting expenditures. Any nOlion that the poorest states are always the most eager to feed at the central government trough is dispelled by the case of Maranhiio. Elites in Ihat state actually feared federal programs, because they might eventually lead to a more diverse society and a political competition that could escape from elite control. Immediate survival meant Qvoiding federa l intervention. Only when a state's elite learned to profit from central government spending, or .vhen its economic interests depended on government support, would dfoMs to :attract such spending ~ come part of its survival strategy. Political clim:lle was important even where individual politi36. Archer bcelmc mini"er Qf Kien~ ~nd t«floology in tM pQlirician from Goi:h who alone rimr huded the Budgn CO n1miu«, 11Ji> took l "national" posirion. Paraib~, tOO, sums full Qf politicians caring linle ~bout loc~1 intu.n•. Ern~ni Sa tim, for m~ny ye~", a member Qf the Budget Commi n« and .rill in the Chan1ber, boasted 10 me that M h~d never gonen anylhin8 for hi •• kelllnl wne .~"'pT o ne wetch Qf road, and even that h~d never been Rnished. adminim~tiQn
.,6
Two Application$ of the Theory
cians saw the: ad va ntages of expendit ures. Parana's i nten ~ interpany conflias reduced the political influence of the state as a whole. Each deputy might actively pursue: his own pork-oriented survival Slrategy. but inlerpany hostility kept Parana's deputies from cooperating and high turnover kept the delegation from attaining much seniority. In Ihis case the structu re of political competition damaged the interests of the whole sta te.
Pork Barrel Politics Under Military Rule When the military came to power in 1964, the role of the Congress in the distribu tion of federa l expendirures changed drarn:nically_ The hostility of th e: generals and their civilian ad visers toward the Congress was not merely rhetorical. Within a few days of the coup the junta issued a decrtt granting the p r~i dent the exclusive right [0 initi:lote all fi nancial bills and preventing the Congress fro m voting any additions 10 expendi tures. The junta soon wro te a new consti tutio n strengthening these rest rictions. The Constitution of 1967, in the words of Ronning and Keith (1976:231 ), ~curbs the ab ility of any congressman to build up his own clientde- via 'pork barrel' kgislatio n. "J' As an institution independently able 10 modify expc-nditurc:s, Congress ceased 10 exist, but the junla went even furthe r, mounting a n intense attack, particularly during the Costa e Silva and Medici administra tions, against nalio nal and state-Ievd ~ profC'S sional politicians." These attacks no twithstanding, clientc1istie pol . ilics survived. The regime had made the conventional lr:lppings of parlia menta ry democracy, including panies and cI« tions. pan of its own surv ival stra tegy. As a result, it had to rely o n mu nicipal authorities 10 deliver the vote in its highly controlled elections. As Paul Cammack (1982) points ou t, once central au tho rities concentrated resources in their own hands, and o nce civilian politicians could make no credible claim to effective ro les in representation or policy formation, municipal autho rities were in a strong position to manipu late resources to benefit the government party. Fo r Congress as a body a nd for mosl individual politieiam, however, budgetary politics disa ppea red. Some of the deputies responding to o ur mail surve-y continued 38. Fo.
~ {oml'r~h~osiv(
<.IUcrlplion of
Ih~
.dorms
~ B~.klioi
(1917).
The Congress Connection
'17
serve after the fall of the piuraiisl regime. Without e){ccption, they agreed that the Congress h3d lost its ability to participate in budget making. Only those individual congressmen who were fa vored by the junta were able 10 make requests that might dicit a positive response. Systematic comparison of the distribution of expenditures before and aher 1964 is not yet possible. Little information was published during the early, reprcssive phase of military rule, and statistical research on the post-1974 period has been ham~red by the economic crisis of the 19805. The fragments of information that do exist suggest that during the early years of military rule rapidly growing Slatts like Sio Paulo improved Iheir relative p0sition in ex~nditurcs. Bttwccn 1964 and 1966, for example, Sio Paulo's ~r capita share of all central government s~nding climbed from 8010 101 percent of the nalional average. Between 1970 3nd 1975 Sao P:mlo's share of IOtal spending continued to grow, and most individual programs benefited from the overall gain. Moreover, the ~r capita distribution across states, which congressional logrolling had tended to even out, now became less even. Not only did the differences betwccn the states in total central government s~nding increase, but in almost every specific program there was more variation among the states. J9 After 1974 the influence of the technocrats declined. With its shocking defeat in the 1974 dection, with civilian pressure for liberalization mounting, and with the increase in oil priccs threatening the "economic miracle," the regime embarked on a gradual political o~ning. Since formal withdrawal from power was never intended 10 end military influence in policy making, the military began to cultivate its regional and social bases much more systematically_ Clientdistic practices increased sharply. But this is the
10
Story of our ncxt chapter.
Conclusion This chapter has focused on the budgetary survival strategies of the Brazilian legislator, a political actor bdow the level of the executive. What h3s been learned? 39. Varianar incrca~ in s~nding in gen~f31 goy""m~nr, 3griculrur~, ~n~rgy. 1fansport, indunry .nd cvmmo:rc~. lab", .nd wdb rc • and ddenS(. It drcl'l:aK<.i in housing and remainal tM same in tGilds., education and culnm. and huhh. 5« Fund.ll~o GftUJio Varpt (19711, 19110).
'J'
Two AppliwtioJlS of
thf
Tht:ury
DepuTies or senators seeking central government spe nding for their electora l bases (o ul d use two stnltcgies. They could gct on
(he Budgct Comminee or the Chamber IC3dcrships, il nd they could pro pose amendments. Congress men adopt ing these smllcgies were less likely 10 be membe rs of the UDN, more lik('ly to plan another run for o ffice, and more likel y 10 concentrate thei r campaigns and votes locally. The na ture of presidential sIT31egics a lso affected slr3tcgic legisbtivc be h3v ior. During Dutra's a UStere administration. the
Budge t Com minee became a watchdog of the Treasury. Vargas's and Kubitsc hek's populist dcveloplm:ntal ism incre3sed opportun-
ities for pork enormously, and the Budget Commin cc became a dispenser of largesse :lnu 3 locus of congrcssion3l power. The aV3il3bility of more resou rces, however, did nOI lu.·cl.'ss:uily t'3se the survival crist's of insecu re deputies. bec3use during these periods more powerful deputies mo nopolized Budget Committee scalS. When more meal wa~ 3Y3ibble. in other words, the older lions ch3S«1 the cubs 3w3y. The budget3ry success of whole Sl31CS (r3lher Ih3n inJividu3l deputies) depended on 3 number of faclors. Not surprisingl y, st3tes were likely 10 do well if 3 brge number of their represent3tiyes sought expenditures_ Moreover, the struCture of legisbtiye organiution affecled St3les_ Reeau$C the rules of the legislalure ensured e3ch party a proportion31 share of seats on each commillei:, srates could bt overreprest'llIed in key decision.m3king silcs like the Budget Commil1tt. and ovcrrepres<'nlation contribUled to expenditure success. Overrepreselliation on the Chamber leadenhip helped as well, and politicians from one's state who occupied ministerial positions were able to bring home more pork. Congressio nal delegations with mixed occup:lIional backgrounds, r:lIher than simply the traditional liberal profcssions. could lobby a bit mo re effectively. Politic31 climate also m:mered. Some St3 te delegations we re so ronu:mious and unst3ble that deputies r:nely had the seniorit y to allain positions of influence in the legislature. Others were so backward th3t survival depended on avoiding fede ral programs, because such programs might increase politic31 compet ition. Fin3l1y, the disappear3nce of the Congress as 3 political force in the post· 1964 military er3 did not eliminate budgetary politics. Innc3d, it elevated olher political figures, especially state gover-
The Congress Connection
'"
nors and municipal authorities, who delivered votes in ("change for allocations. This new political StructUre did change the overall distribution of public expenditures. The pre-1964 Congress had brought together hundreds of deputies into a structUre that made them all political equals. As a result, the structure of policy making they created emphasized stability, logrolling, and compromise. Although some states received more spending than others, almost everyone got a piece of the aClion. When the military eliminated the Congress, the executive branch became much stronger and some stales did much better than others. The executive was able to pursue its own survival strategy at much lower cost and with much greater success when the Congress was no longer a factor.
rho
l~
h
5. When Soldiers Need Friends: The Search for Influence by an Authoritarian Regime in Retreat
Although Chaptu 4 shifted the locus of str:llcgic choice from executives to legislators, it retained the shorl-term natuTc of survival slralcgie5. Presidents sought to stay in office; legislators sought reelection. Fo r both kinds of actors, influence required formal position. Suppose we rebx th e aS$umption$ of immediacy and po sition. Might a political actor Slriv( 10 preserve.' infl uence even
while yidding the formal position which o riginally guarantud that influence? In 1973 and 1974, Brazil's military regime experienced twO shocks. The first hlow was the dram.nic increase in oil prices following the Arab-Israeli conflict. The oil crisis exposed the vulnerability of the "economic miracle" that had legitimized the military junta a nd made Brazil the lOast of the developing world. The second blow was the stunning eleclOral defeat suffered by the military government at the end of 1974. That defeat signified a growing reicaion by Br3Zilians of the military regime itself. The newl y installed administration of General Geisel had already commined itself 10 a gradual democratization, but after these twin shocks it was plain the military would need :1 new cl3im 10 legitimacy if it wanted a ~ ignifi can t role in a pluralist political future . This chapter broadens the nOlion of surviv:iI strategies by examining the social and economic programs General Geisel a nd his successor, General )030 FigueIredo. adopted as part of their efforts to ensure lo ng-range in fluence fo r the military. The fi rst section briefly discusses the background of the abertura ("ojXning"). I argue that even though the dialogue between the regime and civil society revolved primarily around the political alijXcts of the ojXning, the intent to infuse it with social and economic content was
When
SQld~rs
.,
Need Friends
.
soon evident in the public pronouncements of key leaders. Because other scholars have documented the political dimensions of the abertura, I give more attention to the causes of the opening than to specific aspects of liberalization. The next section explores various strategies available to the regime and links alternative group and regional targets to appropriate policy instruments. The third section discusses five broad policy areas: rural development in the Northeast, housing for the poor, the programmatic division of central government expenditures, industrial deconcentration, and wage sctting. This is followed by a comparison of policy evolution in the five areas. Focusing on the deterioration of the economy and the collapse of administrative coherence and credibility, this section asks why the regime could implement certain policies more successfully than others. Throughout the chapter the approach is inductive and comparative, inferring poJicymakers' motivations from the evolution and implementation of the various policies themselves. 1
The "Why" and "What" of Abertura The military conspirators who overthrew joao Goulart in 1964 divide into four groups: the Castelistas (after President Castello Branco), the hard line, the right-wing nationalists, and the military professionals. Leadership of the first military administration fell 10 Castello Branco and his followers, a group of officers trained in Brazil's Higher War College and linked to the conservative National Democratic Union party. Although Castello Branco had an ample range of arbitrary powers at his disposal, there were indications he was commined to returning the nation to democracy. His ruling coalition, however, also included hard-liners and rightwing nationalists. Expecting a prolonged spell of military rule, the hard-liners wanted a more extensive purge of political opposition. The right-wing nationalists were equally authoritarian, but they I. Bcatus.e poli cy making atnnot be directly obse ....ed-Brazil was. aner all, a diCtlltorship-not much atn be learned about the goal. and u.cti(.3 of individual decifion makers. The chapter gives lin)e anention, therefore, to tM "black box· of decision making. Brazilians havc carried out a number of imcreSling srudies of the fonnal struaun: of a:rtaio policy-making or&>."", but none approach the kind of depth achiend, for aample, in Allison', (19711 srudy of the Cuban miMile crisis. 5«, however, Guimaries (1979).
1~",1
Ie
a
'.'
