Unfunded Pension Systems: Ageing and Migration, Volume 264 (Contributions to Economic Analysis)

UNFUNDED PENSION SYSTEMS: Ageing and Migration CONTRIBUTIONS TO ECONOMIC ANALYSIS 264 Honorary Editors: D.W. JORGENS...

Author: S. Uebelmesser

3 downloads 562 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

UNFUNDED PENSION SYSTEMS: Ageing and Migration

CONTRIBUTIONS TO ECONOMIC ANALYSIS 264

Honorary Editors: D.W. JORGENSON J. TINBERGEN†

Editors: B. BALTAGI E. SADKA D. WILDASIN

Amsterdam – Boston – Heidelberg – London – New York – Oxford – Paris San Diego – San Francisco – Singapore – Sydney – Tokyo

UNFUNDED PENSION SYSTEMS: Ageing and Migration

SILKE UEBELMESSER Center for Economic Studies, University of Munich, Munich, Germany, and CESifo

2004 Amsterdam – Boston – Heidelberg – London – New York – Oxford – Paris San Diego – San Francisco – Singapore – Sydney – Tokyo

ELSEVIER B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam The Netherlands

ELSEVIER Inc. 525 B Street, Suite 1900 San Diego, CA 92101-4495 USA

ELSEVIER Ltd The Boulevard, Langford Lane Kidlington, Oxford OX5 1GB UK

ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK

q 2004 Elsevier B.V. All rights reserved. This work is protected under copyright by Elsevier B.V., and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-proﬁt educational classroom use. Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail: [email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www. elsevier.com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent veriﬁcation of diagnoses and drug dosages should be made. First edition 2004 Library of Congress Cataloging in Publication Data A catalog record is available from the Library of Congress. British Library Cataloguing in Publication Data A catalogue record is available from the British Library. ISBN: 0-444-51732-4 ISSN: 0573-8555 1 The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). W Printed in The Netherlands.

Introduction to the Series This series consists of a number of hitherto unpublished studies, which are introduced by the editors in the belief that they represent fresh contributions to economic science. The term ‘economic analysis’ as used in the title of the series has been adopted because it covers both the activities of the theoretical economist and the research worker. Although the analytical method used by the various contributors are not the same, they are nevertheless conditioned by the common origin of their studies, namely theoretical problems encountered in practical research. Since for this reason, business cycle research and national accounting, research work on behalf of economic policy, and problems of planning are the main sources of the subjects dealt with, they necessarily determine the manner of approach adopted by the authors. Their methods tend to be ‘practical’ in the sense of not being too far remote from application to actual economic conditions. In addition they are quantitative. It is the hope of the editors that the publication of these studies will help to stimulate the exchange of scientiﬁc information and to reinforce international cooperation in the ﬁeld of economics. The Editors

This Page Intentionally Left Blank

Preface The main observation of the book is that unfunded, pay-as-you-go pension systems require a well-balanced ratio of members of different generations, old and young, in order to distribute costs and beneﬁts in a mutually advantageous way. Financially self-reliant, individual accounts are per deﬁnition inconsistent with the nature of an unfunded pension system; on the contrary, interactions across generations are a basic condition for unfunded old-age security to function. In this respect, it is interesting to note that the process of writing this book is not so different from the main insights laid down in it. For a book project to be successfully completed, a well-balanced ratio of one’s own ideas and input from outside is essential. Interaction with colleagues, family and friends and their valuable help and constant encouragement are indispensable prerequisites. Thus, I am indebted in different ways to many people who have accompanied me over the last years. This book was completed in July 2003 at the faculty of Economics of the University of Munich. I am particularly grateful to Hans-Werner Sinn for his continuous guidance and constant support at each stage of my work. Special gratitude goes also to Andreas Hauﬂer and Dalia Marin for valuable discussions. This book has greatly beneﬁted from the very productive academic atmosphere at the Center for Economic Studies (CES) and in the economics department where economic research ﬂourishes and new ideas can grow and develop and where a constant process of discussing, rethinking and optimising the arguments and models takes place. In addition, CES offers an unparalleled exchange of ideas with economists from all over the world. I would like to mention especially the lectures and seminars I followed and the discussions I had with Ted Bergstrom, Panu Poutvaara, Assaf Razin, Efraim Sadka, Steinar Strøm and Harrie A. Verbon. I also experienced very inspiring discussions with Kai A. Konrad and Regina Riphahn. The possibility to interact with international experts in the ﬁeld was even further increased by the founding of CESifo, the international platform that links CES and the Ifo Institute for Economic Research. I would also like to thank my former and present colleagues at CES and Ifo—in particular Marcel Thum for his continuous support and his many valuable comments, which greatly helped to improve my work. Special thanks go to Sascha O. Becker, Robert Fenge, Ulrich Hange and Martin Werding with whom

viii

I enjoyed very stimulating discussions. I also want to mention Martina Grass, Christian Kelders, Marko Ko¨thenbu¨rger, Raji Jayaraman, Mikael Priks, Michael Stimmelmayr, Karin Thomsen and Frank Westermann. I gratefully acknowledge valuable research assistance from Christine Kaaz and help with the complicated legal details of the European Union by Christina Ro¨lz and Jesko Ullrich. Rebecca Forwood carefully and skilfully edited the English text for which I am very thankful. Part of this work was written during a research stay at Bocconi Universita`, Milan, Italy. I would like to thank in particular Guido Tabellini and Piedro Garibaldi for the high-quality PhD programme and Alessandra Casarico for our very lively economic discussions. Last but not least, writing a book is a very time- and energy-consuming endeavour which does not work without the emotional support from family and friends. I am very grateful that my parents and my sister have always encouraged and supported my thirst for knowledge from the very beginning. But the person who has accompanied me most closely and most intensively during the last years—through all the ups and downs—is Claudio Thum. Without his academic and emotional support, the book would not have turned out the way it did. I dedicate this book to Claudio and hope that many more joint projects will follow. Silke Uebelmesser Munich, 15 April 2004

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Focus of the analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Structure and overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 2 4

Chapter 2 Unfunded Pension Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Mechanisms of unfunded and funded pension systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Unfunded pension systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2 Funded pension systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3 Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 The concept of implicit taxes and implicit debt . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Implicit taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Implicit debt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3 Other concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A2. Solow– Swan growth model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B2. Derivation of Equation 2.28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 10 11 13 14 19 19 21 26 28 29 29 32

Chapter 3 Projected Development of Fundamental Factors. . . . . . . . . . . . . . . . 3.1 Determinants of the population growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Fertility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Life expectancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.3 Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Development of the total population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Age structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2 Dependency ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A3. Development of the population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B3. Dependency ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33 34 35 36 36 45 47 47 49 51 51

Chapter 4 Country Studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 4.1 Characteristics of pension systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.2 Recent reforms of pension systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

x

4.2.1 Germany and France . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Sweden and Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 UK and Denmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

59 63 65 67

Chapter 5 Welfare Analysis of Pension Reforms . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Intergenerationally efﬁcient reforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Distortions of the saving decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2 Distortions of the labour-leisure decision . . . . . . . . . . . . . . . . . . . . . . 5.2 Intrapersonally efﬁcient reforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Implicit taxes in an intrapersonal context . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Second-best optimality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Empirical evidence—methodology and data . . . . . . . . . . . . . . . . . . . 5.2.4 Individual labour-supply elasticities: the results . . . . . . . . . . . . . . . . 5.2.5 Is the structure of implicit taxation optimal?. . . . . . . . . . . . . . . . . . . 5.2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A5. Standardised work biographies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B5. Derivation of Equation 5.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix C5. Description of the German Socio-Economic Panel (GSOEP) . . . . . Appendix D5. Estimates for the wage equation: men and married women . . . . . . Appendix E5. Results of the Tobit model: smaller sub-groups . . . . . . . . . . . . . . . . . . Appendix F5. Time structure of implicit taxes and wage elasticities . . . . . . . . . . . . .

69 70 71 73 75 76 87 88 97 105 110 113 114 116 117 119 128

Chapter 6 Political Feasibility of Pension Reforms. . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Intergenerational redistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1 Implicit taxes in an intergenerational context . . . . . . . . . . . . . . . . . . 6.1.2 Optimal level of redistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Voting model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Feasibility of pension reforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Status quo. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2 Median age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.3 Indifference age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A6. Calculating the median age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B6. Calculating the indifference age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

131 133 133 136 138 141 144 148 149 153 154 156 157

Chapter 7 Mobility as a Counterforce to Gerontocracy . . . . . . . . . . . . . . . . . . . . 7.1 Voting model with mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Mobility as a commitment device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 The role of education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

159 160 163 165

4.3

xi

7.2.2 Monetary costs of education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.3 Extensions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A7. Social planner problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B7. Monetary costs of education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

173 174 177 178 179

Chapter 8 Qualitative Aspects of Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1 Description of the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1.1 Intention to migrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1.2 Discussion of the intention variable . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Estimation of the intention to migrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A8. Descriptive statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B8. Probit estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix C8. Probit model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

183 185 186 190 193 199 200 200 206

Chapter 9

Sustainability of Pension Systems with Sytems Competition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1 Status quo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1.1 Implicit taxes in an international context . . . . . . . . . . . . . . . . . . . . . . 9.1.2 Calculations for European countries. . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1.3 Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Theoretical results when pension systems are similar . . . . . . . . . . . . . . . . . . . . 9.2.1 Unrestricted mobility—the benchmark . . . . . . . . . . . . . . . . . . . . . . . . 9.2.2 Restricted mobility—the relevant scenarios . . . . . . . . . . . . . . . . . . . . 9.3 Institutional distribution of competence between the national and the European level. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.1 The evolution of social policy issues on the European level. . . . . 9.3.2 Consequences for the distribution of competence and the level of harmonisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Comparison of the theoretical and institutional results . . . . . . . . . . . . . . . . . . 9.4.1 Allocation of labour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.2 Level of redistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.3 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5 Alternative options when pension systems are different. . . . . . . . . . . . . . . . . . 9.5.1 Transfers and individual payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5.2 Home-country principle and delayed integration . . . . . . . . . . . . . . . 9.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A9. Art. 117 ToR and Art. 136 ECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B9. Art. 118 ToR and Art. 137 ECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

207 210 211 213 215 217 219 223 226 226 233 238 239 240 241 241 242 243 245 246 247

xii

Chapter 10

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 Symbol Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269

CHAPTER 1

Introduction In 1889, the ﬁrst publicly administered pension system was launched in Germany by Bismarck, whose social security legislation introduced “invalidity and old-age insurance” for private sector employees. The Bismarckian pension system has served as an example for the “social insurance type” systems which have developed since then, mostly in continental European countries, for example, in Austria, France and Italy. The main characteristics of these systems are that membership is linked to the employment status and beneﬁts depend on contributions paid over the working life and are supposed to enable the recipient to keep his or her standard of living during retirement. Parallel to this development, an alternative approach for old-age security emerged in some Anglo-Saxon countries. This is normally attributed to Beveridge and his analysis in 1942 of old-age security in the UK for the post-war period (“Beveridge-plan”) although a similar system had already been founded in Denmark in 1891 as the preceding institution of today’s “Folkepension”. In contrast to “social insurance type” systems, these “ﬂat-rate beneﬁt type” systems do not limit membership to employees but comprise—in the extreme case—all residents of the country. Beneﬁts are not (closely) linked to contributions and are only supposed to guarantee a uniform, basic level of social protection for old age. While nearly all industrial countries had some kind of old-age security system by the end of World War II, many of these systems were quite limited concerning the part of the population covered and the level of beneﬁts provided (Feldstein and Liebmann, 2002). The second part of the last century, however, saw an expansion of the shape and size of pension systems in all countries. Despite important differences between these two types of old-age security systems and the many variants which have developed since, there is one important common feature. Both types are predominantly organised as unfunded pension systems (Fenge et al., 2002c), i.e. today’s working generation pays contributions which ﬁnance the pension beneﬁts of today’s retirees implying that

2

Unfunded Pension Systems: Ageing and Migration

there is no fund. This is surprising at ﬁrst sight as the type of ﬁnancing is not predetermined in any way—and in fact the German pension system was organised as an investment-based funded system until 1957. Since then, however, the German system and systems in most other countries have been unfunded. This has been mainly motivated by the wish to support the generations most severely affected by the war and its economic consequences. An unfunded pension system makes this possible as pension beneﬁts do not rely on any form of accumulated capital. It can thus be started right away. The coming decades will show whether the ﬁnancing as well as the general organisation of old-age security will change or not. The existing public pension systems face serious challenges with respect to their ﬁnancial sustainability and must be reformed in a way which averts this crisis. During the last decade, reforms of the public pension system have already been made in a rising number of countries, for example, in Germany, Italy and Sweden, and the ongoing discussions about old-age security indicate that further reforms will follow.

1.1. FOCUS OF THE ANALYSIS In this book, we analyse publicly provided old-age security organised as unfunded or pay-as-you-go pension systems and its projected evolution. Our analysis is thus focused on two aspects. First, we concentrate on the ﬁrst, publicly organised pillar for private sector employees and abstract for the most part from privately run company pension plans (second pillar) and privately run individual pension schemes (third pillar). Figure 1.1 shows that for the EU-12 countries, the ﬁrst pillar is by a long way the most important source of old-age income, accounting for 89%, while the second pillar plays only a minor role with 7% and the third pillar is negligible at 1%. This ranking with respect to the relative importance of the ﬁrst pillar, i.e. publicly provided old-age security, also holds on a country level for Italy, Finland, Sweden, Japan and Germany (Figure 1.2). Even for the UK, the US and Japan, almost half of old-age income comes from the ﬁrst pillar. Second, we focus on systems based on pay-as-you-go ﬁnancing. In the academic debate about the optimal conception of old-age security, the question about the optimal choice of ﬁnancing, i.e. the choice between a funded or an unfunded system or a combination of both, receives much attention (Breyer, 2000). But in fact, today, we are not in a position to start all over again and create the most optimal pension system without having to take old claims into consideration. The question as to which ﬁnancing should be chosen was answered,

Introduction

3

Figure 1.1: Relative importance of the pillars of old-age security for total old-age income: EU-12 countries (1994). Source: Following Keller (2000).

USA

First Pillar & Others Second & Third Pillar

Netherlands UK Germany Japan Sweden Finland Italy 0%

20%

40%

60%

80%

Percentage of 100% old-age income

Figure 1.2: Relative importance of the pillars of old-age security for total old-age income— selected OECD countries (mid-1990s). Based on data from the Luxembourg Income Study (LIS) and an OECD Questionnaire, the OECD has calculated the ratio of the average income of individuals from 65–74 years divided into several categories and the average income of individuals over 17 years for the mid-1990s. Source: OECD (2001a).

4

Unfunded Pension Systems: Ageing and Migration

for example, in Germany in 1957 by introducing an unfunded pension system. With the exception of Denmark and the Netherlands, all ﬁrst-pillar pension systems for the private sector are based on unfunded pension systems in countries of the European Union (Economic Policy Committee, 2001).1 Given the pay-as-you-go ﬁnancing of old-age security, generations are thus linked via intergenerational transfers from the young to the old. Factors that change the ratio of old to young affect the budget constraint and—depending on the exact mechanisms of the pension systems—the distribution of the pension burden across generations. We will analyse two important developments which are expected to put particular pressure on unfunded pension systems as they both alter this ratio of old to young: ageing of the population and labour mobility. Ageing as projected for most industrialised countries inﬂuences the ratio of old to young in a twofold way. On the one hand, fertility rates are low. On the other hand, life expectancies are high and even increasing. This leads to a smaller number of young individuals who face a growing number of older individuals. Labour migration intensiﬁes this development if the young react to this growing imbalance by migrating. It is evident that “[t]he crisis of the pension system is a demographic crisis” (Sinn, 2001, p. 77).

1.2. STRUCTURE AND OVERVIEW We analyse three areas which can be distinguished concerning old-age security: (a) efﬁciency considerations and the possibility of Pareto improvements; (b) political economy aspects and the feasibility of reforms; (c) the process of European integration and its inﬂuence on national pension systems. All three areas will be taken up in the following discussion against the setting of publicly provided unfunded pension systems in industrialised countries—with the centre of attention on Member States of the European Union and in particular on Germany—and the crisis they are facing due to ageing of the population and labour mobility. The analysis can be split up into two parts. Chapters 2 –4 serve to introduce the mechanism of unfunded pension systems, to illustrate the projected development 1

In Sweden and Finland, the ﬁrst pillar is partially funded.

Introduction

5

of the ratio of young to old, and to present recent reform experiences. Chapters 5 –9 focus on questions of efﬁciency and feasibility for different reform options for different distributions of political power and for different mobility scenarios. Chapter 2 lays the groundwork for the analysis by detailing the mechanisms of unfunded pension systems compared to funded ones. As in an unfunded pension system pension beneﬁts of the old are ﬁnanced by contributions of the young, the wage sum is essential for the size of these transfers. Depending on the concrete characteristics of the pension systems, the risk of a low growth rate of the wage sum—whether due to low population growth or due to low productivity growth—is carried by contributors, retirees or both. The majority of pension systems is (still) organised such as securing the beneﬁts of the old and shifting the risk to the young. The internal rate of return of unfunded pension systems equals the growth rate of the wage sum. The relation between the internal rate of return and the market rate of return reveals the degree of intergenerational redistribution via the unfunded pension systems from later generations to the early ones. For later generations this is reﬂected in an implicit tax on their contributions to the unfunded pension system to service the implicit debt due to the introductory gains of the ﬁrst generations. We will make extensive use of the concepts of implicit taxes and implicit debt throughout the following chapters to illustrate the pension burden implicit in mature unfunded pension systems. As this implicit debt is given by the introductory gains, questions arise with respect to the distribution of this burden across the life cycle of individual generations and across different generations, as well as the distortions this implies for labour supply and labour allocation. Given the importance of the population growth rate for the size of the internal rate of return and the implicit taxes falling on members of different generations, Chapter 3 illustrates the projected demographic development. In the next ﬁve decades, population in most industrialised countries is projected to decline in a way that changes the age structure fundamentally. Immigration can only alleviate this evolution to a certain extent but is far too low to counterbalance this development on a one-to-one basis. Consequently, the ﬁnancial sustainability of the unfunded pension systems will be seriously endangered as fewer and fewer young face more and more old. This mounts the pressure to reform the pension systems towards more funding, more actuarial fairness and in the direction of a shift of the risk from the young to the old. In Chapter 4, we give an overview of the chosen path of selected countries. Some countries have already initiated more fundamental reforms which increase

6

Unfunded Pension Systems: Ageing and Migration

the funded part in old-age security or which link beneﬁts more closely to contributions. The majority has, however, limited the reform activities so far to parametric changes. For most countries, these reform activities can only be the beginning. We, therefore, consider it to be important to look in detail at reforms which should and could be realised given the ageing of the population and the increasing mobility of at least some groups of the population. It has already been shown that unfunded pension systems are not inefﬁcient per se, implying that a transition to a funded system does not present a Pareto improvement. The implicit debt has to be serviced, which leads in both cases to taxes on the contributions. From an intergenerational perspective, this debt cannot be reduced in a way that makes no generation worse off. However, the taxes might not be optimally distributed over the individual life cycle, thus leading to distortions of the labour-leisure decision. In Chapter 5, we analyse whether the paths of implicit taxes and of labour supply elasticities are optimally related for Germany. We ﬁnd that it is possible to improve efﬁciency by differentiating implicit taxes across age groups, thus reducing the distortions of the labour supply. From an intrapersonal perspective, there are thus ways for a Pareto improvement. For all reform proposals, feasibility considerations have to be taken into account. In a majoritarian voting system, a reform is feasible if the majority of voters incur no loss in terms of utility. If a reform makes no voter of any generation worse off, the majority is ensured. A reform that does not change the intergenerational distribution of the implicit taxes, but increases efﬁciency intrapersonally by distributing the implicit taxes optimally over the life cycle is, therefore, feasible. If, however, a reform involves a change of the intergenerational distribution of implicit taxes beneﬁting some generations and hurting others, feasibility is no longer guaranteed. In Chapter 6, we identify the winners and losers of a reform which initiates a partial transition to a funded system. Contributions and pension beneﬁts are reduced in a way that respects the budget constraint of the unfunded pension system. The implicit taxes are thus lowered for the young, who still have to contribute for many years before receiving beneﬁts, while they are increased for the old, who are close to retirement or already retired. Ageing of the population affects the age and the size of the groups who are for or against the reform. The question is whether the decisive (median) voter belongs to the supporters of the reform and, if yes, for how long, given the development of society towards a gerontocracy. In order to determine the chance for success of reforms against the background of gerontocracy alone, we analyse ﬁrst the case where the young cannot react in any respect—except by voting. We thus exclude any form of exit

Introduction

7

option. This allows us to identify the point in time when reforms are no longer feasible. If the young have lost the majority and no constitutional safeguards have been established, the unfunded pension system might then explode. In the following chapters, we abandon this assumption and explicitly allow for the possibility that the young can react by emigrating. The young have thus either a conditional or an unconditional exit option, even after they have lost the majority. This exit option can serve to counterbalance the power of the old and the system need no longer explode. We start in Chapter 7 with the intermediate case between no mobility and total mobility, where (potential) mobility of the young can be controlled by the old. The young are thus mobile conditional on the old granting them mobility. The old do this strategically in order to commit to a certain level of redistributive taxation. In equilibrium, no young individual emigrates although this would be possible. The exit option of the young is enough to avoid exploitation by the old. By balancing the exit option and gerontocratic power, an intergenerationally redistributive system need then neither explode nor erode. Projections of demographic development show that migration within the European Union is expected to increase slightly in the next ﬁve decades but from a low level. It is certainly possible that migration will settle at some steady-state level after the efﬁcient allocation is reached. This would be in line with our result. Making it easier for the young to emigrate does not necessarily imply that everybody will choose this option. On the contrary, it might be exactly this better exit option that makes migration (partially) redundant—at least for other reasons than efﬁcient allocation. But intra-European migration—even at a relatively low level—might endanger the sustainability of national unfunded pension systems if those who pay contributions to the system leave the country. In order to shed some light on this aspect, in Chapter 8 we estimate who is most likely to emigrate from Germany. We ﬁnd that those individuals consider leaving the country, who are especially needed as contributors to the pension system, i.e. the young with an above-average school level. Reactions in the national pension systems must, therefore, be expected as an attempt to hold back domestic contributors and to attract foreign contributors. Growing integration of the Member States of the European Union can be expected to intensify this competition. We, therefore, look then in detail at different mobility scenarios within the European Union and the consequences for nationally organised pension systems in Chapter 9. It is straightforward to conclude that mobility of the young, which is now unconditional on actions of the old, might result in an erosion of unfunded pension systems. The question which must be addressed concerns the optimal allocation of responsibilities and the optimal degree of harmonisation of old-age

8

Unfunded Pension Systems: Ageing and Migration

security to avoid this potential “race to the bottom” while ensuring an efﬁcient allocation of labour. If pension systems are not sufﬁciently harmonised, the principle of free movement of labour threatens intergenerationally redistributive activities while at the same time the tax implicit in unfunded pension systems potentially distorts the allocation of labour. It can be shown that in order to prevent this from happening, a uniform European pension system is not necessary. All that is needed is co-ordinated or equalised contribution rates. The European Union must then guarantee the binding nature of the harmonisation rules agreed upon by the Member States. At present, however, the Member States are far away from conceding such fundamental responsibilities to the European level. In Chapter 10, the results are summarised. Public unfunded pension systems will face a crisis in the next few decades due to ageing of the population and labour mobility. This crisis can, however, be mitigated if fundamental reforms are initiated and not further postponed which loosen to a certain extent the intergenerational dependencies and thus reduce the pressure from the changing population structure on old-age security.

CHAPTER 2

Unfunded Pension Systems The mechanisms responsible for the projected crisis Wenn wir 700.000 kleine Rentner, die vom Reiche ihre Rente beziehen, haben, gerade in diesen Klassen, die sonst nicht viel zu verlieren haben und bei einer Vera¨nderung irrtu¨mlich glauben, daß sie viel gewinnen ko¨nnen, so halte ich das fu¨r einen außerordentlichen Vorteil […]1 Otto Fu¨rst von Bismarck (1889)2 Abstract Unfunded and funded pension systems display many differences. Focusing on internal rates of return, implicit taxes and implicit debt, we explain the mechanisms of both systems with a special emphasis on unfunded systems. This allows us to determine the factors that are responsible for the crisis of unfunded pension systems projected for the next few decades. If we abstract from all the speciﬁc variants of pension systems, we can basically distinguish two main systems: unfunded pension systems, also called pay-as-yougo systems, and funded pension systems. As will be explained in more detail in this chapter, the most relevant difference is that in unfunded pension systems, no capital stock or funds exist. The beneﬁts of today’s retirees are ﬁnanced by the contributions of today’s employees. Funded systems, on the contrary, are based on

1 I will consider it a great advantage when we have 700,000 small retirees receiving their pension beneﬁts from the state, especially when they belong to those classes which otherwise do not have much to lose by an upheaval and erroneously believe that they can actually gain much by it (own translation). 2 Cited in: Stein (no year, p. 245).

10

Unfunded Pension Systems: Ageing and Migration

individual funds, i.e. individuals pay contributions which are invested in the capital market to ﬁnance their own pension beneﬁts. With the exception of few countries, for example, Chile and Hong Kong, most systems in industrialised countries are organised as unfunded ones today—albeit supplemented by company pension plans and private pension plans whose importance varies from country to country (Fenge et al., 2002c). In this chapter, we explain the mechanisms of unfunded and funded pension systems focusing especially on unfunded ones. The aim is to determine the factors that are responsible for the upcoming crisis of unfunded pension systems. This discussion is very useful as a point of departure when looking closer at the evolution of these factors in the following chapters. In section 2.1, we compare unfunded and funded pension systems on the basis of internal rates of return. Section 2.2 serves to develop the concepts of implicit taxes and implicit debt. Section 2.3 concludes.

2.1. MECHANISMS OF UNFUNDED AND FUNDED PENSION SYSTEMS The description of the two main variants of old-age security systems serves to clarify the mechanisms. We concentrate on the simplest case deriving the results within a two-period overlapping generation (OLG) model and considering labour supply to be completely inelastic. These restrictions are not essential for the conclusions and will be lifted later.3 Each generation lives for two periods of equal length. In the ﬁrst period, the young individuals work, earn labour income and pay contributions to a pension system, and in the second period, the then-old individuals receive pension beneﬁts. We do not consider children explicitly, assuming as Samuelson (1958) has done that they are “part of their parents’ consumption” (p. 468). Because of the overlapping structure of the model, there are two generations alive in each period t: one young generation NtY born in period t; and one old generation NtO born in period t 2 1 (see Figure 2.1). As we assume that no individual dies before the end of the second period, we have O NtY ¼ Ntþ1 :

ð2:1Þ

Individuals want to consume in both periods. The utility function for an individual born in period t comprises ﬁrst-period consumption cY t and second-period consumption cO tþ1 O Ut ðcY ð2:2Þ t ; ctþ1 Þ; 3

See Chapter 5.

Unfunded Pension Systems

Born (young) in period t-1

Period t-1

Period t

Y Nt–1

NtO

Born (young) in period t

NtY

Born (young) in period t+1

Period t+1

11 Period t+2

O Nt+1

NYt+1

O

Nt+2

Figure 2.1: Two-period OLG model.

where U is a strictly monotone and quasi-concave ordinal utility function. Formalising the budget constraints for the two periods, we get for consumption in both periods cY t þ st ¼ vt ð1 2 ut Þ;

cO tþ1 ¼ ptþ1 þ ð1 þ rtþ1 Þst :

ð2:3Þ

In the ﬁrst period, the individual earns vt as wage income and pays ut vt as contributions to the pension system. The net wage income is vt ð1 2 ut Þ if we abstract from taxes and other contributions. The individual spends this income to consume cY t and to save st in the capital market. In the second period, the individual receives ptþ1 as beneﬁts out of the pension system and ð1 þ rtþ1 Þst as his savings plus the market rate of return. The individual uses this to consume cO tþ1 : Consumption in both periods is thus determined by the income in both periods adjusted by saving and desaving. In the lifetime budget constraint, savings just cancel out. But they are needed to transfer resources across periods in order to establish the optimal structure of periodic consumption. So far, we have not been explicit about what pension system we assume, i.e. whether old-age consumption is ﬁnanced via an unfunded pension system or a funded one. We will now discuss each of them in turn. 2.1.1. Unfunded pension systems As the name implies, an unfunded or pay-as-you-go pension system has no funds. Income and expenditure—or in our context contributions and beneﬁts—have to balance in each period. The budget constraint of an unfunded pension system thus requires that Contributionst ¼ Benefitst :

ð2:4Þ

12

Unfunded Pension Systems: Ageing and Migration

Total contributions of the young individuals in period t consist of the contribution rate ut times the wage sum vt NtY Contributionst ¼ ut vt NtY ;

ð2:5Þ

and total pension beneﬁts equal the pension level pt times the number of retirees NtO Benefitst ¼ pt NtO :

ð2:6Þ

With Equations 2.5 and 2.6 inserted in Equation 2.4, we get

ut vt NtY ¼ pt NtO ;

ð2:7Þ

and for the pension level pt ¼ ut v t

NtY ¼ ut vt ð1 þ nt Þ; NtO

ð2:8Þ

where 1 þ nt ¼ NtY =NtO is the population growth factor. 2.1.1.1. Internal rate of return For judging the attractiveness of unfunded pension systems, it is useful to calculate the internal rate of return itþ1 : This rate is deﬁned such that individual contributions and individual beneﬁts are equalised in the following way

ut vt ð1 þ itþ1 Þ ; ptþ1 ;

ð2:9Þ

or 1 þ itþ1 ¼

ptþ1 u v ¼ tþ1 tþ1 ð1 þ ntþ1 Þ; ut vt ut v t

ð2:10Þ

where the last equality follows from Equation 2.8. With 1 þ gtþ1 ¼ vtþ1 =vt for the growth factor of wages and 1 þ mtþ1 ¼ utþ1 =ut for the growth factor of the contribution rates we can rewrite Equation 2.10 as itþ1 ¼ ð1 þ mtþ1 Þð1 þ gtþ1 Þð1 þ ntþ1 Þ 2 1:

ð2:11Þ

Equation 2.11 shows which factors positively inﬂuence the internal rate of return: the growth rate of the population and the growth rate of wages, as both mean a larger wage sum in t þ 1 compared to t. But the internal rate of return also increases when the contribution rate is raised (Breyer, 2000). This is especially relevant for the introductory generation which enjoys—as per deﬁnition—an inﬁnitely high internal rate of return as the contribution rate jumps from zero to

Unfunded Pension Systems

13

a positive level with the start of the system. The same holds qualitatively—albeit on a smaller scale—for every increase in the contribution rate, which is equivalent to the introduction of a new (small) unfunded pension system. Since continuously raising the contribution rate is certainly not a strategy that can be sustained for a long time, we now assume for simplicity that contribution rates stay constant over time, ut ¼ utþ1 ¼ u:4 As stated by Aaron (1966),5 the internal rate of return can then be approximated by the sum of the growth rate of the wages and the growth rate of the population, which corresponds to the growth rate of the wage sum6 itþ1 < gtþ1 þ ntþ1 :

ð2:12Þ

2.1.2. Funded pension systems A funded system is based on a fund, meaning that contributions in period t ﬁnance beneﬁts in period t þ 1: The budget constraint of funded pension systems makes it necessary that Contributionst ð1 þ rtþ1 Þ ¼ Benefitstþ1 ;

ð2:13Þ

where rtþ1 is the rate of return which can be achieved by investing in the capital market. The contributions of the funded system in period t are identical to an unfunded pension system Contributionst ¼ ut vt NtY ;

ð2:14Þ

and beneﬁts are deﬁned as total pension beneﬁts in period t þ 1: O : Benefitstþ1 ¼ ptþ1 Ntþ1

ð2:15Þ

Note that contributions and beneﬁts concern the same individuals. The young in O according to Equation 2.1, i.e. period t are the old in period t þ 1; NtY ¼ Ntþ1 those who contribute in period t are identical to those who receive the beneﬁts in period t þ 1: There is no redistribution across generations or across individuals. 4

In a closed economy, the contribution rates are in fact limited from above by 100% and in an open economy, the opposition of foreign creditors creates an upper bound. 5 Samuelson (1958) has derived this relation in a pure consumption-loan model. In this model without wages, the rate of return equals the “biological percentage growth rate” (p. 472), i.e. the growth rate of the population. 6 For non-constant contribution rates, the internal rate of return corresponds to the growth rate of total social security contributions, respectively.

14

Unfunded Pension Systems: Ageing and Migration Inserting Equations 2.14 and 2.15 into Equation 2.13, we get O ut vt NtY ð1 þ rtþ1 Þ ¼ ptþ1 Ntþ1 ;

ð2:16Þ

and for the pension level ptþ1 ¼ ut vt

NtY ð1 þ rtþ1 Þ ¼ ut vt ð1 þ rtþ1 Þ: O Ntþ1

ð2:17Þ

2.1.2.1. Internal rate of return Although it is straightforward to determine the internal rate of return for funded pension systems, we want to do the exercise for reasons of symmetry. iFtþ1 is deﬁned to equalise individual contributions and individual beneﬁts in the funded pension system,

ut vt ð1 þ iFiþ1 Þ ; ptþ1 :

ð2:18Þ

With Equation 2.17, we get 1 þ iFtþ1 ¼

ptþ1 u v ð1 þ rtþ1 Þ ¼ t t ¼ 1 þ rtþ1 ; ut vt ut v t

ð2:19Þ

and for the internal rate of return iFtþ1 ¼ rtþ1 :

ð2:20Þ

The internal rate of return of funded pension systems is equal to the market rate of return, which is not surprising since the contributions have been invested at exactly this rate. In contrast to unfunded systems, there is no possibility to increase the internal rate of return in funded systems at the expense of other generations via an increase in the population or in the wages, let alone via an increase in the contribution rates. 2.1.3. Comparison Now we compare the internal rate of return of unfunded and funded pension systems in order to be able to say something about the attractiveness of the respective systems. Again, it is important to note that this evaluation is based on the hypothetical case where the introduction of a (new) pension system is discussed. Though this is not the relevant point to start from while discussing reform options given that one system—namely an unfunded pension system—already exists, it is useful as an exercise and helps us to understand the factors that determine the superiority of one system over the other from an ex ante point of view.

Unfunded Pension Systems

15

As we have seen in Equations 2.12 and 2.20, the internal rate of return of unfunded pension systems is equal to the growth rate of the wage sum, whereas contributions to funded pension systems grow with the market rate of return. It is a theoretical as well as an empirical question: which system has the higher rate of return—in the long run and for previous decades? 2.1.3.1. Growth model A theoretical relation between the growth rate of the wage sum and the market rate of return can be established within a neoclassical growth model a` la Solow (1956) and Swan (1956).7 This model derives the conditions for a steady state of the economy, which is deﬁned by a constant capital intensity, i.e. a constant capital – labour ratio k. All investments are thus used to guarantee that capital grows at the same rate as labour. This requires that the returns to both investments are the same. The marginal productivity of capital fk ðkÞ; i.e. the interest rate or what we have called so far the market rate of return, must be equal to the growth rate of labour n which corresponds to the growth rate of the wage sum for constant wages and—if we assume that the wage sum is a constant ratio of the national product— to the growth rate of the national product. Growth then follows the Golden Rule of Capital Accumulation yielding a maximum of consumption fk ðkgold Þ ¼ n:

ð2:21Þ

For a capital intensity smaller than at the Golden Rule Point, k , kgold ; the interest rate exceeds the growth rate of the wage sum (undercapitalisation). This makes it more attractive to invest in capital relative to labour. More capital investment, however, reduces the return to capital and thus increases the return to labour relative to the return to capital—if the production function fulﬁls the normal requirements of positive, but decreasing, marginal productivities. A constant capital intensity is reached when both returns are again equal. For a larger capital intensity than at the Golden Rule Point, k . kgold ; the mechanism is the inverse (overcapitalisation). However, the movement towards the Golden Rule Point from the state of undercapitalisation, k , kgold ; requires that at least one generation reduces the consumption level to save more and to invest more. This shows that this is not a Pareto-improving transition. A capital intensity k , kgold is not Pareto dominated by any other capital intensity. This area is therefore called dynamically efﬁcient. 7

See Appendix A2 for a more formal derivation of the model.