Two Applicatio.,5 of thl! Thtory
opposed the Castdistas' efforts to open the Brazilian economy to foreign investment,l C:lsrello Branco resisted efforts to imensify the political repression. Though in the first year after the coup the hard-liners demanded five thousand purges, only four hundred prop1e lost their polilical rights (Vc:lasco e Cruz and Martins 1983 ). The government set expiration d:ues for its :nbitrary powers, and most observers expected the reslonuiofl of a liberal-democratic regime. In 1965 C3stdlo Branco :lOd his coalition held the prc\';ously scheduled gubern:llorial elections. The outcome was generally favorable for the new government. Of the eleven governorships contested, candidates committed to the revolution won nine. Even though the regime lost only two governorships (and even in those cases the victors were not antimilitary), the t'1t'ction put the president on the defensive and strengthened the hard·lineN. They fo rced Castello Branco to accept new a rbitrary powers through the Second Institutional An (AI·2), a decree Ihal !iel off 3nother W3ve of politic31 purges 3nd suspensions of civil rights. Castello Branco watc hed his W3T minister, Anur da Costa e Silva, accumulate power as a mediator betv.een the Castelislas and the ha rd· liners. Costa e Silva replaced Castello Branco in 1967. In rctrOSpe<:I. COSIa e Silva's assumption of the presidency signaled the prolong3tion of the revolution. As opposition (now including armed insurrections) grew, repression intensified. In early 1968, Student grievances escalated into riots, and major strikes broke out in the industrial centers of QS:ISCO in Sao P:lUlo and Contagem in Minas Gerais. The Congress had begun 10 reassert itself. and it now found itSelf totally at odds with the military rulers. When civilian supponers of the regime (such as presidential hopeful Carlos Lacerda) saw that their own ambitions had no mo re future than those of leftist politicians, they broke away from the regime's suppon coalition and joined [he opposition. Costa e Silva's response, the Fifth Instit utional Act (AI·5), lurned the regime into a more repressive dietatorship. This act could be used to dose [he Congress. replace the administrations of Slates and municipalities. and deprive any eitizen of political rights. These abrogations of democratic libenies were not merely 2. For tM best introduction to the period. sa: SkidmoR (19&1).
When Soldiers Need Friendl
abstract possibilities-they were applied with a vengeance. The Congress was temporarily shut, and the loss of political rights affecled 40 percent of the members of Ihe already docile official opposition party plus some five hundred distinguished people in the universities. the media, and the civil service. Political opponents of the regime disappeared, and torture became commonplace. Censorship was everywhere. In the end, center-left opposition was crushed. The regime's only remaining challenge came from the rightwing nationalists. Led by General Albuquerqut' Lima, they opposed the internationalism of the Castclistas' economic strategy. When a stroke immobiliz.ed COS(3 e Silva in 1969, Lima went to the barracks to campaign for the presidency on a platform of nationalist authoritarianism. The top generals dosed ranks against him, forced the competing factions to compromise, and imposed Garrastaz.u Medici as an "apolitical" candidate. In the words of Velasco e Cruz and Estevam Martins: The compromise excluded no one: it went from the latifundio 10 the mul\inationals, covering alllhe possibilities of exploitation of man by man, from the most modem to the most retrograde, even including those: contrary to the interests of production. The cost would be paid through the wage squeeze, and it would be guaranteed by corporativist unionism, innation, and internal and external indebtt'dness (1983:4041 ).
The Costa e Silva and Medici administrations implemented a brutal repression, but they presided as well over the beginnings of Brazil's economic miracle. Regime propagandists had no difficulty linking rapid economic growth with social peace. With the gross national product increasing mort' than 10 percent annually, the government appeared to enjoy con5iderabk popularity, particularly among the rapidly growing middle and uppt'r middle classes constituting its primary clientele. The Beginnings of Abertura Because they benefited disproportionately from Brazil's prosperity, elites ought to have supplied the regime's srrongest support. Did they respond? The only systematic study of elite: opinion during these years is Peter McDonough's Power and Ideology in Bral;iI (1981). In 1972 and 1973 McDonough and a Brazilian research
""
1~",1
Ie
"'a
'H
Two Applialtions of the Theory
team carned oul an elaborate program of interviews focusing on the leaders of six groups: the opposition party (the Brazilian Democratic Movement or MOB), the church, husiness, lahor, high-level civil servants, and the government party (ARENA). When McDonough's tcam asked these leaders who should receive less .mention from the government, most believed technocrats, the military, and transnational corporations all deserved less anemion. Moreover, the elite respondents supported agrarian reform, reduced military influence in government, redistribution of income, greater political participation by students, and a relaxation of censorship. McDonough's conclusion is unequivocal: "The very legitimacy of the reigning coalition seems problematic" (p. 125). As McDonough shows, opposition to the regime was based nOl just on the absence of civil liberties but also on the consequences of the economic model. By the early seventies most observers believed Brazil's already poor distribution of income had worsened during military rule (Tolipan and Tinelli 1975 ). The regime's economic czars were unmoved: There was no poim, they insisted, in a redistribution of poverty. Brazil had to wait " until the cake grew." Still, by 1972 the GNP had been growing rapidly for five years. Had not the rime for redistribution arrived? Civil society also began to stir. The Catholic church and the National Association of Lawyers became especially active. Protesting against arbitrary arrests, disappearances, and tortures, the bishops and lawyers were symbols of the unwillingness of Brazil's elites to accept what McDonough calls "sustained intrusions into their own spheres of privilege and influence" (p. 125). The choil;e of General Erncsto Geisel to sucreed Medici marked the reascension of the Castelista group, but only a minority of the higher offil;ers supponed the liberal attitudes of the Castelistas. The generals who picked Geisel probably had liule idea of the depth of his commitment to abertura (or distensao, "relaxation," as it was then called), and his selection as presidem may owe much to the faa that his brother, Orlando, was M&l.ici's minister of defense (Lamounier and Fatia 1981 ). Geisel's early proclamations about relaxation were nothing new. His predecessors had talked of a Teturn to democracy, but their rhetoric had always given way to repression. Under Geisel the rhetoric became, haltingly and unevenly, a reality. The opening endured.
r
.... '
l~
u.
When Soldin$ Need Frimd$
'"
The abertura managed to survive for three quite diflerent reasons. Authoritarianism, as we have seen, was only one strand in the idrology of the 1961 coup, and it was a strand that had never been fully institutionalized. If Geisel, as representative of the antiauthoritarian element, wanted to strengthen the Castelistas against the hard·liners and the right-wing nationalist$, he would need the support of elites outside the military. For business and professional leaders. political liberaliution was the minimum requirement for support. Abertura survived as well because Geisel and his chief strategist, General Golbery CoutO e Silva, were enormously clever in controlling its pace in such a way that neither left nor right ever fclt its situation was hopeless. Finally, given that liberaliution fit Geisel's heliefs and his need for allies. it made sense to start the process at the beginning of his term. The lihcr21 opposition was disorganized and quiet, and the militant left had been decimated. The time was right. Why not liberalize?l If Geisel's conception of Qb~rturQ indeed extended beyond palitical liberaliution, his early stalements reveal few such intentions. His remarks at the occasion of the first meeting of his cab· inet were typical: Facile dimibutivism, which Irks 10 reduce individual inequal ities by generosiry in the readjustment of nominal wages, is destined 10 fail, because il generates inflationary tensions and weakens Ihe potential of saving and development. Improvement in the personal distribution of income will have 10 come from ... ed ucation. job crealion, and from perfecting fi scal policy, creating savings funds for workers, and im· proving health and social assistance (Geisel 1974:S2). The minister of planning, )030 Paulo dos Reis Vdloso, echoed the president. Responding in 197) to !Studies crilical of Brazil's
income distribution, Velloso attacked the methodologies of these studies and insisted that growlh deserved continued emphasis. For Velloso, Ihe government's drive for higher median incomes would eventually rai~ the real income of every group.· J. Thi, i, nOf '0 imply thu the a/x"'mJ pr~lkd $l!>OOthly. At k .1$1 until 1977. when Geisel m,n'ged to remOVe .he huds of The Msecurit)'~ opeution.., periodl of intense reprusion (inc luding 10rlllre ~nd ,mpr'loOnmml) were ~ommon. S« Ku~i",ki (1982). 4. Reis VelloloO, lpeed> publisl>ed in Ulri"", Ho,... Mardi 29, 197J.
,.,
Two Appliedtions of th~ Theory
The Abertur3 Broadens Liberalization was [0 begin with the election of 1974. If the eleerion could take place in a n atmosphere frcc from coercion, and i( the gove rnment received a resounding vOle of confidence, the president would ha ve the support his gradualist stra tegy required. Until a few weeks before the eltccion, the government expected an easy vicrory.J Instead, the opposition sWTcd a smashing success, taking 16 of 22 ~natc seats and 160 of 364 Cha mbe r seats.' Why did th e regime fare so poorly when four yea rs earlier it had achieved a great electoral success? Admittedly, the election of 1970 had been held in an a unosphere of intimidatio n, but even w the opposition was dea rl y in disar ray. Moreover, McDonough 's public opinion survey, implemented in 1972- 1973 at the height of the "miracle,R fou nd high levels of support for the regime in all social classes, with only a small falloff among wo rkers (Cohen 1982). Why, then. was the regime so badly defeated in 1974? Why did the same pallfista indust rial workers who had supported the regime in 1972 vote 8- 1 for the opposition ('andidate in the 1974 Sao Paulo Senate race (Cardoso and l..amounier 1975)1 &unom;c collapse did nOI cause Ihe reversal of opinion. True, GNP growth between 1973 and 1974 dedi ned from 14 10 9.H percent, and inflation grew from 12.8 10 27.6 percent. Still. employment was high and real wages were improving. Oider wo rkers knew thei r real incomes had declined since 1964, but thei r memories of the chaos of the Q uadros and Goulart years made them distrustful of open politiCli. Younger workers saw the mselves as beneficiaries of an orde rl y economic growth directed by a mo ralistic military. Mass opinion during the heyda y of growth was supportive. but that support was fragile: . The gove:rn ment bombarded the population with news of its achievements. and opposing views faced an impenetrable: wall of censorship. The 1974 d ectOfal campaign was the first time the opposition was able 10 make il s case abou t the gap between rich and poor, the deterioration of urb:lIl health ('onditions, the rights of workers, and other issues. The military's 5. Sc.:, for tumpk. the )I~temc111~ of one of the ci~,hn k.dc:~ of the- .Jm,,,Pfflonio Por!ell~, in \'t'I'" Oc-lobc. JO, IlI7
jstt~rion.
When Soldiers Need Friends
'47
version of history could not prevail, especially in an atmosphere of economic uncenainty.' By March 1975, Geisel began talking about social policy. Pointing to the recently created Social Development Council (which brought together ministers involved with education, health, housing, and the like), Geisel emphasized the new imponance he would giv~ to social progr3ms. Massive infusions of funds would go to education and health, and a new wage policy would provide the working class with gains above the rate of inl13tion. 1 In a televised speech in August 1975, the president went much funher, leaving little doubt of his intentions: "Rdal
The president went on to link socioeconomic liber31ization to the post-oil crisis economic difficulties, and he boasted of his success in avoiding recession and unemployment and even continuing to innovate in social policy. A few years later, Geisel made the point again, arguing that his goal had been to slow the economy gradually in lieu of a recession. The same theme was elaborated by Reis Velloso: "The maintenance of growth itself, though al rates below those of the earlier phase, would be a facto r in Ihe preservation of internal and external confidence." 10 For Geisel, ex· pansion of soci31 programs had become a necessity. He knew the political opening "could not endure a social crisis of serious proponions." 11 Policy was inseparable from panisan activity. In a speech 10 7. On the subjeC"! of working ..ebM support for 'he military, Stt Step~n (1971:47- 49). 8. Spttch of DKem~r 30, 1974, on Idevi.;on. 9. Spttch of August I , 1975. G~iod also nOl~d thai ~,dn3lion~ i. Moflen presented wilh an (xdusi,·dy political connotation •••• One fcd~ in these propooi tion , ~ con'picuo", no,t~lgia lor the nOi too diU3nt pUt, 3 p~" in which lormal li ~rti .. wu~ emph05ized. Thi •• <suaged Ihe ronsci~nc .. of many, bUI btc:aUst thest liberties weren't rully practiced, they had the opposite elfeC"!, strving the .p~tit'"" and grttd of a few 10 Ih~ d~trimenl of Ihe real notional in!eml.· 10. S~ech of May 18, 1978. In a 'p«<:h on September 24, 1975, Vdloso .Tressed the wage incrc.5t5 that government policy wa, uimularing and cKpre.",d the hope Ihal rca l wages could won return to their levd 01 1960! 11. Sp<"t<:h 0/ IX«m~r 20, 1978, 10 the Economic o..vdopment CounciL
,.,.,.
l~
-
.
te
",
Two Applications
of the Throry
the leadership of the government party in October 1975, the president emphasized his desire for an electoral viclOry by the party. Three elements, according to Geisel, were crucial for an ARENA triumph: the program of the party, ils leadership. and the deeds of the government. In a iuxlaposition that was hardly coincidental, Geisel went on to stress the real wage increases workers had been receiving in 1975. Geisel's hand-picked successor, Joao Figueiredo, had been head of the National Information Service, the Brazilian secret police. That affiliation was an unlikely background for a pn:sident exp«:lcd to continue a political liberalization, bUi Geisel had confidence in Figueiredo :md took pains to help the new president soften his image. Indeed, Figueiredo's achievements in granting :l general political amne~ty and condtl(ting the 1982 elect ion ~ were extremely significant. The president's commitment to the social abertllra was also widely accepted. Even more than Geisel, Figueiredo's speeches Slrl'SSeJ wage increases, expansion of education programs to re:leh the poor (espeei:llly in rural :lreas), :lnd the cemral role of agriculture. U In sum, at least after 1974 and perhaps earlier, abertum included a social component. At bOllom, abert/Ira h:ld to mean more than JUS! politic:l1 liberali"::lIion, because none of the competing factions l'xpeclcd the military to divorce itself from politicallife. 1l The Castelis!:ls kllt·w they wanted to avoid military intervention in futur e governments, and they knew s(Xia l unrest encour:lged intervention. Unless liheralization had a social component, no politic:lI reiaxation could be secure. The desire to prevem socia l unrest implies little more than a short-term modification of economic and social policy while tht' 12. $("(C, 10< ~x:l mplt, hi. mt«ag~ at the openin~ of Ih~ c.::mgr~s in 1981. 1.1. Wald~r d. GOO, [lC'rhap' th~ do,""S! "b.erver 0/ th. G. i",1 adminim.ltiott .od the milituy ditc in &encrdl. ,,:.it~~ that, as ~ given . in the v;tal "'nlc<. of Ihc reg' me, Ihdl Ihe polit;c~ l . mood h~, not btco",~ totaUy ob""lct~ and that tho current c.i.i, is only a Crt'" 01 Ih • •y..em 01 allt,n"". 01 tho reS'n,.. A n.... ",heme 01 partic;pation-Ihw"w> new ,da""n~ bttw(en the Stale anc! dil ... _ anc! a rd",m of doc"ion mechan;,m, wo uld fu"h •• t .....""onSlrun;on of th. pol itica l p.ct and Ih. ,estO'd"O" 01 normality, ... hich ;s understood a. th, ~a""luj acccptdncc "I mil,t:lry·butrau,""i( domina,ion and of ,he di,~~ linkeJ to II. It is only a qu..,rion of maktng slot. corpuratism mOte flex;ble and mOre indus;v. (1978:1l9). It
; ~ I~kcn
coonomi~
,.,.,.
l~ -
Ie
When Soldkrs Need Friends
'49
withdrawal procee"de"d. Why might the" military want long-tenn influe"nce? As we have" seen, the" military was never idC'Ologically monolithic, and the inevitable" fragmentation accompanying the long years of military rule further reduced the unity of military thinking. Still, most officers were strongly anticommunist, proWestern, procapitaiism, and anti populist. 14 Military officers were intenr that a new civilian regime should not reverse the gains of the post-1964 period. They did not want to see the politicians they had overthrown return to power. They definitely did not want too much probing into the Brazilian version of the "diny war" against subversion. The officers rarely knew specifically what they wanted, but they feared a civilian regime could undo all they had achieved. To the degree they could induce the population to support conservative politicians, and to the degree the military commanded respect, civilian rule hcJd less danger.
Strategies of Social Abertura We come now to the four survival strategies available to the regime. Each hypothetical alternative implies a strategic coalition, and each coalition requires a distinct pattern of policy outpUts. The goal of all these survival strategies was "popularity" for the military; the strategies themselves answer the question: "Popularity with whom?'"
Hypothesis 1: Return to tile Barracks The Castelista faction controlling the regime after 1974 sought to ensure its influence in policy making after polirical liberalization. But influence over what? If military leaders simply wanted veto power over decisions affecting their professional interests, they would hardly need broad popularity. No one could forget the coup of 1964. A military with no long-term political interests would still face one problem. Political liberalization, it was said, could not withstand an economic crisis. A recession would hinder the political opening and prevent the military from carrying out its withdrawal. Social turmoil would compel a military presence, weakening the Castelistas and strengthening the hard-liners. Still, if the armed 14. These commcnts an: based on ronvcrs.ariolL5 with high·levd of6cns
~nd
obwrvcrs of the miliury.
""
1~",1
Ie
"'a
Two Applications of the Theory forces merely wanted a clean withdrawal, their objective could be minimal: Avoid economic collapse. This hypothesis, In essence, says the abertura really had no social content. Hypothesis 2: Appeal to Elites Legitimacy, we sometimes assume, is only altained when everyone in society regards the regime positively. Emmanuel Wallerstein suggests, however, that the masses aCtu311y maner very linle: It is doubtful if v~ry many govC'rnm~nts in human history hav~ bccn considered ~Iegitimate~ by th~ majority of th~ exploited, oppressed. and mistreated by their governmenu. Th~ masses may hi: resign~d to their fate, or sullenly restive, or amaud at [h ~ir temporary good formn~, or actively insubordinate. Rut governmentS tend to hi: endured. not appreciated or admired or loved or even supported (1974:143144).
If legitimacy required only the suppOrt of economic elites and region3l politicians, the regime had a number of o ptions. Most industrialists sought a decentralization of decision making and a reduction in government intervention in the economy (de-estatiZaf.10). Nongovernmental elites of all types, according to McDonough (1981), wanted the government to pay less attention to transnational corporations. Right-wing nationalists in the milita ry backed any program favoring Brazilian over foreign firms. The government could also shape traditional social programs to attract upper-status groups. The regime had alre3dy shown, in its post-1968 expansion of the university system, a willingness to accommodate the demands of the upper middle classes (Ames 1973), and since most of the central government's allocation for education already went to universities, elites would appreciate a general expansion of educational funding. In spite of [he seeming anractiveness of these policy options, 3n elite-centered Slrategy would be difficult to implement. Public health programs could not be made attractive to elites, because elites rarely used public health services. Housing projects in the National Housing Bank were 31ready overwhelmingly skewed toward upper-income groups. "Elite interests," moreover, were actually quite diverse. Manufacturers would oppose higher wages for worken, but commercial interests would welcome them. Firms
"
,
,
Wbm Soldins Nud Frimds
• S•
dependent on government purchases would protest against any reductions in the government's role in the economy. Nationalist programs might aid domestic producers, but goods purchased by clites would become more expensive.
Hypothesis 3: S~ek Mau Support If, on the other hand, the regime were to suk the support of the urban working and middle classes, it would have many policy options. It could spend more on health, welfare, and primary education. The Nalional Housing Bank could build low-cost dwellings, and social security programs could broaden their coverage. In terms of macroeconomic management, the regime could pursue income redistribution through wage policy rcfonns. Strategically, however, the prognammatic approach (Iargering specific government programs) was of dubious value. The regime could shin any number of expenditu re progra ms to benefit lowincome groups, but it is questionable how much support such redirections would produce. In 1965, in its fi rst postCOUp budget, the military regime had tried to win over the poor with increased spending on primary education. That effort failed-in part because the Ministry of Education possessed an extraordinary capacity to waste money (Ames 1973) and in part because the regime's leaders feared education might increase popular mobilization. Ten years later, when the hostility of the working class was much deeper, could the regime spend enough to buy support? Significant income redistribution, on the other hand, would certainly elicit a positive response. The problem with redistribution was macroeconomic: Would redistri bution inevitably lead to higher innation and unemployment?1J If the regime tried to go halfway, seeking merel y to prevent further declines in real wages, workers might not respond positively at all. Was a change in policy worth the risk? Hypothesis 4: Strengthen
Regional Support The regime always drew its strongest support from the conservative, predominantly agricultural states of the Northeast. These IS. This;, not 10 .rgue thlt inn~ rion Or ummployment wue r<:g.orded by mO!il economilts., i"",i"ble ronseqllcnca of hi&hcf waga . Inside tM regime, however, that linugc h~d been I f Ihe root of poIl·1964 wage policy.
"
,
Two ApplilaliQr15 of tht Theory
states, especially Bahia, Ceara, Maranhao. Parafba, Piau!, and Rio Grande do Norte, were far less populous (han those of the industrialized Southeast, but the malapportionment of Brazil's clccIOral system exaggerated their influence. '6 Three senators represented each state, and the apponionmcnt rules of the Chamber of Deputies produced a similar (though less pronounced) result. In the election of 1982, for examplr. the nine states of the Northe:lst plus four small states (Rondonia, Ac re, Maw Grosso, and MaIO Grosso do SuI ) had only 26 percent of the voters, but they daimcd 38 percent of all Chamber seats and 49 percent of PDS deputies. A regional survival strategy could target rural areas in all parts of the country, or it could target both urban and rural areas in a selected region. If the Strategy were purely rural, policies setting crop prices or agricultural credit terms would be cem ral Y If the targets included both rural and urban areas in a certain region, then programs such as road building or hOllsing could easily he m:wipubted for political objeClives. Suppose President Geisel had analyzed the results of the 1974 congressional election with the notion of using current budgetary trade-ofEs to maximize future electoral gains. Om' group of states could be written off. Rio de Janeiro, where the opposition MOB won sixty-five of ninety-four seats, and Siio Paulo, where the MOB took forty-five of seventy scats, were hopeless. More money in those states would yield few benefits. A second group of states was politically important but dectorally uncertain. Rio Grande do SuI had supported the opposition in 1974, giving the MDB thirtythree of fifty-six seats, but Rio Grande do Sui was the home base of many active political figures, both in the opposition and in the 16. Thei. influence wa. .1>0 ino, a supportive Stale, was spht into fWO, fhus increasing pwgovemment strength. Th= p.acri~s became w wen kno"o'Il in B.azil [hal C4Uui.mo (KcasuiSlry" ) Me>me an .v.ryday word. 17. Pritt sening provides a powerful moon_ of buying support. In 1980. Dclfim NelO, lhe minisler of planning and al thaI time Ihe le~ding civilian candidale for the p ... idency, incrcued me price the government paid lor sugar. Coincidmully, Dc:16m Neto used the <>ceuion to ask the growers to join Ih. government party. s.c., Jo"",1 do Rrll$il. June J. 1980, p. 1'1.