16

Unfunded Pension Systems: Ageing and Migration

The situation is different for the state of overcapitalisation, k . kgold ; where the capital intensity is higher than at the Golden Rule Point. Saving less reduces overcapitalisation and increases consumption. This brings the economy closer to the Golden Rule Point and to maximal consumption. All generations are able to consume more, which indicates that this is a Pareto-improving transition. This area is therefore called dynamically inefﬁcient. Since we do not expect an economy to be dynamically inefﬁcient in the long run, we either have the case where the economy is in the Golden Rule Point—which is probably a case of academic interest only—or the case where the economy is dynamically efﬁcient. But as we have just seen dynamical efﬁciency, i.e. k , kgold ; implies that the interest rate exceeds the growth rate of the wage sum. Referring to the comparison of internal rates of return of funded and unfunded pension systems, the theoretical results of the Solow – Swan growth model imply that in the long run the internal rate of return of funded pension systems is at least as high as the internal rate of return of unfunded pension systems.8 2.1.3.2. Empirical evidence As dynamical efﬁciency is a long-run concept, it is an empirical question where the economy is right now relative to the Golden Rule Point and where it was during the preceding decades. Let us start with the market rate of return as a benchmark against which the internal rate of return of unfunded pension systems has to be measured. As Breyer (2000) points out, this exercise is complicated by the fact that there is not one market rate of return but many different rates depending on the risk-return combination of the investment portfolio. The market rate of return for German government bonds of 10 and more years was, for example, 4.1% for the period 1970 – 1994 according to the German Council of Economic Experts (Sachversta¨ndigenrat, 1996). For the United States, the President’s Commission (2001) assumes a rate of return net of administrative costs of 4.6%. A portfolio which reﬂects stock indices has yielded rates of return ranging from 2 7 to þ 16% on average for a 10-year investment, depending on the market and particularly on the investment period (Figure 2.2). While the development in Germany (Dax) and in the United States (Dow Jones) is very similar, with rates of return under 5% for 10-year investments ending between 1970 and 1980 and a sharp increase to rates of return mostly between 10 and 15% for investments ending during the 1990s, investing in the stock market was not so favourable in 8

To be more exact, an equilibrium can be efﬁcient or inefﬁcient in the long run. Dynamical efﬁciency in an OLG model can, however, be reached when land is included in the analysis which we implicitly assume here. See Rhee (1991) or Homburg (1991).

Unfunded Pension Systems

17

20%

Rate of return

15%

10%

5%

0%

–5% Dow Jones

Nikkei

Dax 2002

1998

1994

1990

1986

1982

1978

1974

1970

–10% Year

Figure 2.2: Rates of return of stock investments (nominal). Average rate of return for a 10-year investment in the Dax, Dow Jones, or Nikkei Index ending on 1st July of the indicated year. Source: yahoo (2002).

Japan. The Nikkei Index has depreciated since 1989, yielding negative rates of return for 10-year investments falling (partly) in this period. Empirical studies about the internal rate of return of unfunded pension systems show that not even the relatively low rate of return for government bonds can be beaten. When interpreting the results, it is, however, necessary to keep in mind that these studies face the problem that unfunded pension systems are characterised by non-constant contribution rates and involve intragenerational redistribution. It is difﬁcult to calculate the pure internal rates of return in isolation from these two factors. For Germany, Schnabel (1998) ﬁnds internal rates of return for married men and women born in 1930 to be about 3.5%. These rates of return decrease to almost 0% for married men and women born in 1980. More optimistic assumptions with respect to the development of immigration, life expectancy, labourforce participation and retirement age lead to slightly higher rates of return for later born cohorts ranging from 0.5 to 1.0%. The rates of return for single men and women are 1.5 or 0.5 percentage points lower, respectively, for all birth cohorts because of the lack of survivor beneﬁts as part of the pension beneﬁts. Calculations with the pension model developed by CESifo for the Council of Advisors of the German Ministry of Economics (Wissenschaftlicher Beirat,

18

Unfunded Pension Systems: Ageing and Migration

1998)9 also show that the internal rate of return for a German male retiree with a typical work biography has decreased over recent decades. While the cohort born in 1937 receives 3%, the internal rate of return is only about 1.6% for the cohort born in 1980. This downward development closely follows the growth rate of the wage sum.10 The result of the Council of Advisors for Germany is conﬁrmed by Boldrin et al. (1999), who calculated internal rates of return for the average worker in 10 countries of the European Union based on the rules in force in 1999. The birth cohorts concerned thus comprise those who work in 1999. Assuming that these rules stay constant and that workers have the employment history of the average worker, they ﬁnd internal rates of return of 3.1% on average with a minimum of 2.0% (Germany and Italy) and a maximum of 4.9% (Sweden). If we look at the projection for the relevant factors, which we will do in more detail in Chapter 3, it is important to keep in mind that the internal rate of return of unfunded pension systems depends on the growth rate of the wage sum and not on the growth rate of the population alone. A decrease in the (working-age) population can be compensated by an increase in productivity—and in fact, with a production function with positive but decreasing marginal productivities, less labour increases the marginal productivity of labour and correspondingly the wage.11 To sum up, we have seen that during the last decades the rate of return of funded pension systems exceeded the rate of return of unfunded ones—at least on average. The economy was in the dynamically efﬁcient area. The variance of the returns of funded systems, however, was also larger. When considering the introduction of a pension system, this risk – return relation has to be taken into account. Investments on the capital market for old-age provision via a funded pension system not only face a more risky return, but also are not immune against unexpected inﬂation and need to be converted into annuities to counter the risk of unexpected longevity. Unfunded pension systems are also not without risk, even though the risk is of a different nature. Retirees in unfunded pension systems have to rely on the continuity of the pension system, i.e. that the social contract between the young and the old generations is not cancelled or changed in a fundamental way. Even if this is not a very probable option—especially in countries where pension claims are equivalent to property rights12—there is no insurance against 9

For further details see Sinn and Thum (1999), Sinn (1999b) and Thum and von Weizsa¨cker (2000). See Appendix A5 for more details about the assumed work biography of the typical male retiree. 11 This mechanism, however, does not work when a certain wage –pension ratio is aimed at—as is the case in Germany. Then, higher wages automatically lead to higher pension beneﬁts (Keller, 2000). 12 The German Supreme Court, for example, has conﬁrmed in several decisions that pension claims are property rights in the sense of Art. 14 of the Basic Right; this is, however, not the case in the United States. 10

Unfunded Pension Systems

19

reductions of the level of the contribution rate. Although the young might not have the voice option to impose this measure via the political process, there is still the exit option via a reduced labour supply in the form of emigration or self-employed or black-market activities. This could exert a sufﬁciently high downward pressure on contribution rates.13 When the task is to introduce the optimal system and the various risks mentioned are relevant, a mixture of both systems has some merits. Merton (1983) shows, for example, for an economy with uncertainty and nontradable human capital, that a mixed system can be justiﬁed as the unfunded part improves the efﬁciency of risk-bearing in this economy.

2.2. THE CONCEPT OF IMPLICIT TAXES AND IMPLICIT DEBT The internal rates of return seem to be unrelated to the respective pension systems at ﬁrst sight. It can, however, be shown that the difference—or more precisely the size by which the internal rate of return of unfunded pension systems falls short of the market rate of return—depends on the introductory gains of the ﬁrst generation, i.e. the history of the unfunded pension system. We introduce the concepts of implicit taxes and implicit debt to develop this point in more detail before discussing other concepts to measure the burden implicit in unfunded pension systems.14 2.2.1. Implicit taxes The implicit tax is the difference in present value terms between pension beneﬁts and contributions. If the internal rate of return is equal to the market rate of return (funded system), the implicit tax is zero. If, however, the internal rate of return falls short of the market rate of return (unfunded system), part of the contributions is lost or would not be necessary if contributions were invested in the capital market instead. This means that the size of the implicit tax Tt and thus the implicit tax rate tt depends on the difference in rates of return between these two systems. This is also reﬂected in the degree of actuarial fairness linking contributions and beneﬁts. A very weak link, and thus a low rate of return compared to the market rate of return, implies a high implicit tax. In a system with ﬂat-rate beneﬁts without any link, i.e. an unfunded system without intragenerational fairness,15 the entire value of the contributions must be regarded as an implicit tax ðtt ¼ 1Þ: 13

We will focus on these aspects in Chapters 6, 7 and 9. See Lu¨deke (1988) for an early contribution. 15 The terms “fairness” or “intragenerational fairness” (Homburg and Richter, 1990) are meant to say that the implicit taxes are the same for all members of a given age cohort. 14

20

Unfunded Pension Systems: Ageing and Migration

In contrast to this, a very strong link, and thus a rate of return close to the market rate of return, results in a low implicit tax. In a system with a perfectly actuarial link, i.e. a funded system where the internal rate of return is equal to the interest rate, no implicit tax exists ðtt ¼ 0Þ: In addition, there is the intermediate case. Contributions and beneﬁts are linked via the growth rate of the wage sum, i.e. an unfunded system with intragenerational fairness, and the implicit tax rate is tt with 0 , tt , 1: Then the growth rate of the wage sum relative to the market rate of return determines the exact size of the implicit tax rate. In terms of redistribution, a perfectly actuarial system, where the internal rate of return is equal to the interest rate (funded system), does not involve any redistribution among individuals whereas in a system with only partial or no actuarial fairness, there is redistribution among individuals belonging to the same or different generations. In what follows, we concentrate on the implicit tax and the implicit tax rate in an unfunded pension system with an imperfectly actuarial link between contributions and beneﬁts and compare the results to a funded pension system. The size of the implicit tax Tt can be derived by comparing pension beneﬁts and contributions in present value terms. In our simple two-period framework, we get for the implicit tax Tt ¼ ut vt 2

ptþ1 u v ð1 þ itþ1 Þ ¼ ut vt 2 t t ; 1 þ rtþ1 1 þ rtþ1

ð2:22Þ

which is the difference between contributions to the pension system (ﬁrst term) and pension beneﬁts from the pension system (second term)—both discounted to period t: The last equality follows from Equations 2.9 and 2.18, respectively, where i ¼ i; iF depending on whether the pension system is unfunded or funded. Rewriting Equation 2.22, we get 1 þ itþ1 Tt ¼ 1 2 u v ¼ tt ut vt ; ð2:23Þ 1 þ rtþ1 t t where the implicit tax rate tt is deﬁned by the term in parentheses. With the discussion in section 2.1 and Equations 2.12 and 2.20, it is straightforward to conclude that the implicit tax rate tt and correspondingly the implicit tax Tt are zero in a funded system with i ¼ iF ¼ r and positive in an unfunded system with i ¼ i , r: In what follows we concentrate on unfunded pension systems, i.e. i ¼ i , r: The implicit tax rate tt can be written as

tt ¼

Tt r 2 itþ1 ¼ tþ1 : ut vt 1 þ rtþ1

ð2:24Þ

Unfunded Pension Systems

21

The size of the implicit tax rate, i.e. the part of the contributions lost for the contributor, depends again on the difference between the market rate of return and the internal rate of return of the unfunded pension system.16 This allows rewriting total contributions Qt ¼ ut vt by dividing them up into two parts: implicit tax Tt ¼ tt ut vt according to Equation 2.23 and implicit saving St ¼ ð1 2 tt Þut vt ; so that

Q t ¼ Tt þ S t :

ð2:25Þ

Implicit saving St is the part of total contributions which represents savings in the sense that it would be sufﬁcient to yield the pension beneﬁts if the rate of return were the interest rate. This can be seen when rewriting St using Equations 2.9 and 2.24 St ¼ ð1 2 tt Þut vt ¼

1 þ itþ1 ptþ1 uv ¼ : 1 þ rtþ1 t t 1 þ rtþ1

ð2:26Þ

Finally, comparing Equation 2.22 with Equations 2.25 and 2.26 shows that deﬁning implicit taxes as the discounted difference between total contributions and pension beneﬁts is equivalent to deﬁning them as total contributions minus implicit saving. This equivalence also holds for the terms taken one by one. In both cases, the size of the implicit tax and the implicit tax rate depend on the relation between the internal rate of return and the market rate of return. 2.2.2. Implicit debt We have once more pinned down the difference between unfunded and funded pension systems to differences in the respective rates of return. The origin of these differences will now be explained. 2.2.2.1. Theoretical derivation When looking at the history of an unfunded pension system, an unequal treatment of different generations is evident. The ﬁrst generations have greatly beneﬁted from the introduction of the system as they have received beneﬁts while having contributed (almost) nothing. As we have seen above, the internal rate of return for those generations can be very high, thus implying a negative implicit tax rate, i.e. 16

In the following chapters, we will discuss in more detail the factors determining the internal rate of return and thus the implicit tax rate—notably the development of the population growth through fertility, longevity and migration.

22

Unfunded Pension Systems: Ageing and Migration

an implicit subsidy rate, according to Equation 2.24. As has been discussed in section 2.1 the later generations, however, face a relatively low internal rate of return compared to the interest rate and consequently a positive implicit tax rate. The pay-as-you-go pension system involves redistribution from later generations to introductory generations. Pension beneﬁts of the current generation of retirees can be regarded as the debt implicit in the unfunded pension system Dt ¼ pt NtO ¼ ut vt NtY ;

ð2:27Þ

where the last equality comes from Equation 2.7 stating that the pension beneﬁts of each (old) generation—and also of the current generation of retirees—are equal to the contributions of the following (young) generation. Sinn (2000a) shows that the implicit debt can also be written as17 Dt ¼

1 X

Tj NjY Rj ;

ð2:28Þ

j¼t

with Rj as the discount factor for the value of a tax in period t paid in period j j Y 1 for j . t and Rt ¼ 1 for j ¼ t; ð2:29Þ Rj ¼ 1 þ rk k¼tþ1 and assuming that the elements of the sum in Equation 2.28 converge. We ﬁnd that the present value of the implicit debt of the pay-as-you-go pension system is equal to the present value of the implicit taxes of all subsequent generations. This statement is valid for each point in time during the existence of an unfunded pension system, and also for the claims of pension beneﬁts of the ﬁrst generation D0 ¼

1 X j¼0

Tj NjY Rj ¼

1 1 X X rj 2 i j uj vj NjY Rj ¼ tj uj vj NjY Rj ; 1 þ r j j¼0 j¼0

ð2:28aÞ

where the last equalities follow from Equation 2.23. The claims of the ﬁrst generation equal the implicit taxes of all subsequent generations or, to say it differently, the gift to the introductory generation has created an implicit debt that must be serviced by all subsequent generations. The implicit tax of the unfunded pension system is the tax necessary to service this debt. The last equality in Equation 2.28a emphasises again the relation between the size of the implicit tax necessary to service the implicit debt and the difference between the market rate of return and the internal rate of return of the unfunded 17

See Appendix B2 for the derivation.

Unfunded Pension Systems

23

Table 2.1: Rates of return.

Introductory generation Later generations

r>i

r5i

r
þ 2

þ 0

þ þ

Source: Lindbeck and Persson (2003, p. 80).

pension system. The larger the difference, the larger the implicit tax and the implicit debt and thus the intergenerational redistribution from later generations to the ﬁrst. Table 2.1 shows that the introduction of an unfunded pension system always beneﬁts the introductory generation. How later generations are affected depends on the difference between the market rate of return and the internal rate of return. Only if the economy is in the dynamically inefﬁcient area or in the Golden Rule Point ðr # iÞ—something we have ruled out theoretically and empirically for the long run (or considered unrealistic) in section 2.1—are later generations not worse off. For the dynamically efﬁcient case ðr . iÞ; however, later generations lose for the beneﬁt of the introductory generation. Sinn calls the pay-as-you-go pension system a “zero-sum game among generations” (2001, p. 81) which involves redistribution between generations without creating aggregate gains or losses. This result will be important when we discuss the possibility of a Pareto-improving transition in Chapter 5.

2.2.2.2. Empirical evidence In order to calculate the implicit debt, it is best to depart from Equation 2.27, which expresses the implicit debt as the sum of the pension claims acquired up to today. The calculation is relatively easy for those already retired, but more difﬁcult for those currently in the labour force. To arrive at the exact value of the implicit debt, one would have to take into account the detailed rules governing the individual pension systems (Holzmann, 1998). But this means that many different cases have to be considered. For example, what to do if an individual has acquired pension rights with 10 years of contributions but the minimum number of years required in order to qualify for pension beneﬁts are 15 years? And how to proceed when pension beneﬁts are linked to the wage earnings of the best 10 years yet to come? As an approximation, pension beneﬁts are often calculated pro rata on the basis of the actual number of years of contributions relative to the number of years

24

Unfunded Pension Systems: Ageing and Migration

necessary for full pension beneﬁts. The main assumptions needed concern the path of the interest rate, the growth rate of productivity and population and the survival probabilities. Based on the CESifo pension model and the eighth co-ordinated population projection of the Federal Statistical Ofﬁce Germany, the Council of Advisors of the German Ministry of Economics (Wissenschaftlicher Beirat, 1998) has estimated the implicit debt for Germany to be in the range of e5.1 trillion.18 Werding (2002b) conﬁrms this estimate based on the ninth co-ordinated population projection and shows that the implicit debt has remained almost unchanged despite the recent pension reforms. In particular, he ﬁnds an implicit debt of e5.2 trillion based on the rules of the last pension reform, the so-called Riester-Reform (II) of 2001.19 This corresponds to 255% of GDP. For the same scenario, calculations based on the more optimistic population projection of Eurostat (2000) yield an implicit debt of about e4.7 trillion or 232% of GDP. For some major OECD countries, Van den Noord and Herd (1993) and Chand and Jaeger (1996) calculate implicit debts (Table 2.2). They ﬁnd values in the range of 90 –360% for the seven countries examined, with some variations. In particular, Chand and Jaeger report an implicit debt of 221% of GDP for West Germany which exceeds the implicit debt of 157% calculated by Van den Noord and Herd. Despite different values of the calculated implicit debt for some countries, the concept of implicit debt is very useful as it gives an idea about the liabilities in unfunded pension systems. These liabilities, which present ﬁnancial obligations of the state,20 must be taken into account when thinking about fundamental reforms of the pension system—especially a (partial) transition from an unfunded to a funded pension system. Holzmann (1998) proposes a rule of thumb to capture the size of the implicit debt. He illustrates that the implicit debt can be approximated by pension expenditure multiplied by a factor of 20– 30. For Germany with pension expenditure of 11.8% of GDP in 2000 (Economic Policy Committee, 2001) this is conﬁrmed: the implicit debt ranges between the 20-fold and 22-fold of pension expenditure depending on the assumptions made. The implicit debt of the other 18

See also Sinn (2003b). See Chapter 4 for details about this reform. 20 It should be noted, nevertheless, that explicit and implicit debt are different with respect to some aspects. Claims on public pension, for example, cannot be traded and cannot be determined unambiguously because of uncertainties about a number of relevant variables (for example, wage growth rate and population growth rate). Above all, the amount of implicit debt is subject to discretionary choices of the debtor (state). 19

Unfunded Pension Systems

25

Table 2.2: Implicit debt as accrued rights minus existing assets (in % of GDP).

Canada France West Germany Italy Japan United Kingdom USA

Van den Noord and Herd (1993)

Chand and Jaeger (1996)

121 216 157 242 144 156 90

94 265 221 357 166 117 108

Source: Van den Noord and Herd (1993): Table 7 (“gross liabilities-accrued rights” minus “assets-existing”); Chand and Jaeger (1996): Table 16 (“pension liabilities present retirees” plus “pension liabilities present work force”).

countries of the European Union can be deduced accordingly (Table 2.3). Of course, this rule is a rough estimation which is too rough if the impact of pension reforms on the implicit debt is analysed. Ageing of the population, modiﬁcations of the indexation rule or more fundamental changes, for example a partial transition to a funded system, have an impact on pension expenditure which is only fully visible over the course of some decades while it immediately affects the implicit debt. Table 2.3: Pension expenditure (in % of GDP). 2000 Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom

14.5 10.0 10.5 11.3 12.1 11.8 12.6 4.6 13.8 7.4 7.9 9.8 9.4 9.0 5.5

EU

10.4

Source: Economic Policy Committee (2001).

26

Unfunded Pension Systems: Ageing and Migration

2.2.3. Other concepts In the literature, at least two other concepts exist to measure the burden of redistributive activities among generations both including implicit debt: net pension liabilities and generational accounting (Fenge et al., 2002b). Both concepts are based on a longer time perspective which takes into account more generations up to inﬁnitely many compared to the concept of implicit debt which only considers those generations which have already acquired pension claims. We will discuss both in turn. We then give reasons why we consider implicit debt as the concept best suited for our purpose.

2.2.3.1. Net pension liabilities A few years ago, the OECD developed the concept of net pension liabilities (van den Noord and Herd, 1993) as an extension of the concept of implicit debt.21 The essential difference between these two approaches is that net pension liabilities not only incorporate the present value of accrued rights (net of asset, if any) but also the present value of future claims resulting from future contributions plus these future contributions. The assumption is that currently employed individuals continue to pay contributions and acquire pension claims and individuals who are already born or not yet born will also pay contributions and acquire pension claims on the basis of the current rules when entering the labour market in the future. Van den Noord and Herd (1993) calculate the present value of future contributions and pension expenditure for cohorts living up to the year 2160. Based on data from 1990, they ﬁnd for the seven major OECD countries net pension liabilities range from 43% of GDP in the United States to 250% of GDP in Canada. The net pension liabilities in Germany amount to 160%, in France to 216% and in Italy to 233%. The United Kingdom has net pension liabilities of 186% of GDP and Japan of 200% of GDP. For some countries (Germany, France, and Italy) future claims and future contributions cancel in present value terms. Net pension liabilities and implicit debt then coincide (see Table 2.2). In the case of the United States, future contributions even exceed future claims by 50% of GDP. 21 This concept has been further developed in a series of follow-up studies. At the moment, the OECD (for example, OECD, 1998) is co-ordinating a project called “Maintaining prosperity in an ageing society” based on reports from national governments and experts. See Werding and Blau (2001) for the German contribution.

Unfunded Pension Systems

27

2.2.3.2. Generational accounting Generational accounting further extends the concept of implicit debt (Auerbach et al., 1991, 1994). Not only already acquired and future pension claims as well as future contributions are included in present value terms, but also all ﬁscal activities and the taxes and transfers they involve. Implicit and explicit debt are thus both part of the calculations. All aggregate results are broken down to an individual level, i.e. to different age cohorts, so that the part of the implicit and explicit debt falling on individual generations becomes transparent. The idea is that an intertemporal budget constraint is imposed on all ﬁscal activities meaning that all expenditure (transfers, investment, debt services, etc.) must be covered either by government net wealth or must be paid by some—present or future— generations in the form of taxes and social security contributions. Fiscal imbalances show up if the tax and contribution payments net of transfers are negative for the newborn, thus implying that these generations receive more in educational expenditure and old-age security than they pay in taxes and contributions during their working life. As the intertemporal budget constraint must be respected in the end, future generations then have to face a tax burden which exceeds the transfers they get. In many countries, there exist severe intergenerational imbalances between present and future generations (Ja¨ger and Raffelhu¨schen, 1999). To measure this, it is assumed that all future generations face the same ﬁscal policy as the newborn of the base-year generation. If the intertemporal budget constraint does not hold, the present ﬁscal policy is unsustainable and present generations enjoy redistributions at the cost of future generations. The residual to balance this constraint is called intertemporal public liabilities. Implicit debt in the context of generational accounting is then deﬁned as the difference between the intertemporal public liabilities and explicit debt; it is therefore understood in a wider sense similar to the net pension liabilities of the unfunded pension systems as deﬁned above. If nothing changes for present generations, the tax burden for future generations has to double in Spain and to increase by 74% in the United Kingdom to service the intertemporal public liabilities. Future generations in Germany and Italy will see taxes rise by 59 and 53%, respectively, and in France by 34%.22 To be able to arrive at more differentiated results the scope of components included has been enlarged and the number of disaggregated budget items has been increased. When limited to the pension system, the concepts of generational 22

Note that as intergenerational accounting takes implicit and explicit debt into account, it is not possible to deduce unambiguously whether a country has higher or lower implicit debt or pension expenditure (Table 2.3).

28

Unfunded Pension Systems: Ageing and Migration

accounting and implicit debt or implicit taxes, respectively, yield comparable results for already born generations. Generational accounting, however, typically distributes the remaining burden equally across future generations. Although appropriate for some questions, this method is not suited to analyse reforms of the pension system which affect different generations in different ways (Thum and von Weizsa¨cker, 2000). This has been taken into account and corrected in newer studies. Borgmann et al. (2001) take advantage of the more extensive approach of generational accounting to analyse recent pension reforms in Germany with a special emphasis on the interaction between tax and pension systems.23 As the German pension system relies on state subsidies which are partly ﬁnanced by general tax revenues and partly by speciﬁc taxes, for example the newly introduced ecological tax on energy, petrol, oil and gas, generational accounting adapted to this question can produce useful results.

2.2.4. Evaluation We have now discussed three different approaches to measure the burden due to intergenerational redistribution: implicit debt or implicit taxes, respectively, net pension liabilities and generational accounting. All three concepts have their merits as they make the order of magnitude of redistributive activities more transparent for the generations concerned. In addition, they serve to analyse the effect of reform proposals on the total burden and on the burden different generations have to carry. Only if voters are aware of the status quo and know what is at stake when the status quo is maintained in the presence of the ageing of the population is there a chance that reforms ﬁnd the support of a majority in a democratic electoral process. The concepts which are most appropriate depend on the purpose of the analysis. In general, the selected concept should be based on calculations which involve as few individuals, generations or components of the pension budget or public budget as possible and as many as necessary to allow well-founded conclusions. If the focus is on the intertemporal constraint and the ﬁnancial sustainability of the pension budget or of the public budget, the chosen approach must incorporate a long-term and extensive perspective. Net pension liabilities and generational accounting are then the most suited concepts. We are, however, interested in a different aspect: what is the value of the accrued pension rights? This allows us to judge the order of magnitude of intergenerational redistribution via the unfunded 23

See also Bonin and Feist (1999).

Unfunded Pension Systems

29

pension system between introductory generations and later generations. This is especially important to know when considering a (partial) transition from an unfunded to a funded pension system and the costs this involves.24 By costs we mean the debt implicit in the unfunded pension system in the form of accrued pension rights which must be liquidated and converted into an explicit debt. The appropriate approach in this context is thus to calculate implicit debt based on the pension claims acquired by individuals up to the point in time for which the transition is envisaged (Holzmann et al., 2001).

2.3. CONCLUSION We have seen that the size of implicit taxes and implicit debt depends on the difference between the internal rate of return and the market rate of return, and the internal rate of return is determined by the growth rate of the wage sum. The wage sum grows in line with productivity and the population. In Chapter 3, we will look at the projected demographic development in order to evaluate how the ageing of the population will affect the intergenerational distribution of the pension burden as expressed in implicit taxes and implicit debt. In the following chapters, we will then make use of the concepts of implicit taxes and implicit debt in various contexts, which include the intrapersonal and intergenerational distribution of the pension burden. We analyse the consequences this has for the ﬁnancial and political sustainability of the system and for the labour-leisure decision over the life cycle as well as for the allocation of labour in an international context. APPENDIX A2. SOLOW– SWAN GROWTH MODEL25 We depart from the two-period OLG model as introduced in section 2.1. In each period t; two generations are alive: one young generation with NtY members and one old generation with NtO members. The utility function for an individual

24

In fact, every reduction in the size of unfunded pension systems can be interpreted as a (partial) switch to a funded pension system as the saved resources due to lower contribution rates can be invested in the capital market thus starting an individual (complementary) funded system. If we assume that the market rate of return equals the discount rate, the present value of the generated cash ﬂow is zero, which renders the concrete form of this saving irrelevant. 25 See, for example, Wellisch (1999) or Blanchard and Fischer (1993) for an extensive treatment of the Solow–Swan model.

30

Unfunded Pension Systems: Ageing and Migration

born in period t comprises ﬁrst-period consumption cY t and second-period consumption cO tþ1 O Ut ðcY t ; ctþ1 Þ;

ðA2:1Þ

where U is a strictly monotone and quasi-concave ordinal utility function. Let the production function take the form Yt ¼ FðKt ; Lt Þ;

ðA2:2Þ

with capital K and labour L where labour is given in efﬁciency units. In what follows, we assume that all individuals of the young generation work, i.e. Lt ¼ NtY : Output is a homogeneous good that can be either consumed or invested to create new capital K: The function is assumed to be neoclassical, thus exhibiting positive and diminishing marginal productivities and constant returns to scale and fulﬁl the Inada conditions. Constant returns to scale allow rewriting of the production function as Yt ¼ FðKt ; NtY Þ ¼ NtY ·FðKt =NtY ; 1Þ ¼ NtY ·f ðkt Þ;

ðA2:3Þ

yt ¼ f ðkt Þ;

ðA2:4Þ

or where kt ; Kt =NtY is the capital – labour ratio and yt ; Yt =NtY is per capita output—assuming all young individuals of generation t supply efﬁciency units of labour normalised to 1. The marginal productivity of capital—which will be important later—is given by Fk ðKt ; NtY Þ ¼ NtY ·

›f ›k t 1 ¼ NtY ·fk Y ¼ fk : ›k t ›K t Nt

ðA2:5Þ

The slope of the production function thus corresponds to the marginal productivity of capital, which equals the interest rate if we assume perfect competition on the capital market. As already stated, output can be either consumed or invested where investing leads to a higher capital stock of the economy one period later. The net increase in the capital stock over time is gross investment minus depreciation. In what follows, we concentrate on the simplest case where gross investment equals net investment thus abstracting from depreciation. Net investment I is then the difference between the capital stock in t þ 1; Ktþ1 ; and the capital stock in t; Kt It ¼ Ktþ1 2 Kt ;

ðA2:6Þ

which corresponds to the aggregate non-consumption or the aggregate saving respectively in period t: The budget constraint for a closed economy without state

Unfunded Pension Systems

31

activities is then O O FðKt ; NtY Þ ¼ NtY cY t þ Nt ct þ Ktþ1 2 Kt :

ðA2:7Þ

This means that in each period t, the sum of consumption of the young and the old generations plus net investment must equal the output of the economy. By dividing Equation A2.7 by NtY and using NtO ð1 þ nÞ ¼ NtY ; we get f ðkt Þ ¼ cY t þ

cO t þ ð1 þ nÞktþ1 2 kt : ð1 þ nÞ

ðA2:8Þ

A long-run allocation in a dynamical economy is characterised by a steady state. All per-capita variables are then constant over all periods, implying that all variables grow with the same rate n: In particular, the capital intensity stays Y Y constant ktþ1 ¼ kt ¼ k and the consumption when young and old cY tþ1 ¼ ct ¼ c O O O and ctþ1 ¼ ct ¼ c : Equation A2.8 then becomes f ðkÞ ¼ cY þ

cO þ nk: ð1 þ nÞ

ðA2:9Þ

For the efﬁcient allocation which maximises the utility of all generations in the long run, we maximise Equation A2.1 taking the budget constraint in Equation A2.9 into account max u½cY ; ð1 þ nÞðf ðkÞ 2 cY 2 nkÞ: Y k;c

ðA2:10Þ

The maximal consumption is reached for fk ðkp Þ 2 n ¼ 0:

ðA2:11Þ

If we call the value of the steady-state capital intensity that leads to maximal consumption kgold ; then kgold is determined by fk ðkgold Þ ¼ n:

ðA2:12Þ

This condition is called the Golden Rule of Capital Accumulation. According to Equation A2.5, the left-hand side of Equation A2.12 is the interest rate. The Golden Rule of Capital Accumulation thus says that the interest rate has to be equal to the growth rate of labour which corresponds to the growth rate of the wage sum for constant wages and to the growth rate of the national income if the wage sum is a constant ratio of the national income.

32

Unfunded Pension Systems: Ageing and Migration

APPENDIX B2. DERIVATION OF EQUATION 2.28 Following Sinn (2000a), we substitute ptþ1 in Equation 2.26 with Equations 2.7 and 2.1: St NtY ¼

Y utþ1 vtþ1 Ntþ1 Q NY ¼ tþ1 tþ1 : 1 þ rtþ1 1 þ rtþ1

ðB2:1Þ

With Equation 2.25, we can write St NtY ¼

Y Ttþ1 Ntþ1 S NY þ tþ1 tþ1 ; 1 þ rtþ1 1 þ rtþ1

ðB2:2Þ

and similarly for any period j . t; j $ 1: Taking Equation 2.27 and substituting again Qt according to Equation 2.25 gives Dt ¼ Tt NtY þ St NtY :

ðB2:3Þ

Replacing the last summand with Equation B2.2, we get Dt ¼ Tt NtY þ

Y Ttþ1 Ntþ1 S NY ¼ tþ1 tþ1 : 1 þ rtþ1 1 þ rtþ1

ðB2:4Þ

The last summand can again be replaced by Equation B2.2 for t þ 1: Continuing this operation ad inﬁnitum, we get Dt ¼

1 X

Tj NjY Rj ;

j¼t

which corresponds to Equation 2.28 if Rj is deﬁned as in Equation 2.29.

ðB2:5Þ

CHAPTER 3

Projected Development of Fundamental Factors The path of factors inﬂuencing demography in the next few decades Demography is destiny

Auguste Comte (1798 – 1857)

Kinder kriegen die Leute immer.1

Konrad Adenauer (1957)2

Abstract In the next 50 years, the population in most industrialised countries is expected to decline. Life expectancies will rise and fertility rates will fall short of the replacement rate. Immigration can only alleviate this development to a certain extent as the migration volumes needed to counterbalance the low fertility rates and the high life expectancies on a one-to-one basis would be enormous. The decreasing total population will fundamentally alter the ratio of (young) contributors to (old) recipients. This will have serious consequences for the ﬁnancial sustainability of unfunded pension systems. If Auguste Comte is right, then not only demography, but also the future of the pension system follows an inevitable path. We have seen in Chapter 2 how population growth—beside productivity growth—affects the implicit tax and the implicit debt of unfunded pension systems. Although other factors also matter and demographic trends can change, the status quo and the projected development of 1 2

There will always be children (own translation). Cited in: Su¨ddeutsche Zeitung, 4 January 2003.

34

Unfunded Pension Systems: Ageing and Migration

the population deserve a detailed analysis to get a better idea about what can be expected in the next few decades. In 1967, Paul Samuelson observed3: The beauty of social security is that it is actuarially unsound. Everyone who reaches retirement age is given beneﬁt privileges that far exceed anything he has paid in […] Always there are more youths than old folks in a growing population. More important, with real incomes growing at some 3% a year, the taxable base upon which beneﬁts rest in any period are much greater than the taxes paid historically by the generation now retired […] A growing nation is the greatest Ponzi game ever contrived. In this chapter, we ask whether the conditions for a Ponzi-game-like pension system are still fulﬁlled and discuss the prospects for the future. Since the 1970s, real GDP per capita growth has been mostly below 3%—in Germany, for example, it is expected to grow only by 0.7% in 2002 according to projections of the Ifo Institute for Economic Research—and is expected to stay below 3% for most of the next few decades. A similar but even more dynamical development is projected for population growth. Therefore, the focus of this chapter lies on the ageing of the population in industrialised countries and in particular in Germany. Ageing will be considered as the simultaneous realisation of two phenomena: higher life expectancy and lower fertility—both increasing the number of “old folks”, i.e. potential recipients of pension beneﬁts, relative to the number of “youths”, i.e. potential contributors. The ageing process will be weakened or intensiﬁed by migration ﬂows which therefore need to be included in the analysis. In section 3.1, we present data and discuss the evolution of fertility, life expectancy and migration ﬂows. In section 3.2, we use the projections to derive the development of the total population and of the age structure. Section 3.3 concludes.

3.1. DETERMINANTS OF THE POPULATION GROWTH In this section, we analyse the evolution of the main determinants of the population growth, i.e. migration, fertility and life expectancy and evaluate the importance of each of them. 3

Cited in: The Economist, 14 February 2002.

Projected Development of Fundamental Factors

35

3.1.1. Fertility In 2000, the fertility rates in all 15 countries of the European Union (EU) were well below the replacement rate of 2.1 children per woman needed to keep the population constant—with a signiﬁcant variance across countries, however (Table 3.1). Among the ﬁve largest countries, Spain and Italy bring up the rear with fertility rates of 1.19 and 1.22, respectively. Germany is only doing slightly better with an average of 1.4 children per woman. France and the United Kingdom stand out against these countries with fertility rates of 1.73 and 1.72, but also miss by far the replacement rate of 2.1. On average, the smaller countries are doing better, with Ireland on top with an average of 1.9 children per woman followed by Denmark, Finland, Luxembourg and the Netherlands with fertility rates in the range of 1.7 – 1.8. Although the fertility rates of all countries are expected to converge over the next 50 years to levels of 1.5 –1.8, this will still lag behind the replacement rate of 2.1. The United States experienced a similar development until the 1990s. Since then, the fertility rate has increased to about 1.9 and is projected to reach the replacement rate of 2.1 by 2050. The fertility rate in Japan follows more closely the European development, with an average of 1.3 children per woman in 2000 and an expected increase to 1.8 in 2050 (United Nations, 2001a).