,
When Soldiers Need frit.'nds
'SJ
military regime, so it could never be written off. Economically important Minas Gerais gave the government thirty-seven of sixtyone seats, but Minas was surely no government bailiwick. Minas boasted an extremely active political elite, one that would respond quickly to signs of neglect (Fleischer 1973). A third group of states offered better long-term prospects. Northeastern states had smaller concentrations of blue· collar workers and lower levels of classbased opposition. Moreover, clientelistic relationships between landowners and their dependent tenants guaranteed a heavily proregime vote in rural areas. These factors point to the Northeast as the key to a regional strategy. BUI one caveat remained: Why invest resources where the lead is already secure? If ARENA elects thirty·seven out of fifty deputies in Bahia, thirty of forty in Ceara, and nineteen of twenty-four in Piaui, these states are safe. Why not put resources in politically important but elcclOrally marginal areas? A strategy that favors marginal cases, ignoring both unyielding opposition and unwavering support, runs into one major obstacle. For local elites, survival depends on securing government largesse for their regions. The greatest pressure on the regime came from states with the largest progovernment delegations. Because Bahia strongly supported the regime, its leaders expected rewards com· mensurate with that support. The regime, on the other hand, should treat Bahia as a "safe" region and devote its efforts elsewhere. Thus the regime's political interests were directly opposed 10 the claims upon it. II Recapitulation
Four distinct strategies of long-range survival Were available to Geisel and Figueiredo. It would be surprising, however, if they had implemented only one strategy, because few administrations can act as single-mindedly as the preceding hypotheses demand. If pressure from loyalist states was too strong to ignore, the n::gime might have opted for a region-wide approach, putting money everywhere in the Northe~st. It might have adopted multiple strat18, In an interesting pi«c on Mex ican presidems, Coleman and Wanat (1973 ) diocoveroo that one pre>idem appeared to be putting resource, wh.r. th. PRJ was weakest. whil. his sucr"nor rewarded those Slaies thai were strongly pro-PRI. Such uctical choiCfi would have Mf.n somewhat tasiu in Mexico, with its more insti _ tutionalized 3ulhorituianiim.
"
"
,,.
Two AppliCllti{}lfs of t~ Theory
cgies or evc:n inherently contradictory combinations. Combining a Northeast regiona l approach with social spending, fo r example.
would h;n'C been coumc:rproductivc:. Silla' most of the beneficiaries of social programs wert located in indusuial arcas. highe r social spending would have: inc:vitably reduced the: share of the budget going to the Northeast. Strategies are truly conlradiclOry o nl y in the long run. In th e shoft run, the irreconci lable can be rceon eilC'd through cxp:msion-that is, by do ing everything simullanoously and lening successors pay the cost. Sy 1983, as we shall $tt, the ttOnomic decline: was so Sleep thai th e con~ucnas of such c:xpansion were: ftll almost immediately, and the military regime had so degenerated that it could no lo nger make choiet·s. II could only Mpush with its stomach (empum,r com Q barrigaj-that is, postpom: wit hout deciding. M
Five Policies The five policies discussed here are not a random sample of the univerK of policy areas, (lOr do they include every important iss ue. They were chosen because IOgeth er they affect th e vital inte rests of all potential regime targets. Put another way, each regional or class targel cared inte nsely about at least one of these policies.
Agricultllre The No rtheast supported th e military regime, but its support was not the response of a conservatism based on wealth, for the Norlheast, especially its interior, is aClllal1y Brazil's pooreSt region. Concentration of landownership is extreme, agricultural productivity is very low, and per capita inco mes are less than one-fifth those of Ihe prosperous SOUlh and SOlltheast. As a resul t, migration is very high from the interior to cities in the Northeast and to o th er regions. The close relationship between the military regim e and NortheaSlern elites was like the relationship between right-wing regimes and backward regions in many parts of the world, but in Brazil a dominant ecological factor, the periodic drought sufferid by the arid serrao region, conditioned the linkage as well. Occurring about every seven years, these droughts led to a long history of dam conSlruction and river-valley development. In the late 19505,
Whtn Soldius Nerd Ffiends
'"
under the leadership of Northeastern economist Celso Furt:Jdo, region:ll pbnning took on :I more comprehensive tone with the creation of [he Superintendency fo r the Develupment of the Northeast (SUDENE). SUDENE's gO:lls included greater agricultural productivity, crop diversifica tion, and even colonization, but its main objective was industria lization. By the middle sixties, it was obvious that SUD ENE had failed. I[s resources were insufficient, its management was tOO influenced by the region's conservative oligarchy, it was unable to enforce compliance with its pbns, and its industrialization strategy was technically deficient. The industrial park of the Northeast had grown, but industry created small numbtn of fobs, developed few links to agriculture, and magnified intraregional disparities by locating new plants in the large cities. While successive Brazilian governments tried various schemes to integrate the Northeast into the nation, the structure of the national c<:onomy was changing. In the fifties and early sixties, Brazil's civilian administrations emphasized import-substitution industrialization at the expense of agricuhure. The military government dismantled many industry- promoting cont rols as pa rt of a general economic libtr:llization, but agriculture still needed compt'nsation for the distortions of food price controls, overvalued exchange rales, and export I:lxes. Agricultural policy :lfter 1964 responded to twO factors. One was poverty and the cyclical problem of drought. The olhermore a nalional than a regional problem-was the need to create an environment conducive to the expansion of capitalism in the countryside. The regi me's attempts to promote agricultural exports had led to the creation of firms producing inputs for modem agriculture. Unless farmers had access to credit to invest in new technologies, the new fi rms would have no customers (Muller 1982). The regime made many efforts to de:aJ with the twin problems of rural poverty and agricultural modernil.3fion." This section evaluates three such programs: a nationwide system of ru ral credit; a land reform scheme btgun after the drought o f 1970; and a program of integrated rural development projects.
.,6
Two Appliwtimrs .../ t/'r TIJcury
Rural Credit. The N:l1ional System of Rural C red it was cre:lIcd in 1965. Cornposcd of the Central Bank , th e Bank of Braz il :lod o th er official fin;!.n ci:!1 insti tutio ns, and th e p ri vate commercial b:ani.:s, the system was based on subsidi zed imcT!;'SI r:ltt:s. Until 1979, nominal inlc reSi r;lteS were fixed;1I ]J to 2 1 percent with no monetary correctio n- th at is, without adjustment for inflation. Sinct' inflation never descended to Ih:1I level, real imerest rates were a lwa ys ncginiyl.'. Commercial banks were n:uu ra ll y uncmhusiastic about lending to ru ral borrowers at such rales. so the Central Bank req uired the banks 10 reserve JO percent of thei r passhook deposits for fur al credit. Demand fo r rural credit was intense b ccaus~ r~ a l interest rates were quite favorable. From a le vel of - 46 percent in 1965, re31 rates moved as high (that is, became less f3vo rab le) a ~ - I percent in 1973. Bill in 1974, the first full year o f the oil shock, real interest rates declined to - 14.5 percent, reaching - 27 percent in 1977 :lUd - 69 percent in 19110. T he result was a SUbS\;>n\;al tra ns· fer of real income to agricu ltu re. Before 1974 such transfers were never mo re th3n 3 pereem of agricuhura l income, hili in 19741975 they averaged 9 percent 3nd between 1976 and 1980 the a verage was 16.6 percent fDa Mata 1982).zu By the late 1970s the credit system was comi ng under consider3ble aU3ck.11 The World Bank (198 1:23 ) offerro proof tha i credit was skewed towa rd 13rger loans and. b)' implication. loward larger farmers. In 1976, 3ccordin g to th e World Ba nk. 72 pc:rcent of all loans were below fifty minimum salaries in siu . Ev~n though these loans accounted fo . nearly three- fo urths o f all individual contracts. th~y absorbed o nly 11 p( rc~n t o f all credit . TI,e hugest loa ns. those over five hl',ldred mi ni mu m S31aries, accounted fo r only 4 percent of all loan contracts, but they 3bsorbed 52 percent of all credit. The lTend over th e course of the ~m iracic H loward lending to lO . "lVfcuhur,1 credit dttlincd simply in 1977
;1.~ ~
_ ull uf
d,~
dlum 01
T,..,~,u.,. Min i"tr Simon$<"n 10 r..,."":1 lhe mu~ supply U Infi.auon ~c(dt"lw. "flit ,..,~1 volulIK of ruul CRdll drd i...-d conllnuou~y in $u bu-qU ....1 YUr$, ~Ithough
by qui lt v~ri~blt ~mounfL 2 1. OM crilicism. ooting Ihal land prien tOSC I~sl"'! " 'hen ,..,~l lnltfac tAlt!! ..~,," ncptivc, suggactd thai l ub.idiud t rWII Itndtd 10 drivt up lhe prier ul brtd (Conudot 1975). Bm Rn.... dc ( t 9&lb ) . tbuncd Ihi, o;r;tlqtJC by :\ho" 'lng thai l~nd priCe!; tOSC d uring gtllCr.lJ «OIlOmic do,,·ntum. 'f"\Prd)«s 01 In!C'~! ~UbsKhulioo bK:tust land "'~s rtp.rdcd ~. a hedge- ~p,m;t «OnomIC dcd llle.
WImI Soldim Need FrimdJ
'57
big farmers was even more suiking. In 1969, the largest loans had represenu~d only about 20 percent of to[al credit and abom 1 percent of loan comracts. B)' 1976 such loans had grown fourfold in number and more Ihan doubled in volume. Brazilian officials also faced wong cridcisms of Ihe regional distribution of loans. According 10 the B:lnk of Brazil, the Northeast contribUied 22 percenl of the value of IOtal 1977 crop output but received only 12 percent of crop credits. By 1979, however, the situ:ltion had changed considerably. In documems (Sampaio and others 1980) and in interviews con· dUCled as part of this siudy, Brazilians reported thai the World B3nk h:ld pressured them 10 redireCl their efforts toward sm:ll1er farmers and toward the Northeast. Loans to "mini" and small producers, loans that had accounted fo r 39 percent of contracts and 14 percent of credit in 1977, now accounted for 71 percent of contraClS and 26 percent of credit volume. In ilS 1981 report on Brazilian agricultural policies, the World Bank noted approvingly that Ihe share of total crop credit going to the Northeast had risen from 12 to 14 percent between 1977 and 1979. Table 29 allows a more del ailed examination of Ihe regionalization of credit. In both the Northeast and North, more farmers received loans. It happens, however, that the redistribution of credit began bero,t the World Bank applied pressure. For Amazonas, Pad, and M:lTanhiio, 1975 was the first year of uninterrupted gain in Iheir shares o f credil. For Piaui .. nd Rio Grande do Norte, the improvement began in 1974. Paraib.. , Alagoas. and Bahia began their climb in 1973. while uani started to rtceive more in 1976. The loser in this redistri bution was Sao Paulo, whose share of lotal credit declin~d from 27 to 19 perccnt between 1974 and 1981. Did Ih~ redistribution of total credit put money in the hands of more farmers. or did big farmers simply get bigger loans? Smaller fa rmers did benefit . but their rewards came laler. Of the ten states in the North and Northeast showing gains in total credit. the lag befort a higher percenlage of fa rmers began to benefit averaged almost four years. In other words: The regime began shifting credit to the Northeast around the beginning of the Geisel administration, but uOlil Figueiredo's presidency credit failed to reach more. small farmers. Can we conclude that Geisel sought to recruit larger land-
h
1
,.,
Tab le 29. SUlf'
A ...,UOIIaJ Valli( Num~'
L OOII$
G rllllltd
10
Produu TS
f/lld
Cooptrlllillt S by the Ru,al Credit System : 1970- 1981
.1170
11171
11172
. 1I7J
1\17~
1975
1976
11177
1978
1979
198 0
1981
0.11 0.18
0.18 O.J O
0.27
0.2 1 0.22
(0.1S1 (0.161
0. 17 0. 1'
0.28 0.28
O.H 0.37
0. ~5
o.~o
0.45
0.62 0.7 1
1.0J 0."