Table 3.1: Fertility rates.

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom

2000

2010

2020

2030

2040

2050

1.54 1.77 1.40 1.34 1.19 1.73 1.89 1.22 1.72 1.71 1.31 1.53 1.73 1.50 1.72

1.68 1.76 1.47 1.45 1.34 1.79 1.83 1.36 1.75 1.79 1.41 1.64 1.69 1.61 1.75

1.74 1.79 1.50 1.52 1.42 1.80 1.82 1.43 1.79 1.79 1.45 1.69 1.70 1.70 1.79

1.77 1.80 1.50 1.56 1.48 1.80 1.81 1.48 1.80 1.79 1.48 1.70 1.70 1.77 1.80

1.80 1.80 1.50 1.60 1.50 1.80 1.80 1.50 1.80 1.80 1.50 1.70 1.70 1.80 1.80

1.80 1.80 1.50 1.60 1.50 1.80 1.80 1.50 1.80 1.80 1.50 1.70 1.70 1.80 1.80

Source: Eurostat (2000)—baseline scenario.

36

Unfunded Pension Systems: Ageing and Migration

3.1.2. Life expectancy The 15 countries of the EU show a more homogenous picture with respect to life expectancy. Men born in 2000 will live on average between 74 and 76 years—except for Portuguese (72) and Swedes (77). Until 2050, life expectancy is projected to increase by 4 –6 years to 79– 81 years. Portuguese will then live 78 years and Swedes 82 years (Table 3.2). Women of most European countries born in 2000 can expect to live between 79 and 81 years. In Spain, Sweden and Italy, life expectancy is 82 years and in France even 83 years. In 2050, women will live between 83 and 85 years—except for Italians, Austrians and Swedes with 86 years and French with the highest life expectancy of 87 years (Table 3.3). The development of life expectancy in the United States is very similar—increasing for men from 75 to 80 years over the next 50 years and for women from 80 to 85 years. Japanese men and women born in 2000, however, will live much longer, with 78 and 85 years, respectively. Life expectancy will even continue to increase strongly until 2050 to 84 years for men and 92 years for women (United Nations, 2001a). 3.1.3. Migration After analysing the mainly domestic determinants of population growth, we now want to look at the status quo and the projection for migration—with a special Table 3.2: Life expectancy (men).

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom

2000

2010

2020

2030

2040

2050

74.8 74.2 74.7 75.9 74.9 74.8 74.0 75.5 74.4 75.5 75.0 72.0 73.9 77.3 75.2

77.1 76.0 76.6 77.7 75.9 76.8 75.8 77.4 77.1 77.0 76.1 73.8 75.7 78.2 77.0

78.7 77.4 78.1 79.1 77.0 78.3 77.2 79.0 78.8 78.2 77.3 75.4 77.4 79.1 78.3

79.6 78.4 79.2 80.2 78.0 79.3 78.2 80.1 79.7 79.2 78.5 76.8 78.7 80.0 79.3

80.0 78.9 79.8 80.8 78.8 79.8 78.8 80.7 80.0 79.8 79.7 77.7 79.6 81.0 79.8

80.0 79.0 80.0 81.0 79.0 80.0 79.0 81.0 80.0 80.0 81.0 78.0 80.0 82.0 80.0

Source: Eurostat (2000)—baseline scenario.

Projected Development of Fundamental Factors

37

Table 3.3: Life expectancy (women).

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom

2000

2010

2020

2030

2040

2050

80.9 79.0 80.8 81.0 82.1 82.8 79.4 82.0 80.8 80.9 81.2 79.2 81.1 82.0 80.0

82.8 80.0 82.3 82.4 83.3 84.2 81.0 83.4 82.5 82.0 82.1 80.7 82.5 82.8 81.7

84.0 81.1 83.5 83.5 84.2 85.4 82.3 84.5 83.7 83.1 83.0 82.0 83.6 83.5 83.1

84.7 82.1 84.3 84.3 84.7 86.3 83.2 85.3 84.5 84.1 84.0 83.1 84.4 84.3 84.1

84.9 82.8 84.8 84.8 85.0 86.8 83.8 85.8 84.9 84.7 85.0 83.8 84.9 85.1 84.8

85.0 83.0 85.0 85.0 85.0 87.0 84.0 86.0 85.0 85.0 86.0 84.0 85.0 86.0 85.0

Source: Eurostat (2000)—baseline scenario.

focus on Germany as an example of an industrialised country with an ageing population, i.e. with low fertility rates and rising life expectancy. In the EU, mobility is not signiﬁcant. Each year, only 0.2% of the population migrate to another European country (OECD, 2001b), and only 1.2% of the population move to another region within a country (Gros and Hefeker, 1999). In the United States, mobility between States is about three times as high as mobility within Germany or within England and Wales (Eichengreen, 1991). The numbers for Europe are especially low if one takes into account the still persisting differences between the national as well as regional labour markets—especially with respect to wages—thus hinting at fundamental structural problems. The present level of mobility within the EU is certainly not sufﬁcient for an efﬁcient allocation of labour. The picture is slightly different when commuters who work in a foreign country without moving their place of residence are included. 3.1.3.1. Migration in Europe Tables 3.4 and 3.5 display migration inﬂows of EU citizens in EU countries. With 41%, Germany is by far the most popular destination for EU migrants, followed by the United Kingdom with 21%. An 8% of migrants move to Belgium, 7% to Spain and 6% to the Netherlands. Only a very small number of EU migrants choose one of the other EU countries. If looked at the other way round, it is also possible to identify those EU countries where most EU migrants come from. Of course, the bigger EU countries

38

Unfunded Pension Systems: Ageing and Migration

Table 3.4: Intra-European mobility of EU citizens (major countries). EU immigrants by nationality (in %)

EU emigrants by nationality (in %)

Receiving country (data from year) Germany United Spain Italy France (1999) Kingdom (1998) (1998) (1999) (1998)

Austria Belgium Germany Denmark Spain Finland France Greece Ireland Italy Luxembourg Netherlands Portugal Sweden United Kingdom Total inﬂow of EU citizens Percentage of all EU immigrants

8.8 1.5 – 1.8 6.1 2.1 11.3 13.0 2.0 25.8 0.5 4.8 10.9 2.5 8.9 100 40.6

0.1 1.2 13.3 3.8 9.8 2.4 22.0 18.3 2.8 14.2 0.0 4.2 3.6 4.4 – 100 20.5

1.5 5.8 31.9 1.4 – 3.3 12.1 0.2 0.9 8.9 0.1 4.9 6.4 2.4 20.4 100 6.7

4.6 3.5 24.2 2.1 10.6 2.0 19.6 7.3 1.6 – 0.1 4.5 3.6 3.0 13.3 100 2.8

1.0 6.7 10.7 1.4 9.2 1.1 – 1.4 2.0 13.8 0.3 3.1 31.9 2.5 15.1 100 1.8

4.2 2.6 11.5 2.4 6.2 3.1 15.0 9.9 2.0 16.4 0.3 5.8 7.7 3.5 9.5 100 100

Inﬂows of EU citizens as percentage of total inﬂows of EU citizens received by each EU country. Source: OECD (2001b), p. 35.

dominate intra-EU migration, led by Italy with 16% of total EU migration and France with 15%. Germany (12%) and the United Kingdom (10%) come next. Given their relatively small size, the share of Greece (10%), Portugal (8%) and the Netherlands (6%) is remarkably high.4 The analysis of intra-European mobility by nationality shows great differences across countries, which can be partly explained by taking into account cultural and historical ties as well as climatic preferences (OECD, 2001b). Afﬁnities with respect to culture and language can be observed in the case of Germans who move to Austria (making up 53% of all EU foreigners5 to Austria), French to Belgium (28%) and Luxembourg (27%), Finns to Sweden (36%) and Swedes to Finland 4

Compared to France, Italy, and the United Kingdom, Greece and Portugal only have one-sixth of the population and the Netherlands only one-fourth, while Germany is eight times bigger than Greece and Portugal and ﬁve times bigger than the Netherlands (see Table A3.1). 5 EU foreigners are EU citizens who migrate from one EU country to another, i.e. they are foreigners of EU nationality with respect to their EU destination country.

Table 3.5: Intra-European mobility of EU citizens (smaller countries). EU immigrants by nationality (in %)

EU emigrants by nationality (in %)

Receiving country (data from year)

Austria Belgium Germany Denmark Spain Finland France Greece Ireland Italy Luxembourg Netherlands Portugal Sweden United Kingdom Total inﬂow of EU citizens Percentage of all EU immigrants

0.9 – 11.0 1.4 4.2 1.5 28.3 2.2 1.2 9.3 0.7 22.1 4.7 2.0 10.8 100 8.4

1.8 9.7 23.8 2.0 5.8 2.5 10.3 3.4 2.7 6.9 0.1 – 3.7 3.3 23.8 100 6.0

– 1.4 52.7 1.7 2.4 2.6 5.1 4.0 0.9 10.4 0.3 4.2 3.2 3.4 7.6 100 3.6

0.5 6.4 8.5 2.0 1.3 1.0 26.6 1.0 1.3 6.7 – 2.7 25.1 1.7 5.2 100 2.5

1.1 1.1 13.7 13.4 3.4 35.9 7.2 2.4 1.6 3.5 0.0 4.2 0.8 – 11.8 100 2.5

2.1 1.9 20.9 – 6.4 5.0 9.6 1.5 1.7 6.8 0.0 7.6 1.2 18.4 16.8 100 2.4

1.2 3.7 22.0 0.9 18.7 1.0 15.7 0.4 0.7 7.6 0.3 6.9 – 2.3 18.7 100 0.9

3.6 3.2 26.2 3.6 0.9 4.1 14.7 – 1.0 9.1 0.1 6.6 0.3 7.1 19.5 100 0.9

1.8 1.1 12.4 4.5 3.1 – 7.0 2.0 1.7 4.9 0.0 3.8 0.3 44.6 12.9 100 0.5

4.2 2.6 11.5 2.4 6.2 3.1 15.0 9.9 2.0 16.4 0.3 5.8 7.7 3.5 9.5 100 100

Projected Development of Fundamental Factors

Belgium Netherlands Austria Luxembourg Sweden Denmark Portugal Greece Finland (1999) (1998) (1998) (1999) (1998) (1998) (1998) (1998) (1999)

Inﬂows of EU citizens as percentage of total inﬂows of EU citizens received by each EU country; data for Ireland as a receiving country are missing. Source: OECD (2001b), p. 35.

39

40

Unfunded Pension Systems: Ageing and Migration

Table 3.6: Stocks and ﬂows of foreigners (major EU countries). Data from year

Receiving countries Germany

United Kingdom

Flows (Percentage of total inﬂows) EU inﬂows 20.1 Stock (Percentage of total population) Foreigners 8.9 EU foreigners 2.3

Spain

47.5

38.8

3.8 0.7

2.0 0.7

Italy

2.2 0.3

France 5.6

1998a

5.6 2.0

1999 1998

Source: OECD (2001b), p. 35 and Table A1.5. a

Data for Germany are from 1999.

(45%). Historical ties play a role in the case of Portuguese in France (32%) and Italians in Austria (10%). British and German retirees prefer to spend their retirement years in Southern European countries, i.e. Spain (20%) and Portugal (19%) in the case of British retirees and Greece (26%) in the case of German ones. Immigration ﬂows of non-EU citizens, however, are more important, as can be seen in Tables 3.6 and 3.7. Only in two European countries do foreigners from another EU country make up more than half of the total inﬂows—namely, Luxembourg (70%) and Portugal (51%)—with Belgium (49%) and the United Kingdom (48%) coming next. In Germany, in contrast, only 20% of the immigrants come from another EU country and in France only 6%. Immigrants make up a signiﬁcant share of the population in Luxembourg, at 36%, and to a lesser but still large extent in Belgium, Austria and Germany at 9%. As a consequence of the low level of migration ﬂows of EU citizens compared to non-EU citizens, the share of EU foreigners as a percentage of the total population in EU countries is not very signiﬁcant. EU foreigners make up about 31% of the Luxembourgian population and 6% of the Belgian population. This can be explained by the institutions of the EU with seats in these countries. In all other countries, the share of EU foreigners as a percentage of the total population ranges from 2% in Germany to close to 0% in Finland and Italy. For most countries, total migration does not present a very important phenomenon if migration is deﬁned as a change of the place of residence. This neglects, however, commuters who work in a country different from the country where they live. Data for the border area of Luxembourg with Germany, France and Belgium show that the inﬂows to Luxembourg are underestimated by a large extent when based on foreigners with a place of residence there. In 1998, 70,800 commuters worked in Luxembourg and added to the 153,000 immigrants (OECD, 2000).

Data from year

Receiving countries Belgium

Netherlands

Austria

Luxembourg

Sweden

Denmark

Portugal

Greece

Finland

Flows (percentage of total inﬂows) EU inﬂows 48.5 24.4

20.2

69.7

23.4

27.7

50.9

22.9

19.2

Stock (percentage of total population) Foreigners 8.8 4.1 EU foreigners 5.5 1.2

9.2 1.2

36.0 31.0

5.5 2.0

4.9 1.0

1.9 0.5

Source: OECD (2001b), p. 35 and Table A1.5.

1.7 0.3

Ireland 1999 3.1

1998 1998

Projected Development of Fundamental Factors

Table 3.7: Stocks and ﬂows of foreigners (smaller EU countries).

41

42

Unfunded Pension Systems: Ageing and Migration

Of course, this effect is largely concentrated in border areas and therefore much smaller for larger countries. Overall, the volumes of intra-European but also inter-regional migration are rather small. We will now look at some aspects in more detail for Germany. 3.1.3.2. Migration to and from Germany Focusing ﬁrst on EU immigration to Germany, we can observe that a higher degree of integration within the EU over the last few decades has not increased the number of immigrants to Germany from other EU countries (see Figure 3.1). With the exception of Italy, the levels have never been very high and have been even decreasing. The accession of new members to the EU has not inﬂuenced the number of EU immigrants to Germany. We now want to look in more detail at the development of migration from and to Germany over the last 10 years as the process of European integration has been accelerated during this period. In Figure 3.2, we have inﬂows of all foreigners and for the sup-group of EU foreigners net of outﬂows from 1990 to 1999. Total net inﬂows peaked in 1992 with 590,000 more immigrants than emigrants. Net inﬂows

Figure 3.1: EU immigrants to Germany. Source: OECD (1999)—EU countries with migrants to Germany of more than 5000 per year. Belgium, France, Germany, Italy, Luxembourg and the Netherlands have been members of the EU or its preceding organisations, respectively, since the beginning.

Projected Development of Fundamental Factors

43

Figure 3.2: Inﬂows and outﬂows for Germany. Source: OECD (2001b)—Tables B1.1 and B1.2.

decreased and even became negative in 1996 and 1997. In 1998, net inﬂows amounted to 118,000. The development is qualitatively similar for EU foreigners. Until 1996, there were 20,000 to 40,000 more EU immigrants than EU emigrants. In 1997 and 1998, this ratio also became negative.6 The number of Germans who have left Germany for another country has also been at a very low level and has not changed signiﬁcantly during recent years. Up to now, the volume of emigration of Germans from Germany is negligible compared to the number of foreigners who leave Germany to return home or to move to another country. In 1998, for example, only 116,000 Germans emigrated (Statistisches Bundesamt, 2001).7 In total, less than 0.15% of the German population has left Germany each year compared to between 7.5 and 10.5% of the foreign population. The majority has chosen another European country as the destination country, while about 20% of the German and 5% of the foreign emigrants have moved to the United States and about 10% of both groups have migrated to Asia and Australia (Statistisches Bundesamt, 2001). To conclude, for Germany as well as for the other countries of the EU, intraEuropean migration does not (yet) happen on a large scale. Only a small share of 6

Data on inﬂows of EU foreigners are missing for 1999. For an analysis of emigration from and immigration to Germany from 1945 to 1994, see Mu¨nz and Ulrich (1996). 7

44

Unfunded Pension Systems: Ageing and Migration

the EU population is involved compared to the larger migration ﬂows of non-EU foreigners. One could expect that the more integrated markets of the EU will reduce the costs of migration in a signiﬁcant way and will lead to more intraEuropean mobility—perhaps at the expense of non-EU migration. We will therefore look at migration projections for the next ﬁve decades to see whether they reﬂect these assumptions. 3.1.3.3. Projections Table 3.8 presents the estimated development of net migration according to projections from Eurostat (2000). The number of people who immigrate to one of the countries of the EU net of those who emigrate varies strongly from country to country. In 2000, Germany by far had the biggest volume, with 300,000 immigrants net of emigrants, followed by the United Kingdom with 90,000 and France and Italy with 50,000 each. Net migration is predicted to increase over the next 20 years in most EU countries and then stay constant. The volume is supposed to double in Spain, Austria and Portugal and to increase between 30 and 60% in Sweden, Belgium and Italy. For other countries, net migration is expected to fall. Ireland will see 70% fewer immigrants net of emigrants, while the volume will decrease by between 20 and 35% for the United Kingdom, Germany and Luxembourg. Table 3.8: Net migration (in 1000s). 2000

2010

2020

2050

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom

10 11 300 22 31 50 17 50 3 33 10 12 6 15 90

13 10 250 23 46 50 10 65 2 34 15 19 5 18 80

15 10 200 25 60 50 5 80 2 35 20 25 5 20 70

15 10 200 25 60 50 5 80 2 35 20 25 5 20 70

EU total

661

640

622

622

Source: Eurostat (2000)—baseline scenario.

Projected Development of Fundamental Factors

45

According to these projections, net migration will decrease over the next 50 years despite the enlargement of the EU. It will fall by 6% for the EU as a whole and will go down for most countries—including Germany despite its common borders with Poland and the Czech Republic. These data, however, do not reveal the path of immigration and emigration and where the immigrants will come from. In fact, the Ifo Institute for Economic Research (Sinn et al., 2001) has estimated a total inﬂow to Germany from the ﬁve new and future EU Member States with the largest population (Poland, Romania, Slovakia, Czech Republic and Hungary) of 3.2 –4 million in the ﬁrst 15 years. Depending on the temporal distribution of the migration ﬂows, between 200,000 and 250,000 migrants are expected to come in the initial years. It can be expected that net migration for Germany as projected by Eurostat (2000) would, then to a larger extent, be made up by immigrants from the new member countries indicating a rise of intra-European immigration at the expense of immigration from non-EU countries.8 Overall, the total volume of migration is not very high and will certainly not be high enough to counterbalance the low fertility rates. It is nevertheless possible that the “wrong” individuals emigrate, i.e. those who are most needed as contributors to the unfunded pension systems. In Chapter 8, we will therefore discuss in more detail the characteristics of those who think about migrating to another country. Given the relatively low volumes, the importance of migration with respect to reducing demographic imbalances should not be overstated. Slow or even negative population growth due to low fertility rates can only be offset by net immigration to a certain extent. Over a long period, natural increases due to excess of birth over death are more important for population growth in most countries (OECD, 2001b). Net immigration might prevent a decrease in population for a limited time; but it certainly cannot on its own provide an answer to population ageing.

3.2. DEVELOPMENT OF THE TOTAL POPULATION We have seen in section 3.1 how fertility, life expectancy and migration are supposed to develop over the next 50 years. Fertility rates will increase, though they will still fall short of the replacement rate; life expectancy of men and women will rise and net migration to countries of the EU will decrease.

8

Of course, part of the formerly non-EU migration will be EU migration after the enlargement.

46

Unfunded Pension Systems: Ageing and Migration

Taking the evolution of these three factors together, total population in the 15 countries of the EU will decline by about 4% with a large variance between individual countries (Table A3.1): the biggest winners will be some of the smaller countries, notably Luxembourg (28%), Ireland (25%) and the Netherlands (11%). Concerning the major EU countries, France and the United Kingdom will gain by 4 and 3%, respectively, while Germany will lose about 8% and Spain about 12% of the population and the number of Italians will decrease by almost one-ﬁfth (Figure 3.3). The demographic development in Japan shows some parallels with bad performers among the European countries with an expected decline of the population of 15%, while the United States outperforms all European countries with a—comparatively—astonishing population increase of 40% (United Nations, 2001a). Between 2030 and 2040, the population of the United States will surpass that of the 15 countries of the EU. In order to draw some conclusions about how unfunded pension systems will be affected by the projected demographic development, the age composition of the population must be analysed in more detail. A decreasing total population must not necessarily endanger the ﬁnancial sustainability of the pension system if the ratio of “youths” to “old folks” stays approximately the same. If fertility and life expectancy, however, change without counterbalancing migration in a way that alters this ratio fundamentally, serious consequences for the pension systems are

Figure 3.3: Population (major EU countries). Source: Eurostat (2000)—baseline scenario.

Projected Development of Fundamental Factors

47

inevitable. We will, therefore, have a closer look at the projected change in the age structure and the dependency ratio in the following sections. 3.2.1. Age structure The projections for fertility rates and life expectancies taken together with expected net migration make clear why the age structure of the population has already departed from the pyramid shape of the 1950s. At the moment, the age structure resembles more a pine tree, and will approach the shape of an ironically apt sarcophagus in most countries by 2050 (Figure 3.4). In the same way as a growing—pyramid-like—population makes unfunded pension systems a highly proﬁtable Ponzi game for all participants, a shrinking population where “youths” no longer necessarily outnumber “old folks” involves great losses for the latecomers. While in the United States, the youngest individuals will still be more numerous than any other age group in 2050; in Japan and Germany this is not even the case in 2000, let alone in 2050. In 2050, 75 to 80-year-olds will constitute the largest age group in Japan while 60 to 65-yearolds will dominate in Germany. It goes without saying that a system that involves transfers from the young to the old—as pay-as-you-go pension systems do—ﬁnds itself in deep crisis given the development of the number of potential contributors to potential recipients. This becomes even more clear when looking at the evolution of the dependency ratio or potential support ratio which depicts precisely this relationship. 3.2.2. Dependency ratio We have seen that the population is ageing because of low fertility rates and increasing life expectancies which are not (sufﬁciently) offset by immigration. The old become a proportionally larger share of the total population. For an analysis of the ﬁnancial sustainability of unfunded pension systems, the old-age dependency ratio, i.e. the ratio of those 60 years and older to those between 20 and 59 years, is of particular importance as the former constitute those who, in general, receive pension beneﬁts while the latter, in general, ﬁnance these beneﬁts by paying contributions. Figure 3.5 shows the projected development of the old-age dependency ratio for the ﬁve major European countries. Today, the situation is comparable for the countries considered with a dependency ratio of about 0.40. This ratio will, however, increase strongly over the next half century, reaching about 0.70 for the United Kingdom and France, 0.80 for Germany and a dramatic 0.90 for Italy and Spain. This means that while today, 10 individuals of the working generations

48

Unfunded Pension Systems: Ageing and Migration

Figure 3.4: Age pyramid. Bars consist of 5-year groups from 0–4 years of age to 95– 99 years of age. Source: United Nations (2001b)—middle variant; Eurostat (2000)—baseline scenario.

Projected Development of Fundamental Factors

49

Figure 3.5: Old-age dependency ratio (major EU countries). Ratio of those 60 years and older to those between 20 and 59 years. Source: Eurostat (2000)—baseline scenario.

support four retirees, they will have to ﬁnance pension beneﬁts for 7– 9 retirees in 2050. Or, to express it in terms of the potential support ratio which is the inverse of the dependency ratio: in 2050, for every retiree there will be only 1.4 working-age individuals in the United Kingdom and France, 1.25 in Germany and no more than 1.1 in Spain and Italy, compared to an EU average of 2.5 in 2000. The smaller countries will again outperform the larger ones with dependency ratios of between 0.6 and 0.7 in 2050. Only Austria and Greece will approach the level of Germany with ratios of 0.8 (Table B3.1). Unsurprisingly given the path of fertility rates and life expectancies, the United States has a much more favourable ratio between those 60 years and over and those 20 –59 years. The dependency ratio is expected to increase from only 0.30 in 2000 to 0.55 in 2050. In Japan we ﬁnd the inverse picture, with a dependency ratio of 0.40 corresponding to the European average in 2000 and a dramatic increase to 1.05 in 2050 (United Nations, 2001b).

3.3. CONCLUSION Immigration can only alleviate the demographic development to a certain extent. The migration volumes needed to counterbalance the low fertility rates and

50

Unfunded Pension Systems: Ageing and Migration

the high life expectancies on a one-to-one basis would require enormous numbers of immigrants. Based on the 1998 revision of the population projections, the United Nations (2000) have calculated how many migrants would guarantee a constant total population and a constant dependency ratio deﬁned as the number of individuals 65 years and older to the number of 15 to 64-year-olds. The countries of the EU could reach a constant population with a total of 47.5 million net immigrants over the period 2000 – 2050 and a constant dependency ratio with a total of 700.5 million net immigrants. The medium variant of the population projection of the United Nations assumes 16.4 million net immigrants for the period until 2050, which makes clear how absurdly high these required migration volumes are. For the United States the population can be held constant more easily, with 6.4 million net immigrants between 2000 and 2050, which is even much less than the net migration of 41.8 million on which the medium variant of the population projection is based. A constant dependency ratio, however, requires a huge net inﬂow in the range of 592.8 million. In Japan, the medium variant of the population projection assumes zero net migration. Compared to this, 17.1 million net immigrants over the period 2000 – 2050 are necessary for a constant population and 554.5 million net immigrants for a constant dependency ratio. It is clear that it is not realistic to assume such migration ﬂows in the future. In many countries, migration volumes of the size needed to keep the population or the dependency ratio constant are very likely to face serious social and political objections. We, therefore, must conclude that the coming years will be marked by a drastic ageing of the population if fertility rates do not change in a fundamental way. In what follows, we will focus on the impact of the projected demographic development on the ﬁnancial sustainability and the political survival of unfunded pension systems. By doing so, we will abstract from a deeper analysis of the inverse effect, i.e. the effect of the unfunded pension system on the growth rate of the population—in particular on the decision to have children. Unfunded pension systems socialise children as the contributions paid by the children beneﬁt all old individuals covered by the pension system and not only the parents. This can be justiﬁed as a means of insuring individuals who remain childless against their will (Sinn, 2004). But this form of old-age security also provides incentives to remain intentionally childless as having one’s own children is no longer necessary for consumption in old age: it is sufﬁcient if others invest in children. This creates a moral hazard effect.9 The negative relation between unfunded pension systems

9 For empirical evidence for the negative impact of old-age security on fertility, see Cigno and Rosati (1996) and Cigno et al. (2000).

Projected Development of Fundamental Factors

51

and the decision to have children, however, need no longer hold with endogenous fertility (Razin and Sadka, 1995, amongst others).10 The following chapters analyse the future of unfunded pension systems given the projected development of population growth and in particular of the ratio of “youths” to “old folks”. This will put high pressure on pay-as-you-go ﬁnanced pension systems. Reforms are thus necessary before the crisis has completely materialised, i.e. before the ﬁnancial basis and the political support are eroded. In Chapter 4, we will present reforms of the unfunded pension systems which have been realised in some countries in recent years. APPENDIX A3. DEVELOPMENT OF THE POPULATION See Table A3.1.

APPENDIX B3. DEPENDENCY RATIO See Table B3.1. Table A3.1: Development of the population (in 1000s).

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom EU total

2000

2010

2020

2030

2040

2050

10,233 5351 82,339 10,564 39,472 59,424 3816 57,562 439 15,957 8096 10,028 5185 8870 59,676

10,367 54,854 83,463 10,783 39,875 61,554 4173 57,204 474 16,754 8153 10,339 5274 8963 61,001

10,494 5563 83,226 10,798 39,449 62,951 4451 55,810 503 17,323 8169 10,542 5316 9136 62,308

10,534 5647 81,781 10,699 38,504 63,710 4631 53,821 532 17,749 8099 10,695 5282 9258 63,211

10,369 5624 79,339 10,538 37,158 63,368 4735 51,217 550 17,843 7890 10,771 5120 9204 62,825

10,073 5547 75,584 10,188 34,872 61,990 4754 47,679 560 17,655 7577 10,648 4933 9200 61,647

377,010

383,862

386,039

384,152

376,551

362,906

Source: Eurostat (2000)—baseline scenario. 10 See Schweizer (1995) for an analysis of the similarity between overlapping-generation models as the basic framework for endogenous demographic change and local public goods models.

52

Unfunded Pension Systems: Ageing and Migration

Table B3.1: Dependency ratios. 2000

2010

2020

2030

2040

2050

Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom

0.41 0.37 0.45 0.44 0.39 0.39 0.29 0.45 0.36 0.34 0.40 0.38 0.38 0.43 0.39

0.47 0.47 0.49 0.48 0.44 0.46 0.32 0.53 0.42 0.43 0.44 0.42 0.50 0.52 0.44

0.59 0.53 0.59 0.56 0.53 0.57 0.40 0.62 0.51 0.53 0.54 0.48 0.62 0.57 0.51

0.71 0.62 0.76 0.67 0.69 0.67 0.47 0.82 0.63 0.67 0.73 0.56 0.69 0.66 0.65

0.73 0.64 0.78 0.81 0.90 0.71 0.59 0.94 0.63 0.67 0.77 0.69 0.69 0.66 0.67

0.72 0.60 0.79 0.83 0.91 0.73 0.65 0.93 0.60 0.65 0.81 0.68 0.71 0.71 0.69

EU total

0.41

0.47

0.56

0.70

0.76

0.77

Source: Eurostat (2000)—baseline scenario.

CHAPTER 4

Country Studies Recent reform activities in selected countries The process of globalisation, demographic developments and European integration itself have brought […] intense pressure to reform. There is a broad agreement that the welfare state ‘patient’ is in urgent need of comprehensive therapy; however, there are widely diverging views on the causes of the illness and, therefore, on what would constitute the correct treatment. Martin Kolmar (1997, p. 56) Abstract The projected demographic crisis endangers the ﬁnancial sustainability of public unfunded pension systems in most European countries. The majority of countries, however, have so far only enacted parametric reforms, which are not sufﬁcient to cope with the rising pension burden. Some countries have opted for more fundamental reforms, such as, Germany, which has initiated a partial transition to a funded system, and Italy and Sweden, which have switched from a deﬁned beneﬁt to a deﬁned contribution system. With the exception of a few countries, notably the United Kingdom, the reform activities undertaken to date can only be the beginning if the collapse of the pension system is to be avoided. The distribution of the burden of unfunded pension systems across generations, which is reﬂected in the generational-speciﬁc implicit tax rates, depends on the difference between the market rate of return and the internal rate of return. Factors which affect the internal rate of return also affect the intergenerational burden sharing. As we have seen in Chapter 2, the internal rate of return is determined by the growth rate of the wage sum, which can be approximated by the sum of

54

Unfunded Pension Systems: Ageing and Migration

the productivity growth rate and the population growth rate. Thus, the mechanisms behind unfunded pension systems make them very vulnerable to changes in these two factors. The ageing of the population, which is already visible in some countries and which will hit all industrialised countries in the next few decades, puts particular pressure on unfunded pension systems, which rely on a “healthy” ratio of retirees to contributors, the so-called dependency ratio. Chapter 3 has shown that rising life expectancies and falling fertility rates lead to more old individuals and fewer young individuals and thus to a deterioration of the dependency ratio. This endangers the ﬁnancial sustainability of unfunded pension systems and makes some comprehensive therapy necessary. In this chapter, we discuss reform options in general and with respect to recent reform activities of selected countries. In section 4.1, we analyse more generally how pension systems with different characteristics will be affected by the demographic crisis. In section 4.2, we then describe some countries in more detail with a special emphasis on recent reforms in order to evaluate whether the national pension systems are already sufﬁciently prepared for the demographic crisis. In section 4.3, we conclude.

4.1. CHARACTERISTICS OF PENSION SYSTEMS To compare pension systems of different countries in general and to analyse the direction of the most recent reforms in particular, we discuss three types of characteristics, some of which have already been mentioned in earlier chapters (Lindbeck and Persson, 2003): the degree of actuarial fairness, the degree of funding and the structure as either a deﬁned beneﬁt (DB) or a deﬁned contribution (DC) scheme. Although a strict categorisation is difﬁcult as pension systems of “Bismarckian” and “Beveridgean” origin have further developed and moved closer to one another, the main features of the status quo and of the expected evolution can be highlighted. When speaking about pension systems a` la Bismarck or Beveridge, i.e. systems with a more or less pronounced link between contributions and beneﬁts, the essential difference is the degree of actuarial fairness. The closer the relation between contributions and beneﬁts, the more actuarial the pension system is. While a pure “ﬂat-rate beneﬁt type” version of a Beveridgean pension system is completely non-actuarial, “social insurance type” systems in the tradition of Bismarck are more actuarial. As we have already mentioned, pension systems with different degrees of actuarial fairness can be organised as predominantly funded or unfunded systems.

Country Studies

3

4 DK

D

1 UK

55

S 2

Sp Fr G I

DC DB Mixed

Degree of actuarial fairness Figure 4.1: Categorisation of pension systems. Source: Lindbeck and Persson (2003), categorisation following Fenge et al. (2002c).

While funding implies that contributions are invested in the capital market and yield the market rate of return, ﬁnancing via pay-as-you-go results in an internal rate of return which equals the growth rate of the wage sum. The third type relates to the general structure, which can be based either on DBs or DCs:1 as implied, the difference concerns the relation between contributions and beneﬁts. In a DC system, contributions are set and beneﬁts adapt such that the budget constraint holds, while in a DB system, it is the other way round. Figure 4.1 displays the three types. To summarise, systems in quadrant 1 have a low degree of actuarial fairness and funding in contrast to quadrant 4, where the degree of actuarial fairness and funding is high. Quadrant 3 characterises systems, which are funded but not actuarial, while quadrant 2 refers to systems, which are unfunded and actuarial. As the growth rate of the wage sum normally falls short of the market rate of return, at least when the economy is dynamically efﬁcient,2 the bottom right corner in Figure 4.1 is depicted as somewhat less actuarially fair than the top right corner. At least in theory, each degree of actuarial fairness and funding can be combined with either DBs or DCs. Real-world pension systems mostly fall into categories which can be described by a subset of these combinations—or at least have done so before the most recent reforms. Until the mid-1990s, pension systems in most industrialised countries were predominantly unfunded pay-as-you-go systems organised on the basis of DBs. The main difference between national pension systems concerned the degree of actuarial fairness. “Flat-rate beneﬁt type” systems displayed a low 1

Often, one also ﬁnds the name “notional deﬁned contribution systems” (NDC) which hints at the fact that individual accounts—at least within an unfunded pension system—are normally only a ﬁction. 2 See Chapter 2.

56

Unfunded Pension Systems: Ageing and Migration

degree of actuarial fairness. This applied especially to the pension systems in the UK, Ireland and the Netherlands. “Social security type” systems in contrast had a higher degree of actuarial fairness. To this group belonged, in particular, Austria, Belgium, France and Germany. The most recent reforms have been characterised by a move towards more actuarial fairness (see Italy, Sweden and Denmark) and more funding (see Sweden and Denmark) which has been combined with a switch to a DC system in some countries (Italy, Sweden). To give an idea of how to categorise the existing public pension systems, the major countries of the European Union, plus Sweden and Denmark, are shown in Figure 4.1. Before discussing some of the most recent reforms in detail, we analyse each of the three characteristics in the light of the projected demographic development and determine the expected direction of reforms if ﬁnancial sustainability is the centre of attention. We depart from an unfunded, partially actuarial system with DBs. Financing pension systems on a pay-as-you-go basis implies intergenerational transfers in the sense that the young working generations pay contributions which ﬁnance the pension beneﬁts of the old retired generations in the same period. If, in addition, old-age security is organised as a DB system, it is evident that the ageing of the population and the rising dependency ratio will result in a higher pension burden for the young generations. The growing pension burden of the coming decades would be more equally shared between the generations and the sustainability would be more easily maintained if the pension systems were characterised by more funding and DCs. First of all, a higher degree of funding reduces the amount of intergenerational transfers and thus decreases the dependencies across generations with respect to old-age security. Second, a DC system in which contributions grow with the wage sum implies that a lower growth rate of the wage sum due to ageing leads to lower beneﬁts. The risk of a negative demographic development is thus mainly carried by the old generation: ageing of the population no longer leads to an unlimited increase in contribution rates, but results in lower pension beneﬁts. Actuarial fairness is less concerned with the intergenerational distribution of the pension burden and only indirectly linked to the issue of ﬁnancial sustainability if we think about funded and DC systems, which are normally more actuarially fair. Actuarial fairness above all implies more efﬁciency: the closer the link between contributions and beneﬁts, the smaller the taxes implicit in the contributions and the smaller the distortions of the labour-leisure decision resulting from these taxes.3 As we will see in the following chapters, as long as reforms lead to more efﬁciency for all in the sense of smaller distortions, they could and should be 3

See Chapters 2 and 5.

Country Studies

57

implemented. More actuarial fairness should thus be an element of every reform proposal. The transition from an unfunded to a funded system as well as the switch from DB to DC systems does not, however, imply an improvement for all generations. In most cases, the older generations will be burdened more heavily while the younger generations will be relieved to some extent. If reforms contain such elements to maintain the ﬁnancial sustainability, they need to be justiﬁed on other grounds—for example, based on intergenerational fairness. It is then, however, important to consider explicitly the political economy aspect, as the feasibility of reforms which redistribute the burden across generations is not guaranteed.4 In what follows, we outline some of the recent reforms and sketch the impact of the demographic development on pension expenditure to underline the need to act further. We then determine whether the changes concern mostly the degree of funding, the degree of actuarial fairness or the DB/DC dimension.

4.2. RECENT REFORMS OF PENSION SYSTEMS In the 1990s it became clear that the ﬁnancial sustainability of the unfunded pension systems in most countries was endangered by the projected demographic development. Almost all governments reacted and passed legislation to reform the public pension systems. Very often, however, these reforms have only led to parametric modiﬁcations and have postponed more fundamental changes. If we focus our attention on the Member States of the European Union, we ﬁnd that Austria, Belgium, France, Greece, Ireland, Luxembourg, the Netherlands, Portugal, Spain and the UK have so far only enacted minor reforms, most of which have included measures to tighten eligibility criteria and to increase actuarial fairness.5 The public unfunded pension systems of some of these countries will be hit comparatively little by the demographic crisis. In the UK and in Ireland, for example, the size of the public pension systems is quite small and focuses mainly on providing a basic pension, while most earnings-related pensions are covered by company and private pension plans, and in Denmark, the major part of the system is already based on fully funded, DC pillars. However, other countries, whose public unfunded pension systems will be affected more heavily by the crisis, have not yet sufﬁciently prepared themselves.

4

See especially Chapter 6. For details, see, for example, Gruber and Wise (1999), Feldstein and Siebert (2002) or Fenge et al. (2002c).

5

58

Unfunded Pension Systems: Ageing and Migration

Figure 4.2 shows the growth of pension expenditure between 2000 and 2050 for the Member States of the European Union. It is evident that already in 2000, pension expenditure as a percentage of GDP differed widely. Among the ﬁve major countries, the UK enjoyed comparably low pension expenditure at 5% of GDP, while pension expenditure in Spain amounted to 10% of GDP, in France and in Germany to 12% and in Italy to 14%. The latter three countries also exceeded the EU average of 10%, as did Austria, Greece, Finland and Denmark. When looking at the evolution of pension expenditure over the next ﬁve decades, one can distinguish three cases. Expenditure will not rise by much in Italy, Luxembourg and Sweden and will even decline slightly in the UK. In contrast, Spain, Greece and the Netherlands will have to face almost a doubling of pension expenditure relative to GDP while France, Germany and most other countries will see an increase of about 4% points. This underlines the extent to which the ﬁnancial sustainability is endangered. (Further) pension reforms are required in most countries to limit the increase of the pension expenditure necessary to meet the pension claims under the present pension regimes. The Economic Policy Committee (2001) has analysed which factors are driving the increase in pension expenditure and distinguishes between demography,

EU UK Sweden Spain Portugal Netherlands Luxembourg Italy Ireland Greece Germany France Finland Denmark Belgium Austria

2000 2050

0

5

10

15

20

Percentage 25 of GDP

Figure 4.2: Development of pension expenditure (in % of GDP). Source: Economic Policy Committee (2001). Data for France are only up to 2040.

Country Studies

59

employment, eligibility and beneﬁts. Ageing of the population is by far the most important reason for the worsening of the expenditure side of unfunded pension systems. More old individuals receive more pension beneﬁts, which must be shouldered by fewer contributors if the present regimes are not changed. But there are ways to counteract this tendency. One possibility is to reform the labour market in order to increase the share of the working-age population who are employed. Another possibility concerns more fundamental reforms of the pension system itself—in particular tightening up eligibility (for early retirement) and reducing pension beneﬁts. Some countries have already started to reorganise old-age security more fundamentally. We want to review the reform activities of some of these countries and point out the different ways chosen to prepare for the demographic crisis. The aim is not to give a complete description but rather an idea about the sensibility of different countries as to the projected crisis of old-age security. Given the important role public pension systems (still) play in most countries, we focus on the development and the recent reforms of public pension systems which apply to private sector employees.6 As the “social insurance type” systems have always been quite generous and have aimed at securing the standard of life after retirement, these systems will, in general, be much affected by the rising dependency ratio. Germany has reacted by initiating a partial transition to a funded system. Italy and Sweden have opted for a switch from a DB to a DC system. France, in contrast, has so far only made some parametric changes. “Flat-rate beneﬁt” countries in contrast have never been as generous, so that the ageing of the population affects pension expenditure on a comparatively smaller scale. We discuss the public pension systems of the UK and Denmark. The low importance of the public system for old-age security in the UK and the high degree of funding in Denmark are responsible for the comparatively small impact which the demographic development is expected to have. 4.2.1. Germany and France Between 1957 and 1969, the German public pension system was introduced as an unfunded, deﬁned-beneﬁt “social insurance type” system where beneﬁts depended on gross wages. Contributions and beneﬁts were linked, which established some degree of actuarial fairness and served to enable the retirees to maintain their standard of living. In the following years, the German public pension system 6

Details can be found, for example, in Gruber and Wise (1999), Feldstein and Siebert (2002) or Fenge et al. (2002c).

60

Unfunded Pension Systems: Ageing and Migration

expanded both with respect to the number of individuals covered and to the generosity of beneﬁts. In 1962, 25.9 million individuals were members of the pension system (Gesetzliche Rentenversicherung) while in 1982 the number had increased to 30.3 million and in 1992 to 44.6 million due to German reuniﬁcation (Verband Deutscher Rentenversicherungstra¨ger, 2001).7 Between 1957 and the late 1970s, the replacement ratio of beneﬁts relative to net wages grew by about 5% points and has stayed around 70% since then (Figure 4.3) corresponding to a replacement ratio relative to gross wages of about 50%. This increase in individuals covered and in beneﬁts went hand in hand with contribution rates rising from 14.0% in 1957 to 19.1% in 2002—equally shared between employees and employers (Figure 4.4). In addition, state subsidies also rose. Until the late 1980s, no serious problems as to the ﬁnancial sustainability of the German pension system occurred (Ru¨rup, 2002). However, a report by Prognos (1987) concerning the overall economic development against the background of a shrinking population and the report of the German Council of Economic Experts (Sachversta¨ndigenrat, 1988), both made it clear that this development could not continue and that reforms would be needed. In fact, the projected demographic evolution with more old and fewer young individuals would put heavy pressure on the German pension system in the coming decades due to its pay-as-you-go nature. But only in the last 10 years have serious attempts at a fundamental reform been undertaken. The reform of the pension system in 1992 was thus the ﬁrst major alteration of the pension system. By replacing pension beneﬁts tied to gross wages with beneﬁts tied to net wages as one of the main parts of the reform, an increase in the contribution rate to 40% of wages was avoided. German reuniﬁcation took place just before the reform, bringing with it the expansion of the pension system to East Germany. This made further reforms necessary. The so-called Blu¨mReform, which wanted to introduce a demographic factor, was, however, abolished by the newly elected government before becoming effective. Instead, the Riester-Reform (I) was passed in 1999. By restricting the growth of pension beneﬁts in 2000 and 2001 to the inﬂation rate—in addition to an increase in state subsidies ﬁnanced by the newly introduced ecological tax on energy, petrol, oil and gas—the rise of the contribution rate would have been slowed down. But the increase of the contribution rate to 26% (or 28%, depending on the speciﬁc assumptions) would have been still ominously high. This reform was, however, suspended after the ﬁrst year. So in 2001, pension beneﬁts grew again in line with net wages. 7

These numbers comprise employees and workers but exclude miners.

Country Studies

61

Pension benefits relative to net wages

75%

70%

65%

60%

55%

1997

1992

1987

1982

1977

1972

1967

1962

1957

50% Years

Figure 4.3: Development of the replacement ratio in Germany. Source: Verband Deutscher Rentenversicherungstra¨ger (2001, p. 238).

Contribution rate relative to gross wages

22% 20% 18% 16% 14% 12%

1997

1992

1987

1982

1977

1972

1967

1962

1957

10% Years

Figure 4.4: Development of the contribution rate in Germany. Source: Verband Deutscher Rentenversicherungstra¨ger (2001, p. 243).

62

Unfunded Pension Systems: Ageing and Migration

Experts and many politicians subsequently agreed that the increase in the contribution rate had to be further slowed down. The Council of Advisors of the German Ministry of Economics (Wissenschaftlicher Beirat, 1998)8 advocated very forcefully the reduction in the generosity of the unfunded system and the partial transition to a complementary funded system which every contributor would ﬁnance via additional savings, and this was at the core of the RiesterReform (II) passed in 2001. From 2002, a partial transition to a funded pension system combined with a decrease of the non-funded part of the pension system came into effect. (Voluntary) private savings for old age will be subsidised with about e10 billion a year from 2008 onwards in order to make it attractive for young individuals to build up a funded supplement to their reduced non-funded pension claims.9 At the same time, recommended private savings of 4% of gross wage from 2008 onwards will lower what is called “net wages” in the pension formula and hence reduce pension beneﬁts and contribution rates. The aim is to keep the contribution rate below 22% until 2030 while at the same time pension beneﬁts must not fall below 67% of net wages. If one of these two thresholds is endangered, the government is obliged to act. It is not clear, however, which variable will then be used for the adjustment (Sachversta¨ndigenrat, 2001). But already in 2003, the projected contribution rate of 18.7% (Bundesministerium fu¨r Arbeit und Sozialordnung, 2001) will be exceeded by a considerable amount reaching 19.5% as decided by the German parliament on 15 November 2002. Such a fundamental correction at such an early point casts serious doubt on the assumptions on which the predictions of this reform are based. The reforms in Germany until the most recent one have been mainly parametric reforms aimed at a reduction of beneﬁt levels in order to prevent contribution rates from rising to unsupportable levels, and have not fundamentally changed old-age security in Germany. The latest reform has been more far-reaching by introducing a voluntary, but subsidised, funded element. When looking at the development of the pension expenditure in Figure 4.2, it must, however, be doubted whether the ﬁnancial sustainability of the pension system can be guaranteed without further—more radical—reforms. In France, the projected ageing of the population and the consequences for pension expenditure have so far not led to fundamental changes. The French public pension system comprises many unfunded pension schemes, which are different across sectors and occupations. For the private sector, they consist of two 8

For further details see Sinn and Thum (1999), Sinn (1999b), and Thum and von Weizsa¨cker (2000). The system will be phased in over the next 6 years starting with a subsidisable sum of 1% of gross wage in 2002 and reaching 4% of gross wage from 2008 onwards.

9

Country Studies

63

parts, a general system (Re´gime ge´ne´ral) and a supplementary system (Re´gime comple´mentaire) where the latter is based on compulsory company pension plans. In what follows, we focus on the general system. In 1993, the ﬁrst major reform to the system aimed at changing the calculations of pension beneﬁts. The main feature was to change from wage to price indexation. In addition, the minimum number of years necessary to qualify for pension beneﬁts as well as the reference period for calculating pension beneﬁts were increased (Gern, 2002). This rather parametric reform has, however, not fundamentally affected the development of the French unfunded pension system. In 1997, it was decided to introduce the possibility for employees and employers to contribute to private funded pension systems in addition to the unfunded pension system. Similar to the change that Germany put into practice 4 years later with the Riester-Reform (II), it was planned that the contributions should be voluntary but generously subsidised, but a change of government prevented this reform from coming into effect. Funded pension components still play a minor role. It remains to be seen whether further, more far-reaching reforms will be initiated in France to avoid a further increase in the contribution rate (which currently stands at 16.4%) and to limit the projected growth in pension expenditure. 4.2.2. Sweden and Italy A similar pressure stemming from the projected demographic development has induced Sweden and Italy to opt for more fundamental reforms of their respective pension systems. The Swedish pension system was organised as a mixed system—typical for Nordic countries—until the 1990s with a tax-ﬁnanced basic pension (Folkpension) for everybody and a mainly employer-ﬁnanced additional pension for employees (Allma¨n Tilla¨ggspension) on the basis of an unfunded, DB system. In the early 1990s, the ﬁnancial sustainability became more and more endangered. Three principles ﬁnally emerged from the discussions in Sweden (Palmer, 2002). First, beneﬁts should be based on contributions paid. Second, beneﬁts should grow in accordance with the growth rate of the wage sum, and third, annuities even within an unfunded pension system should reﬂect rising life expectancies. In 1994, the Swedish parliament enacted a reform on the basis of these principles, which comprised an unfunded notional DC pension system as the ﬁrst pillar, supplemented by an obligatory funded second pillar. So, both pillars are based on individual accounts. The difference is that for the ﬁrst unfunded pillar (Inkomstpension), the accounts are notional, i.e. there is no fund, and the internal

64

Unfunded Pension Systems: Ageing and Migration

rate of return corresponds to the growth rate of GDP, while for the second pillar (Premiepension), contributions are invested in the capital market and yield the market rate of return. In both cases, however, the accounts are illiquid in the sense that the beneﬁts can only be claimed at the time of retirement. Then, beneﬁts are converted into an annuity which takes into account changes in life expectancy. The contributions are ﬁxed at 18.5%, with 16% going to the unfunded part and 2.5% to the funded part. This level is supposed to remain unchanged in the future. Risks of ﬁnancing the beneﬁts when individuals approach their retirement age have to be carried by the members of each generation themselves. This reform, which became effective from 2000 for all individuals born after 1953, thus replaced the mixed system by a “social insurance type” pension system with a quasi-actuarial link between contributions and beneﬁts. This is supplemented by a funded part and from 2003 onwards by a means-tested basic pension (Garantipension). Probably inspired by the Swedish reform, Italy has chosen a similar way to reform its pension system though it departs from an even more problematic starting point. Not only is the projected demographic development more dramatic, but pension expenditure as part of GDP is higher than in most countries. The discussions about reforms, which began in the early 1990s, started from an unfunded DB system ﬁnanced by contributions and by generous state subsidies (Sistema retributivo). In 1992, the Amato Reform was enacted, which aimed at limiting the ratio of pension expenditure to GDP to its 1992 level (Franco, 2002). The main features included a shift from wage to price indexation, a rise in the retirement age and in the minimum years of contributions necessary to qualify for pension beneﬁts, and a longer reference period for calculating pension beneﬁts. This reform achieved the cancellation of a quarter of net pension liabilities. In addition, the main differences between the pension rules for different sectors and occupations were eliminated, thereby making the formerly much segmented old-age security in Italy more homogeneous (Brugiavini, 1999). Despite these different measures, neither pension expenditure nor contribution rates could be expected to stay at supportable levels in the next few decades. In 1995, the Dini Reform was passed to reform the Italian pension system more fundamentally. The main objectives were again to stabilise the ratio of pension expenditure to GDP. But in addition, the reform aimed at reducing labour market distortions and at increasing the actuarial fairness of the system (Franco, 2002). Linking contributions and pension beneﬁts closer together reduces the implicit tax and thus the distortionary effects on the labour supply. Similar to changes in Sweden, this reform initiated the shift to an unfunded, notional DC system based on individual accounts (Sistema contributivo). Pension beneﬁts are related to contributions paid over the whole working life and no longer to the wage income of the last ﬁve working years. These contributions yield an internal rate of return

Country Studies

65

equal to the growth rate of GDP and are converted into an annuity at retirement, which depends on retirement age and life expectancy. The minimum years necessary to qualify for pension beneﬁts were reduced. In total, the Italian “social insurance type” pension system is now marked by a quasi-actuarial link between contributions and beneﬁts. The reform, however, will be much more gradually effective than the Swedish reform. Only from 2015 onwards will the new system be relevant for most individuals, and not before 2050 will all pension beneﬁts be calculated according to these new rules (Fenge et al., 2002c). The growth of pension expenditure will not be signiﬁcantly limited and contribution rates are ﬁxed at a very high level of 32.7%. This reform process was accompanied in 1993 and 1995 by legislation which aimed at increasing the role of funding for old-age security—so far, however, without much success. The funds should come from tax-favoured contributions from employers and employees and for the larger share from the gradual channelling of contributions which have so far gone to severance payment funds. Sweden and Italy have thus opted for a shift towards notional DC systems. These systems have some advantages over DB systems, which make them more apt to guarantee the ﬁnancial sustainability of old-age security with an ageing population (Feldstein, 2002). In particular, the actuarial fairness is increased by linking contributions and pension beneﬁts more strongly while at the same time future beneﬁts are limited to amounts, which can be ﬁnanced by the contribution rates. This implies that demographic or other risks to the amount of future pension beneﬁts can no longer be shifted to the next generations but have to be carried by each generation itself. 4.2.3. UK and Denmark Old-age security in the UK originates from the Beveridgean idea of a comprehensive system with ﬂat-rate beneﬁts ﬁnanced by ﬂat-rate contributions. In the years since World War II, this has developed into a two-part system with ﬂat-rate beneﬁts and additional earnings-related beneﬁts. Both types of beneﬁts are organised on the basis of unfunded pension systems. The ﬁrst part, the State Basic Pension, with its earnings-related contributions and ﬂat-rate beneﬁts, was intended to guarantee pension beneﬁts for all employees and their dependents. Although not so much affected by most of the recent reforms, the beneﬁt level has been successively reduced over recent years (Fenge et al, 2002c). The second part, the State Earnings-Related Pension Scheme (SERPS), was introduced in 1978 to replace an older system. One special feature since 1988 has been the contractingout principle, which enables individuals to choose an approved company pension plan instead and to be granted a reduction in contributions to SERPS. The following

66

Unfunded Pension Systems: Ageing and Migration

years saw further reforms of SERPS, which have made the system less attractive, for example, a shift to price indexation and a decrease in beneﬁts. This development, together with a further widening of the contracting-out principle— including also individual retirement accounts beside company pension plans—has signiﬁcantly reduced the number of individuals covered by SERPS to only about one third of those concerned (Gern, 2002). A reform in 1995 included an increase in the retirement age for women and a revision of the calculations of pension beneﬁts which abolished a guaranteed minimum pension. These measures were intended to reduce pension expenditure. Reforms by the Blair government have again introduced a guaranteed meanstested minimum pension, which is linked to wage earning. In 2002, SERPS ﬁnally was replaced by the State Second Pension (S2P), which puts more emphasis on low-income earners and provides them with a more generous additional pension. From 2007 onwards, these pension beneﬁts will be ﬂat-rated. This development shows that the public pension system in the UK focuses once again on providing some basic old-age security and does not aim at securing the standard of living after retirement. On the whole, the UK does not face a pension crisis as serious as most other industrialised countries (Gern, 2002). The public unfunded pension system (SERPS or S2P) is not very important for most individuals as most employees are covered by company or private pension plans and the pension beneﬁts for those covered are very low by European standards. In contrast, the private funded pension system is well established and has been further developed by the introduction of a new individual DC system in 2001. The demographic development is also less negative. For those who are not contracted out, contribution rates are now 22.2% above some deﬁned earnings thresholds from which the contracted-out rebate is subtracted if applicable. The Danish pension system is similar to the system of the UK in the sense that it reﬂects the Beveridgean idea of old-age security; but at the same time, this is complemented by other pillars, thus constituting a mixed system which is typical for the Nordic countries (Fenge et al., 2002c). The ﬁrst pillar (Folkepension), which is fully tax-ﬁnanced, guarantees everybody basic old-age security in the form of means-tested ﬂat-rate beneﬁts which amount to about 38% of the average wage and can increase subject to an income test. This pillar is supplemented by two further pillars. One pillar (Arbeitsmarkedets Tillaegspension) is based on ﬂatrate contributions and ﬂat-rate beneﬁts depending on the years of contribution. But contributions are low and the share of beneﬁts in total retirement income will also be low when the system matures in the coming years (Gern, 2002). The other pillar (Saerlige Pension), which has been effective since 1999, consists of earningsrelated contributions and completely actuarial beneﬁts. Both complementary

Country Studies

67

pillars are organised as a DC system with full funding. Earnings-related contributions are quite low, amounting to 1% for the Saerlige Pension as the other components are either tax-ﬁnanced or based on ﬂat-rate contributions. With a major part of the public pension system based on fully funded DC pillars, the impact of ageing on the pension system is less pronounced than in other industrialised countries. In addition, most employees contribute to individual accounts established through collective bargaining and again organised as fully funded DC systems.

4.3. CONCLUSION The most recent parametric and fundamental reforms of unfunded pension systems can be seen as responses to the ageing of the population. Given the mechanisms of unfunded pension systems, which rely on contributions of the young generation to ﬁnance the pension beneﬁts of the old generation, it is evident that a rising dependency ratio endangers their ﬁnancial sustainability. To guarantee that the budget constraint in the present form holds, contribution rates must rise or beneﬁt levels must fall or both. Alternatively, it is possible to cope with the demographic challenge by changing the pension system and thus the budget constraint more fundamentally. The more or less fundamental reforms in all countries have aimed at reducing pension beneﬁts in order to obey the budget constraint without having to increase contribution rates by too much. Some countries, such as France, have so far only reformed minor aspects of their unfunded pension systems, while other countries have gone much further. Germany has initiated a partial transition to a funded system—even though on a voluntary basis—to complement the reduction in the size of the system as to contribution rates and beneﬁt levels. Sweden and Italy have reformed their unfunded pension systems by switching from DB to DC systems. Above all, contributions and beneﬁts are linked more closely, which increases actuarial fairness and reduces distortions of the labour supply. The UK—starting from a less problematic status quo—has focused on reforming pension beneﬁts for low-wage earners. Denmark, which will also not be affected so much by the ageing of the population—though more than the UK, has strengthened the funded part of the public pension system even further. Nevertheless, the most recent reforms can only be the beginning. In fact, individuals are aware of the unsustainability of the public pension systems and of the further need for reforms (Boeri et al., 2002). A large share of individuals in Germany (85%) and Italy (63%) agrees with the statement that “the pension system will face a crisis in the next 10 –15 years”. And this despite the pension

68

Unfunded Pension Systems: Ageing and Migration

reforms which have taken place in both countries in the last decade but which are judged as ineffective by 40% of the Germans and Italians and as only a ﬁrst step to stabilisation by about 50%. So, a majority of Germans (81%) and Italians (58%) believes that “in the course of the next 10 years there will be another pension reform reducing signiﬁcantly the amounts of public pensions”.10 In the next chapters, we will deal with various aspects of unfunded pension systems against the background of the imminent crisis. Efﬁciency of reforms will be discussed as well as feasibility of reforms. Pension systems must be constructed in a way such that the ﬁnancial sustainability and the political support are guaranteed. If this is endangered, reforms are needed which will restore sustainability and are supported by the majority.

10 A similar survey conducted in France, Germany, Italy and Spain in 2000 led to comparable results (Boeri et al., 2001). Eighty percent of the respondents in Germany and France thought that in 10 –15 years the national pension systems would no longer be able to keep the pension beneﬁts at the 2000 level. Almost three quarters of those in Italy agreed with that statement, but only a minority in Spain. Consequently, 70% in France, Germany and Italy expected a reform decreasing their pension level signiﬁcantly, while in Spain not even half of the respondents thought that this would be likely.

CHAPTER 5

Welfare Analysis of Pension Reforms The perspective of efﬁciency Doesn’t this mean that […] we should privatize on simple efﬁciency grounds? It can seem like a compelling argument. But it turns out that there is a catch […]. Paul Krugman (2002, p. 1) Abstract Unfunded pension systems do not waste resources. They are efﬁcient, which implies that a transition to a funded system does not present a Pareto improvement. The reason for this is that the debt implicit in unfunded pension systems due to the introductory gains of the ﬁrst generation has to be serviced. This can be either in the form of implicit or explicit taxes. In both cases, the labour-leisure decision is distorted in an identical way. For a Pareto-improving reform, we need to identify possible sources of inefﬁciencies. The size of the implicit debt and thus of the implicit taxes necessary to service this debt is determined. But the implicit taxes might not be optimally distributed over the life cycle leading to distortions of the labourleisure decision. We compare the path of implicit taxes over the life cycle to the path of labour-supply elasticities on the basis of the inverse compensated elasticity rule. For men and married women, the distortions are already small. Efﬁciency could, however, be improved by differentiating implicit taxes across some age groups. In many countries, the forthcoming demographic crisis has initiated reform activities of different shapes. Some reforms have been mainly concerned with reducing pension beneﬁts in order to guarantee the ﬁnancial sustainability of the unfunded pension system. They are often combined with a partial transition to

70

Unfunded Pension Systems: Ageing and Migration

a more funded system and involve the redistribution of the pension burden across generations. Other reforms aim at a higher degree of actuarial fairness in order to minimise the distortions of the labour-leisure decision. While reforms, which involve changes to the implicit debt falling on different generations cannot be Pareto-improving, reforms which increase efﬁciency on an individual level can. In this chapter, we analyse a reform, which increases efﬁciency by distributing the implicit taxes optimally over the life cycle of individuals without changing the overall burden of implicit taxes falling on individuals of different generations. In section 5.1, we give an overview of the literature on the efﬁciency of intergenerationally redistributive reforms. In section 5.2, we focus on the efﬁciency of intrapersonally redistributive reforms. We derive theoretically the optimal ratio of implicit taxes to labour-supply elasticities over the life cycle and evaluate empirically for Germany whether reforms can still improve the optimality by further reducing distortions.

5.1. INTERGENERATIONALLY EFFICIENT REFORMS Whether or not a Pareto-improving reform from a pay-as-you-go pension system to a funded pension system is possible has been a matter of debate for more than a decade. More speciﬁcally, there are two aspects (Breyer, 1989). The ﬁrst one is about the Pareto optimality of the introduction of an unfunded pension system. The question is of whether an unfunded system Pareto-dominates a funded system, i.e. whether the decision to introduce an unfunded system is in accordance with efﬁciency principles, even if one can anticipate that the internal rate of return of the unfunded pension system is lower than the market rate of return— at least in the long run. The second aspect concerns the question of whether there is a Pareto-improving transition from an already established unfunded system to a funded one. Some authors have constructed models that show that there are efﬁciency gains from a transition from an unfunded to a funded pension system—and other authors have argued that a Pareto-improving transition is not possible. Before discussing the literature in detail, it is important to note with respect to the ﬁrst question that the introduction of a pay-as-you-go system amounts to pure redistribution among generations.1 This means that “the burden imposed on the participants of an existing PAYGO system cannot necessarily be attributed to the inefﬁciency of the system” (Sinn, 2000a, p. 395). The fact that the rate of return of a pay-as-you-go 1

This is the case at least when the market interest rate is used as the intergenerational discount rate.

Welfare Analysis of Pension Reforms

71

pension system is smaller than the rate of return of the capital market—as is the case today in all pay-as-you-go pension systems in industrialised countries—is not sufﬁcient to call the system inefﬁcient. The difference between the two rates of return is simply a tax—in this case an implicit tax—needed to service the implicit government debt which results from the introductory gains of the ﬁrst generation.2 Breyer (1989) shows that an unfunded system with lump-sum contributions and lump-sum pension beneﬁts is efﬁcient. Inefﬁciency—if any—can be diagnosed if there is a Pareto improvement via a transition to a funded system. This is the case if at least one generation is better off with a funded system without making any generation worse off or—if we abstract from the simplifying assumption of homogeneous individuals within one generation—without making any member of any generation worse off. This means that the (explicit) debt, which is necessary for servicing the remaining claims against the unfunded pension system, must be reduced to zero within a ﬁnite number of periods and that no generation in the transition periods is worse off compared to a continuation of the unfunded pension system (Breyer, 2000). But for this, an unfunded system must generate additional distortions which would be reduced by the transition to a funded system and which could be exploited for achieving a Pareto-improving transition. Simply replacing an implicit debt with an explicit debt is not enough for this to be true, and this is exactly the “catch” of a transition which must be taken into account. Basically, two different areas of inefﬁciencies have been analysed by various authors: distortions of the saving decisions with negative effects on capital accumulation and income growth; and distortions of the labour-leisure decision. We will show that for the ﬁrst kind of potential inefﬁciency the question of funding vs. non-funding is crucial, whereas for the second kind, the aspect of quasi-actuarial vs. non-actuarial fairness matters.3 5.1.1. Distortions of the saving decision In an unfunded pension system there is no capital accumulation—except for a security stock. The contributions of the working generation in one period serve to pay the pension beneﬁts of the retired generation in the same period. In a funded system, in contrast, the contributions of one individual over his working life— invested in the capital market for the market rate of return—constitute his pension beneﬁts when retired. 2 3

See Chapter 2 for a discussion of the concept of implicit taxes. See Lindbeck and Persson (2003) for an extensive discussion of these two dimensions.

72

Unfunded Pension Systems: Ageing and Migration

Mackenroth (1952) has claimed4 that both systems are equivalent stating “…, dass aller Sozialaufwand immer aus dem Volkseinkommen der laufenden Periode gedeckt werden muß” (p. 42).5 This means that the consumption of the old generation always necessitates deferred consumption of the young generation, no matter whether ﬁnanced via an unfunded or a funded system. This statement, however, can be attacked from various sides (see, for example, Homburg, 1988). First of all, it is always an option for a small open economy to transfer consumption possibilities from the future to the present or vice versa by combining negative and positive current account balances. There are even ways for a closed economy to shift consumption possibilities to the future through an increased production of durable goods (Breyer, 2000). The most important objection, however, concerns the neglected aspect that the social expenditure can be covered more easily with a larger national product due to the increased saving and capital formation involved in a funded system.6 As Feldstein (1977) and Homburg (1988) amongst others show, each generation builds up or holds a larger capital stock in a funded system compared to an unfunded system, which results—for a given employment of labour—in a larger national product. Some authors, notably Feldstein (1974), argue that the transition to a funded system in a closed economy has the advantage that the higher capital accumulation and larger national product make it easier to repay the debt. This, however, says nothing about the transition from an unfunded to a funded system. Feldstein (1996b) himself notes that “[s]hifting to a funded program cannot reverse the crowding out of capital that has already occurred” (p. 0). To show the advantage of a transition, Feldstein (1996a) and Feldstein and Liebman (2002) assume that the market rate of return exceeds the discount rate, which implies that the capital intensity is below the welfare maximising level. Additional investments will then have a positive present value hence a net income gain. The reasons held to be responsible for the difference between the two rates, such as a corporate income tax, are not convincing, however, because there are other ﬁscal policy measures which could be used to reduce these distortions. Breyer (1989) shows that if one abstracts from the assumption of a difference between the market rate of return and the discount rate, this result can no longer be sustained. In a closed economy, the public debt burdens the domestic young generation. For an additional capital accumulation, this generation must increase 4

See Homburg (1988) for similar statements in the American literature. …that all social expenditure must always be covered by the national income of the current period (own translation). 6 It is also possible to cover the social expenditure by consuming the capital stock, i.e. by not replacing depreciated investments. 5

Welfare Analysis of Pension Reforms

73

the saving rate at the cost of expenditure for consumption. The compensation has to be made according to the intertemporal rate of substitution, whereas the additional product of the extra capital corresponds to the marginal productivity. Given that both were equal before the transition, the intertemporal rate of substitution now exceeds the marginal productivity of capital. This makes it necessary to raise an additional credit to compensate the young, which in turn makes it impossible to reduce the debt to zero. Or as Homburg (1997, p. 237) has put it: “Saying that it would be proﬁtable to have more wealth is different from saying that it would be proﬁtable to form more wealth”.7 5.1.2. Distortions of the labour-leisure decision Some authors who argue that a Pareto-improving transition from an unfunded to a funded pension system is possible have focused on distortions of the labourleisure decision that result from contributions levied as a proportional tax on wage earnings. A change of the system to a funded one would remove this deadweight loss and would allow use of the corresponding surplus for a Pareto-improving transition. The idea is to substitute—in a transitional phase—the proportional tax by a lump-sum tax which is then gradually reduced to zero by exploiting the efﬁciency gains. For a system with lump-sum pension beneﬁts, Homburg (1990) shows this for a closed economy, while Breyer and Straub (1993) prove a conjecture made by Homburg and Richter (1990) for an open economy. One ﬁrst objection is that it is questionable whether a lump-sum tax really is an available instrument. Brunner (1994, 1996) advances a second point, saying that it only makes sense to introduce an income tax in the ﬁrst place if there are heterogeneous individuals who cannot be distinguished but who should be treated differently using income as the observable characteristic. The income tax then also serves to distribute income between individuals. Any change in the system which is supposed to reduce the deadweight consequently involves intragenerational redistribution which is in conﬂict with a Pareto-improvement. However, Fenge and Schwager (1995) show that for heterogeneous individuals, an abolition of an unfunded system can be Pareto improving if either one of the following two conditions is fulﬁlled: the income differential between the poor and rich individuals must be small (so that the redistributive part of the pension system is small); or the deadweight loss caused by the distortions of the labour-leisure decision by the contributions must be large. To say it differently, a Paretoimproving transition to a funded pension system is possible if the eliminated 7

For a similar statement see Sinn (2003b).

74

Unfunded Pension Systems: Ageing and Migration

labour-leisure distortions are big or at least outweigh the newly created distortions of the intragenerational redistribution. Therefore, one could be prompted to conclude that without intragenerational redistribution, the abolition of the pay-asyou-go pension system is Pareto-improving as this reduces the labour-leisure distortion without the side-effect of leading to intragenerational distortions. Fenge (1995), however, proves that in a model with identical individuals and thus no intragenerational redistribution, a pay-as-you-go system is Pareto-efﬁcient if the system is intragenerationally fair,8 i.e. if the pension beneﬁts are a proportion of the contributions paid. An equivalent tax will be needed in the transition periods in order to service the (implicit) public debt which is inherited from the incumbent pay-as-you-go system, as long as lump-sum payments are ruled out as a means to ensure an efﬁcient reform. As a consequence, there is no instrument left to compensate the members of any generation involved in the transition process for their potential losses. Thus the essence of this debate can be summarised as follows: compared to fully funded systems, pay-as-you-go pension systems involve an implicit or, sometimes, explicit tax—usually a wage tax—which is due to the partial or complete absence of actuarial fairness. This is, therefore, true for “ﬂat-rate beneﬁt” schemes where there is no link between beneﬁts and prior contributions and for “social insurance type” schemes where beneﬁts are closely linked to prior contributions and establish some degree of intragenerational fairness at least. As long as the system is not quasi-actuarially fair, efﬁciency gains are possible by establishing a (closer) link between contributions and beneﬁts. Lindbeck and Persson (2003) calculate the change in the implicit taxes over the entire life span by a shift from a non-actuarial to a quasi-actuarial system and ﬁnd a reduction in the implicit taxes of 10.8 percentage points (from 20.0 to 9.2%). The distortions of individual labour-leisure decisions in a quasi-actuarial system, however, which are induced by the remaining implicit taxes on wages, cannot be exploited for a Pareto-improving transition path towards a fully funded system. As Lindbeck and Persson (2003, p. 91) put it: “A shift from a quasi-actuarial PAYGO system to a fully funded, actuarially fair system, where the old PAYGO claims are ﬁnanced […] is equivalent to no reform at all”. To put it differently, the pay-as-you-go pension system is efﬁcient which implies that an intergenerationally efﬁcient reform via a transition to a funded system is not possible. The debt implicit in the unfunded pension system— because of the introductory gains of the ﬁrst generation—has to be serviced in 8

Here, the term “intragenerational fairness” (Homburg and Richter, 1990) is meant to say that the implicit taxes of annual contributions are the same for all members of a given age cohort. This is, for example, a distinctive feature of the German public pension system.

Welfare Analysis of Pension Reforms

75

the form of either implicit or explicit wage taxes. In both cases, the labour-leisure decision is distorted identically. It is, therefore, not an option to reduce the implicit debt to zero in a Pareto-improving way. From the point of view of an individual, however, it is interesting to see whether the implicit taxes are optimally distributed over the life cycle to minimise the distortions of the labour-leisure decision and thus the deadweight loss. The possibility of an intrapersonally efﬁcient reform will, therefore, be analysed in the following sections.