0.64 0.64
0.62 0.28
0.70
1.0J 0.75
0.96 0. 79
10.671
0.48
o.n
0.88 (0.721
1.02 0.76
1.10 0.8J
1.26 0.90
1.40 1.1 7
1.33 1.44
1.1 2 1.42
0.38 0.62
0.14
O.H
0.57
0.97
O.SS
10.401 0.88
0.61 0.9S
0.59 (0.861
0.67 I.!S
0.7 1 1.28
0.9J 1.89
LS5 3.JS
1.J9 3.65
0.)0 0.98
0.49 1.4'
o.n
10.281 1.02
0.J5 1.08
0.4 9 1.20
0.48 1.22
0.«> (1.111
0.52 1.4 I
0.7J 1.83
0.57 2.6 1
1.10 1.94
Pa'd V~J...,
Numbe r Ma,.."h~
Valuf Number
P"'''''
Valuf Number Ua," Val"" Number
o.ao
1.43
1.48
1.55
1.4J
1.76
2.09
2.42
l .52
2. H
12·341
2..94
2.3 9 1.97
2.45
l .1Il
1.78 2.76
1.85
3. 1~
1.49 2.65
( 1.301
1.68
0.35 0.57
0.62 I . I~
0. 60 1.03
10.571 0.78
0.59 0.98
0.68 0.93
0.76 0. 98
0.8 7 1.0 1
0.93 (0.981
0.85 1.09
1.02 1.29
1.19 1.86
0.74 1.15
1. 19 1.77
(0.8lj 1.68
0.86 1.66
0.9J 1.96
0.'.12 \.66
1.05 1.85
1.08 I.S 4
1.01 (1.651
1.0 1 1.75
I. t l
I.S3
1. 15 2.49
2.93 2.27
J.72
3.48 1.SJ
J .0 3
2.H l .H
2.45
2.48 l . 15
2. 3 1
(l .071
2.2J 2.I S
2.28
3.46
I."
~.7J
Rio G,,.,,dt
tk
NO.tf
VaJuf Number
Pa,,,,b,, ValUf Number
Prmam bu(o
,
VaJuf N umber
2.14
1.02
2.55
2.76 2.93
l .93 4. 15
.60
Two ApplicatioY/$ of the Thro ry
owners while Figueiredo appealed to Ihe rural middle class and the poor? Regardless of his political objectives, Geisel knew that if the South was to become the source of Brazil's export agriculture, the NortheaSI would have to playa bigger role in domestic food production. Small farmers, however, are difficult 10 reach with loans. Frequently they lack dear land tides, and their economic viabiliry may be limited. Moreover, until [he B:lnk of Brazil built its "advanced posts" in the back country, credit could not be physically put in the hands of small and medium farmcrs Y Furthermore, the World Bank- an institution with considerable leverage-undoubtedly had an impaci on policymakers. In the end, rural credit flows responded more to technical and external constraints than to political strategy. Finally, is there any evidence that credit flows rdleCled the disproportionate political power of small states? Figure 10 traces the interstate distributions of credit volume and individual COntraClS. Both distriburions became more equal, but the number of contracts per state evened out much more. States with small numbers of farmers must therefore have received more credit per fanner. In this caSl' the political importanc!.' of small, rural siaies helped them artraCl a greater subsidy for their farm populations. Land Reform. Though elile fears of a radical agrarian reform connibuted to the 1964 coup, the concept of land reform remained attractive to the military government, at least symbolically, as a tool for rural modernization. Certainly a well-designed agrarian reform could increase both produClion capacity and rural incomes. Even so, the r!.'gime did practically nothing during its first six years in power. Bur in 1970 a drought hit the Northeast, and after a trip to inspect the affeCled region, President Medici created the Program for Redistriburion of Land and Support for Agroindustry in the Northeast (PROTERRA). PROTERRA's obieClives included "the promotion of better access to land, the creation of better opportunities for employment 22. Mo.wv~ •• spm~l loan programs ....·e.e ",t up /0' the pu.ch.", 0/ inpu", bUI th~y inc,""as.ffi cf~dit conc~nlfation in crop. such a. suga., a h~. vy use. of fe.lili te. hUI • crop inn r.si ngly gtoW" in large. Un;IS. Similarly, when auernptS were mad~ in the cody 1~7o. to ofiert erroil toward employmrnt-gcncrating aCtiviti ", (~s a r... ult of the drought 0 / 1~70), erNit shi/ted to wa.d li"TStock, ~.n · lu~lIy ,"u sing grtat~1 la nd concentratio n. Nun]x,rg (1~86).
s.-.
"
"
I I I / / / /
I I I I
,I \ \
\
I I
I I I I I
I I I I
\ \
\ I I / / /
.6,
Two Applications of Ihe Theory
of unskilled labor, and the development of agro-industry in the
Northeast." Specifically, PROTERRA could: Acquire or expropriate (with indemnific:uion) bnd required for the public good. This land could be sold to small and medium rural producers to achieve better distribution of culti"able land. Make loans to small and medium rural producers for the acquisition of farmable land and the expa nsion of holdings to a profitable size.
Finance projects destined for the expansion of agro-industry, including sugar and the production of inputs for agriculture. Provide financial assistance to organize and modernize rural properties and to organize and broaden agricuhural research, warehousing, transport, electric energy, and so forth. Provide subsidies for the use of modern inputs and guarantee minimum prices for expo rts. The accomplishments of PROTERRA in land reform are besl exhibited by disaggregating ils budgets. FUNTERRA, the subprogram responsible for land acquisition, was budgeted in 1972 at 2.6 percent of the PROTERRA budget. In 1973 FUNTERRA's budget climbed to 18.7 pereen! of the total, and in 1974 it received 15.8 percent. In 1975, however, land acquisition programs absorbed only 7.9 percent of the total, and they dropped to 3.5 percent and 3.0 percent in 1976 and 1977. By 1978 FUNTERRA had almost disappeared, attracting only 0.8 percent of the PROTERRA total. Funding for agro-industrial projects, by contrast, grew to almost three-fourths of PROTERRA's total budgelY The failure of land acquisition programs was 31so reflected in PROTERRA's system of implementation. Land acquisi ti o n and expropriation were placed under the command of the National institute of Agrarian Reform (INCRA). INCRA's initial regul atiom 23. The ctwit Jines of Ihe Bank of Ihe NortheaSI provide a pJrJnd leS! of the i ..... plementation of bnd ..dorm. Lind "cquisitions received 3 1"""'''1 01 :In [>RO-
TERRA crwi! in ]972 and SA percenl in 1973 but only 0.35 percent of all PRO· TERRA credit in \974. ]n lal(( yurs no loans were mde for land uquililions at all. The lending of the Bank of Brazil also illustrate> the powerful (o"«nltation of PROTERRA capital ~hind 3gm·industry. l\erv;""n I ':17 1 and 1':178, morc than 97 percent of the bank's loans wem fot land acqui~ilion, but morc than 62 percell! of the loaned funds wenl to agro-industry.
r
.... '
l~
u.
Whm
Soldi~s
Nud Frimds
'"
exempted farms of less than a thousand hectares from expropriation. Larger landowners who submined "lenus of adherence" (0 the program became responsible for its implementation. The landowners themselves would dete rmi ne which portions of their land would be bought by INCRA and which tena nts would receive land. Moreover, adhering landowners received extremely favorable compensation for the expenses of relocation and resettlement. In August 1974, when it was dear that nOl hing was happening, INCRA assumed sole control of the project, giving the landowners thirty days to accept INCRA's proposals or face immediate expropriation. INC RA's proposals, however, were simply grants of credit to the recipients of individual parcels, and INCRA compensated landowners generously for quite marginal pieces of land. Very little redistribution ever took place. By 1975 I'ROTERRA credit had been used (0 purchase only about 1.5 percent of the (Otal farm area in the "priority zone" of Pernambuco. No expropriations had occurred. Only 840 families had settled on acquired land, and after 1975 the number of new owners declined steadily (Carvalho 1982). Why was PROTERRA so unsuccessful? PROTERRA's land acquisition policy was barely an improvement over mere credit. Land recipients had (0 pay for their land at market prices, fo rmer landowners were well compensated, and the parcels of land were tOO large. 1 ' PROTERRA was also administratively weak. At the inception of the program INCRA relied on the landowners themselves. When that failed, it lacked the personnel to do the work itself. Most importa nt, however, was the absence of political will. President Medici may have believed in PROTERRA, but economic CLafS Ddfim Ne(O, Reis Vdloso, and Simonsen did not. According t o informants, Ddfim New had told the World Bank, which supported agrarian rdorm in the Northeast, that he had nothing against land reform. Since the World Bank had the money, Ihe World Bank could pay fo r redistribution. But acqui6Cence was not support. PROTERRA needed active backing, because the land24. INCRA w~nttd to ~void tht CTunon of a mm'fundiJtll dau, 50 ir man · duro minimum , ius fot I~nd par",b. That p~r,(ls hid to nfnge ISO h«t~r<s in CUr3 and 60 hecrar« in Ptrn~mbuco. Pcoplt living 01\ su,h potuls wllo Were not I~nd recipients n~ fU rally lWf<: unhappy with the ntW owneR and with tNCRA, .nd rhei r reloc.arion became J Source of tension.
"
,6,
Two Applications of the Theory
owners, for whom redistribUlion embodied the specter of radic:JI change, were the mainstay of the SI3te oligarchies.
Rural Development. By the middle 1970s, the context of rural policy had changed. Rural development specialists had 3(:quircd much greater technical expertise, and the World Bank had begun pursuing Roben McNam:lra's "Basic Needs" Str;ucgy. With PROTERRA an acknowledged failure, it was time for a new beginning. In late 1973 the central government created an imerministerial commission to develop a program of "integrated rural dC\'clopment" for the Northeast. Integrated projects would combint: infrastructure. extension, marketing, land title assistance, and social projects. After McNamara's famous Nairobi speech stressing basic needs. and after a number of World Bank rural development missions had visited Brazil, the World Bank began to receive funding requests. One of the earliest, coming from the Brazilian state of Rio Grande do Norte. invited the bank to evaluate an ongoing project called Serra de Mel. The bank eventually rejected Serra de Mel but offered to consider new requests. Projects initiated by other states followed. first in Ceara and then elsewhere in the Northeast. At the end of October 1974, these incipient projects were brought under the control of the central government by the creation of POLONOROESTE. Officially, POLO NOROESTE sought to promote the "development and modernization of agricultural and livestock activities in priority areas of the Northeast." POLONORDESTE would include feeder roads, rural electrification, crop storage, agricultural research and extension, and rural credit. These acti\'ities were to be concentrated in specific areas called Integrated Rural Development ProjectS (PORls) . Ultimately there were forty-three projects, nine supported by the World Bank. Every Northeastern state had at least one, and between 8 and 9 million people were included as project beneficiaries. Funding for POLO NORDESTE projects never reached the levels originally expected. After peaking at almost US $2 billion, Brazilian funding declined 8.7 percent (i n real terms) in 1979-1980 and another 12 percent in 1980-198\. According to World Bank informants. funding levels would have declined even more after the credit crunch of 1977, but the bank threatened to suspend its own disbursements unless the Brazilians maintained their contributions.