5.2. INTRAPERSONALLY EFFICIENT REFORMS In a recent contribution, Wrede (1999) has shown the condition under which the result of Fenge’s (1995) model carries over to the case with two and more working periods. His analysis is based on a simple, but straightforward, extension of the Fenge (1995) model. Instead of dividing the relevant life span of an individual into two periods—a period of economic activity and a period of retirement—he considers a three-period setting with separate labour-leisure choices for the ﬁrst two sub-periods. Given the different horizons for discounting future pension claims in order to relate them to current contributions, it is obvious that, ceteris paribus, the implicit tax rate that individuals face differs between these two subperiods. Quite generally, it will be higher for young workers than for older workers. Forced saving at a lower rate of return than the market rate of return is more costly earlier in life as contributions are then longer locked in at this low rate of return. From an analytical point of view, the stylised model employed by Wrede (1999) sufﬁces to conclude that the original result of Fenge (1995) can be generalised to a multi-period setting only if the internal rate of return to the contributions made by young and by middle-aged workers differs only by the interest factor—thus, from an ex ante perspective, the implicit tax rates imposed on their periodical wages are the same. In any other case, the question of whether or not there are opportunities for Pareto-improving reforms of existing pension systems on an intrapersonal level is open again. We build on the results of Wrede (1999) in order to analyse the question of whether the path of implicit taxes over the life cycle of an individual or an agecohort is optimal given the path of labour-supply elasticities. We are, therefore, interested in possible efﬁciency gains on an intrapersonal level within the unfunded pension system. The path of implicit taxes might not be optimal, thus distorting individual decisions regarding periodic labour supply. Thus, a new kind of potential inefﬁciency enters the picture. We investigate the intertemporal structure of annual implicit taxes that arise within the life cycle of each age-cohort,

76

Unfunded Pension Systems: Ageing and Migration

i.e. on an intragenerational or intrapersonal level, given the labour-supply elasticities. This point may shed new light on the debate regarding the efﬁciency of unfunded pension systems. In what follows, we will discuss the problem both theoretically, applying optimal taxation results to an overlapping-generation (OLG) model, and empirically, using micro-data from Germany for an econometric analysis of individual labour supply decisions of men and married women.9 The reason why we consider only married women is that unmarried women present a very heterogeneous group which includes women who live with a partner as well as single women—both with and without children. Controlling adequately for these qualitative differences would require creating many sub-samples. By focusing on married women we avoid this problem. The rest of this section is organised as follows. In section 5.2.1, we derive implicit tax rates in an intrapersonal context for an unfunded pension system with intragenerational fairness. In section 5.2.2, the second-best optimum for the path of implicit taxes is discussed and in section 5.2.3, the methodology for the empirical approach for estimating labour-supply elasticities is outlined. The database we use is the pooled time-series data of the German Socio-Economic Panel (GSOEP) for West-German employees from 1988 to 1998.10 In section 5.2.4, we report the results of the estimations and in section 5.2.5 we evaluate the optimality. Section 5.2.6 concludes with a discussion of the policy issues that arise. 5.2.1. Implicit taxes in an intrapersonal context The path of implicit taxes over individual life cycles is highly dependent (i) on the effective employment record; (ii) on the precise set of rules that govern the way in which pension claims are built up; (iii) on changes in either contribution rates or (expected) beneﬁt levels. First, we concentrate on unfunded pension systems in order to highlight the intertemporal structure of implicit taxes within the framework of a three-period OLG model. The pension system that is modelled here follows the German example with a (not perfectly) actuarial time-invariant link between contributions and beneﬁts and constant contribution rates. We then turn to the particular problems involved in a potential “gender tax-gap” that arises from the treatment of married couples in many public pension systems. Finally, we look at the path of

9

This section is based on Fenge et al. (2002a). See also Fenge et al. (2002b). For a brief description see Burkhauser et al. (1997).

10

Welfare Analysis of Pension Reforms

77

Figure 5.1: Three-period OLG model.

implicit tax rates over the life cycle of men and married women in the German pension system. 5.2.1.1. The path of implicit taxation—the basic model We employ a model similar to the one used in Chapter 2 when we introduced unfunded pension systems—with two changes which will be crucial for the analysis in this section. First, we allow for elastic labour supply. Second, we concentrate on a three-period OLG model following Wrede (1999). Individuals who are born in period t; work in period t and period t þ 1 when M they are young ðNtY Þ and middle-aged ðNtþ1 Þ:11 Disposable time per period is Y Y M normalised to 1 with labour supply lt and lM tþ1 ; 0 # lt ; ltþ1 # 1: Per unit of labour, they earn gross wage Wt and Wtþ1 net of contributions to the unfunded O Þ; they retire and are pension system u: In period t þ 2; when they are old ðNtþ2 entitled to receive pension beneﬁts ptþ2 (see Figure 5.1). M O Periodic consumption is denoted by cY t ; ctþ1 and ctþ2 : In the ﬁrst two periods of life, individuals consume part of their wages, while cO tþ2 is ﬁnanced by their oldage pensions and, if appropriate, by pure life-cycle savings—assuming that there are no bequests. The overall setting for our analysis is that of a small open economy where in each period the interest rate rt is exogenously determined. For simplicity, we assume that the interest rate remains constant across periods, rt ¼ rtþ1 ¼ r: We assume that all individuals born in period t are identical to abstract from intragenerational redistribution. Ex ante, each of them faces the following

11

Subscripts indicate the period concerned while superscripts denote the life period, i.e. young, middle-aged or old.

78

Unfunded Pension Systems: Ageing and Migration

problem: M O Y M Ut ðcY max t ; ctþ1 ; ctþ2 ; 1 2 lt ; 1 2 ltþ1 Þ X X c ;l

ð5:1Þ

subject to the intertemporal budget constraint12,13 cY t þ

cM cO ð1 2 uÞWtþ1 lM ptþ2 tþ1 tþ1 Y tþ2 þ þ ¼ ð1 2 u ÞW l þ : ð5:2Þ t t 2 ð1 þ rÞ ð1 þ rÞ ð1 þ rÞ ð1 þ rÞ2

Wages may differ due to technological change, i.e. Wt – Wtþ1 is possible. In what follows, we write lifetime consumption as cY t þ

cM cO tþ1 tþ2 ¼ ct ; þ ð1 þ rÞ ð1 þ rÞ2

ð5:3Þ

thus replacing periodic consumption by the composite consumption good ct as the relative price of consumption in each period is determined solely by the interest factor ð1 þ rÞ which will not be affected in the following analysis. In the lifetime budget constraint, life-cycle savings that are used to transfer resources across periods in order to establish the optimal structure of periodic consumption just cancel out.14 After these preliminary remarks, we now turn to the derivation of implicit taxes in an intrapersonal context with a focus on the intertemporal, i.e. life-cycle, aspect. We assume, for simplicity, that the structure of periodic labour supply is uniform Y M M across subsequent generations, i.e. lY t ðWt Þ ¼ ltþ1 ðWtþ1 Þ and lt ðWt Þ ¼ ltþ1 ðWtþ1 Þ; etc. For deriving the implicit tax rates, it is important to know how much of the pension beneﬁts can be attributed to contributions made in both working periods. We ﬁrst consider the general case before discussing two speciﬁc versions. In unfunded pension systems with an imperfectly actuarial link between contributions and beneﬁts, the periodic internal rates of return i to earlier contributions can be written in terms of the population growth rate and the productivity or wage growth rate, i.e.

itþ1 ¼ iðgtþ1 ; gtþ2 ; ntþ1 ; ntþ2 Þ; itþ2 ¼ iðgtþ2 ; ntþ1 ; ntþ2 Þ: 12 13

14

Note that Wt lt corresponds to vt in Chapter 2. X takes the value ( Y; M; O; if the variable is defined for all three periods ðe:g: cÞ X¼ Y; M if the variable is defined only for the working periods ðe:g: lÞ:

Yet, they are needed—assuming that the optimal degree of consumption smoothing is not accomplished through public pensions alone.

Welfare Analysis of Pension Reforms

79

Here gtþ1 is the growth rate of wages from period t to t þ 1; and ntþ1 is the growth rate of the population from period t to t þ 1: Thus, i is determined by the ingredients of the Aaron (1966) condition, i.e. the factors constituting the growth rate of the wage sum. Pension beneﬁts can then be written as M ptþ2 ¼ ð1 þ itþ1 Þð1 þ itþ2 ÞuWt lY t þ ð1 þ itþ2 ÞuWtþ1 ltþ1 :

ð5:4Þ

With Equations 5.3 and 5.4, Equation 5.2 reduces to ct ¼ W t l Y t ð1 2 tt uÞ þ

Wtþ1 lM tþ1 ð1 2 ttþ1 uÞ ; ð1 þ rÞ

ð5:5Þ

where ð1 þ rÞ2 2 ð1 þ itþ1 Þð1 þ itþ2 Þ ; ð1 þ rÞ2

tt ¼ ttþ1

and

ð1 þ rÞ 2 ð1 þ itþ2 Þ ¼ ð1 þ rÞ

ð5:6Þ

are the implicit tax rates. Let us now assume that ntþ2 ¼ ntþ1 ¼ n and gtþ2 ¼ gtþ1 ¼ g for the convenience of the presentation when we discuss two versions of the general case. We assume for both versions that contributions in each working period lead to the acquisition of 1/2 pension claims. In the ﬁrst version, pension claims depend on the period in which contributions have been made. With 1 þ idtþ1 ¼ ð1 þ gÞð1 þ nÞ and 1 þ idtþ2 ¼ ð1 þ gÞð1 þ n=2Þ; we get for the implicit tax rates

tdt ¼ tdtþ1

ð1 þ rÞ2 2 ð1 þ gÞ2 ½ð1 þ nÞð1 þ n=2Þ ; ð1 þ rÞ2

ð1 þ rÞ 2 ð1 þ gÞð1 þ n=2Þ : ¼ ð1 þ rÞ

ð5:7Þ

Note that in a three-period model—in contrast to a two-period model—pension beneﬁts in any period are ﬁnanced by contributions from individuals of two generations, i.e. from the young generation and from the middle-aged generation. There are thus ð1 þ nÞ þ ð1 þ nÞ2 more young and middle-aged contributors than d and the implicit tax rate old recipients. This explains the difference between ttþ1 derived within the two-period model in Chapter 2. In the second version, we assume—as is the case in most pension systems, for example, in Germany—that pension claims are independent of the timing of the contributions. The internal rates of return are then 1 þ ind tþ1 ¼ ð1 þ gÞ and

80

Unfunded Pension Systems: Ageing and Migration

1 þ ind tþ2 ¼ ð1 þ gÞð1 þ nÞð1 þ n=2Þ and the implicit tax rates become

tnd t ¼ tnd tþ1

ð1 þ rÞ2 2 ð1 þ gÞ2 ½ð1 þ nÞð1 þ n=2Þ ; ð1 þ rÞ2

ð1 þ rÞ 2 ð1 þ gÞð1 þ nÞð1 þ n=2Þ : ¼ ð1 þ rÞ

ð5:8Þ

We see that the different rules do not change the implicit tax rate on contributions paid in the ﬁrst working period tdt ¼ tnd t ; but do lead to different implicit tax rates in the second working period tdtþ1 . tnd tþ1 : For the two versions, we ﬁnd tt . ttþ1 . 0:15 In fact, the implicit taxes can even be calculated on an annual level, thus yielding a full lifetime proﬁle of implicit taxes. With constant contribution rates, using reasonable assumptions on annual interest rates, the growth rates of the number of contributors and of the individual wages, this proﬁle will be downward sloping over the life cycle (Beckmann, 2000). Before giving the intuition for this, let us ﬁrst summarise the result. Proposition 1 (Wrede, 1999). In a three-period setting, the implicit tax rates are positive and decrease over the two working periods if it , r ;t and if contributions in each working period lead to equal acquisitions of pension claims. The enlargement of the simple model to a three-period setting makes clear which insights are lost with respect to the intertemporal path of implicit taxes when concentrating on a two-period model.16 The intuition can be seen in the following way (Beckmann, 2000). Let the pension claim of a person who has paid contributions for one period be called a pension unit. If the acquisition of the pension unit is postponed by one period—by varying the labour supply accordingly,17 the contributions necessary to acquire 15

To be more exact, Beckmann (2000) shows for a multi-period model that it is possible—albeit not very likely—that it . r for contributions made in late working periods for the case of time-independent pension claims. This implies a negative implicit tax rate. For an illustration consider tnd tþ1 in Equation 5.8. In the following, we abstract from this theoretical possibility of a negative implicit tax rate and assume it , r throughout. 16 The difference of the implicit tax rates tt between different periods can be really substantial. For example, let both working periods comprise 20 years and further assume that the growth rate of the population is zero, the contribution rate is 20%, the annual growth rate of wages is 2% and the annual interest rate is 4%. t1 would then be 54% and t2 32% of the contributions. 17 The labour-supply decision against the background of different implicit tax rates over the life cycle will be the focus of the analysis of the rest of this chapter.

Welfare Analysis of Pension Reforms

81

this pension unit will rise by ð1 þ itþ1 Þ: At the same time, e1 of contributions paid tomorrow is only worth 1=ð1 þ rtþ1 Þ euros today. The ratio rtþ1 ¼ ð1 þ itþ1 Þ=ð1 þ rtþ1 Þ is thus the relative price of a pension unit acquired in the next period. If the pension beneﬁts are independent of the timing of the contributions, the growth rate of the population ntþ1 affects the present value of the pension claim but not the costs of obtaining the claim. If the timing, however, is important, the growth rate of the population ntþ1 inﬂuences the present value of the pension claim as well as the costs. If rtþ1 , 1; individuals prefer to postpone the acquisition of a pension unit until the next period. Continuing this argument, it is obvious that the implicit tax rates must fall over the individual life cycle. This result shows that it is too rough an approximation to take the implicit tax rates averaged over the life cycle, as this hides the different impacts on different age groups. The above maximisation problem can be solved to determine the marginal effect of the implicit tax rates on utility. Departing from Equations 5.1 and 5.2, we obtain the indirect utility function Vðð1 2 tt ÞWt ; ð1 2 ttþ1 Þð1 þ rÞ21 Wtþ1 ; 1 þ r; IÞ;

ð5:9Þ

where the net wages, ð1 2 tj Þð1 þ rÞ Wj ; j ¼ t; t þ 1; and the interest factor, 1 þ r; are the relevant prices, and I ; ð1 2 tt ÞWt þ ð1 2 ttþ1 Þð1 þ rÞ21 Wtþ1 is full income for lt ¼ ltþ1 ¼ 1: For our purpose, the most important result is that t2j

›V ›V ¼ 2lð1 þ rÞt2k Wk lk , 0; ¼2 ›t k ›ð1 2 tk Þ

for k ¼ t; t þ 1

ð5:10Þ

by Roy’s identity, l . 0 being the marginal utility of leisure, goods consumption, or—more generally—of income. Irrespective of their precise timing, taxes tk on wages (or, subsidies for leisure) decrease the welfare of all individuals V as they distort the ﬁrst-best allocation. If the excess burden of tax payments cannot be avoided, one should at least attempt to minimise it. In order to do so, the main instrument is given by the time structure of t: We will, therefore, focus on the implicit tax rates and the labour-supply elasticities of individuals of different age groups when determining the distortions of the labour-leisure choice. 5.2.1.2. Gender differences In our basic model, we look at individuals who pursue a full-time working career over the working periods. If, instead, we focus on individuals with more fragmented employment records, i.e. mostly women, we have to take into account how spells of unemployment, disablement or maternity leave are treated when

82

Unfunded Pension Systems: Ageing and Migration

individual pension beneﬁts are calculated. For individuals with many years of low contributions—possibly due to long part-time careers—regulations become important, for example, with regard to minimum pensions. In many existing public pension systems, there is a potential “gender tax-gap” which implies systematic differences between men and women as to their implicit lifetime tax rates and their implicit annual tax rates. One reason why both t and tt might be lower for women is that in industrialised countries average life expectancy is considerably higher for women than for men. Almost nowhere in current pension systems, is this reﬂected by gender-speciﬁc beneﬁt formulae, which would follow from actuarial principles. However, for married women who are not working on a full-time basis during their working periods, this effect can be more than off-set by three aspects (Feldstein, 1996a,b). First, in some countries (for example, the USA and Japan), there are noncontributory beneﬁts for spouses of retirees if the former do not have an employment record of their own. If a married woman takes up work—temporarily or part-time, and usually at a lower wage rate than average men—she will forego (part of) the beneﬁts that are linked to her marital status. Second, a woman who stays out of the labour market most of her life is at least entitled to receive survivor beneﬁts once her husband is deceased. Again, a married woman will see (part of) these beneﬁts be reduced if she holds pension claims of her own so that part of her contributions is lost on beneﬁts foregone. Third, living longer implies that the expected value of survivor beneﬁts that the husband will receive is very low—in fact, it will mostly be zero. As long as the earnings of women do not exceed a relevant threshold, the implicit tax rates can be much higher than those for men—in fact, t can be up to 100%. The basic problem is that in many countries working women are subjected to rules that were designed for the case of non-working housewives and mothers. Today, these rules are no longer appropriate for the majority of married women as they create disincentives for women with (less than) average earnings capacities to extend their labour force attachment even if they want to.18 If we want to state these problems more formally, our simple three-period model meets certain limitations. In its basic form, the model allows neither for variations in life expectancy, nor can it easily be extended to cover an additional period of time for surviving spouses. For the case of married women with earnings that fall in the range where reductions of their individual pension entitlements 18 It should be stressed here that we are primarily concerned with non-contributory beneﬁts that are linked to just being married, not to being a mother. In the context of pay-as-you-go pension systems, beneﬁts of the latter type may serve a speciﬁc function in rewarding investment in human capital, i.e. in future workers and contributors (see Sinn and Werding, 2000, and Werding, 2002a).

Welfare Analysis of Pension Reforms

83

matter, we, therefore, consider the following scenario: we abstract from differences in life expectancy for an instant and focus on the existence of dependants’ allowances and survivors’ beneﬁts. Building on a simple “male-chauvinist” model of household labour supply, we include pension beneﬁts that are granted to dependants and survivors and are linked to the employment record of the “ﬁrst earner” in the household in the husband’s budget set.19 For his wife, these beneﬁts will then be exogenously given as a beneﬁt component p t ; while earnings-related beneﬁts are subject to a special (say, linear) discount rate d; with 0 , d # 1; such that M;f f pft ¼ p t þ ð1 2 dÞpt ðuwft22 lY;f t22 ; uwt21 lt21 Þ;

ð5:11Þ

where the superscript “f ” denotes variables relevant for the woman’s own pension beneﬁts—notably, the labour supply and the wage in both working periods. Substituting pft 2 p t in the woman’s lifetime budget constraint—subtracting p t to avoid double counting on the household level—we obtain cft ¼ ½1 2 uðtt þ dÞWtf ltY;f þ

f ½1 2 uðttþ1 þ dÞWtþ1 lM;f tþ1 ; ð1 þ rÞ

ð5:12Þ

which is equivalent to Equation 5.5. tt þ d and ttþ1 þ d are the adjusted implicit tax rates. It is easy to see that, in this version, implicit taxes for married women are always higher than those for married (“ﬁrst-earner”) men as d . 0: Given this particular tax structure, married women then decide whether or not to participate in the labour market, either on a full-time or a part-time basis.20 Of course, differences in life expectancies are ignored here. As already mentioned, living longer, however, has two effects on the implicit taxes for married women which might offset one another. On the one hand, the expected value of an individual’s pension beneﬁts is higher as pensions incur in more periods. On the other hand, the expected value of survivor beneﬁts for husbands is mostly zero. The diverging effects considered here—several types of reductions in the pension beneﬁts of married women modelled here, as well as the higher life expectancy of women with the two counteracting effects—need not cancel out. Ultimately, the existence and size of what we call a “gender tax-gap” is an empirical issue to which we turn now for the case of the German pension system.

m Y;m m M;m In other words, they will be included in the pension beneﬁts pm t ¼ pt ðuwt22 lt22 ; uwt21 lt21 Þ: Here, we abstract from more complex models of household time allocation incorporating other decision rules, like those related to joint optimisation or mutual altruism, or introducing household production as a third option for using time. Qualitatively, these models would lead to the same result.

19 20

84

Unfunded Pension Systems: Ageing and Migration

5.2.1.3. Empirical illustration—the structure of implicit taxes for the case of Germany In Proposition 1, we have stated that implicit tax rates in unfunded pension systems will constantly decrease over the life cycle provided that (i) contribution rates are constant over time and (ii) the beneﬁt formula (or, rather, the contribution –beneﬁt link) remains unchanged. Obviously, these conditions are not fulﬁlled when looking at real-world pension systems. As has been shown in Chapter 4 for Germany, the contribution rates have increased from 14% in 1957, when the system was introduced, to 19.1% in 2002 and, given the current legal framework, are expected to approach 22% by 2030. Over the same period, the replacement ratio relative to net wages has increased from some 60% to 70%; it has then been roughly constant until very recently and is now expected to decrease again to about 68% by the year 2030. Given these instationarities, the time structure of implicit taxes for each age cohort, if evaluated at an annual level, may be rather different from the simple proﬁle predicted in our three-period OLG model. Using the CESifo pension model,21 we are able to calculate the relevant time proﬁles over a full working-age period of 45 years for all age cohorts born between 1929 and 1999.22 Using lifeexpectancy data for male individuals in order to determine the length of the retirement period and including survivor beneﬁts for the remaining life span of their widows, the results can be taken to represent the lifetime proﬁles of implicit taxes for representative (married) men in each cohort.23 We conﬁne our attention to the eight birth cohorts born in 1929, 1939, etc. in Figure 5.2. It is easy to see that older age cohorts, i.e. those born between 1929 and 1949, have beneﬁted from the introduction or expansion of the German pension system that took place in the past-war period. The negative implicit tax rates observed for the ﬁrst years of employment of these individuals clearly reﬂect the introductory gift, which always arises if an unfunded pension system is started or expanded. For the age cohort born in 1959, we have a very simple pattern of implicit annual tax rates that are slowly, but constantly, declining over time. All age cohorts born in 1969 and later are negatively affected—though each at a different stage of their life cycle. The upswing of contribution rates and the downturn of beneﬁt levels, which are projected to take place in the coming decades, peak at around 2035 21

For details see Sinn and Thum (1999), Sinn (1999b), Thum and von Weizsa¨cker (2000), and Wissenschaftlicher Beirat (1998). 22 For the assumptions regarding the standardised work biographies, see Appendix A5. 23 Thus, our calculations conform to the simple “male-chauvinist” model of household labour supply suggested above.

Welfare Analysis of Pension Reforms

85

Figure 5.2: Implicit tax rates over the life cycle: German men. Source: CESifo pension model.

when the demographic crisis will reach its climax.24 In each case, however, we also observe the fundamental downward trend that was predicted in our simple model, once the parametric changes have been made and the system settles into a new “equilibrium”. Our theoretical predictions are thus conﬁrmed by this empirical example. If we look at the 10-year period from 1988 to 1997 for a smaller set of cohorts born between 1929 and 1977, which are the cohorts and years that effectively feature in our following empirical analysis, things turn out to be clearer. Figure 5.3 illustrates the path of annual implicit tax rates for German men and married women of different ages for this period. We calculate the time structure of tt for all birth cohorts considered. Building on these proﬁles for each birth cohort, we construct an artiﬁcial life cycle spanning 45 years, which is effectively made up of a series of overlapping sub-periods, each with a maximum length of 10 years.25 The ﬁgure shows the result of a non-linear regression that is ﬁtted to the series of implicit tax rates.

24

See Chapter 3. In Figure 5.3, the oldest cohort is born in 1924 and assumed to be in their last year of employment in 1988, then aged 64. In our empirical model, we dropped individuals aged 60 and over, thus starting with the birth cohort of 1929.

25

86

Unfunded Pension Systems: Ageing and Migration

Figure 5.3: Implicit taxes in Germany for the period 1988– 1997: men and married women. Source: CESifo pension model.

If we look at the results for men ﬁrst, the implicit tax rates decrease from about 72% in the ﬁrst year of employment to about 17% in the year before retirement. This clearly conforms to our theoretical results within a three-period model. Turning to similar calculations for women, we have to take into account that women have a higher average life expectancy than men. However, their own pension claims rarely extend to beneﬁts for surviving spouses and are reduced if they coincide with survivor and/or spouse beneﬁts that originate from their husbands’ pensions. The net effect of these diverging trends is an empirical matter. The implicit tax rates are at about 78% in the ﬁrst year of employment and fall to about 26% in the year before retirement. This also conforms to our theoretical results within a three-period model. The results displayed in Figure 5.3 show that, at each point in time during the active period of life, implicit tax rates falling on female workers (who are married) are higher than for their male counterparts.26 This is mainly what we expected based on our theoretical considerations.

26

For women who stay single over their entire lifetime, the effects of higher life expectancy and the absence of survivor beneﬁts their spouses would be entitled to receive nearly cancel out. Thus, their lifetime proﬁle of implicit tax rates is rather similar to that of men. Since we will not deal with this very heterogeneous sub-group in the following, we abstract from discussing the results in detail.

Welfare Analysis of Pension Reforms

87

We now want to derive an optimality rule linking the paths of implicit taxes and labour-supply elasticities over the life cycle for men and married women. We then evaluate the path of implicit taxes for the estimated labour-supply elasticities on the basis of this optimality rule. 5.2.2. Second-best optimality In our basic model for full-time earners, two goods (leisure ð1 2 lÞ consumed in periods t and t þ 1; respectively) are being taxed, while one (goods consumption c) is not.27 We have seen that the implicit taxes imposed on periodic wages differ across periods. This relates our analysis to a series of standard results from the theory of optimal taxation (Sandmo, 1974, and Atkinson and Sandmo, 1980). In the tradition of optimal-taxation theory, public authorities are assumed to maximise utility V (Equation 5.9) subject to the additional constraint that tax revenues have to meet a given amount G

Gt ¼ tt uWt lYt þ

ttþ1 uWtþ1 lM tþ1 : 1þr

ð5:13Þ

Here, Gt is the effective tax to be levied from individuals in a given generation across their life cycle—discounted at period t values—in order to keep the public debt implied in the pay-as-you-go pension system on an equilibrium path. In other words, Gt is an implicit tax on actuarial returns to earlier contributions. This tax is required to make the periodic budget of the pension system including the implicit debt just grow by the growth rate of the wage sum—making sure that u can be held constant over time. It is straightforward to derive28 that optimal taxation requires implicit tax rates on period t wages to be inversely related to the compensated price elasticities of labour supply in period t; modiﬁed by compensated price elasticities with respect to period s – t so that

stþ1;tþ1 2 st;tþ1 ttþ1 tt ¼ ; 1 2 tt st;t 2 stþ1;t 1 2 ttþ1

ð5:14Þ

where sts are the compensated price elasticities. This is closely related to the well-known Ramsey rule for the case of one good and two periods (see Sandmo, 1987, and Atkinson and Stiglitz, 1980). In order to minimise distortions, the set of

27 28

Alternatively, c may be taxed at a ﬁxed rate. See Appendix B5 for the derivation.

88

Unfunded Pension Systems: Ageing and Migration

taxes should essentially be chosen so as to make the substitution effects as small as possible. To sum up, we may state Proposition 2 (“Ramsey rule”). Wages should be taxed more heavily in those periods of an individual’s working life where (a) the compensated elasticity of labour supplied with respect to the same period’s net wages (i.e., the own-price effect) is low compared to that in other periods of life and; (b) the compensated elasticity of labour supplied with respect to all other periods’ net wages (i.e. the cross-price effect) is low. An important special case captured by Equation 5.14 is obtained if cross-price effects are assumed to be absent. If sts ¼ 0 with s – t; stt then can be abbreviated to read st and Equation 5.14 reduces to

tt s ttþ1 ¼ tþ1 : 1 2 tt st 1 2 ttþ1

ð5:15Þ

In this case, we can state Proposition 3 (“Inverse compensated elasticity rule”). In the absence of cross-price effects, the optimal tax rate in each period is inversely proportional to the (compensated) labour-supply elasticity of the same period with respect to net wages. Building on the theoretical discussion of the intertemporal structure of implicit tax rates, we will now follow Sandmo’s (1974, p. 705) advice that “[e]mpirical explorations of optimal tax structures will be valuable contributions to further study of this problem” in applying his original results to our particular problem. In the next sections, we will develop an econometric framework with which to evaluate the optimality. As intertemporal cross-price effects are very hard to handle empirically, we focus on a test of the optimality rule stated in Proposition 3 with sts ¼ 0 thus focusing on a partial equilibrium framework. 5.2.3. Empirical evidence—methodology and data In this section, we build on the theoretical analysis provided in the last sections. We assume that individuals maximise utility with respect to their consumption of goods and leisure observing a budget constraint. The form of the utility function

Welfare Analysis of Pension Reforms

89

and the budget constraint are assumed to be known, thus determining the structure of the individuals’ labour-supply function. The parameters of the utility function turn into a number of coefﬁcients for all the determinants of individual labour supply that are captured by the model. These coefﬁcients can then be estimated empirically. We evaluate whether the timing of annual implicit tax rates in unfunded pension systems conforms to the “inverse elasticity rule”. Assume that labour supply for young individuals is less elastic than for older individuals. The decreasing path of implicit tax rates over the life cycle as calculated in section 5.2.1 for the German pension system would then be optimal as labour-supply elasticities and implicit tax rates would vary in an inverse manner. And assume in addition that the labour-supply elasticity of men exceeds the labour-supply elasticity of women. The observed gender-speciﬁc differences in the level of implicit tax rates would then go in the right direction. In what follows, we check whether the paths of labour-supply elasticities and implicit tax rates are related in an optimal way. For this, we estimate the elasticity of labour supply for different periods of a typical life cycle with respect to taxes imposed on wages or net wages. Empirical investigations into the time structure of wage elasticities of individual labour supply are largely lacking. Existing econometric studies usually show that, based on static models of labour supply, labour force attachment declines with age, controlling for a number of other socio-economic variables.29 Where age-related regressors are also included in a quadratic form (or as a polynomial of higher order), estimated time patterns of individual labour supply are a little more complicated but, as a rule, labour-force participation and/or the amount of labour supplied are monotonically decreasing starting from some year of age—usually around the late 20s or early 30s. Given our data constraints, we proceed by estimating a static model assuming that labour supply in one period of life is completely separated from labour supply in other periods, i.e. intertemporal cross-price effects are zero, sts ¼ 0:30 This will allow us to evaluate the results on the basis of the inverse compensated elasticity rule within a partial equilibrium framework (Proposition 3).

29 For studies for Germany, see Franz (1985), Strøm and Wagenhals (1991), Untiedt (1992), Buslei and Steiner (1999) or Kaltenborn (2000). International surveys can be found in Pencavel (1986), Killingsworth and Heckman (1986), and Blundell and MaCurdy (1999). Note that the ﬁrst three studies concentrate on the labour supply of women. 30 For a dynamical (i.e. life-cycle) model, see, for example, Heckman and MaCurdy (1980), Browning et al. (1985), Mulligan (1998), and Ziliak and Kniesner (1999).

90

Unfunded Pension Systems: Ageing and Migration

5.2.3.1. The research strategy In the following, we will investigate how labour supplied by men and married women in different age groups reacts to net wages and, hence, to implicit taxes imposed on wages. We thus focus on the question of how many hours of work an individual supplies in response to wage taxation given the labour-supply elasticity and abstract from the participation decision.31 Before estimating labour-supply functions in terms of hours worked and calculating the relevant labour-supply elasticities, however, we have to take a number of preliminary steps. For individuals who are employed, gross wages are easily observable. For non-employed individuals this information is lacking. Excluding the latter from a randomly assigned sample would involve a sample selection bias since the selection between those employed and those non-employed is not exogenous with respect to wages. In order to cure this problem, we follow Heckman’s (1979) two-stage procedure to estimate a function of (hourly) gross wages. We then use the estimated and observed gross wages in order to simulate the corresponding net wages for all individuals. For the simulation of net wages and corresponding (changes in) net household income, we rely on a simpliﬁed model of the German tax-transfer system, based on the tax simulation tool provided by Schwarze (1995). Effectively, our simulation tool covers the main types of taxes and social security contributions as well as the most important transfer programmes that are in operation in Germany.32 Of course, we only include the implicit-tax part of contributions in the relevant deductions; the saving part does not show up. Here, two problems arise. First, note that in the presence of progressive taxation and transfers that are both determined at the household level, hourly net wages are not deﬁned at an individual level. Instead, increments in net household income

31

Participation decisions are inﬂuenced by a number of institutional details other than implicit tax rates arising from the speciﬁc beneﬁt formula of a given pension system. For instance, if the focus were on participation vs. non-participation, special rules applying to early retirement would be very important. At the same time, other regulations that may inﬂuence the choice between an extended period of training and labour market entry would have to be taken into account. Here, however, we try to avoid these peculiarities by concentrating on individuals in those age groups (20–59) where these aspects can be expected to be of minor importance. 32 For an overview, see Haisken-de New (1997). The Schwarze (1995) simulation model includes taxes levied on income (Einkommensteuer, Solidarita¨tszuschlag) and contributions related to public pensions (gesetzliche Rentenversicherung), public health care insurance (gesetzliche Krankenversicherung), long-term care insurance (Pﬂegeversicherung) and unemployment insurance (Arbeitslosenversicherung). We included in addition the transfers provided in form of social assistance beneﬁts (Sozialhilfe) and housing beneﬁts (Wohngeld).

Welfare Analysis of Pension Reforms

91

which result from an increase in the labour supply are inﬂuenced by public transfers (or transfer reductions) and the progressive nature of taxes imposed on wages (including wages earned by the partner). We, therefore, use the information on hourly gross wages obtained from our wage estimation in order to calculate marginal increases in net household income for the hypothetical case that the individual is working one additional hour per year. The second problem follows immediately from this procedure. If, due to progressive (household-level) taxes and transfer reductions, net wages wit are given by non-linear increases in the household income, they are in fact no longer exogenous with respect to the number of hours actually worked. In other words, estimates that use marginal net wages in the sense laid out above as an explanatory variable for labour supply hit observed at an individual level can be seriously distorted. In order to take care of this potential endogeneity problem, we, therefore, use mean values of hours worked h it when simulating (changes in) household net income that may be relevant for decisions to work one additional hour, given the individual’s gross wage. Mean values h it are cell averages taken from appropriate sub-groups of individuals (formed by age, qualiﬁcation, and the number of children living in the household), thus re-establishing exogeneity of wit to the extent needed. We can then turn to estimating a labour-supply function. Since reliable information on the “desired” number of hours worked is rarely existent, it is conventional to use hours actually worked instead, which are easily observable. At the same time, actual working hours are clearly a censored (i.e. non-negative) variable. The standard procedure applied to investigating labour supply in terms of hours supplied is, therefore, given by the Tobit model (suggested by Tobin, 1958). Since observable values hit are censored, the marginal impact of changes in the individual- and time-speciﬁc variables xit are not equal to the coefﬁcients btj [ b: Consequently, the elasticities of labour supply with respect to net wages wit [ xit are calculated from

hwit

0 ›Eðhit lxit Þ wit b xit wit w ¼ ¼ bt F ; ›wit hit sh hit

ð5:16Þ

where F represents the distribution function for the standardised normal distribution with Nð0; s2h Þ: Of course, elasticities derived from the estimated coefﬁcient for net wages just represent “uncompensated” wage elasticities. Therefore, they have to be decomposed into income and substitution effects in a way that will be explained below. The latter will then give us the compensated elasticities we are interested in. As a last step, individual elasticities can then be aggregated using the relevant

92

Unfunded Pension Systems: Ageing and Migration

schedule of (cross-section and two-period longitudinal) weights in order to obtain representative estimates.33 5.2.3.2. Data and variables Our econometric analysis is based on micro-data taken from the GSOEP for the years 1988 to 1997.34 For the purpose of our study, we will pool the series of annual data provided by the GSOEP to form a number of appropriate sub-samples. We will conﬁne our attention to West-German households of German nationality. Thus, we want to ensure that our results are not distorted by the process of social and economic change that is going on in East Germany in the aftermath of reuniﬁcation, or by idiosyncratic features in the behaviour of foreigners and immigrants. Since our focus is on labour-force participation, we select individuals aged 20 – 59, looking at men and married women in turn. Finally, we exclude the self-employed and civil servants because in Germany these individuals are subject to a very different treatment with respect to old-age provision. We also exclude those who are already in retirement. The general idea of all these restrictions is to have a sample of individuals who are as homogenous as possible without greatly reducing the number of observations. In our estimates, there are two endogenous variables. First, the hourly gross wages, which are needed for imputing gross wages to persons who are not employed, as the point of departure for simulating net wages and net household income (Heckman procedure). Second, the actual number of hours worked, which is used as a proxy for the “desired” amount of labour supply and is accounted for on a weekly basis (Tobit model). Exogenous variables that are assumed to determine labour supply can be grouped by the time-variant and individual effects that they are expected to cover. Basically, period-speciﬁc effects that are common to all individuals are captured by a dummy variable “year of survey”. Potential business cycle effects are represented by annual rates of unemployment which can also be interpreted as a proxy for changes in labour demand and other features of the labour market situation that may vary over time. Individual determinants can be traced back to variables like completed levels of schooling and professional training, years 33 For each observation—i.e. for each individual in each year—we are effectively using information on annual amounts of income that are collected retrospectively in subsequent waves of a panel survey. Therefore, we have to correct the weights attached to individuals in a cross-section perspective by the probability of “survival” from period t to period t þ 1: Note that, here, individuals with hit ¼ 0 have to be excluded, for the reason that their elasticity hw will be inﬁnite given the way it is calculated. 34 Some additional information regarding the GSOEP data-set is included in Appendix C5.

Welfare Analysis of Pension Reforms

93

Table 5.1: Descriptive statistics: men by age groups. Characteristics

Age No. of children Children aged 0–3 Children aged 4–6 Living with a partner Secondary degree University degree Vocational training Net household income Interest þ Rent income No. of observations Employed as of total sample Job experience (years) Hours worked per week Gross hourly wages Net household income Interest þ Rent income Marginal net hourly wages

Men aged 20–39

D D D D D D

Men aged 40 –59

Sample means

Std. dev.

Sample means

Std. dev.

28.970 0.656 0.060 0.035 0.723 0.899 0.113 0.652 16,148 722

5.431 0.971 0.238 0.183 0.448 0.302 0.316 0.476 9869 4353

49.418 0.618 0.013 0.018 0.920 0.964 0.117 0.804 22,894 1095

5.885 0.962 0.115 0.132 0.272 0.187 0.322 0.397 13,295 3509

9369 76.1% 8.115 37.001 16.828 17,745 624 7.996

5720

5.728 8.952 9.148 9969 4293 6.352

84.2% 27.292 38.604 19.924 24,786 1141 11.176

7.922 7.046 10.177 13,405 3491 6.601

D denotes dummy variables with D ¼ 1 if characteristic is present. Incomes are in year-1995 Euros.

worked in full-time employment (as a proxy for actual job experience), living with a partner and having children. With regard to the individual budget constraint, two further variables are clearly important: net household income (including income derived from wages earned by the spouse, capital income, public transfers, etc.) and the net wage earned if the individual’s labour supply is extended by 1 h on an annual level.35 The descriptive statistics grouped by the larger age groups of young (20 – 39 years) vs. older (40 –59 years) individuals are summarised in Tables 5.1 and 5.2. Some of the facts that are evident for men and married women are clearly artefacts of our deﬁnition of age groups. This is not only true for the result that individuals in the higher age group are older than individuals in the lower age group. For similar reasons, the latter are much more likely to live with small children, while the former have longer job experience. Nonetheless, among 35

As already explained above, we evaluate both income and wages at the cell averages of hours actually worked calculated from appropriate sub-groups of individuals but given the individual’s gross wage and household income in order to avoid a potential endogeneity problem regarding net wages and the number of hours worked.

94

Unfunded Pension Systems: Ageing and Migration

Table 5.2: Descriptive statistics: married women by age groups. Characteristics

Age No. of children Children aged 0–3 Children aged 4–6 Secondary degree University degree Vocational training Net household income Interest þ rent income No. of observations Employed as of total sample Job experience (years) Hours worked per week Gross hourly wages Net household income Interest þ rent income Marginal net hourly wages

D D D D D

Married women aged 20–39

Married women aged 40 –59

Sample means

Std. dev.

Sample means

Std. dev.

31.841 1.438 0.129 0.081 0.924 0.058 0.749 20,353 676

4.547 1.089 0.335 0.272 0.265 0.234 0.433 11,484 3561

48.296 0.514 0.006 0.010 0.954 0.037 0.651 22,160 1120

5.534 0.826 0.080 0.099 0.210 0.189 0.477 12,614 3345

4388 50.7% 6.703 26.993 11.766 21,878 646 5.323

4028

4.759 11.906 6.347 11,798 4381 4.180

53.9% 12.363 25.982 11.724 23,964 984 5.234

9.350 11.177 5.805 11,656 3424 3.854

D denotes dummy variables with D ¼ 1 if characteristic is present. Incomes are in year-1995 Euros.

the other variables considered, differences over the age groups may be relevant for the individual probability to enter employment and to work for the given hours per week. One may note that young men have about the same number of children living in their households as older men, while on the side of married women there is a pronounced difference regarding the number of children in the household across the age groups. Young women are more likely to live with children, while for older women this probability is close to that for men. The levels of education completed do not show much variation across age groups and across gender. The slightly higher share of older men with a secondary or university degree compared to younger men and the larger part of older married women with a secondary degree than younger married women can be explained by the fact that some of the individuals in the younger age group are still about to ﬁnish their secondary or university degree. That younger age groups acquire on average a higher degree when this effect is taken into account becomes clear when looking at the share of young married women with a university degree. Even though some young women might still be at university, the share of young women with a university degree exceeds the share of older women. Thus, the numbers given in Table 5.1 need not contradict, and those in Table 5.2 can be taken to

Welfare Analysis of Pension Reforms

95

conﬁrm the well-known long-term shift towards higher levels of skills that are both offered and demanded on the labour markets in industrial countries. In total, most individuals have completed a secondary degree and between 65 and 80% have followed vocational training. There is, however, an across-gender difference concerning academic education, with twice as many men having a university degree than married women. The participation rate of women is clearly lower than that of men. The fraction of individuals in employment is constant for women across age groups whereas there is some difference for men across age groups. The latter can be explained by the fact that, given our deﬁnition of age groups, many of the young men may not have completed their training or education yet, whereas older men certainly have. One would also expect a lower participation rate for young women than for older women across age groups. In addition to the same reasons, which apply to young men, young women should also be more likely to be on maternity leave than older women. Controlling for this, the fraction of young women working may, therefore, have considerably increased over the time. From the number of hours worked per week it is easy to see that, as in many other countries, married women who are employed work less on average than men. At the same time, there are no major differences across age groups regarding this variable. Without many exceptions, men participate in the labour market on a full-time basis, while women in employment work on average 26 – 27 h per week. There is clearly a wage gap across gender if we look at gross hourly wages, but it is much smaller for young individuals than for older ones. With the exception of academic degrees, young married women have (more than) caught up with young men in terms of their qualiﬁcations, while older women are less qualiﬁed than their male counterparts. Wages increase for older men compared to young men. By contrast, gross hourly wages of young married women exceed those of women in the higher age group by a small margin. Of course, part of the explanation lies with the higher qualiﬁcations of young married women. The most important variables for our analysis are, however, the net household income and net wage variables. Net household income comprises all sources of income (own wages and wages of the spouse, rent and interest income, etc.) on an annual basis. To see the importance of non-work income, interest and rent income is also listed separately. Marginal net hourly wages per week are then calculated as the additional net wage earned through an additional hour of work per year. The values obtained correspond to reasonable assumptions. In Germany, marginal tax rates can easily exceed 50%. They turn out to be lower for older individuals than for the young— among other things because the implicit tax rates involved in the pension system decline substantially over the life cycle. At the same time, marginal tax rates are

96

Unfunded Pension Systems: Ageing and Migration

much higher for married women than for men. This reﬂects mainly the effects of progressive taxation at the household level as well as the higher implicit tax rates. The increase of net household income of men and married women over the life cycle results from higher wages of the individual and the spouse due to seniority and life-cycle effects concerning interest and rent income. It is interesting to see that interest and rent income is slightly higher for older employed men than for non-employed, whereas it is lower for younger employed men and for employed married women compared to the respective non-employed individuals. The latter is consistent with the usual assumption of a negative income effect with respect to labour-force participation, while similar effects are seemingly absent in the case of older men. 5.2.3.3. The sub-sample design Our basic approach is to split a full sample of individuals (aged 20 –59) into two or more broad age groups (young workers aged 20– 39 and older workers aged 40 –59) and to consider a series of standard, static regressions regarding the labour-supply behaviour of each of these groups in separation. One major problem involved in this estimation strategy is given by potential cohort effects. When pooling data that cover a time span of 10 years and splitting the pooled data into two broad age groups for a ﬁrst round of estimates, the “life-cycle” effects on the wage elasticity of labour supply that will show up in our results may effectively be driven by the fact that young and older individuals belong to different birth cohorts. These cohorts may differ with regard to labour force attachment from the very ﬁrst day of their working life. Clearly, this potential distortion must be taken into account when interpreting our results. Given the long period of time spanned by our data, however, there may be a possibility to separate life-cycle from cohort effects.36 Note that in a pooled sample which is based on annual survey data ranging from 1988 to 1997 the two sub-groups deﬁned by the age groups 20– 39 and 40– 59 will overlap. Older workers are made up by the cohorts born between 1929 and 1957, while young workers are born between 1949 and 1977. As a consequence, nine cohorts will be contained in both age-related sub-samples, moving from the young to the older group of workers during the survey period. This overlapping structure of our data can be exploited to trace back cohort effects in detail—even though the overlap is too small to be used for controlling potential cohort effects in a systematic way. 36

For an empirical treatment of cohort effects, see Boockmann and Steiner (2000).

Welfare Analysis of Pension Reforms

97

Table 5.3: Wage elasticities of labour supply: structure of the results.

Building on our basic estimates for the two sub-samples of young and older workers, we will then proceed in the following direction. We split the pooled sample into two sub-samples I and II, which cover the waves of 1988 – 1992 and 1993 –1997, respectively. In addition, we consider a larger number of age groups, consisting of ﬁve birth cohorts each. Given these operations, all birth cohorts belong to different age groups in the two sub-samples: in sample I, they are aged 20 –24, 25 – 29, etc., while in sample II the same individuals are aged 25 – 29, 30– 34, etc. At the same time, all age groups are represented by two different groups of birth cohorts, one from sample I and another one from sample II. The structure of the two-dimensional classiﬁcation which we obtain in this way is illustrated in Table 5.3. Here, age groups are listed in columns and birth cohorts in rows. The shaded cells indicate which of the potential combinations are covered by our data. Within the above structure, the procedure of estimation is essentially the same as before. We concentrate on smaller groups of individuals. Given that we have ﬁlled the table with all the results we are looking for—i.e. with estimates for labour-supply elasticity for each birth cohort and age group marked by a shaded cell—we may then interpret differences that show up in the vertical direction as cohort effects, while differences in the horizontal direction can be attributed to life-cycle effects. This approach shows the importance of cohort effects relative to life-cycle effects and helps us to isolate the life-cycle effects we are interested in. 5.2.4. Individual labour-supply elasticities: the results The focus of our investigation is on typical life-cycle patterns of the labour-supply elasticity, i.e. on differences between young and older workers. We will now treat

98

Unfunded Pension Systems: Ageing and Migration

the case of men and married women belonging to different age groups in separate estimates. 5.2.4.1. Estimates As explained before, information on gross wages is available only for individuals who are actually employed. However, restricting attention to these individuals in order to estimate the labour-supply elasticity involves a potential sample selection bias and may thus lead to distorted results. We, therefore, employ the Heckman (1979) procedure to estimate gross wages.37 We use these gross wages to calculate corresponding net wages and to derive net household incomes and marginal net hourly wages. We then move on to estimating the Tobit model by addressing the number of hours worked on a regular basis, now including marginal net wages and net household incomes as additional regressors. Furthermore, we now have to distinguish between the age groups deﬁned before. For ease of exposition, we limit the presentation to the estimates obtained for the larger sub-groups of cohorts aged 20 –39 and 40– 59, respectively, in Tables 5.4 and 5.5.38 In the case of men, coefﬁcients estimated for the partner variable and the number of children in the household are signiﬁcant and positive throughout. Living with children aged less than 3 years has a signiﬁcantly negative impact on the amount of hours worked for older men, in contrast to a positive impact for young men. Job experience and levels of qualiﬁcations become less important in the case of older individuals where holding an academic degree even has a negative impact on the amount of labour supplied. Another observation is that the (negative) role of unemployment rates is much more important for older individuals than for young ones. Turning to the estimates for married women, the picture is slightly different— the main similarity being that the model again seems to ﬁt better for young women than for those in the higher age group. This time, all child-related variables have a negative, but not always signiﬁcant coefﬁcient, which is easily explained through child-care obligations and scarce supply of institutional child-care facilities in Germany. Apparently, both aspects are more relevant for labour supply decisions of married women than of men. Surprisingly, the coefﬁcients related to all levels of qualiﬁcations—with one exception—now also exhibit negative signs, though only university degree and secondary degree for young women have a signiﬁcant 37 38

The results of the wage equation can be found in Appendix D5. See Appendix E5 for the results for the smaller sub-groups.

Welfare Analysis of Pension Reforms

99

Table 5.4: Results of the Tobit model: men. Variable

Men aged 20 –39 Coefﬁcient

Living with a partner Children aged 0–3 Children aged 4–6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2.760*** 2.143** 2 1.156 0.694*** 2.786*** 3.181*** 3.534*** 6.675*** 2 0.621*** 0.018*** 2 0.339*

Net yearly household income E Marginal net hourly wages w

1.73 £ 1025 0.085***

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

9369 2242 0.000 6.71%

Std. err. 0.440 0.941 1.110 0.246 0.627 0.638 0.457 0.223 0.028 0.001 0.177 0.000 0.000

Men aged 40–59 Coefﬁcient 1.549** 2 4.451** 0.546 2.472*** 0.923 2 2.025** 1.456** 0.601* 2 0.025 0.000 2 1.090*** 8.34 £ 1025*** 0.354***

Std. err. 0.720 2.228 1.948 0.292 1.495 0.899 0.633 0.355 0.016 0.000 0.214 0.000 0.000

5720 903 0.000 3.89%

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level. Incomes are in year-1995 Euros.

effect. As in the case of men, the overall impact of current job experience is more important for young women. From an economist’s perspective, two major determinants of individual labour supply should be the wage and household income. The coefﬁcients estimated for the marginal net hourly wage w are both highly signiﬁcant and positive in all subsamples, which is precisely what one would expect. Taxing wages should reduce labour supply at least with regard to the “uncompensated” effect—although theoretically the effect can go either way. The coefﬁcient of household income E is highly signiﬁcant in all sub-samples except for younger men. Considering the importance of early years in a work career for lifetime income of men, this may be largely plausible. Against a theoretical background, a result that is more striking is that household income has a positive effect for men in both sub-samples—implying a positive income effect. This result is not uncommon in empirical studies of labour supply for men—and in fact, the effects are insigniﬁcant for young men and not very large in both cases.39 39

For broader surveys, see the references made in footnote 29.

100

Unfunded Pension Systems: Ageing and Migration

Table 5.5: Results of the Tobit model: married women. Variable

Married women aged 20– 39 Coefﬁcient

Children aged 0–3 Children aged 4–6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

2 17.620*** 2 0.817 2 9.302*** 2 4.268*** 2 5.511*** 2 1.338 1.855*** 2 0.211*** 0.009*** 1.304*** 2 1.0 £ 1024*** 1.173*** 4388 2163 0.000 7.72%

Std. err. 1.436 1.532 0.389 1.553 1.678 0.959 0.461 0.060 0.002 0.354 0.000 0.052

Married women aged 40–59 Coefﬁcient 2 1.820 2 4.494 2 2.951*** 2 2.232 2 12.813*** 0.283 0.171 0.017 2 0.000 0.257 2 1.0 £ 1024*** 1.757***

Std. err. 3.722 3.470 0.462 1.908 2.074 0.741 0.241 0.017 0.000 0.309 0.000 0.052

4028 1855 0.000 8.13%

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10 level. Incomes are in year-1995 Euros.

An increase in household income by e1000 would lead to an increase in the number of hours worked per week of 0.02 for young men and of 0.08 for older men. For married women, household income has a negative effect in both subsamples, which indicates a negative income effect. We may conclude that labour-supply decisions of married women are more in line with the assumption of optimising behaviour for given wages and household income than those of men. 5.2.4.2. Wage elasticities of labour supply Building on our estimates for the Tobit model, we can then determine the elasticities of labour supply with respect to net wages. In order to see whether periodic implicit tax rates, tt and ttþ1 ; are related to the wage elasticities observed for young vs. older individuals as required by the “inverse elasticity rule”, it is the compensated wage elasticities s that we have to determine. Compensated elasticities of labour supply can be derived by splitting the observable, uncompensated elasticities h into an income effect and a substitution effect

Welfare Analysis of Pension Reforms

101

Table 5.6: Wage elasticities of labour supply: young vs. older individuals. Men aged 20–39 Compensated wage elasticity s Income effect Uncompensated elasticity h

0.041*** 0.000 0.042***

Men aged 40–59 0.236*** 0.002*** 0.238***

Married women aged 20 –39

Married women aged 40 –59

0.543*** 2 0.001*** 0.542***

0.809*** 2 0.001*** 0.808***

Compensated elasticity s Income effect Uncompensated elasticity h

***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level.

according to

h¼

›h w ¼ ›w h

›h E wh ›h w wh þs þ ¼ eE E ›E h E ›w h S |ﬄﬄﬄﬄ{zﬄﬄﬄﬄ} |ﬄﬄﬄ{zﬄﬄﬄ} Income effect

ð5:17Þ

Substitution effect

by the Slutsky decomposition. In theory, the substitution effect should be positive, while the income effect should be negative. The direction of the combined effect and, therefore, the sign of h is undetermined. As we have seen in the previous section, we cannot expect all results to be well behaved in the above sense in an empirical context with detailed regressions for men and married women in different age groups. In order to see what happens within our econometric model, we calculate the uncompensated effect h from the coefﬁcient for w and the income effect e E wh=E from the coefﬁcient of E: The substitution effect s can then be determined as a residual. The ﬁnal results are estimations for men and married women and displayed in Table 5.6. Table 5.6 shows that (compensated) wage elasticities obtained for older men and older married women are higher than for those in the younger age group. In fact, this is the result expected beforehand, based on the casual observation that labour force attachment gets weaker near the end of the working-age period for a number of reasons.40 In both cases, the (within-gender) difference in elasticities 40

Note that those already receiving disability beneﬁts or early retirement pensions are excluded from our sample. At the same time, we are unable to control for the fact that some who would be entitled to receive the same types of beneﬁts may continue to work, nor can we deal with some other forms of (partial) “early retirement” that are more hidden—for example, that are effectively ﬁnanced through unemployment beneﬁts.

102

Unfunded Pension Systems: Ageing and Migration

appears to be substantial. At the same time, we observe remarkable differences in wage elasticities across the two gender groups for both age categories. We will return to the latter observation in the discussion of our results, concentrating ﬁrst on an investigation of the relevant life-cycle effects. However, in order not to arrive at premature conclusions, we should take into account the fact that both periods considered here are highly aggregated. As a matter of fact, the group of young individuals aged 20 – 39 mixes those who are still students (at best, working on a limited basis in order to expand their current budget) with individuals who are just entering the labour force and with others who are in the middle of their career. Similarly, the group of older individuals aged 40 –59 includes those at the height of their working career as well as those who are already approaching retirement. To work out the potential differences in labour-supply behaviour across all of these different groups in detail, we, therefore, move on from our basic ﬁndings by looking at a larger number of age groups that are more narrowly deﬁned, at the same time attempting to disentangle pure life-cycle effects from potential cohort effects. Tables 5.7 and 5.8 contain the values estimated for the compensated elasticity s for the case of nine groups of birth cohorts which move from one 5-year age group to another over two sub-samples I and II, covering the panel waves of 1988 – 1992 and 1993 –1997, respectively. Within the structure explained in Table 5.3, the procedure of estimation is essentially the same as before. We only concentrate on smaller groups of individuals now. Having ﬁlled the table with estimates for the elasticity of labour supply with respect to wages for each of the shaded cells, we may interpret any differences that show up in the vertical direction as cohort effects, while horizontal differences can be attributed to life-cycle effects. Table 5.7: Compensated wage elasticities of labour supply over the life cycle: men. Birth cohorts

Age groups 20 – 24

69 – 77 64 – 72 59 – 67 54 – 62 49 – 57 44 – 52 39 – 47 34 – 42 29 – 37

0.123*** 0.064*

25– 29 0.071*** 0.008

30 – 34

0.012 2 0.065**

35 – 39

2 0.005 2 0.034

40 – 44

0.003 2 0.004

45 – 49

0.185*** 0.200***

***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level.

50 – 54

0.158*** 0.143***

55 – 59

0.298*** 0.442***

Welfare Analysis of Pension Reforms

103

Table 5.8: Compensated wage elasticities of labour supply over the life cycle: married women. Birth cohorts

Age groups 20–24

69–77 64–72 59–67 54–62 49–57 44–52 39–47 34–42 29–37

25 –29

30–34

35–39

40–44

45 –49

50–54

55–59

0.369*** 0.486*** 0.578*** 0.614*** 0.448*** 0.602*** 0.498*** 0.546*** 0.580*** 0.747*** 0.581*** 0.773*** 0.835*** 0.939*** 0.754*** 0.857***

***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level.

Although we observe in our results some irregularities—such as negative wage elasticities mixed with positive ones—and movements from one cell to another that are seemingly erratic, we should not put too much emphasis on insigniﬁcant estimates. Keeping in mind the distinction between life-cycle effects, to be derived from adjacent cells in a horizontal direction, and cohort effects which show up in a vertical direction, we can ﬁnally detect a meaningful life-cycle pattern of wage elasticities, as shown in Figure 5.4. In Figure 5.4, we have two estimates of the compensated wage elasticity of labour supply for each age group of both men and married women. Each of these pairs is taken from different birth cohorts who we follow from sub-sample I to subsample II. Values for two successive age groups that are derived from the same set of birth cohorts are connected by a line. Building on our earlier considerations, circles or rhombi linked by a line represent the life-cycle effects for one sub-sample of birth cohorts, while jumps in the vertical direction indicate cohort effects.41 We ﬁnd that for men, the compensated elasticity of labour supply with respect to net wages is slightly higher for young individuals, aged 20 –29, than for those aged 30 –44. In addition, the estimates obtained for the very young appear to be rather reliable in terms of statistic signiﬁcance whereas most of the estimated negative elasticities for men aged 30 – 44 turn out to be insigniﬁcant. These results are not too surprising: these are the age groups of men who are characterised by participation rates which, in many industrial countries, are very close to 100%. They are very likely to work anyway, irrespective of the economic incentives 41

Horizontal movements can be also affected by period effects—i.e. differences between the subperiods 1988 to 1992 and 1993 to 1997—which may be imperfectly controlled by our “year of survey” dummies and by aggregate unemployment rates for each year.

104

Unfunded Pension Systems: Ageing and Migration

Figure 5.4: Compensated wage elasticities of labour supply over the life cycle. Filled circles and rhombi indicate signiﬁcance at the 10% level. Source: GSOEP (waves 1988–1997); own estimates.

to do so. For men aged 45 and over, elasticities of labour supply go up. Again, these results are highly signiﬁcant for all the six groups that fall into this category. In other words, men who are approaching retirement are more likely to respond to economic incentives than younger men. As a consequence, implicitly taxing the wages of these individuals must be expected to have a stronger impact on the amount of labour supplied than taxing the wages of younger individuals. For married women, the labour-supply elasticities that follow from our estimates are highly signiﬁcant across all age groups. For younger women, aged 20 –39, they exhibit a slightly decreasing life-cycle trend. However, as in the case of men, the elasticity starts to increase from the age group of 40– 49 and does so over all subsequent age groups. Furthermore, across all age groups considered here the elasticity of labour supply with respect to wages is at a higher level for married women than for men. This observation reﬂects a common result in many econometric studies indicating that female labour supply reacts more strongly to changes in net wages than male labour supply. In addition to these observations it should be noted that in Figure 5.4, vertical differences between the estimates of labour-supply elasticities are close to zero in the case of men. Cohort effects are thus negligible. For married women, we see cohort effects for some age groups. However, they do not disturb the general life-cycle path.

Welfare Analysis of Pension Reforms

105

Table 5.9: Time structure of implicit taxes and wage elasticities: young vs. older individuals. Men

t20 – 39 1 2 t40 – 59 5 1:55 1 2 t20 – 39 t40 – 59

Married women

t20 – 39 1 2 t40 – 59 5 1:35 1 2 t20 – 39 t40 – 59

s40 – 59 ¼ 5:78 s20 – 39 s40 – 59 ¼ 1:49 s20 – 39

5.2.5. Is the structure of implicit taxation optimal? As the ﬁnal step we need to discuss whether the actual pattern of implicit tax rates is optimal according to the “inverse elasticity rule” as stated in Proposition 3. More precisely, we have to evaluate the intertemporal structure for men and married women against the background of our theoretical and empirical results. The ratio of implicit tax rates for young vs. older individuals should be inversely related to the ratio of (compensated) wage elasticities.42 The same condition should also hold for the ratio of implicit tax rates for men vs. married women. Both cases require that a welfare comparison across individuals of the same or of different generations is possible. We will now look at each of these issues in turn. Concentrating ﬁrst on the intertemporal structure of t for the two broad age groups, the results are summarised in Table 5.9. Once again, the relevant tax rates tt are calculated using the CESifo pension model, while st is known from Table 5.6. Let us start by looking at the results for men. For the “inverse elasticity rule” to hold, the ratio of the implicit tax rates and the inverse ratio of the labour-supply elasticities should be equal. This is clearly not the case, as the former is 1.55 and the latter is 5.78—thus almost four times as large. In order to increase the ratio of implicit tax rates, taxes on younger individuals should be higher and/or taxes on older individuals should be lower. Although the path of implicit tax rates of the German public pension system is already falling over the two periods, the implicit tax rates for young individuals could be increased even more relative to the rates for older individuals given the relatively low labour-supply elasticity for young individuals compared to older individuals. Optimally, labour supply should be taxed more in that period, when it reacts less strongly and vice versa.43 Turning to married women, we can conclude from Table 5.9 that the overall structure of implicit tax rates across the life cycle is roughly consistent with 42

We will comment further down on how the results would change qualitatively if we took the total (marginal) taxes into account, and not only the implicit taxes. 43 See section 5.2.6 for a discussion about how this differentiation of implicit tax rates across age groups could be realised.

106

Unfunded Pension Systems: Ageing and Migration

Figure 5.5: Compensated wage elasticities of labour supply over the life cycle: men. Filled circles and rhombi indicate signiﬁcance at the 10% level. Source: See Figures 5.3 and 5.4.

the “inverse elasticity rule”. The ratio of the implicit tax rates is 1.35 and thus only slightly smaller than the inverse ratio of the labour-supply elasticities, which is 1.49. As in the case of men, the tax rates imposed on young women could be even increased relative to the rates for older women. However, considering the small difference between the ratios of implicit tax rates and inverted labour-supply elasticities, the implicit taxes are already quite optimally distributed over the life cycle. Again, however, we should take into account that within the two broadly deﬁned age groups wage elasticities vary much more than that shown in Table 5.9. In order to evaluate the “inverse elasticity rule” in detail, we, therefore, look at the life-cycle structure of implicit taxes and labour-supply elasticities for the larger number of age groups deﬁned over 5-year intervals.44 In Figure 5.5, we have repeated the path of implicit tax rates from Figure 5.3 and the path of labour-supply elasticities from Figure 5.4 for men. One can note that, in general, the paths are inversely related. However, one can also see that some modiﬁcations of the path of implicit taxes as indicated by the dotted arrows would further decrease distortions of the labour supply. Individuals who are at 44

See Appendix F5 for the detailed results.

Welfare Analysis of Pension Reforms

107

the early stages of their working life—those aged 20 to 29—react slightly more elastically to changes in net wages than workers who are in the middle of their working career. The implicit tax proﬁle should thus be lower for these age groups relative to those who have settled in their jobs. In the absence of adjustments of this type, the path of implicit taxes of unfunded pension systems may effectively create a barrier to fully participating in the labour market for job entrants. We also observe that the level of implicit tax rates imposed on men aged 30 –44 does not matter too much from an optimal taxation perspective. Here, the elasticity of labour supply with respect to wages is close to zero, so that excess burdens of taxation are very small. Things are different with respect to the implicit tax rates placed on men aged 55 –59. Our results imply that according to the “inverse elasticity rule”, the relatively low implicit tax rates for these individuals are still much too high given their high labour-supply elasticities. Otherwise, besides choosing one of the various routes into early retirement,45 older workers are likely to avoid working overtime hours and try to reduce their workload through part-time work if their tax burden is too high. Figure 5.6 shows the path of implicit tax rates from Figure 5.3 and the path of labour-supply elasticities from Figure 5.4 for married women. A general impression is that the wage elasticity of labour supply of married women decreases slightly for young age groups and increases for those aged 40 and over. Thus, on the whole, implicit taxes should be slightly lower for the 25 to 34-year-old as indicated by the dotted arrow. For the older age groups, the declining tax proﬁle, however, is largely appropriate. If we put these ﬁndings together, we can conclude the following: Proposition 4. According to the “inverse elasticity rule”, the optimal time structure of implicit taxes for men and married women should follow an inversely “U-shaped” pattern. The current timing of tt is roughly consistent with optimality conditions for the case of middle-aged individuals. But the high taxes imposed on young individuals and the excessively slow decrease of tax rates falling on men who are about to enter retirement constitute problems in terms of efﬁciency.

45

This option is not considered in our empirical model where we concentrate on labour supply in terms of hours worked. Studying participation decisions taken by older workers in the presence of early retirement programmes, which imply an increase in implicit tax rates to more than 100% if the alternative is to retire now or later without any corresponding adjustment in the level of annual pension beneﬁts, is clearly an important issue in itself (see, for instance, Brinch et al., 2001).

108

Unfunded Pension Systems: Ageing and Migration

Figure 5.6: Compensated wage elasticities of labour supply over the life cycle: married women. Filled circles and rhombi indicate signiﬁcance at the 10% level. Source: See Figures 5.3 and 5.4.

For men and married women, an optimal path should at least start at lower levels of implicit tax rates, then increase for some years and ﬁnally decrease over the entire working life. By contrast, abstracting from discretionary changes in contribution rates or beneﬁt levels that are rarely intended to actively shape the life-cycle structure of implicit tax rates, implicit tax rates in many existing unfunded pension systems are constantly declining over the life cycle. So far, we have focused on the optimal timing of the implicit tax rates, treating men and married women in isolation. As a last step, we should also spell out our ﬁnal results regarding the (non-)optimality of the “gender tax-gap” that exists in many public pension systems. In order to do so, we look at cross-gender differences between implicit tax rates and labour-supply elasticities rather than at their structure over time. We, therefore, restrict our attention to the structure of tt and st for men and married women based on the two broad age groups considered in Table 5.9.46 The results are shown in Table 5.10. 46

From the results presented in Tables 5.7, 5.8, E5.1 and E5.2, it would be easy to derive a richer picture for the case of men vs. married women building on our 5-year age groups. We abstract here from this option.

Welfare Analysis of Pension Reforms

109

Table 5.10: Time structure of implicit taxes and wage elasticities across genders: men vs. married women. Young individuals

women men t20 1 2 t20 – 39 – 39 ¼ 1:13 women men 1 2 t20 t 20 – 39 – 39

Older individuals

women men t40 1 2 t40 – 59 – 59 ¼ 1:29 women men 1 2 t40 – 59 t40 – 59

men s20 – 39 women ¼ 0:08 s20 – 39 men s40 – 59 ¼ 0:29 women s40 – 59

The results are already immediate from Figure 5.4. Since married women in both age groups are much more responsive to taxes, implicit tax rates imposed on them should always be lower than for men. However, given the usual “secondearner” status of many women and the widely used reductions in their pension beneﬁts if the latter coincide with dependents’ or survivors’ beneﬁts, implicit tax rates for women are much higher than the tax burden falling on men. In this case, the “inverse elasticity rule” of optimal taxation is obviously violated. Proposition 5. The implicit taxes should be higher for men than for married women when considered together according to the “inverse elasticity rule”. This optimality requirement is clearly violated. We have focused so far on analysing the optimality of the path of implicit taxes over the life cycle given the labour-supply elasticities. This has allowed us to see the distortions of labour supply due to implicit taxes in isolation—an aspect which has so far not received much attention in the debate about where to ﬁnd and how to exploit the inefﬁciencies of unfunded pension systems for a Pareto-improving transition towards more funding. We want to conclude by discussing how the results have to be modiﬁed when including explicit (marginal) taxes in the analysis. In contrast to implicit taxes, the path of which depends on the age group of the individual, explicit taxes are a priori age-neutral and, given their progressive character, the rates rise instead with wage. If we assume that wages reﬂect productivity or are based on seniority, then wages—and consequently explicit taxes—can be expected to increase on average over the life cycle. If we thus add the rising path of explicit tax rates to the falling path of implicit tax rates, some of our conclusions become even stronger. We have found so far that implicit taxes should be lower for young men and married women and for old men compared to middle-aged individuals. While our results are modiﬁed with respect to the young age groups for which the comparatively low explicit taxes might sufﬁciently counterbalance the high implicit taxes to minimise the distortions of the labour supply, this is not the case for the older age groups. Here, our results hold a fortiori.

110

Unfunded Pension Systems: Ageing and Migration

The labour supply of men is even more strongly distorted if explicit taxes are taken into account and the labour supply of married women is now also likely to be distorted. This implies that implicit taxes have to be reduced to an even larger extent for older individuals to establish optimality with respect to labour-supply elasticities. To evaluate the impact of explicit and implicit taxes on labour-supply decisions on an individual level, it is, however, necessary to consider the personal work biographies. They might in fact diverge from the representative individual on whom the calculations of the explicit taxes are based with respect to the importance of productivity and seniority for the wage determination. In what follows we, therefore, restrict our attention again to implicit taxes for the concluding remarks because the reasoning is more straightforward and—as we have seen—the results can also be expected to hold qualitatively when explicit taxes are included in the analysis. 5.2.6. Conclusion What have we learned from our investigation into the structure of implicit taxes that are involved in virtually all unfunded pension systems? And what are the policy implications of our ﬁndings? If we restrict our attention to the scenario which corresponds to the simple three-period model employed in the theoretical analysis, our conclusions are as follows: since in real-world pension systems implicit tax rates must be expected to decrease over the life cycle, support for the optimality of this structure can be based on the observation that the labour-supply elasticity increases over the same period of time for both men and married women. The paths of implicit tax rates and labour-supply elasticities are, therefore, inversely related in accordance with the “inverse elasticity rule”. At the same time, our results point to three potential sources of inefﬁciencies entailed in many existing unfunded pension systems. First, when investigating the time-structure of implicit taxes in detail, one problem is given by the fact that the highest level of implicit tax rates usually falls on very young individuals. Second, another problem is constituted by the excessively high level of implicit tax rates falling on individuals near their retirement age.47 A third problem arises from the higher level of implicit taxation which is relevant for a typical second earner and thus mainly for married women. 47

The problem here is not that this level of taxation is lower than for younger individuals, but that it may still be too high given the increase in the labour-supply elasticity.

Welfare Analysis of Pension Reforms

111

The latter problem falls in the same class as a number of other well-known distortions of similar types that are created through progressive household-level taxation, child-related beneﬁts that are inversely related to household income or decrease with the number of hours worked by the mother, etc. Therefore, it may not rank highest on the agenda for strengthening the incentives for women to participate in the labour market. Nonetheless, there may be reasons to redeﬁne rules that govern the treatment of married couples in existing unfunded pension systems and, in particular, to reconsider the widespread use of (non-contributory) survivor or spouse beneﬁts. In general, the key to solving these problems will lie in increasing the degree of actuarial fairness by determining individual pension claims much more on the basis of individual accounts. For example, mandatory contributions could be paid to spouses who are not working. It can be expected that for single-earner couples the burden involved in public pension systems would go up. For all other individuals—two-earner couples as well as single men and women—it may go down, because the link between contributions and pension beneﬁts would become more actuarial, ceteris paribus. Let us turn to the ﬁrst two potential inefﬁciencies mentioned above. The ﬁrst problem concerns the excessive burdens created through high implicit tax rates for those just entering the labour force. Young people may prefer to stay in the educational system for a longer period of time than is necessary to invest in future productivity or to complete a degree for signalling purposes.48 They may try to work on a part-time basis only or avoid work subject to contributions to the pension system by entering into the shadow economy or choosing selfemployment if the latter is a strategy for not paying contributions as is the case in Germany. From an economic point of view, this may put unfunded pension systems under substantial ﬁnancial pressure. The second problem refers to the implicit tax rates that may be too high for those—notably men—just about to enter retirement. In practice, there are several routes for older workers who want to exit from the labour market. Income-support programmes sometimes allow for a partial exit. Similarly, there are programmes in which the part-time work of older individuals is subsidised through unemployment insurance beneﬁts or even out of the pension budget. These various beneﬁts increase the disincentives to work prior to the earliest age at which full old-age pensions become available.

48

If looked at it the other way round, one might interpret the higher taxes falling on young individuals as an indirect way of subsidising higher education since they reduce the opportunity cost of extended periods of training. Nonetheless, there should be better ways to produce a similar effect using more targeted instruments outside the public pension systems.