"
"
When
Soldi~s
,6,
Need Friends
Did funding decline because of the change in presidents? Infonnants from the World Bank admitted they worried when Figueiredo replaced Geisel, but they anributed me slide in POLONORDESTE's funding to the growing economic crisis rather than a deliberate attempt to sidetrack the program. At lower levels of policy making, support for me program had actually increased because many state-level agriculrutal experts who were enthusiastic about POLONORDESTE had moved up 10 federal jobs when Figueiredo assumed the presidency. Recapitulation. Clearly redistribution had little support. PROTERRA's only serious backing came from the World Bank, but even the bank realized that significant agrarian reform was a dream. An agrarian reform would be financially practical in Brazil only if it deferred compensation or simply confiscated land. Since the landed elite would never accept deferred compensation, no consensual agrarian reform was possible. A confrontational policy was simply inconceivable-a regime backed for ten years by the upper strata was hardly likely to change horses in mid-f1ood. In the end, agricultural credit replaced the zero-sum politics of PROTERRA with an expanding-sum game. Conscious of the need to increase production of food crops, both post-1974 administrations shifted credit to the Nonheasl. Geisel backed bigger farmers, while a slight shift toward smaller farmers was evident under Fi· gueiredo. Whether this shift came from a deliberate search for allies is uncertain, because the technical and external constraints affecting lending were at least as important as political motivations. If the regime had a long-range survival strategy targeting the poor in the rural Northeast, the best evidence might be POLO· NORDESTE. In the economic crisis POLO NORDESTE suffered but endured. But without the money and encouragement of the World Bank, and withom the emhusiasm and technical skill of agricultural experts in the states, POLONORDESTE would have died, because the political support of high-level policymakers was quite weakY POLO NORDESTE might produce support for the 25. The bureau cratic infighting thai hampered implementation of POLO· NORDESTE was a sign of Ihe absence of support in Bra.ilia. Fonnal responsibility fnr POlONORDESTE was loe,neO U Ihe federal level, in a coordinuing council consisling of rcprt$Cma!ive~ from Ihe Planning Ministry (SEPLAN), Ihe Minismes
,.,.,-
l~ -
.
Ie
.66
Two Appliwtio1ls of the Theory
regime in the future, but its development was apparently nO( the consequence of a deliberate search for a rural base. Housing
At the time of the 1964 coup, the Brazilian housing industry was near collapse. Fixed interest rates combined with rapidly climbing inflation 10 produce negative real rates of return on ca p-
ital applied in housing. The social security pension funds were almost the only source of finance available 10 the middle class and the upper working class, but they built few dwellings, and the high degree of subsidization implicit in the system made it subject to political favoritism. The deterioration of the housing inclusuy could not continue. Housing was too imponant-nOI JUSt to construction companies and their employees hut to the regime's middle-class supporters and to the leadership itself_ Brazil's extraordinary urbanization in
the fifties and early sixties had left the country short millions of units of housing, and somt key figures in the regime believed the mushrooming squaner settlements were potential hotbeds of social upheaval and political radicalism (Ames 1973: 16-26). If the government wanted to eradicate squaner senlements, it would have to house their residents. The regime also sought to increase its support among the urban middle and working classes. In a letter to Presidem Castello 01 Agrirolrur~ and lnt~r;or, and the unrral Rank. A< th~ rq;;onal l..... tI POLO· NORDESn projects were comrolled by SUD ENE. II was originally rxpected that SUDENE would ~ a monitoring and coordinating agency between th~ Itat"" and the govemmem in 8 .... j)ia. The lancr, how ..... er, would nor yield its financial control to SUDENE. Rather than trandcrring a block of money to SUDENE for the whole of POLONORDE.Sn, the government in Bra.i1ia parceled out tho money Stale by stale. The Wotld Bank followed suit, refusing to give SUDENE control of ito block grant. A. Ihe slares developed mOre experti~, they ~n 10 bypass SUD ENE and deal dir.cdy wilh Ih. CC'"Olral government. Unable 10 provide logis· tical support to Ihe states and unwilling to relinquish comrol over Iheir acliv ities, SUOENE became a bottkn«k in the program. In the urly 19~0s. a debate took pbce over the relative emphasis on social vu,;u< inlustructuul components in individual POLONORDE.Sn projttlS. According to World Bank informanto, this debale was nJOinly carried on inside th. bank and among the Siale agliculrurll experts nother Ihan 31 the level of Ihe central administralion . llras~ia may h3'·e linkcd POLONORDESTE 10 polcnlial politica! supporters in ,he northuSl, bUI ito ConCern did nor
....'
l~
u.
Whm So/dim Nud frimtb
'"
Branco, Sandra Cavalcanti, the first president o f the new National Housing Bank, wrote that: The masses a~ dnrinne and hun , and we will have 10 make an efion 10 give them $Orne pleuure. I think tha t a solution to the: problems of meller ... will have a mollifying and JOOthing effect on these civic wounds (Auvedo and Gama de Andr.ade 1981:108). In a simil3r vein, Ro~no Campos, perhaps the leading civilian in the early post-1964 leadership, argued that: Ownership of housing contributes much more than ~nlal propeny 10 social stability. 1be owner of a private dwelling thinb twice bdo~ getting involved in strttC br.awls Of pillaging other peopk" property and becomes an all y of order (AlC"Yedo and Ga ma de Andrade 198 1:109).
If a restored housing industry was a necessity for the regi me, its challenge was the reconstruction of the sector on a footing congruent with its overall ~onomic program. Housing constru ction would have to offer positive rates of return without large subsidies. The regime's solution was the creatiOn of the National Housing Bank (BNH) a nd the Housing Finance System. All tra nsactions in this system were subject to quanerly monetary correction, assu ring positive rates o f retum.U The acti vities of the BNH encompasstd. a broad range of urban projects, includin g finance for dwellings, sanitation, site improvements, and transporta tion. Progra ms directly aiding the urba n poor we: re o rganized un der the bank's Social Operatio ns Portfolio. Under this program the bank built some residences directly, while o thers we:re built by the bank-financed housing companies '(COHABs) o rgan ized at the state and local level. The Sodal Operations Portfolio was desisned to construe! housing fo r fa milies earning fro m one to th re:c minimum salaries, 26. When the fi,.t budgctary allocation of the RNH proved in adequale, il was given the U~ of the Stnioril}' Sccuriry Fund. or FeTS (FundQ d.: Garant;a do Tempo tit: Sct-vi~o). This fund, Rnanced by deposits of 8 ~tcem of each employcc's wagCl, had been created 10 provide employers Ihe frccdom 10 fire wo.kel"$ by giving the lam r the right to lump·sum payments upon retircIMnI, dismisul, 01 illness. The FGTS funds grew rapidly-so rapidly, in faer, lhat by 1972 their net inereue amounted to I percent of the GOP (Ca~nt" and Reynolds 1975). BNH funds also calM from volumary s.avinga Ihrough Ihe pusbook accounts of thrift inlliNIK>ns. By 1972 theiC OUlSlrippro the FGTS and became the largest &O!Jrce oJ BNH income.
,68
Two Applica tions of thf ThfOry
or US $65 10 $195 ptr month in 1980 dollars. With twenty-year financing (rather than the usual len or fifteen ) and interest of 4 to 6 percent plus monetary correction, the borrowing conditions were much hctu:r than those governing middle and upper-middleclass units. By 1970 the COHABs had financed over 170,000 dwellings, more than 40 percent of all units developed by the H ousing Finance System (Azevedo and Gama de Andrade 1981 ), The National Ho using Ba nk's low-inco me programs slaried promisingly, but they soon entered a period of decay. While the GNP soared and middle-income housing boomed, low-income programs collapsed. Between 1970 and 1974 the govt:rnment constructed fewer than 77,000 "popular" seelOr units, a decline of 57 percent fro m the IOtal reached hetween 1964 and 1969. As a percentage of all BNH progra ms, low-income ho usi ng declined from 40 10 12 percent. With aberwra, however, came revitalization. From a low in 1974 of 7,83 1 uniTs, low-income housing grew rapid ly, peaking in 1980 at almost 200,000. The rise, as Figure I I illustrates, was steady and almost unbroken. Over the whole period, then, low-income ho using foll owed a cycle: expansion under the relativel y liheral Castello Branco government, near-total decl ine during the repression-cu m-growth of Mi dici, restoration during the gradual democratization of Geisel and Figueiredo. Wh y did these programs wane so quickly in 1970- 1974 and recover so dramatically afterward? Officially, popular housing collapsed because an extremely high percentage of its residents fell behind in their payments. By December 1973 more than 36 percent were at least three months behind, and the percentage of vaca nt units reached 6.7 percent (BNH 1979). Medici apologists blamed poor tenant selection, bad building sites, and low-quality materials. These 3re valid reasons, but the underlying problem was simpler: The wa ges of poor Brazilians were rising more slowly than their house payments." The BNH was having difficulty recouping its costs, and since it was not in business to subsidi ze the poor, low-income programs had to be reduced. The posl- 1974 resuscitation of low-income housing followed a series of technical improvements in the program. Payments we re 27. II Itas b«n
~[i m~[ed
lhal
~«n Oe~mkr
1970 and
~~mkr
[974
Ihe number of hours of [ ~bQr required 10 urn Ih. min imum food ration 1'Qse from [03 10 158. ~e DlfESf (19 75).
"
"
.6,
Whm Soldier5 Need Friends
" "38
" " 3.' ,., 32
" " ",., e8
e,
"..,
..,
' .8
"
" ",~~~~~~~~~~~~ _'"
~
'0
~
_'"
'!I ..O .. '
'l- '!>
~
.,
\:)
'\
~
q
0.'
~~~~~~~~~~~~~~~~~'!I
""""''''''''''''''' SOURCE: BNH (1981). 1969 .. 1. Author's estimate tor 1981. Figure
I I.
Index of Loans Approved by Social Operations Portfolio: 1964-1981
lowered by extending the mortgage term from twemy to twemyfive years and by reducing the imerest rate three percentage points. COHABs were also given the right 10 construct and finance housing for a wealthier section of the lower middle class, those earning three 10 five minimums. U With these changes the vacancy rateIS. Ikcau.e .peculation had driven up housing prices, such buyers kc,a"", mud! rna,", ;nt~=tcd in low-income dwelling •. In a comparison of projects com-
'7 0
Two Applicati()lu of the Theory
6.7 percent al the end of 1973-dropped to 0.3 percent in September 1978, and the percentage of residenls at least three months behind in payments-36.3 percent in 1973-droppcd to 12.6 percent. Did these changes in housing programs reAcet a class strategy or a regional str;ltegy? If the regime was appealing directly to actual or potential housing recipients, the interstate distribution of government housing per urban dweller should have hecome more
equal. n If the regime's str:ltcgisTs were assuming that each state had equal importance as a target, then the distribution between states, regardless of the states' urban populations, should have become more equal. In Brazil such a regional strategy would ino rdinately favor the Northeast, because its states are mostly smaller and more rural. Figure 12 plots the distribution of COHAB units per srale, both in absolute terms and in relation to the percentage of Brazil's urban population TC'siding in the stat.... ~o Both distributions be<;ome more even. Because the distribution prior to Gdsel's administra· tion was heavi ly skewed toward Sao Paulo and Rio, putting money in Northeastern cities resulted in a distribution that was more even both in regional and in class terms. Figure 13 traces the distribution of all BNH investments (sewage, water, electrification, and the like). Here the picture changes. The distribution per state evens out dramatically during the aber· tUTa, but the distribution per urban dweller docs not. In the pro· vision of infrastructure, therefore, the regime appealed to states as states rather than to urban dwellers as individuals. In effect, the regime was appealing to political elites and commercial interests in conservative regions. pl~t~d
in Minas ~rai. in 1968 and 1978, Auvwo and Gama d~ Andrade (1981: 121) found IhaT 94 percenT 011968 borrowers earnw Une tu Ihree minimum ubrics; only 40 percent of 1978 borrowe,," were in this <::Itegory. Those e~rning thr~ to live minimum. rose from 4 to 5 I per<;'rnt. Richer borrowe .. were I.... lik.ly 10 be in Urearl on Iheir p.ymcnl<- 01 cuurse, so th hank would r..:oup ilS loao •. 29. nu. ilIaTcgy nsum ... that regardl .... of Ih. ">Ie of .... idencc, urban dwdJ· en re.pond equally to eQual amounts of hou~ing, so the units provi
"
"
.,.