112

Unfunded Pension Systems: Ageing and Migration

If we take these problems seriously, they should be solved by redistributing the burden of implicit taxes over the individual life cycle as to minimise these distortions. Lindbeck and Persson (2003) conclude that in an optimal taxation framework, a decreasing path of implicit taxes over the life cycle might require “the contribution rate… [to] increase with age in a quasi-actuarial system, in order to bring about tax smoothing” (p. 85). The analysis of this section, however, has shown that this argument ignores the importance of the relation between implicit taxes and labour-supply elasticity. The “inverse elasticity rule” demands that to minimise distortions, taxes should be highest where the labour supply is least elastic and vice versa. Given the “U-shaped” path of laboursupply elasticity, smoothing the path of the implicit taxes cannot be optimal. In contrast, we have seen that the implicit tax rates of unfunded pensions should be lower for young and very old individuals—notably for men—than for middleaged ones. In theory, this can be done by differentiating either of the two instruments across age groups that are mostly relevant for the level of tt : annual contribution rates or annual rates of return. The annual contribution rates needed to acquire one pension unit could be differentiated by age, for example, by increasing the contribution rate for those age groups with a rather inelastic labour supply and decreasing it for the young and older individuals who react more strongly to changes in net wages. This would correspond to a higher implicit tax rate falling on middle-aged individuals compared to young and older ones. Alternatively, annual rates of return could be set differently for different age groups. Keeping the contribution rate constant across age groups, individuals would then acquire more pension units by paying contributions when young or close to retirement than when middle-aged. The 30 to 44 year-old with a rather inelastic labour supply would only get a low rate of return on their contributions or, to put it differently, would have to pay a high implicit tax rate. In both cases, contribution rates and rates of return would have to be set such that the annual budget constraint of the pension system and the total implicit debt carried by the individual would not change: income and expenditure of the unfunded pension system must be balanced in each period and the present value of individual contributions and beneﬁts over the full life cycle must not be altered. Still, these two constraints leave some degree of freedom for manipulating the level of annual implicit tax rates. In practice, differentiating contribution rates by age may be harder to accomplish than varying rates of return. The reason for this is that administrative costs for employers who would then have to consider the age structure of their employees to determine the contribution rates matters much more than similar effects resulting from changes in the beneﬁt formula, which could be handled by

Welfare Analysis of Pension Reforms

113

the social security administration.49 In fact, Austria has already made a step in this direction. By attributing higher pension claims to contributions made at the age of 31 –45, it has shown that such a form of differentiation across age groups is feasible. But as we have seen, pension claims should have been decreased not increased for these age groups, given their rather inelastic labour supply, in order to establish an optimal path of implicit tax rates over the life cycle. Another difﬁculty could involve predicting the proﬁles of the implicit tax rates for all the individuals affected. Figure 5.2 illustrates that any changes in the system, expected or not, can create a disturbance that hits many cohorts at a time, each being in a different stage of the life cycle. In order to install an optimal structure of implicit tax rates, this has to be avoided. In fact, the structure of optimally differentiated contribution rates or rates of return must not only hold within a given period of time. Instead, it must be designed to hold over the full period of labour-force participation plus retirement of a given age cohort. It is obvious that optimality conditions of this kind are not easily implemented in realworld pension systems. All in all, our results do not make a strong case against the efﬁciency of existing pay-as-you-go pension systems from an intrapersonal point of view. Yet, they highlight some details where improvements are necessary, in particular with respect to the excessive implicit taxation of married women and to the suboptimally high taxation of very young and very old individuals, notably men. While women should be subjected to an individual treatment, and not just to rules applying on a household level, all individuals should be taxed conditional on their age, for instance by applying lower annual contribution rates or higher rates of return to those at both ends of the age distribution. Thus, our results contradict a standard proposal that suggests smoothing the proﬁle of implicit tax rates across the individual life cycle and advocates a path of implicit tax rates that takes labour-supply elasticities into account. APPENDIX A5. STANDARDISED WORK BIOGRAPHIES We construct an individual with a stylised biography and working career (see Table A5.1), which we do not alter over time, i.e. across generations, as a representative agent in each age cohort. 49

It is also apparent that adapting public pension systems to optimal taxation rules may involve a huge loss in ﬂexibility regarding short-term adjustments, which is often cited as being one of the main advantages of unfunded pension systems when compared to the rigidities involved in funded systems. On the other hand, inasmuch as short-term ﬂexibility effectively means susceptibility for (myopic) political manipulations, this need not be a serious drawback.

114

Unfunded Pension Systems: Ageing and Migration

Table A5.1: Basic assumptions of the representative agent. Age 20–52 Age 53–64 Age 65–74 Age 75(–86)

Full-time employment with average earnings Contributions paid on full-time earnings Reduced probability of full-time employment 83.4%: contributions paid on full-time earnings 16.6%: (full) disability beneﬁts received Period of retirement Old-age pension beneﬁts based on prior earnings Death at age 75 Survivor beneﬁts payable to the surviving spouse

33 years 12 years 10 years 11 years

We consider a male (blue- or white-collar) worker who enters his active period of life at the age of 20 and earns the average wage throughout his career. He is fully active until the age of 52 when he is regarded as being disabled with some positive probability. With what is left of his working capacity, he continues to work until the age of 64. Upon retirement, he is entitled to pension beneﬁts accruing to him and, where appropriate, to his spouse. When he dies, his widow will receive a widows’ pension for some more years if this is contained in the respective pension system.50 As a result, the three main types of pension beneﬁts—disability pensions, old-age pensions, and survivor beneﬁts—are included in our model.

APPENDIX B5. DERIVATION OF EQUATION 5.14 Formulating the problem of optimal taxation stated in Equation 5.9 in terms of a conventional Lagrangean equation leads to the following set of ﬁrst-order conditions: ›V ›l ›l 2 m tt Wt t þ ttþ1 ð1 þ rÞ21 Wtþ1 tþ1 þ ð1 þ rÞ12k Wk lk ¼ 0; ðB5:1Þ ›t k ›t k ›t k k ¼ t; t þ 1; and m . 0 being the Lagrange multiplier associated with the revenue constraint of Equation 5.13. Using Equation 5.10 and rearranging terms, we can rewrite these conditions as

tt W t

›lt ›l lþm þ ttþ1 ð1 þ rÞ21 Wtþ1 tþ1 ¼ 2 ð1 þ rÞt2k Wk lk m ›t k ›tk ¼ 2nð1 þ rÞ12k Wk lk :

ðB5:2Þ

50 Mortality assumptions are based on conditional life expectancy for men and women at relevant ages. For details, see Thum and von Weizsa¨cker (2000) who did earlier calculations for the case of Germany.

Welfare Analysis of Pension Reforms

115

With Equation B5.2 we can now derive the optimal structure of tk ; i.e. the optimal timing of implicit taxes. First of all, differentiating the individual budget constraint (Equation 5.2) with respect to tk yields

›c t ›l ›l 2 ð1 2 tt ÞWt t 2 ð1 2 ttþ1 Þð1 þ rÞ21 Wtþ1 tþ1 þ ð1 þ rÞt2k Wk lk ¼ 0: ›tk ›tk ›t k ðB5:3Þ Solving Equation B5.3 for ð1 þ rÞt2k Wk lk and substituting the result into Equation B5.2, we obtain ðtt ð1 2 nÞ þ nÞWt

›l t ›l ›c þ ðttþ1 ð1 2 nÞ þ nÞð1 þ rÞ21 Wtþ1 tþ1 ¼ 2n t : ›tk ›t k ›t k ðB5:4Þ

We allow for ›ct =›tk being of arbitrary form. If, in fact, we impose no further restrictions at all on the basic optimal taxation problem, we can rewrite Equation B5.2 as tþ1 X

ts ð1 þ rÞt2s Ws

s¼t

›l s ¼ 2nð1 þ rÞt2k Wk lk ; ›t k

ðB5:20 Þ

(see, for example, Atkinson and Stiglitz, 1980). Note that the term ›ls =›tk ; capturing both “own-price” and “cross-price” effects, is equal to ›l s ›l s ›l k t2k ð1 þ rÞ Wk ¼ lk 2 Ssk ð1 þ rÞt2k Wk ¼2 ›t k ›I ›ðð1 2 tk ÞR12k Wk Þ by the Slutsky expansion, where

›ls Ssk ¼ ¼ Sks t2k ›ðð1 2 tk Þð1 þ rÞ Wk Þ u

are the (symmetrical) substitution terms, evaluated at a compensated level of utility. Applying these intermediate results to Equation B5.20 and rearranging yields ! tþ1 tþ1 X X ›ls t2s t2s l: ts ð1 þ rÞ Ws Sks ¼ n þ ts ð1 þ rÞ Ws ðB5:5Þ ›I k s¼t s¼t The term in brackets on the right-hand side is independent of k and can, therefore, be replaced by j: Deﬁning the compensated price elasticities

sks ¼

ð1 2 ts Þð1 þ rÞt2s Ws Sks ; lk

116

Unfunded Pension Systems: Ageing and Migration

and switching the notation to tax rates imposed on wages net-of-taxes, t=ð1 2 tÞ; leads to tþ1 X s¼t

or, written in matrix form, St ¼

st;t

st;tþ1

stþ1;t

stþ1;tþ1

ts s ¼j 1 2 ts ks

!

tt =ð1 2 tt Þ

ðB5:6Þ

!

ttþ1 =ð1 2 ttþ1 Þ

¼

j j

! ¼ j:

ðB5:60 Þ

Using Cramer’s rule we can then solve for the optimal tax rates, which are given by

tt j ðs ¼ 2 st;tþ1 Þ 1 2 tt det S tþ1;tþ1 ttþ1 j ðs 2 stþ1;t Þ; ¼ 1 2 ttþ1 det S t;t

and ðB5:7Þ

respectively (taking for granted that det S . 0). The optimal timing of tax rates is thus stþ1;tþ1 2 st;tþ1 ttþ1 tt ¼ ; ðB5:8Þ 1 2 tt st;t 2 stþ1;t 1 2 ttþ1 which is identical to Equation 5.14.

APPENDIX C5. DESCRIPTION OF THE GERMAN SOCIO-ECONOMIC PANEL (GSOEP) The GSOEP is a longitudinal survey, organised in four sub-samples, covering a total of about 8000 households and 15,000 individuals. Meanwhile, a maximum of 17 waves (1984 – 2000) is available for evaluation. The four sub-samples of the GSOEP are –

–

Sample A (“West Germans”) includes private households living in West Germany with a head of the household of German nationality or of a foreign nationality that is not part of sample B; Sample B (“Foreigners”) includes private households living in West Germany headed by a person of Italian, Greek, Yugoslavian, Spanish or Turkish nationality (i.e. the main foreign nationalities present in Germany);

Welfare Analysis of Pension Reforms – –

117

Sample C (“East Germans”) includes private households in East Germany with a German head of the household; Sample D (“Immigrants”) includes a small number of private households in which at least one person has immigrated into West Germany since 1984.

In 1998 an additional sample E has been introduced in order to overcome panel mortality and to expand the data base. It includes about 800 private households in West Germany and 200 households in East Germany. Three survey instruments are essential for the GSOEP: –

–

–

One questionnaire, which has to be ﬁlled in on an annual basis, addresses a number of questions that are relevant at the household level (housing, wealth, transfers, etc. including information on children in the household, who cannot answer the questionnaire by themselves). A second questionnaire, which has to be answered annually by each person in the household who is at least 16 years old at the beginning of the survey year, covers many individual socio-economic features including labour-force participation and earnings. In addition, the full biography and employment record of each individual aged 16 and over has to be reported once a person enters the survey. Afterwards, this type of information is updated in each survey year.

In some of the panel waves, additional questions are included in the survey, focusing on speciﬁc issues. In our analysis, however, we rely on the information that is collected on a regular basis.

APPENDIX D5. ESTIMATES FOR THE WAGE EQUATION: MEN AND MARRIED WOMEN Information on gross wages is only available for individuals who are actually employed. However, restricting attention to these individuals in order to estimate the wage elasticity with respect to hours of work supplied involves a potential sample selection bias and may, therefore, lead to distorted results. Building on the Heckman (1979) procedure, we, therefore, impute gross wages in two stages. At the ﬁrst stage (selection equation), participation probabilities for all individuals (both employed and non-employed) are estimated. This gives us a probability for the observation of a wage, which can be used to correct for the sample selection bias when estimating a wage function over all individuals in employment (Table D5.1). At the second stage (wage equation), the wage function

118

Unfunded Pension Systems: Ageing and Migration

Table D5.1: First stage– selection equation (Heckman). Variable

Men Coefﬁcient

Married women Std. err.

Living with a partner Children aged 0–3 Children aged 4–6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Interest and rent income l

0.132*** 0.288*** 0.052 0.111*** 0.047 0.400*** 0.243*** 0.190*** 2 0.010*** 0.000*** 2 0.041** 0.024 2.742

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

15,089 4298 0.000 17.05%

0.048 0.111 0.128 0.026 0.064 0.067 0.051 0.011 0.001 0.000 0.018 0.022 0.046

Coefﬁcient 2 0.202 2 0.464*** 2 0.109 2 0.062*** 2 0.021 2 0.095 0.078* 0.045*** 2 0.001 0.000 0.038** 0.092*** 15.770

Std. err. 0.184 0.086 0.073 0.016 0.066 0.090 0.045 0.013 0.001 0.000 0.018 0.026 1.456

8416 4569 0.000 3.91%

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level. ‘Interest and rent income’ is divided by 10,000.

Table D5.2: Second stage –wage equation (Heckman). Variable

Men Coefﬁcient

Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 l (sample selection) No. of observations Censored observations Prob ðx . 0Þ

0.978 23.550*** 3.214*** 1.662*** 2 0.054*** 0.001*** 2 0.790 15,089 4289 0.000

Married women Std. err.

Coefﬁcient

Std. err.

1.462 0.824 0.672 0.153 0.009 0.000 0.454

3.531*** 12.188*** 4.492*** 0.893*** 2 0.005 2 0.000 15.770

1.046 1.838 0.806 0.240 0.015 0.000 1.456

8416 4569 0.000

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level.

Welfare Analysis of Pension Reforms

119

is estimated and used to impute wage rates for the non-employed based on the fact that they share characteristics with the employed (Table D5.2). The selection bias is captured with the selection variable l and enters the wage equation. For reasons of identiﬁcation, we exclude variables from the wage equation which can be supposed to inﬂuence only the participation decision but not the wage: family situation (partner, number and age of children), other household income (interest and rent income) and the unemployment rate.

APPENDIX E5. RESULTS OF THE TOBIT MODEL: SMALLER SUB-GROUPS Tables E5.1 and E5.2 display the results of the Tobit model for the smaller sub-groups. Table E5.1: Results of the Tobit model: men (smaller sub-groups). Variable for birth cohort 1929– 1937

Men aged 55 – 59 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Dropped 33.630 2 5.671 2 12.596** 0.915 2 2.050 2 5.323*** 0.174** 2 0.001 2 4.571***

Net yearly household income E Marginal net hourly wages w

1.5 £ 1024** 0.594***

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

843 273 0.000 7.42%

Std. err. 32.599 3.645 5.906 4.458 2.101 1.691 0.074 0.001 1.244 6.8 £ 1025 0.104

(continued)

120

Unfunded Pension Systems: Ageing and Migration

Table E5.1: Continued. Variable for birth cohort 1934– 1942

Men aged 50 – 54 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2 18.947* 9.158 0.476 2 1.127 8.679*** 2 0.994 2 18.281*** 0.808*** 2 0.010*** 2 2.835***

Net yearly household income E Marginal net hourly wages w

8.2 £ 1025*** 0.175***

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

1045 137 0.000 8.42%

Variable for birth cohort 1939– 1947

11.241 16.377 0.717 3.238 1.757 1.187 1.587 0.071 0.001 0.678 2.7 £ 1025 0.053

2 16.237 2 33.479*** 0.572 2 3.305 2 1.605 7.536*** 2 2.848 0.078 2 0.000 2 0.710

Net yearly household income E Marginal net hourly wages w

6.9 £ 1025*** 0.266***

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

724 96 0.000 5.71%

Coefﬁcient 2 68.344 2 7.968 0.552 2 3.674 6.714** 1.881 2 15.209*** 0.567*** 2 0.060*** 2 2.188 1.9 £ 1024*** 0.534***

Std. err. nd 21.324 2.109 3.919 2.959 1.660 2.549 0.103 0.001 0.623 4.6 £ 1025 0.074

1247 355 0.000 6.07%

Men aged 45 –49 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Std. err.

Men aged 55 – 59

Std. err. 10.817 9.251 0.721 7.932 2.769 1.564 2.027 0.111 0.019 1.031 2.4 £ 1025 0.059

Men aged 50 – 54 Coefﬁcient 2.727 2 9.791 1.680** 2 1.473 2.388 2 2.750* 2 17.305*** 0.748*** 2 0.010*** 2 0.659 8.5 £ 1025*** 0.250***

Std. err. 6.563 6.860 0.754 3.378 2.033 1.580 2.048 0.091 0.001 0.450 2.2 £ 1025 0.053

1101 137 0.000 4.38%

(continued)

Welfare Analysis of Pension Reforms

121

Table E5.1: Continued. Variable for birth cohort 1944– 1952

Men aged 40 – 44 Coefﬁcient

Std. err.

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2 1.485 4.430 0.552 11.267*** 2 0.094 5.717*** 2 2.140** 0.174*** 2 0.003** 0.500

5.693 3.900 0.412 2.880 1.241 1.628 0.874 0.062 0.001 0.758

Net yearly household income E Marginal net hourly wages w

7.6 £ 1025*** 2 0.004

2.7 £ 1025 0.053

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2 Variable for birth cohort 1949– 1957

927 70 0.000 6.37%

Std. err.

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2 2.106 2 1.756 0.261 9.132*** 9.141*** 3.883*** 4.562*** 2 0.265*** 0.006*** 0.187

2.446 2.391 0.318 2.158 1.178 0.956 0.698 0.064 0.002 0.536

Net yearly household income E Marginal net hourly wages w

5.6 £ 1025*** 2 0.048

2.2 £ 1025 0.035

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

1107 95 0.000 8.47%

Coefﬁcient 2 1.201 2 2.843 0.630 2.548 3.724** 2 0.036 1.157 2 0.092 0.002* 2 1.073** 4.9 £ 1025** 0.326***

Std. err. 4.955 3.562 0.572 3.265 1.277 2.119 1.255 0.072 0.001 0.497 2.2 £ 1025 0.051

1038 105 0.000 2.72%

Men aged 35 – 39 Coefﬁcient

Men aged 45 – 49

Men aged 40 – 44 Coefﬁcient 2 1.371 0.173 1.160*** 5.502*** 9.638*** 0.919 0.392 0.053 2 0.001 2 0.978*** 5.3 £ 1025** 0.007

Std. err. 2.018 1.731 0.325 1.683 1.303 0.998 0.718 0.052 0.001 0.320 2.5 £ 1025 0.043

1315 90 0.000 3.33%

(continued)

122

Unfunded Pension Systems: Ageing and Migration

Table E5.1: Continued. Variable for birth cohort 1954– 1962

Men aged 30 –34 Coefﬁcient

Std. err.

Men aged 35 – 39 Coefﬁcient

Std. err.

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2 0.661 2 2.126 2 0.009 0.185 7.859*** 4.863*** 7.687*** 2 0.657*** 0.018*** 2 0.214

2.777 2.844 0.420 1.472 1.291 1.068 0.638 0.087 0.003 0.671

1.131 2 0.599 0.186 6.016* 6.978*** 1.520 3.818*** 2 0.195*** 0.004** 0.582*

1.246 1.349 0.343 1.304 1.203 0.994 0.657 0.065 0.002 0.319

Net yearly household income E Marginal net hourly wages w

2.7 £ 1025 2 0.117***

2.8 £ 1025 0.045

2 3.2 £ 1026 2 0.009

2.3 £ 1025 0.037

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2 Variable for birth cohort 1959– 1967

1376 162 0.000 6.79% Men aged 25 – 29 Coefﬁcient

Children aged 023 Children aged 426 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

1595 128 0.000 2.84%

13.328*** 7.542 0.795 5.746*** 7.675*** 4.238*** 14.967*** 2 2.201*** 0.101*** 0.045 1.8 £ 1025 0.016 1860 503 0.000 10.43%

Std. err. 3.202 4.903 0.692 1.525 1.306 0.972 0.807 0.168 0.010 0.623 3.3 £ 1025 0.054

Men aged 30 –34 Coefﬁcient 2.221* 0.713 0.366 1.822 5.070*** 1.499* 7.773*** 2 0.737*** 0.023*** 2 0.414 7.5 £ 1025*** 0.028

Std. err. 1.305 1.496 0.429 1.148 1.138 0.888 0.601 0.081 0.003 0.323 2.8 £ 1025 0.044

2058 226 0.000 3.59%

(continued)

Welfare Analysis of Pension Reforms

123

Table E5.1: Continued. Variable for birth cohort 1964– 1972

Men aged 20 – 24 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Std. err.

2 2.568 2 296.003 3.584*** 2 0.945 1.996 0.639 15.076*** 2 4.045*** 0.324*** 2 0.770 3.7 £ 1024*** 0.125*

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

1783 806 0.000 5.56%

2 1.8 £ 1025 0.179***

Std. err. 2.150 3.474 0.804 1.250 1.349 0.891 0.776 0.169 0.010 0.364 3.6 £ 1025 0.057

2263 603 0.000 7.67% Men aged 20 – 24 Coefﬁcient

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

3.523 2 11.427*** 1.611** 3.870*** 8.531*** 4.209*** 15.339*** 2 2.223*** 0.010*** 0.074

6.9 £ 1025 0.069

Variable for birth cohort 1969– 1977

Net yearly household income E Marginal net hourly wages w

Coefﬁcient

28.009 nd 1.330 1.891 12.957 1.531 2.054 0.823 0.086 1.220

Net yearly household income E Marginal net hourly wages w

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Men aged 25 – 29

0.658 2 14.834 0.558 4.402** 2 11.517 0.123 17.168*** 2 5.083*** 0.454*** 2 0.734 2 1.0 £ 1024 0.491***

Std. err. 9.871 16.151 1.334 2.042 9.413 1.726 2.500 1.111 0.128 0.709 8.7 £ 1025 0.104

1626 647 0.000 2.73%

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level. Incomes are in year-1995 Euros. nd indicate no data.

124

Unfunded Pension Systems: Ageing and Migration

Table E5.2: Results of the Tobit model: married women (smaller sub-groups). Variable for birth cohort 1929– 1937

Married women 55 – 59 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Dropped Dropped 0.114 11.686 2 10.031 0.119 0.909 0.114** 2 0.003** 5.740**

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

1.5 £ 1024 0.303

358 233 0.000 13.15% Married women 50 – 54 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

6.548 13.597 17.097 2.967 0.979 0.069 0.001 2.230

2 5.8 £ 1024*** 3.039***

Net yearly household income E Marginal net hourly wages w

Variable for birth cohort 1934– 1942

Std. err.

Dropped Dropped 2 12.444*** 23.866** 0.469 0.927 0.695 2 0.012 0.000 2 1.561 2 2.7 £ 1024*** 1.772*** 735 344 0.000 7.85%

Std. err.

2.644 1.964 7.588 1.656 0.597 0.041 0.001 1.295 4.8 £ 1025 0.133

Married women 55 – 59 Coefﬁcient 15.688** 2 130.864 9.757*** 10.081* 2 131.384 2 0.124 2 0.440 0.011 0.000 1.571 2 3.1 £ 1024*** 3.040***

Std. err. 7.361 nd 3.635 5.909 nd 2.154 0.693 0.045 0.001 1.024 7.7 £ 1025 0.209

612 360 0.000 10.75%

(continued)

Welfare Analysis of Pension Reforms

125

Table E5.2: Continued. Variable for birth cohort 1939 –1947 Children aged 0 –3 Children aged 4 –6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w

Married women 45 – 49

Married women 50 – 54

Coefﬁcient

Coefﬁcient

Dropped 2 128.630 2 8.058*** 2 6.239 2 13.690*** 2 4.501*** 0.307 2 0.012 0.001 0.140 2 3.7 £ 1025 1.736***

Std. err. nd 1.424 8.939 4.091 1.742 0.693 0.054 0.001 1.240 3.4 £ 1025 0.130

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

584 261 0.000 11.19%

Variable for birth cohort 1944 –1952

Married women 40 –44 Coefﬁcient

Children aged 0 –3 Children aged 4 –6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

2 4.372 7.476 2 4.959*** 3.981 2 8.525** 3.497** 0.102 2 0.005 0.001 2 0.429

Net yearly household income E Marginal net hourly wages w

2 1.2 £ 1024*** 1.434***

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

842 333 0.000 9.45%

2 73.994 2 128.102 2 11.691*** 2.788 2 13.351*** 2 4.300*** 0.418 0.008 2 0.000 0.160 2 1.6 £ 1024*** 2.018***

Std. err. nd nd 2.123 5.653 4.411 1.462 0.538 0.037 0.001 0.666 3.5 £ 1025 0.120

840 395 0.000 10.47%

Std. err. 13.281 9.134 0.734 3.968 3.775 1.533 0.681 0.068 0.002 0.979 3.1 £ 1025 0.095

Married women 45 – 49 Coefﬁcient 2 27.429** 6.891 2 3.841*** 2 0.973 2 16.650*** 2 1.074 0.457 0.012 2 0.000 0.714 2 6.1 £ 1025** 1.490***

Std. err. 11.365 9.082 1.010 3.574 3.495 1.608 0.645 0.053 0.001 0.693 3.0 £ 1025 0.115

844 330 0.000 7.39%

(continued)

126

Unfunded Pension Systems: Ageing and Migration

Table E5.2: Continued. Variable for birth cohort 1949– 1957 Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w

Married women 35 –39

Married women 40 – 44

Coefﬁcient

Coefﬁcient

2 2.434 6.913 2 4.943*** 2 6.045 2 19.037*** 1.027 2 0.740 0.012 0.004 0.816 2 1.0 £ 1024*** 1.095***

Std. err. 6.632 6.151 0.783 4.704 4.325 2.058 0.883 0.108 0.004 1.229 2.3 £ 1025 0.106

2 20.030*** 2 4.820 2 4.578*** 2 8.530*** 2 3.393 3.330** 1.537*** 2 0.191*** 0.007*** 2 0.337 2 6.2 £ 1025*** 1.298***

Std. err. 6.432 3.491 0.633 2.618 3.323 1.499 0.591 0.061 0.002 0.547 2.2 £ 1025 0.083

No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

841 383 0.000 6.55%

1072 409 0.000 7.66%

Variable for birth cohort 1954– 1962

Married women 30 – 34

Married women 35 –39

Coefﬁcient

Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

2 11.022* 2 1.044 2 9.279*** 2 11.604*** 2 4.394 2 1.520 0.877 2 0.177 0.012* 1.760 2 6.7 £ 1025** 0.971*** 925 461 0.000 7.29%

Std. err.

Std. err.

5.996 6.450 0.877 3.950 3.025 1.958 1.158 0.175 0.007 1.635

2 9.935*** 2 3.803 2 5.116*** 2 4.863* 2 4.576* 0.078 0.685 2 0.141 0.008** 1.471**

2.845 2.494 0.682 2.888 2.709 1.720 0.755 0.096 0.003 0.665

3.1 £ 1025 0.096

2 9.5 £ 1025*** 0.931***

2.0 £ 1025 0.087

1240 547 0.000 5.06%

(continued)

Welfare Analysis of Pension Reforms

127

Table E5.2: Continued. Variable for birth cohort 1959– 1967

Married women 25 –29 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2 Variable for birth cohort 1964– 1972

2 14.887** 16.995*** 2 16.969*** 2 9.583** 2 7.537 2 2.716 3.092 2 0.226 0.005 2 0.537 2 9.2 £ 1025 1.599***

Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

5.930 5.535 1.273 3.741 5.832 2.694 2.239 0.436 0.025 1.621 5.8 £ 1025 0.140

799 422 0.000 11.82%

2 0.080 Dropped 2 25.420*** 2 1.986 19.105 2 0.790 2 3.639 0.957 2 0.037 2 1.850 2 2.7 £ 1024** 1.637*** 228 103 0.000 16.15%

Coefﬁcient 2 18.338*** 2 2.777 2 10.877*** 2 7.656*** 2 3.272 2 2.845 4.189*** 2 0.494*** 0.018** 2 0.281 2 9.5 £ 1025** 1.148***

Std. err. 2.271 2.360 0.829 2.598 2.943 1.748 1.185 0.175 0.007 0.615 3.8 £ 1025 0.105

1272 681 0.000 10.02%

Married women 20 – 24 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate

Std. err.

Married women 30 – 34

Std. err. 8.078 2.573 4.634 15.466 3.087 4.531 1.559 0.153 2.138 1.2 £ 1024 0.307

Married women 25 –29 Coefﬁcient 2 14.505*** 12.625*** 2 13.555*** 0.531 2 10.074* 2 3.657 3.122 2 0.242 0.004 1.958** 2 4.0 £ 1025 1.566***

Std. err. 2.744 3.297 1.132 2.984 5.468 2.234 2.046 0.413 0.024 0.848 5.0 £ 1025 0.129

906 435 0.000 11.27%

(continued)

128

Unfunded Pension Systems: Ageing and Migration

Table E5.2: Continued. Variable for birth cohort 1969– 1977

Married women 20 – 24 Coefﬁcient

Children aged 0– 3 Children aged 4– 6 No. of children Secondary degree University degree Occupational training Job experience (yrs.) Job experience (yrs.)2 Job experience (yrs.)3 Unemployment rate Net yearly household income E Marginal net hourly wages w No. of observations Censored observations Prob ðx . 0Þ Pseudo-R2

2 1.177 Dropped 2 23.343*** 2 2.925 28.995* 1.842 2 1.064 2 0.503 0.192 2 2.304 2 2.7 £ 1024* 2.010***

Std. err. 7.449 3.242 5.333 17.375 3.762 5.930 2.421 0.270 1.882 1.5 £ 1024 0.380

193 96 0.000 14.77%

Results for constant and ‘year of survey’ dummies are not reported. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level. Incomes are in year-1995 Euros. nd indicate no data.

APPENDIX F5. TIME STRUCTURE OF IMPLICIT TAXES AND WAGE ELASTICITIES Table F5.1 shows the results for men. Since the estimates of labour-supply elasticities are mostly insigniﬁcant for the middle age groups, we concentrate on the extremes. The changes in implicit tax rates imposed on men aged 20 –29 are close to and those of men aged 45 – 54 are in the range of the inverse changes in the labour-supply elasticities and thus rather optimal according to the “inverse elasticity rule”. Taxes should, however, fall even more strongly for the 55 to 59-year-old relative to the 50 to 54-year-old. Table F5.2 reports the results for married women. The tax rates are constantly declining over the life cycle, which implies a ratio slightly bigger than one. The inverse ratio of wage elasticities is sometimes smaller than one, thus indicating that for an optimal tax schedule, implicit taxes should decrease for some age groups. More precisely, implicit tax rates should be lower for married women aged 25– 29 and 30 – 34 when compared to those aged 30– 34 and 35– 39. One explanation is that these women respond more strongly to economic incentives given the outside option of bearing and raising children.

Welfare Analysis of Pension Reforms

129

Table F5.1: Ratios of implicit taxes and wage elasticities: men. net Implicit taxes tnet t =tt11 ; ½tt =ð1 2 tt Þ½ð1 2 tt11 Þ=tt11 ]

Wage elasticities

net tnet 20 – 24 =t25 – 29 ¼ 1:06

s25 * *– 29 * =s20 * – 24 ¼ 1:11

net tnet 25 – 29 =t30 – 34 ¼ 1:07

s30 – 34 =s25 – 29 ¼ 1:50

net tnet 30 – 34 =t35 – 39 ¼ 1:09

s35 – 39 =s30 * *– 34 ¼ 0:08

net tnet 35 – 39 =t40 – 44 ¼ 1:11

s40 – 44 =s35 – 39 ¼ 20:09

net tnet 40 – 44 =t45 – 49 ¼ 1:14

s45 * *– 49 * =s40 – 44 ¼ 246:25

net tnet 45 – 49 =t50 – 54 ¼ 1:18

s50 * *– 54 * =s45 * *– 49 * ¼ 0:78

net tnet 50 – 54 =t55 – 59 ¼ 1:27

s55 * *– 59 * =s50 * *– 54 * ¼ 2:08

***, ** and * indicate statistical signiﬁcance at the 1, 5 and 10% level.

Similar conclusions apply to women in two of the subsequent age groups who are beyond their fertile age. The labour-supply elasticity is slightly higher for those aged 40– 44, who might think about taking up work again following an extended period of maternity leave, relative to those aged 45– 49. We also ﬁnd a higher labour-supply elasticity for married women aged 50 – 54 compared to those aged 55 – 59. This is in contrast to the results for men, where labour-supply elasticity increases signiﬁcantly for the highest age group. Table F5.2: Ratios of implicit taxes and wage elasticities: married women. net Implicit taxestnet t =tt11 ; ½tt =ð1 2 tt Þ½ð1 2 tt11 Þ=tt11

Wage elasticities

net tnet 20 – 24 =t25 – 29 ¼ 1:04

s25 * *– 29 * =s20 * *– 24 * ¼ 1:19

net tnet 25 – 29 =t30 – 34 ¼ 1:05

s30 * *– 34 * =s25 * *– 29 * ¼ 0:73

net tnet 30 – 34 =t35 – 39 ¼ 1:06

s35 * *– 39 * =s30 * *– 34 * ¼ 0:83

net tnet 35 – 39 =t40 – 44 ¼ 1:07

s40 * *– 44 * =s35 * *– 39 * ¼ 1:06

net tnet 40 – 44 =t45 – 49 ¼ 1:09

s45 * *– 49 * =s40 * *– 44 * ¼ 0:78

net tnet 45 – 49 =t50 – 54 ¼ 1:11

s50 * *– 54 * =s45 * *– 49 * ¼ 1:08

net tnet 50 – 54 =t55 – 59 ¼ 1:16

s55 * *– 59 * =s50 * *– 54 * ¼ 0:80

***, ** and * indicate statistical signiﬁcance at the 1, 5 and 10% level.

This Page Intentionally Left Blank

CHAPTER 6

Political Feasibility of Pension Reforms The perspective of political economy when the young have the voice option Securing the long-term sustainability of pension systems requires [us…] to maintain the ﬁnancial sustainability of pension systems, so that the future impact of ageing on public ﬁnances does not […] lead to an unfair sharing of resources between the generations. Social Protection Committee (2001, p. i) Abstract The impending demographic crisis calls for fundamental reforms of the pension system. In a democracy, however, reforms require the support of the majority of the electorate. We assume that the young have only the voice option to defend their interests—thus excluding any form of exit option. A reform that aims at reducing the size of the unfunded pension system is supported by the younger individuals and opposed by the older ones. As long as the young have the majority, this reform is feasible in a democratic voting process; as soon as society becomes a gerontocracy, this reform is no longer feasible and there is risk that the unfunded pension system will be further expanded. We determine for Germany, France and Italy the latest point in time at which a majority is in favour of reforms of the pension system. For this, we calculate for each year the “indifference age” as the age of the cohort that is not affected by the reform and the “median age” as the age of the politically decisive cohort. In Germany, a reform can be democratically enforced until 2012. France becomes a gerontocracy in 2014 and Italy as early as 2006.

132

Unfunded Pension Systems: Ageing and Migration

The Social Protection Committee (2001) has formulated three requirements to guarantee the sustainability of unfunded pension reforms. Pension systems must have the capacity and the ability to meet their social aims and to respond to the changing needs of society and individuals. In addition, ﬁnancial sustainability is important and in particular the fair sharing of the pension burden among generations. In Chapter 5 we analysed the possibility of Pareto-improving reforms of unfunded pension systems based on the concept of implicit taxes. Implicit taxes are a special feature of unfunded pension systems where the ﬁrst generation has received an introductory gift which constitutes an implicit debt for all following generations. To service this implicit debt, implicit taxes are needed. Our ﬁrst result was that from an intergenerational perspective, this debt cannot be reduced in a way that makes no generation worse off—or to put it differently, transforming this implicit debt into an explicit debt serviced by explicit taxes is nothing more than a semantic change. Our second result demonstrated that given this implicit debt and the implicit taxes entailed, there are ways for a Pareto improvement from an intrapersonal perspective. The burden of implicit taxes each generation has to carry must be distributed optimally over the life cycle of the individual members. Optimality implies that the distortions of the labour-leisure decision be minimised and thus requires that the labour supply elasticities be taken into account. In this chapter, we shift our focus from aspects of efﬁciency to the question of the feasibility of reforms. For this we assume that the young cannot react in any way other than by voting. We thus exclude any form of exit option (Hirshman, 1970). The young can neither emigrate nor evade the implicit taxes by working less or working in the shadow economy. This allows us to analyse the chance for success of reforms against the background of gerontocracy alone and to determine the point in time when reforms are no longer feasible. In Chapters 7 and 9, we will contrast this result with a situation in which the young can react by emigrating. We then explicitly consider the interaction of mobility of the young and the interests of the old. In a majoritarian voting system, a reform is feasible if the majority of voters incurs no loss in terms of utility. The reform which we have discussed in detail in section 5.2 fulﬁls this requirement. All contributors to the unfunded pension system beneﬁt if the rules of the distribution of implicit taxes over the individual life cycles are changed in order to minimise distortions on an individual level, while non-contributors and retirees are not affected. This reform is thus feasible now and later, independent of the demographic development. But there is a strong point in favour of a reform as soon as possible in order to reduce the distortions for a maximum of generations.