Whm So/dim Nttd FrimdJ
.-"
~
" "" '""
, I , • ,, ,I
."
I ,
,,
•~~ " "
" ~ " ~
".
I I
.ji ','
~ ~ ~ , ~ :
I
, I
I I
, , '
, , , ," , , ,, ,
I I
,
I I
,
f
"
,
)
,, , \
~
\,,
, \ ,\
~
~
f
L _ _"
to'
-'\
,, '- _...
-
By state By state percentage 01 urban population SOURCE: BNH (1981)
Figure
1
t.
Distribution of COHAB Housing per Siaic and per Urban Dwclkr: 1967- 1982
While the Geisel and Figueiredo administrations sought to cnsure the influ ence of the mililary in the coming pluralist regime, some of their lower-level officia ls had much more personal survival straltgies. Mario A ndreaz~ sought the presidential no mination of the government party (the Democratic Social party, or PDS) from the time he became minister of the interior in 1979 un til he fina lly lost the nomi nation in September 1984. In an attempt (0 gamer
Two App/i(lItions of tilt Theory
'7'
" "
"
•
."
~!!1
•• -.0' ~n~
.0° !!::oc 3;;"
"'":i =. .. ~!O · 0 0
z.
-' o· z
8
7
0.8
•g'O!! t
0 .7
";;:;:: E "J
., .. ;; 0 ocr.,,,,
0.6
..,111"'21 ~ii'=-
0 .5
:;~ii'ii
'C-J:~ c lll ;;!: _
!!oro
"
6
& OA
SOURCE: leGE ( 1970-1981)
figure
I}.
Distribution of BNH InvCSlmCnlS: 1970- 198 1
support from the governors of tbe Nonhcast, Andreall.3 \fied to put a BNH proj('Ct in every Nonhc:lm'rn municipality. He kill'\.\' that the Northeaste rn governo rs wo uld co mmand small but unifil'd delegations 3t the PDS nomi nating ..:onvenlion, and bt'c:\use the PDS itsdf was disproportionately No rtheastern, these delega tions
When Soldiers Need friends
would heavily influence the convention. Andreazza lost the PDS nomination, but without this strategy he would not have been a viable candidate. His ambitions also produced, without question, an increase in housing for the poor. Trade·offs AmOltg the Ministries If indeed Generals Geist!! and Figueirt!do were attempting to ensure the long-term survival of military influence, we would cer· tainly expect changes in allocations to the central government ministties.l' Our fir~t step is to examine actual expenditures. Table 30 presents the share of the budget spent by each ministry during each administf:ation. Election years are separated in order to assess shorHerm electoral effects. H Both presidents reduced the budget share of the three military ministries and the Ministry of Transport.)J Both presidents increased the shares going to agriculture, education, and welfare. Health programs grew under Geisel, while housing programs (in the Interior Ministry) expanded under Figueiredo. H The "deviation" columns in Table 30 measure the changes in expenditures during the election year from each administration's 31. In order to uncover policym~kers' ""rceptions of the li nkage b"two<:n u· penditurn and b"Mficiaries. I carried OUt a series of interviews in the Pbnning Secretarial, ,he civilian policy staff of Ihe presid."t (Ihe Casa Ci"i l), and in nine mini.tri ••. Th~ interviews wert co nducted in Ihe summer of 1981 and b"rwttlL March and Augu", 1983. Wheth.r r•• pondrnts Were technocratic or political in orientation, OT high or low in Tank, they thought certain program s ~id .""cific classes. Hnlth ~nd .:mitation b"nefit t..... urban poor. Agricultural programs aid fanners, naturally, csl""ci. lly prosperous farmers. Welfare helps Ihe urban middlt and working classes. The dfects of cduCltional programs del""nd on ' ..... iT levd: The higher the grade. the higher th • .social da .. of the b.."eficia'l'. Public work.. proiects provide corulruction iobs, bUI most policymakcts s~w Ihc$( programs as lesponses to inf•• structural nttds and to the dout of local business and political leaders. 32. No.e .h", Ihe 19H2 dec';"n wa. guberna.orial •• welt as legi.luiv•. 33. The decline in spending on tr~nsportalion had been ca lled for by the So<:" ond Nation.1 Development Pl an. issued in 1974 alteT Ihe oi l shock. Transpon w's still the I3rgnt single mini.r'l'. 34. A complcte evaluation of rrgional and d.ss eXl""ndirure strategies would require ~n analysis of cx""ndiNres for nch program in each state. Official gOY" ernment repons have b..gun to cross-tabul'le programs wilh statoo, but th"SC' reports va.dy overtsrimale the amounl of sl""nding in thc fed.ral capilal. The program lotal•• re Ihus aCCUratC, but the SflIte total. arc nor. The Gcn11io Vargas Foundation, a SC'mipllblic Brazilian reSC'uch innifUtion, i. currently eng.>gcd in an accurate rcgionaliution of expenditures by program al five-yur intervals, but only the rurs t970 and 1975 haye been compleled. The limited dua that are available. however, arr sufficient to provide a sen.., of the rrgime's .t"'logy.
1~",1
Ie
a
..
00 , •, -, , 00 , + + ... , ,
~.~ ~ "....... ""
~o::;~:t
~
..;~
~
~
,
.., ~g
:g
'1l
'" .5
.;
.
.,
~
0.., "'..,.. ... 0 '"10 0 , 0 , 0 , 0 ,0 + 0, 0.. .... 00 .., '" '" '" '"
-~
,
~;!
...
~
.... · " .... ..,0...., _00 _
.E .5
'"
.
- - .., .... 01-8
" ' 0 .... ..,
"'!"l"":"!
:;::~ ~i t:;!; c:i..,:..:..;~..:
."
When Soldiers Need Friends
trend.J~ Figueiredo's election year deviations were much greater than those of Geisel; indeed, the average difference between predicted and actual shares in 1982 is more than three times the same difference in 1978.j~ Different programs also deviated in each administration. Geisel boosted agriculture and welfare in the election year and made extra cuts in the already declining military ministries. , J7 These shifts fit a strategy, centered on agricultural aid to Northeastern farmers and welfare for the urban middle and working classes. By contrast, Figueiredo's election year deviations seem confused. His adminisnation had emphasized social programs, but in 1982 he switched toward infrastructure (transport and communications) and away from education, housing, health, and welfare. Well before the election, however, the economy had begun to deteriorate, and economic conditions were the paramount issue of the election. Public works spending creates private sector jobs that can be targeted and controlled by local elites. Social programs disproportionately benefit urban groups unlikely to vote for the government party in any event. Figueiredo's clection year expenditures reflect more than just a short-term survival strategy. The Army Ministry, profiting from the insecurity of an administration faced with restive officers and strident opposition politicians, managed to reverse the fall of its budget share. The army's gain, along with the increased volatility in ministerial spending in 1982 (as compared to 1978), is our first indication of Figueiredo's loss of control of his administration. The programmatic shifts of the Geisel and Figueiredo years represent a broadening of the regime's social base, but regional shifts are less dear. Table 31 shows that central government expendi-
35. Rates of change were calculated for each program - lim iM:twan 19711973 and 1975_ 1977 and then iM:twan 1975_1977 and 1979_1981. The first rate of change wa. e>lt.apolatro to 1978 and the second rale of change was e>Itrapola~d to 1982. The differcnces between the~ predj~ions and the actual share'S for earn progr:am durinS the dn:rion yurs were then ",lculaled. The dection of 1974 was nol included becau~ Gei~l had little cOll1rol oycr spending during his finl y~r in office. 36. The muns of the absolu~ values of Ihe diffnences equal 0.41 for Geisel and 1.34 for Figueiredo. 37. Gei~l's conlidence in his military suppan mUSI have incrusN afrer 19n, when he remoyed wenl Frat:o, a dissidml memiM:r of Ihe high command. See GOes (1978). Why the txpendiNtes of the Interior Ministry suffered is uncertain. Housing progr:arm arc one componenl of ln~rior, bUI ,ina: the National Housing Bank itself Stew in 1978, it may be thai the nonhousing componenlS of Ihe InICI;OI Minislry bott most of the shon-Ienn reduction.
1~",1
Ie
a
",
Two Appircalicms of ,h,.
TI~ory
lures going to the Northeast declined during the adrninisU3tio ns of both Geisel and Figueiredo. These reduct io ns, however, might simply have been by-products of the increases in social programs, because social programs weigh morc hea vil y in expenditures into the industrialized Southca st. j~ If a regional STrategy was pursued. it might sho w up in the movements of direct transfers 10 st ate go vC'rnrn ents. Altho ugh some transfer payments arc subject to long-Ierm arrangementS and thus cannot easily be manipu13tcd from year 10 year, transfers also have a "controllable" component that includes grams in such areas as management of health care facilities, support of univc rsitits. and provision of agricultural scrviccs .J~ Overall, l a rg~' in· creases in transfers benefited the Northeast during the Figueiredo government, and in 1982 (;an cleo;tio n year) the steady upwMd trend rulminated in the delivery to the Northeast of ;almost half of alluans(ers 10 slates (Rclende 1982a). In sum, then, in the industri;al Southeast Figuciredo chose to suess direct programs like education a nd he!l.lth, but in the North· cast the real objects of political recruitment were state and local elites. In that clientelistic realm, state a nd local clites were patrons 10 program beneficia ries but clients of the central adminiSlr;ation. 40
Proposed vs. Acwal Budgets. After 1981 the "dceomposition of the Figueiredo adminisu a tion became a common theme in the M
38. Wh(lher lhe rqoonal shim W C(( 1("~lIr an anifaCl of progl~ m (h~"gu cannor be okfiniri"dy anSWnN unrilthe completion of the 1 ~80 region~liurion projeCI. Uling the ~. li ... rioM .v. iI~bk for 1~70 and 1975, there wa' ..... n · idm~ of a bi • • (ither row,,,d prov~m, or rc-gion•. Thi, was okw on;oW by lim calculating. for u ch J>fOYaln in (ach SUtc, tho: ohn c of tOla l CCIIIf. 1 govcrnmm r <""nd,ng ren-ivcd in 1~70. The po-ranlage- change- in th. t shan bnwffn ,,70 .oW " 75 was lhen u.kulatw . If the chango . noss program' for individu. l SlatCS varied ks. than the changcs aCrm5 "atcs for inJi ~idual procrams. then program lhim dominated ngjonal .bitts and via versa. Calculating the Slanda,d dtvia nons of each program across "01.. and euh ,tate 3(1"..... pmgrami, .hen tak ing .he lnC~ns of the sund~rd dcvi~lions for program' and for Sial ... OM finds thai ,he lnCanS ate almost idcnriul. Thu. nather program. nOf ' UtCS dominated the shi ft~ th.t OIXUrrN. 3~. Publi~ed rcpom of Tranden do nO! allo ... a K pai-arion of therr components, bue year-to-yur ( hang.. in""iu bly rdlcr:! (hang~ in the controllable com· ponem. Acwrding 10 Reunde (I ~82.a). 1oa1 and Slate cl ites were dttply ;nvol~ed in determining the , iu and di.rribution of rontrOlbble transfen.. 40. A exoolent diK"llWi on of cl ientdisrie pra~cn durin! the Mklici ~a f"l i< found in Cammack (I~82).
1
,,8
Two Appliwtions of the Theory
Brazilian media and in intellectual circles,"' The regime, it was claimed, had become dircctionbs. Its plans were made not for years ahead but for months or even weeks. Ministers like Mario Andreazza saw their programs as springboards to the presidency; other officials lined their own pockets Of those of their friends. Figueiredo's decline is reflected in the way ministerial allocations changed from original budget proposals to actual spending. Suppose we compare thc proposed-to-aclUal changes that occurred in the election year 10 changes in other years of the same administradon. 4l Of all Geisel years, the proposed-3CIul] ch:mge w;}s the smallest for fifteen of sixteen ministries in the election year of 1978. Of all Figueiredo years, however, proposed-actual differences were smallest in the election year of 1982 for only two ministries. For seven ministries, in fact, the 1982 change was th ~ largest of any Figueiredo year. These midcourse deviations reflect Figueiredo's intense electoral effoTt, an effort greater in vigor than Geisel made either in 1974 or in 1978. A comparison of the differences between the proposed budgets of the election year and the year just prio r to the election supports this interpretation. The average change between Geisel's 1977 and 1978 budget proposals was 42 percen!. Figueiredo's propos;lls ;lV· eraged only a Z3 percent shift from 1981 to 1982. In ot her words, Figueiredo's plans, illustrated by the proposal·to·proposal jumps, changed little in election rears. but his actual expenditures changed greatly as the election itself unfolded.·) Did Figueiredo modify budget proposals simply because he changed his mind, deciding 10 restructure spending once the year was under way? l'robably not; the president rarely exercised leadership. Rather, the pulling and pushing of ambitious ministers led to the sharply chanGing all(}(:ations of 1982. 41. 5« Vti~ and /51" i during ,h= rurs; <~e Abo A$sis 1I98J) 3nd I.cs"" (1980) . 42. Pro~d budgets, publi,hed in Dccember 0 / th. pr.ceding ulendu )·.~ r. rcprcsc:nr ~ sta •• men. of inlen •. Everyone kn",," .hal .hey cha nge, and in &en ..31 .he changes
Jr
.... '
l~
u.
Wbm Soldiers Need Frimds
'79
Expenditures in Education. Figure 14 disaggregates the expenditures of the Ministry of Education and Culture by level of eclucation .... Under Geisel higher education did very well. even surpassing the peak attained by Medici. 4 $ Figueiredo, on the other hand, sought to redistribute education spending. 46 His planners viewed university allocations as subsidies to the upper middle class, and as the economy declined the number of university graduates seemed excessive. Literacy rates, moreover, had been growing very slowly in the 19705. Many of the regime's more conservative economists, especially those with backgrounds in "human capital" theories, began urging cuts in higher education spending. Figueiredo's political strategists, hoping for a substantial popular response to spending on primary schools, agreed. Primary schooling's share of education spending began to grow under Figueiredo, and by 1981 its growth had become significant. 4 7 Still, the spending of the Ministry of Education and Culture needs to be put in perspective. The ministry represented less than 40 percent of total education expenditures, and the president had much less control over the foundations and aUiarchies dominating educational finance than over the ministry itself. Figueiredo managed to increase primary expenditures even outside the ministry, but the increase was smaller. Moreover, secondary schooling, not the universities, paid for the gain in primary education. Once again we see the ~decompos ition" of the Figueiredo administration, at least in relation to the centralized decision making of Geisel. Under Geisel, the losers and winners in education were the same both in the ministry and in the decentralized educational institutions. Figueiredo had less control, so programs 44. Figur~ 14 ~how$ only th"", u""nditur .. of th. ministry th~t can},., divilkd iOlo hi"'~r, ""condary. and primary level •. Such ~x~nditur •• amoun.cJ '0 about SO ""r~nt of all the mini~try's expenditu=. 45. Apologists for th. G.;""I admini ~ml!ion f,equemly memion the oo.rall in· c,.a"" in education .pending realized during his pr.,iden,")". Rarely do they giv~ eq ual prominen~ to the .hih, in .""nding bctwun the various le,·d. of education. Se. Albuqu.rqu~ (n.d.) and R.i, Vdloso (1978 ). 46. In Brazil', federal sy>l(m, public hi"'.r eduution is the ,(Sponsibiliry of th. ~ntul gm"ernment. ~nd about 50 ""r~nt of all fed~ral ,pending on education goes to the universiti(S. Ptimary and ~condary education nt b.:uicaUy Slate and locaL Th. f.deral sha re in alll~vd, has grown rapidly in r"""m years as th. c~mral gov.rnn'~nt has inctea,ingly pteentpted the tax po"·.r~ of stat.. and municipalities. 47. It is wonh noting th~t in 1984 univ.... iti(S aU m".r th~ country went on a I'rolonsed strike to prot(St inadNJuate funding. lbe strike was partly a =uh of th .. ~ ,hilts in budS.t ~har...
"
"
,'0
Two Applications of the Theory
"
Higher education
Secondary edl.lCalion Primary educahon
SOURCE: IBGE (1970-1981).
figurt Lt. Percentage Division of Actual Expenditures in Ministry of Educa tion and Culturt
moved in opposite directions. To a certain extent, then, Figueiredo's political weakness ncg:ncd his social strategy.
Recapitulatioll. Geisel's :md Figucirrdo's expenditure policies reveal similarities and differences. They both cut the shares of the military and transport:n ioll rniniSlrics, and they both increased spending on agriculture, education, and welfare. Figueiredo also
.... '
l~
u•
,. ,
When Soldien Need friend5
aided housing, a program nicely attuned to the presidential ambitions of Interior Minister Mario Andreazza_ The elections of 1978 and 1982 illustrate the difference between the two presidents_ Expenditure shifts in Geisel's last year (1978) reflected a continuation of the strategies that had motivated his administration earlier. Figueiredo, on the other hand, had the bad luck to preside over a deteriorating economic and political situation. His expenditures in 1982 represent a midcourse correction just before a crucial election: more spending to pacify the military and more spending on public works to cut unemployment and reward state and local clites. The regional distribution of expenditures initially seemed confused. The Northeast lost from changes in the distribution of min· isterial expenditures but gained from changes in the payment of transfers. Regional political differences, however, explain the apparent contradiction. In the clientelistic Northeast, transfers increased the political clout of powerful elites. In the industrialized Southeast, direct programs were appeals to the middle and working classes. The priorities of the NoO administrations arc also reflected in the changing distribution of education spending. Geisel reached out to all segments of civil sociery, but his prime target was the upper middle class, a group for whom access to higher education was paramount. Figueiredo sought to widen the regime's base by increasing spending on primary education at the expense of universities. The effort met with only partial success: Figueiredo conlrol1ed the Ministry of Education and Culture, but less than half of al1 educational expenditures were under its control, and the bureaucracy outside the ministry resistt:d his efforts and reduced the overall shift away from higher education.
The Politics of Industrial Deconcentration Industry locates near markets, cheap transponation, and raw materials. The primary "growth pole'" in Brazil is the Rio de Janeiro-Belo Horiwnle-Sao Paulo triangle. especially the ciry and state of Sao Paulo. In 1980 Sao Paulo state contained 21 percent of Brazil's population, produced 54 percent of its industrial product, and consumed 39 percent of all energy. With its surrounding area, the eiry of Sao Paulo represents one of the largest industrial
rho
l~
h
,8,
Two Appliwtions of the Theory
concentrations in the world. Because the pace of industrialization has been so rapid and the influx of newcomers so large, the delivery of urban services has never caught up with demand. Transportation is woefully in3dequate; infant monality extrcmelr high; environmental damage already substantial. Sao Paulo, il may be said, is Oiler-industrialized. Official recognition of the problem came in the basic planning document of the Geisel administration, the Second National Dcvelopment Plan (PNDII). PNDII called for a reduction in the interregional inequalities crc:ued by industrialization: To avoid contin uin g the tendency for indusl rial activity to concentra te in a single metropolitan arca, stimulus will Ix: given {Q a better balance in the Sao Paulo-Rio-Belo Horizonte triangle. The objective will be
a decemralization compadble with the preservation of economics of scale (Brazil, PNDIi:40).
Industrial deconcentration was by no means the sole concern of PNDII. It primarily emphasized the dilemma of expensive oil, a dilemma it sought to resolve through a quantum leap in Brazil's level of development. Central to this leap would be the development of basic indUStry. including capital goods, electronics. and such industrial inputs as metals, petrochemicals, fertilizers, paper and cellulose, raw materials for pharmaceuticals, and construction materials. PNDJI stressed exports of manufactured goods and effons to reduce the cost of technology and increase the production of foodstuffs . Whether central to PNDII or not, industrial deconcentration was now on the regime's policy agenda. What were its strategic implications? Deconcentration would strengthen the regime in states benefiting from new industry. Though it would diminish the political weight of Sao P:lUlo, in the long run it would help the paulista working class, because social services could catch up with demand. But if the regime wanted the paulista industrialists to move to the himerlands. it would have to till toward domestic industry and away from transnational corporations seeki ng investment opportunities. PNDII m3de promises to domestic entrepreneurs, but its promises were vague: creation of strong entrepreneurial structures, use of the development banks to aid domestic firms, formation of national conglomerates, and help for small and medium-sized companies.
r
.... '
l~
u.
·~
•
,8,
When Soldiers Need friends
One of the most important tools for implementing decancennation was the fiscal incentive program,OI Fiscal incentives were exemptions from taxes on income and on imported equipmem, They were gramed by the lndumial Development Council (COl), an organ of the Ministry of Industry and Commerce, An industrialist desiring a tax exemption would present a project to the COl. If the project met the council's criteria, it would gram a tax exemption. The annual distribution of projectS approved for fiscal incentives, by value and absolute number, is shown in Tablts 32 and 33. The Northeast, expected to be the major beneficiary of industrial deconcemration, did increase its percentage of incentives, and the Southeast and Siio Paulo received smaller shares. But the changes were rather small, and until 1981-1982 (when a few enormous projects absorbed a huge share of incentives), Siio Paulo still attracted a substantial portion. However we judge the interregional shifts in incentives, their magnitude pales before the huge dedine in the total value of the incentives granted. After peaking at 53 percent of all direct federal receipts in 1976, incentives declined to 10 percent of taxes in 1978 and only 2 percent in 1980. The Northeast, in other words, may have gained in the battle for incentive shares, but the virtual disappearance of the incentive program itself makes the victory rather hollow. Deconcentration was also implemented through the lending of the two major official devdopment banks, the Bank of Brazil and the National Economic Development Bank (BNDE). As Tables 34 and 35 demonstrate, their lending pattern differed from the fiscal incentives of the CDL The distribution of loans from the Bank of Brazil (basically a commercia l bank) shifted toward the Northeast after 1974, although the region had actually been gaining since 1969. Sao Paulo's share, along with the shares of the Federal District, Rio Grande do Su i, and Parana, declined after 1974 . The geographic distribution of BNDE loans is harder to characterize, because its emphasis on a smal( number of very large projects produced a volatile loan distribmion. The Northeast's share of BNDE loans did not grow until the beginning of the Figueiredo 48. Two other p"'gram~ ,,/ d"':"",enltuion are ignorw here: tht regional di· ,·i.ion of 1"""Uy collected tax •• ~nd the Fund lot th. Developmem of Inlegr~led Progr~m •.
"
"
,"
Two Applicatiolls of the Theory
years, and except for Bahia the region fared worse under Gcisd than under th e Medici government.
Taken together, the incentive and loan programs reveal little evidence of a substantial shi ft away from the traditional industrial centers. Why was there so linle ch:mgc? Consider th t Industrial Development Council, the body ch:uged with formubling and implementing deconccntration_ The C DI was headed by SCVtrO Gomes, the minister of industry and commerce. Go mes began as an enthusiast, arguing that: Tht indi:;criminalc localion of industrial p lants is a major cauS(C of the disorder ly and quasi-chaotic growt h of some Brazilian urban c.. m ... rs. For a resource-poor country with the cXICnsil'c territory of Brazil, the <'Conomic and sodal (OSIS Ihal r~suh from ~xc~ssi\"c urban concentra· lion are heavy. From the purely economic poinl of view, the most rationa l road rests, generally, in Ihe prOCl,'s,~ of accumulation and con· cent ration of wealth. But th is proceSS, if conduct ed withoU! limits, bt~!ly and j