Political Feasibility of Pension Reforms

133

These conclusions for the feasibility of intrapersonal reforms are not at all straightforward for reforms that involve redistribution of the burden of the implicit taxes between generations, thus making some generations better off and some worse off. If there are winning and losing generations, the question of feasibility needs some careful analysis. This chapter focuses on the feasibility of an intergenerational reform which does not necessarily imply a Pareto improvement. In section 6.1, we determine the pension burden of each generation, or more precisely of a representative of each generation. Reforms which aim at changing the distribution of the burden across generations must be constructed such that they are preferred by a majority in a voting process. In section 6.2, we therefore brieﬂy discuss a simple voting model. Section 6.3 is then devoted to the question of whether a stylised reform that redistributes the pension burden more equally across generations ﬁnds a majority. For this, we determine for France, Germany and Italy the development of the age of the decisive (median) voter and of the voter who is just not affected by the reform. Comparing the age of these two voters tells us whether the stylised reform still ﬁnds a majority and, if yes, for how long. Section 6.4 concludes.

6.1. INTERGENERATIONAL REDISTRIBUTION Before discussing a stylised reform which aims at redistributing the pension burden across generations, we want to determine this burden for each generation. This amounts to deriving the part of the contributions that constitutes implicit taxes. This gives us an idea about the importance of the burden for different generations and about the necessity to reform the system towards a more equal sharing of this burden. We then deﬁne criteria for and discuss justiﬁcations of a stylised reform that reduces the size of the unfunded pension system—inducing lower contribution rates and lower pension beneﬁts. 6.1.1. Implicit taxes in an intergenerational context We start by analysing the size of the implicit taxes that fall on members of different generations and especially their variation across generations. 6.1.1.1. The path of implicit taxation—the basic model In Chapters 2 and 5, we have analysed how implicit taxes can be deﬁned within a two- or three-period overlapping-generation (OLG) framework. We now want to

134

Unfunded Pension Systems: Ageing and Migration

determine the implicit taxes an individual of a certain generation has to carry over the life cycle and to compare them for different generations. By studying how implicit taxes are affected by the demographic development—particularly the evolution of the dependency ratio—we can then make projections of the development of implicit taxes for future generations. As we have seen in Chapter 5 for a three-period model, the sum of implicit taxes paid in both working periods is given for an individual of generation t by

Gt ¼ t t uW t þ

ttþ1 uWtþ1 ð1 þ rÞ

ð6:1Þ

where t is the implicit tax rate, u the contribution rate, W denotes gross wages and r the interest rate. We assume that labour supply is normalised to 1. Since implicit taxes are deﬁned as the difference between contributions and pension beneﬁts in present value terms, we can rewrite Equation 6.1 for the case where pension claims do not depend on the period in which contributions have been made as follows1

Gt ¼ u W t þ u W t

Y M 1 þ gtþ1 ð1 þ gtþ1 Þð1 þ gtþ2 Þ ðNtþ2 þ Ntþ2 Þ 2 uWt 2 O 1þr ð1 þ rÞ Ntþ2

ð6:2Þ

X with 1 þ gtþi ¼ Wtþi =Wtþi21 indicating the growth factor of wages and Ntþi denoting the size of the young (Y), middle-aged (M) and old (O) generation in period t þ i: The ﬁrst two terms indicate the contributions made to the unfunded pension system in the two working periods and the last term presents the pension beneﬁts received in the retirement period—all terms discounted to period t values. Equations 6.1 and 6.2 can easily be generalised to k þ 1 working periods and d periods in retirement so that

Gt ¼ u W t

k X i¼0

ttþi

k kX þd X 1þg i 1þg i 1þg j 1 ð6:3Þ ¼ uW t 2uWt 1þr 1þr 1þr D i¼0 j¼kþ1

where we assume gt ¼ g;t for the convenience of the presentation. The dependency ratio D is deﬁned as the ratio of those receiving pension beneﬁts to those contributing to the pension system: Xd ð1 þ nÞi 1 i¼1 ð6:4Þ D¼ Xkþ1 j ð1 þ nÞd ð1 þ nÞ j¼1 1 See Chapter 5. The qualitative results do not change for the second case considered in Chapter 5—namely that pension claims are dependent on the timing of the contributions paid.

Political Feasibility of Pension Reforms

135

for the case of a constant growth rate of the population n: If due to a shock in period t fertility falls and life expectancy increases, the dependency ratio increases. It is then straightforward to show that total implicit taxes increase: dX þk ›Gt 1 1þg j ¼ 2 uW t .0 ð6:5Þ 1þr ›D D j¼kþ1 The implicit tax burden, which members of generation t have to carry, rises. 6.1.1.2. Empirical illustration—the structure of implicit taxes for the case of Germany We have seen that given the demographic development of recent years and the projections for the coming decades, the dependency ratio for Germany has constantly increased and will continue to do so in the coming years at an even faster rate.2 This is reﬂected by the development of the implicit tax part of the contributions: as future generations will be smaller, the per capita burden of implicit taxes will be higher than expected some decades ago. Calculations with the pension model developed by CESifo for the Council of Advisors to the German Ministry of Economics (Wissenschaftlicher Beirat, 1998)3 based on the eighth coordinated population projections of the German Federal Statistical Ofﬁce have found that the implicit tax rates have risen steadily since the introduction of the unfunded pension system in Germany in 1957. Those who were born in 1937 and entered the pension system in 1957 at the age of 20 incurred implicit taxes of about 35%, while those who were born in 1980 and start their working career today face implicit taxes of 50%. The tax share of contributions is expected to rise even further in the next few decades (Sinn, 2000a). These results are conﬁrmed by Schnabel (1998), based on slightly different demographic projections. He ﬁnds an implicit tax of 38% for the birth cohort of 1940, which increases to 66% for the birth cohort of 1980. Beckmann (2000) calculates the implicit tax for several combinations of the relevant growth rates. For plausible parameter values, the implicit taxes are in the range of 55 –65% of total contributions. In the United States, the implicit tax which young workers have to carry as members of the pension system (OADSI) amounts to 50% and is thus not very different from the tax burden implicit in the German unfunded pension system (Murphy and Welch, 1998). 2

See Chapter 3. For details see Sinn and Thum (1999), Sinn (1999b), Thum and von Weizsa¨cker (2000), and Wissenschaftlicher Beirat (1998). 3

136

Unfunded Pension Systems: Ageing and Migration

The increase in the implicit tax rate for future generations endangers the ﬁnancial sustainability of the unfunded pension system and erodes its political support amongst those who have to carry the burden. In section 6.1.2, we discuss how a reform should be constructed, i.e. on which criteria a reform aiming at redistributing the burden more equally across generations should be based. The issue of political economy—especially concerning the feasibility of the reform—will be dealt with in detail in sections 6.2 and 6.3. Section 6.4 concludes. 6.1.2. Optimal level of redistribution We have already mentioned that a (partial) transition from an unfunded to a funded system, which involves shifting the burden of implicit taxes across generations, does not present a Pareto-improving policy. Such a reform, however, might be justiﬁable on other grounds. It is, therefore, important to specify criteria to compare losses and gains across generations (Breyer, 2000). These include minimising aggregate distortions and achieving an intergenerationally fair distribution of the pension burden. 6.1.2.1. Efﬁciency In general, smoothing taxes over generations raises efﬁciency—if we assume that a welfare comparison across generations is possible—as the welfare loss increases progressively with the level of taxation (Barro, 1979). This reasoning equally applies to implicit taxes if all generations are identical with respect to their labour supply elasticity.4 Given that the total size of all future implicit taxes is determined by the gift to the introductory generation, the only aspect that can be inﬂuenced is the distribution across generations.5 The ﬂatter the path of implicit taxes over all generations, the smaller the distortions are on an aggregate level (Thum and von Weizsa¨cker, 2000). An efﬁcient intergenerational redistribution is thus deﬁned as the minimum of total distortions over all generations. Smoothing the path of implicit taxes is, however, not possible in a Pareto-improving way—at least if we exclude side payments—as the burden for early generations will increase. Kifmann and Schindler (2001) show that smoothing across generations can be realised with funded elements and a demographic factor. By considering explicitly 4 For this we have to assume that the labour supply elasticity of later generations will not signiﬁcantly increase because of better outside options due to, for example, higher mobility. 5 We abstract here from a dynastical perspective considering all generations as independent from one another.

Political Feasibility of Pension Reforms

137

the development of fertility and life expectancy, this is an attempt to distribute the burden in accordance to the responsibility each generation has for the evolution of the population and in particular of the dependency ratio. As Breyer (2001) states, choosing the particular distribution of the burden across future generations is then society’s task in reaction to the crisis. 6.1.2.2. Intergenerational fairness As we have seen in section 6.1.1, the burden of implicit taxes for individual members of different generations depends on the number of members of these generations. The size of a generation, however, is determined by the fertility decision of the parental generation. By not investing sufﬁciently in children, a parental generation causes the per capita burden of implicit taxes to increase while saving at the same time on expenditure for raising children.6 Without reforming the rules for pension beneﬁts, the pension claims of the parental generation are unaffected by this demographic change—at least in a deﬁned beneﬁt system. Oksanen (2002), therefore, proposes that generations with the same level of fertility and longevity should also face the same conditions as contributions and pension beneﬁts, i.e. the same per capita implicit taxes.7 In general, taxes should be conditioned on individual economic performance. Applied to implicit taxes, this means that they should be related to the investments of one generation in the human capital of the next generation (Thum and von Weizsa¨cker, 2000). Old-age consumption is only possible if at least one of two investment options is carried out (Sinn, 2000a): investment in real capital or human capital. If investment in human capital is not sufﬁcient to guarantee the ﬁnancial sustainability of an unfunded pension system, investment in real capital must be increased correspondingly. A reform which aims at partial funding can thus be seen as a “pragmatic solution” of the crisis of unfunded pension systems due to the lack in human capital (Sinn, 1999a, p. 1). Thus, the smoothing of the burden of implicit taxes across generations cannot only be advocated in order to minimise total distortions, but can also be justiﬁed on the grounds of intergenerational fairness. In fact, Breyer (1989) has stressed that it is distributional aspects which matter for reforms as “the problem of equity among generations inevitably arises” (p. 655). We have found two related criteria, both of which can serve to justify a reform proposal that aims at reducing the size of the unfunded pension system—possibly 6

We abstract for the moment from higher life expectancy as the second reason for an increasing dependency ratio as the parental generation cannot directly inﬂuence its average life span. 7 See also Sinn and Werding (2000).

138

Unfunded Pension Systems: Ageing and Migration

complemented by a partial transition to a funded pension system. It now remains to be seen how a reform can be constructed that fulﬁls these criteria and is also supported by the majority of the electorate in a majoritarian voting process. For this, we turn to the political economy aspect.

6.2. VOTING MODEL Starting from Aaron (1966), a country should adopt an unfunded pension system if the growth rate of the wage sum exceeds the market rate of return—at least if we assume that countries choose the system which guarantees the best way of saving for retirement. All parametric changes to the system concerning contribution rates, pension beneﬁts, etc. should also be decided with this in mind. Breyer and Craig (1997), however, argue that this view is too “naı¨ve” as it neglects the aspect of political economy,8 and that voting over the shape and size of pension systems needs to be taken into account. In fact, Browning (1975) shows how the preferred contribution rate increases with the age of the individual. The intuition for this result is that in calculating the marginal cost of a higher beneﬁt level, past contributions are sunk costs which do not inﬂuence the trade off between the marginal costs of future contributions and the marginal beneﬁts of future pension beneﬁts. The older the politically decisive individual, i.e. the median voter, the higher is the realised contribution rate as a relatively short period of paying contributions is dominated by a relatively long retirement period. We build on a model from Sjoblom (1985) to reproduce the result that the preferred contribution rate increases with age within a three-period OLG model.9 In the simplest version, this model is based on the following assumptions: decisions are taken by simple majority rule by which each individual has one vote. They are assumed to be binding forever. Individuals are selﬁsh and homogeneous. Factor prices are exogenous and labour supply is completely inelastic.10 The members of the young and the middle-aged generation work and pay contributions to the unfunded pension system and the members of the old generation receive pension beneﬁts. As we have seen in Chapter 2, individuals 11 maximise utility by choosing the optimal level of consumption in all periods cX t ; 8

See Breyer (1994a) and Galasso and Profeta (2002) for comprehensive surveys of the literature. See also Homburg (1988). 10 See Breyer (1994a). 11 X takes the value X ¼ Y, M, O for the young, middle-aged and old generation, respectively. 9

Political Feasibility of Pension Reforms

139

and the optimal contribution rate u so that M O max Ut ðcY t ; ctþ1 ; ctþ2 Þ X c ;u

ð6:6Þ

subject to the budget constraints Y cY t þ st ¼ ð1 2 uÞWt ; M Y cM tþ1 þ stþ1 ¼ ð1 2 uÞWtþ1 þ st ð1 þ rÞ;

ð6:7Þ

M cO tþ2 ¼ ptþ2 þ ð1 þ rÞstþ1

This implies that life-time consumption must equal labour income, i.e. gross wages W net of contribution rate u in the two working periods, plus pension beneﬁts p in the retirement period. Savings s cancel out in present-value terms— with r as the interest rate—but are necessary for consumption smoothing. Pension beneﬁts can be written as

u ð6:8Þ D where the dependency ratio D is given by Equation 6.4 for k ¼ 1 and d ¼ 1. Differentiating Equation 6.6 with respect to cX t and u subject to Equation 6.7 yields with Equation 6.8 ptþ2 ¼ Wtþ2

UcXi þ li ¼ 0;

for i ¼ t; t þ 1; t þ 2

ð6:9Þ

and

lt Wt þ ltþ1 ltþ1

Wtþ1 Wtþ2 2 ltþ2 ¼ 0; for a young individual in t ð1 þ rÞ ð1 þ rÞ2 D Wtþ1 Wtþ2 2 ltþ2 ¼ 0; for a middle-aged individual in t þ 1 ð1 þ rÞ ð1 þ rÞ2 D 2ltþ2

Wtþ2 ¼ 0; for an old individual in t þ 2 ð1 þ rÞ2 D ð6:10Þ

with lt ; ltþ1 ; ltþ2 ; as the Lagrange multipliers and with r as the interest rate. Periods before the vote are not affected by the chosen contribution rate. The number of terms for the young, middle-old and old individual in Equation 6.10 thus equals the number of periods remaining in the individual’s life as individuals only take into account those costs and beneﬁts which are due in this and the next period(s). Using Equation 6.9, we can rewrite Equation 6.10 as

140

Unfunded Pension Systems: Ageing and Migration

follows: UcYt Wt þ UcMtþ1

Wtþ1 Wtþ2 2 UcOtþ2 ¼ 0; ð1 þ rÞ ð1 þ rÞ2 D

UcMtþ1

Wtþ1 Wtþ2 2 UcOtþ2 ¼ 0; ð1 þ rÞ ð1 þ rÞ2 D 2UcOtþ2

ð6:11Þ

Wtþ2 ¼0 ð1 þ rÞ2 D

If we assume utility to increase with consumption at a decreasing rate and normal goods, old-age consumption must increase for older individuals to satisfy the ﬁrstorder conditions in Equation 6.11. This implies that utility-maximising pension beneﬁts and—according to Equation 6.8—contribution rates must rise with age. Older individuals consequently vote for higher contribution rates than younger individuals, who can be expected to vote for no contribution rate at all if the internal rate of return falls short of the market rate of return. This holds equally within a model with more than three generations.12 The realised contribution rate is the contribution rate preferred by the median voter. To put it differently, there is no majority for another contribution rate—in particular there is no majority for a reform of the pension system which would result in a change to the contribution rate. The size of the contribution rate depends on the age cohort the median voter belongs to. If the median voter gets older due to the ageing of the population as discussed in Chapter 3, the chosen contribution rate can be expected to rise as well.13 In section 6.3, we will rely on this result when analysing a reform of the unfunded pension system that is supposed to smooth the burden of implicit taxes of the pension system across generations in a way that increases overall efﬁciency and intergenerational fairness. We focus on the question of whether this reform is supported by the median voter and, if yes, for how long, given the ageing of the population and of the median voter. 12

The result that the age of the median voter has a positive impact on pension beneﬁts as a fraction of GNP has been conﬁrmed by Breyer and Craig (1997) for OECD data—even though the effect on beneﬁts per pensioner is not signiﬁcant. 13 In fact, there are two effects which have to be taken into account. On the one hand, an older median voter has to pay contributions for fewer years. This makes a more generous pension system more attractive. On the other hand, contributions increase over-proportionally given that ageing implies that fewer contributors have to ﬁnance the pension beneﬁts of more retirees. This at least partially offsets the preference for a more generous system. The preferred contribution rate only increases with the age of the median voter if the former effect dominates the latter.

Political Feasibility of Pension Reforms

141

6.3. FEASIBILITY OF PENSION REFORMS In the last decade, many reforms of the unfunded pension systems were passed in industrialised countries.14 But further reforms are needed. It is, however, not clear a priori whether further reforms will ﬁnd a majority in a democratic voting process, given the way they intend to redistribute the rising pension burden across generations. The model developed in section 6.2 can be used to show that an unfunded pension system with an internal rate of return below the market rate of return may be supported by the majority if those close to retirement or already retired outnumber the younger voters. This can even lead to the result that a rising dependency ratio decreases the internal rate of return, while raising at the same time the number of supporters of the pension system (Marquardt and Peters, 1997). These two effects of population ageing—the (potential) medianvoter effect and the rate-of-return effect—will be carefully analysed in what follows. For this, two cases have to be distinguished with respect to the age cohort the median voter belongs to: ﬁrst, the decrease in population growth is so small that the identity of the median voter does not change. Contributions have to increase or beneﬁts have to decrease as fewer contributors have to ﬁnance the pension beneﬁts of more retirees. This results in a lower rate of return and induces the median voter to prefer a smaller system. Second, the decrease in the population growth is sufﬁciently large to change the identity of the median voter. Then there are in fact two effects which have to be taken into account. On the one hand, there is again the negative impact on the rate of return. On the other hand, an older median voter has to pay contributions for fewer years. This makes a more generous pension system more attractive. This at least partially offsets the preference for a smaller system due to the rate-of-return effect. The system only grows or—in our framework—the preferred contribution rate only increases if the latter effect dominates the former. This reasoning makes it clear that the number of overlapping generations matters. If we start from a two-period OLG model with ageing but positive population growth, the median voter remains a member of the young generation. There is only the rate-of-return effect. The median voter thus prefers a lower contribution rate. Based on this reasoning, Casamatta et al. (2001) warn that population ageing may leave retirees of the transition generation worse off and lead to the elimination of the pension system. In the following, we assume single-year periods for every year of life. This enables us to allow explicitly for both effects and to determine which effect dominates for the three countries considered. 14

See Chapter 4.

142

Unfunded Pension Systems: Ageing and Migration

We analyse the feasibility of an idealised reform similar to the Riester-Reform (II) as explained in Chapter 4. We assume that the whole time path of the contribution rate without a reform is shifted in a parallel way. Starting with the year 2002, each year’s contribution rate is set one percentage point below the contribution rate which would have resulted without a reform. Pension beneﬁts fall correspondingly. The missing pension claims can then be offset by private savings in a way that corresponds to a partial transition to a funded system.15 This results in a smoothing of the pension burden across generations. We want to see whether this reform ﬁnds a majority although it favours generations which are not yet born and therefore cannot vote and burdens some of the present voters. And there are no subsidies included to make the partial transition to a funded system more attractive.16 In more detail, we abstract from risk issues and price changes induced by both demographic changes and policy reforms. We also assume that there is no heterogeneity within an age cohort; voters only differ with respect to age. This simpliﬁcation can be justiﬁed for France, Germany and Italy as the pension systems in these three countries are in the tradition of Bismarck with rather stable replacement ratios across wage levels and thus no signiﬁcant intragenerational redistribution. This enables us to concentrate on the issue of intergenerational redistribution and allows us to model the choice of the contribution rate as onedimensional voting. The focus of the analysis is thus on the distributional effects for different age cohorts; efﬁciency aspects of a partial transition to a funded system are not considered as they have already been discussed in Chapter 5.17 This reform leads to changes in the behaviour of the retired and the contributors. One can expect an increasing labour supply and reduced incentives for early retirement. However, these behavioural changes will not affect the voting behaviour because they are second-order effects which—within the framework of our marginal reform—only lead to negligible utility changes. The calculations, therefore, do not need an economic optimisation model but can be made—without any loss of generality—by assuming a given behaviour. 15

The present value of the cash ﬂow generated by private savings equals zero if we assume equality of the market interest rate and the discount rate which equals the rate of time preference. Thus, the concrete form of the funded system does not have any impact on the calculations of the distributional effects. 16 In this respect, the stylised reform is different from the Riester-Reform (II). 17 See Besendorfer et al. (1998), who also neglect efﬁciency considerations and focus on intergenerational redistribution effects of various reform proposals. They, however, do not particularly analyse the feasibility of the proposals.

Political Feasibility of Pension Reforms

143

This model is thus in the tradition of Browning (1975), who was the ﬁrst to analyse the choice of contribution rates based on present value calculations and without considering behavioural reactions.18 Several authors have dealt with this question within the framework of more and more extensive simulation models in order to take into account the numerous interactions and interdependencies within and around the social security system. A special focus is on changes in behaviour as a reaction to reforms but also to a lesser extent on questions of feasibility. Fehr (2000) and Hirte (1999, 2000), amongst others analyse the economic effects of different reform proposals for Germany within a dynamical computational general equilibrium (CGE) model. Fehr quantiﬁes the distributional and efﬁciency effects of some recently discussed reforms in Germany without, however, analysing the political feasibility of the proposals. Hirte includes the question of the political feasibility into his retrospective analysis of reforms enacted during the 1990s. He also emphasises this aspect when looking at privatising old-age insurance, where he does not ﬁnd a majority for a partial or full transition to a funded system against the alternative of the pension system of 1992. This result changes only when introducing a pre-announcement period between 15 and 25 years, thus altering the timing of the reform and shifting the burden of the transition to later generations. Galasso and Profeta (2004) focus on demographic, economic and political aspects of the reform debate in six countries, determining the social security system for different scenarios that would arise as an equilibrium outcome of an election. By adopting this equilibrium concept they, however, abstract from transitional aspects. Bu¨tler (2000) looks at reform proposals for Switzerland given the demographic situation of 2005. She analyses the feasibility of different reforms for selected adjustment mechanisms as to the governmental budget constraint by identifying the losers and winners of the reforms. In this respect, our approach is similar. However, we do not stop after 1 year. Instead, we are interested in determining the path of the majorities for or against the stylised reform of the pension system in order to ﬁnd out when the majority shifts. It is important not only to know whether a reform is feasible today or in the near future, but also to have an idea of the time period within which the reform has to be enacted. To calculate the time of shifting majorities, we choose a present value approach with a particular focus on the political feasibility of the stylised reform in the next few decades. This allows us to concentrate on the demographic development and the consequences of having fewer children and more retirees for the political feasibility of reform proposals.19 18 19

See section 6.2. This section is based on Sinn and Uebelmesser (2000, 2002). See also Uebelmesser (2003a).

144

Unfunded Pension Systems: Ageing and Migration

6.3.1. Status quo For the feasibility analysis, we select three major European countries: France, Germany and Italy. In these countries, a sufﬁcient majority is covered by the public pension system so that it is possible to assume that all voters are affected by the stylised reform or, alternatively, that the same share in each age group is affected.20 France serves as an example of a country in which no fundamental reforms have taken place so far, while Germany has initiated a partial transition to a funded system and Italy has started a transition from a deﬁned beneﬁt to a deﬁned contribution system.21 The different historical evolution of the national pension systems—and in particular, the impact of the most recent reforms—leads to different, countryspeciﬁc contribution rates and replacement ratios relative to gross wages. Figure 6.1 shows the projected evolution in France, Germany and Italy for the coming decades.22 France will see a signiﬁcant rise in the contribution rate from 16 to 27% between 2002 and 2050 without any further reforms while the replacement rate will stay more or less constant at 30%. In Germany, the contribution rate will increase from 19 to 26% over the next ﬁve decades and the replacement ratio will fall from 45 to 38%. In Italy, the contribution rate remains at 33% in the same period of time while the replacement ratio will fall quite strongly from 53 to 39%. This underlines the different approaches chosen so far by the three countries in order to cope with the projected demographic crisis. If nothing changes, a doubling of the dependency ratio—which is approximately projected for France, Germany and Italy23—must lead to a doubling of contribution rates, a halving of pension beneﬁts or a combination of both. In fact, we can observe versions of all three strategies when looking at the three countries. While in Germany, both the young and the old have to carry part of the rising burden—the former via increasing contribution rates and the later via decreasing pension beneﬁts, 20 This is a problem concerning, for example, the United Kingdom. Only about a third of all employees are covered by the public earnings-related pension system (SERPS). It is, however, not possible to assume a constant share of individuals covered in each age group as the contracting-out principle has been effective only since 1988 with a wider opening towards company and private pension plans during the last few years. 21 For a short description of the most recent pension reforms in France, Germany and Italy see Chapter 4. 22 These projections concern the contribution rates and replacement ratios of the “Gesetzliche Rentenversicherung” in Germany, the “Re´gime Ge´ne´ral” in France and of the “Fondo Pensioni Lavoratori Dipendenti” in Italy. 23 According to the Eurostat (2000) baseline scenario, the dependency ratio will increase between 2000 and 2050 from 0.39 to 0.73 in France, from 0.45 to 0.79 in Germany, and from 0.45 to 0.93 in Italy. See also Chapter 3.

Political Feasibility of Pension Reforms

145

Figure 6.1: Facts about the national pension systems in France, Germany and Italy. Source: CESifo pension model: calculations of the Ifo Institute for Economic Research (see Fenge et al., 2002b).

146

Unfunded Pension Systems: Ageing and Migration

the situation is different for young and old in France and in Italy. In France, the ageing of the population leads to a higher burden for the contributors while the replacement ratio will stay approximately constant and in Italy, the demographic development results in a lower replacement rate while the burden of the contributors will not change. The evolution of the national pension systems can also be illustrated by the path of the implicit tax rates as a percentage of lifetime earnings (Figure 6.2). In Italy, the implicit tax rate has risen very sharply for cohorts born between 1940 and 1975 to reach about 17% of lifetime earnings and stays at this level for later born cohorts. In contrast to the evolution of implicit tax rates in Italy, the implicit tax rates in France and Germany have risen constantly since 1940 and will continue to rise caching up with the Italian implicit tax rates. But these relative differences in the evolution of the implicit tax rates must not obscure the fact that the implicit tax rates might already have reached levels which are no longer supportable. Further reforms which smooth the burden more equally across generations are thus necessary in all countries to avoid this evolution. We, therefore, analyse the feasibility of the stylised reform in which the contribution rates in all countries are shifted downwards by one percentage point and the pension beneﬁts are adapted accordingly.

Figure 6.2: Implicit taxes in Germany, France and Italy. Source: CESifo pension model: calculations of the Ifo Institute for Economic Research (see Fenge et al., 2002b).

Political Feasibility of Pension Reforms

147

Smoothing the burden over the generations will be supported or rejected depending on personal interests. The reform is only feasible in a democratic country when the majority of the voters will beneﬁt from the changes. Whether one belongs to the losers or winners of the reform depends on age. In general, older individuals will be among the losers and younger individuals among the winners. As long as the young are in the majority, reforms which lead to a partially funded system can be implemented. As soon as the elderly are in the majority, such reforms are no longer feasible. The question is whether a majority still favours a partial transition to a funded system and, if yes, for how long. On the basis of a simple, theoretical approach, we use the CESifo pension model based on the demographic projections of Eurostat to answer these questions. This model has already served as the basis for the calculations of the Council of Advisors to the German Ministry of Economics (Wissenschaftlicher Beirat, 1998)24 and has been adapted by the Ifo Institute for Economic Research to pension models of other countries.25 The model allows us to calculate the gains and losses for alternative age cohorts and to project the political feasibility of reforms and the point of time at which the societies in France, Germany and Italy become gerontocracies. The projection describes the time frame for the transition of the unfunded pension system to a partially funded system. To do so, we calculate a “median age” and an “indifference age”. The median age is deﬁned as the age that splits the voters into two equally large groups when they are arranged in ascending order as to their age. One half of the voters is older than the median age and the other half is younger. In the case of pension reforms which distribute between age cohorts, the median age plays an important role. A reform will be feasible if and only if the median voter votes in favour of it. The indifference age is deﬁned in a way that the cohort with this age is not affected by the stylised reform. Older cohorts lose and younger cohorts win. In present value terms, the indifferent cohort loses as much in pension claims as it saves in contributions. For a median age below the indifference age, the young have the majority and the stylised reform can be democratically enforced. The situation is different though if the old have the majority. We calculate the median age and the indifference age for alternative calendar years to see how the chances for a reform may change in the course of time in France, Germany and Italy.

24 25

See also Sinn and Thum (1999), Sinn (1999b), and Thum and von Weizsa¨cker (2000). See Fenge et al. (2002b) for an example.

148

Unfunded Pension Systems: Ageing and Migration

Figure 6.3: The median of the age distribution of the eligible voters: France, Germany and Italy. The calculations are based on citizens of the year 2000, their descendents and naturalised foreigners. Source: Eurostat (2000), OECD (2001b), and Bundesausla¨nderbeauftragte (2002), own calculations.

6.3.2. Median age For the median age, we use the baseline scenario of Eurostat (2000). The projection contains information about the distribution of the population over different age cohorts for every calendar year. We assume that the minimum voting age remains at 18 and that identical shares of voters of all age cohorts participate in the election and decide non-altruistically. With the help of this information it is possible to calculate the median age which splits the distribution of voters into two equally large parts.26 As we have already seen in section 6.3.1, the three countries are differently affected by the demographic development. This is also reﬂected by the evolution of the median age (Figure 6.3). According to our calculations, in 2000 the median age of voters was 46 in France and Italy and 47 in Germany, which implies that there were equally many voters younger and older than 46 or 47, respectively. Of course, the median age is changing rapidly because of ageing, i.e. decreasing fertility rates and increasing life expectancies. This demographic development will shift the median age in all three countries to higher and higher levels in the course of the next few decades. The development will be similar in Germany and France. In Germany, the median age will be 50 in 2010 and will go up to 55 in 2025. In France, the median age will reach 50 in 2015 and increase to 53 in 2029. The Italian median voter will 26 See Appendix A6 for details about the assumptions made with respect to immigration and naturalisation.

Political Feasibility of Pension Reforms

149

be 50 in 2010 and 55 in 2023. The Italian population, however, will age much more rapidly from 2025 on. In 2030, half of the voters will be 58 and older. In all the three countries, the median age will thus have increased by about 4 years in about one decade; and in three decades, the decisive age cohort will be between 7 and 12 years older than today. This will strengthen the position of those who are in favour of an extension of the present unfunded pension system and weaken the position of those who have to carry the burden.

6.3.3. Indifference age To calculate the indifference age, we again use the CESifo pension model based on the baseline scenario of Eurostat (2000). In addition, we assume a productivity growth rate of 1.75% per capita and a strict equivalence between contributions and pension beneﬁts. For the stylised reform, the contribution rates in this and in all succeeding calendar years are one percentage point below the contribution rates as shown in Figure 6.1. Pension beneﬁts—including already acquired pension claims—are cut accordingly. The retired and almost retired are against the reform. But the younger voters are in favour of the proposed changes because these cuts reduce the implicit taxes to be paid by them and allow them to save and build up their own pensions. They gain from the reform because the necessary savings to compensate for the reduced pension beneﬁts are lower than the contributions in the present unfunded pension system.27 We calculate the effects of a cut in contributions for alternative age cohorts and calendar years. Of special interest for us is the effect of the cut on a generation’s cash ﬂow in present value terms which accrues from the chosen calendar year until the expected year of death.28 For a positive present value we assume a vote in favour of the reform and for a negative present value we expect a vote against the reform. Until retirement age, this reform eases the ﬁnancial burden of the contributors. After retirement, however, the retirees face lower pension beneﬁts. For a young cohort, the positive effect on the contributions dominates. For a cohort close to the retirement age, the negative effect on pension beneﬁts is more important. The cohort for whom the present value of the changes in the cash ﬂow is closest to zero 27

The German government seems to have realised this. The Riester-Reforms (I) and (II) have aimed at easing the burden for the younger members of the working generation by reducing the value of the already acquired claims. See Chapter 4 for details. 28 See Appendix A5 for some details about the assumed biography of the representative individual.

150

Unfunded Pension Systems: Ageing and Migration

Figure 6.4: The indifference age: France, Germany and Italy. Source: CESifo pension model, Eurostat (2000), own calculations.

is the indifferent cohort. The age of this cohort is the indifference age of the respective calendar year. Repeating these calculations for alternative calendar years yields the time path of the indifference age as shown in Figure 6.4.29 In 2003, the indifference age is 48 in Germany and 47 in France and Italy. Younger individuals should demand a reduction of pension beneﬁts. The lower these beneﬁts are, the easier it is for them to get rid of the implicit tax burden—at the expense of the older individuals. Conversely, individuals older than the indifference age should be in favour of extending or at least maintaining the present level of the unfunded pension system. The higher the contribution rate, the higher the pension beneﬁts are and the higher the part of the implicit tax burden which is shifted to the younger contributors. In Germany, the indifference age will rise to 50 in 2011 and to 51 in 2021. The increase is much slower in France and in Italy. In France, the indifference age will reach 49 in 2014 and remain at this level and in Italy it will stay at 48 from 2005. The number of retirees keeps increasing relative to the number of contributors, which deteriorates the situation of the contributors and shifts the borderline of contributors who prefer to abolish the unfunded pension system upward. The increase in the indifference age slows down at around 2012 in Germany and even earlier in France and Italy. It can easily be seen that the evolution of the indifference age depends only indirectly on the demographic development. The ageing of the population in general and the rise in the dependency ratio in particular are most dramatic in 29

See Appendix B6 for an illustration of the procedure.

Political Feasibility of Pension Reforms

151

Italy, followed by Germany and France.30 If this implied a one-to-one increase in the implicit tax rates falling on lifetime income of the contributing generations, one would expect that more age groups would support the stylised pension reform in Italy than in Germany and France. But above all, the burden that the younger generations have to carry and correspondingly their attitude towards the stylised reform are affected by the rules of the existing unfunded pension system. With respect to the rules, the pension systems in France and Germany differ in one essential aspect from the Italian one as already outlined in Chapter 4: in a deﬁned contribution system as in Italy, contributions are set and beneﬁts adapt such that the budget constraint holds, while in a deﬁned beneﬁt system as in France and Germany, it is the other way round. The growing pension burden of the coming decades will be better shared between the generations, if the pension systems are characterised by more funding and deﬁned contributions. First of all, a higher degree of funding as resulting from the stylised reform reduces the amount of intergenerational transfers and thus decreases the dependencies across generations with respect to old-age security. Second, a deﬁned contribution system in which contributions grow with the wage sum and beneﬁts adapt implies that a lower growth rate of the wage sum due to ageing leads to lower beneﬁts. The risk of a negative demographic evolution is thus mainly carried by the old generation: ageing of the population no longer leads to an unlimited increase in contribution rates, but results in lower pension beneﬁts. A smaller pension system thus frees resources for funded old-age provision, beneﬁting younger generations and hurting older generations. This ﬁrst effect shows up in the indifference age for all three countries and is constant over time. The question then is how the growing burden due to ageing of the population will be distributed across generations and how this distribution is changed by the stylised reform. For this second effect, the rules of the pension systems are essential. In a notionally deﬁned contribution system as in Italy, the old already carry the risks from ageing. The possibilities to shift the burden even more to these generations are thus very limited. In a deﬁned beneﬁt system as in France and Italy, however, there is still scope for an intergenerational redistribution of the pension burden favouring the young at the expense of the old. Consequently, the number of age groups which prefer to abolish the unfunded pension system stays almost constant in Italy—the small increase at the beginning

30

See Chapter 3.

152

Unfunded Pension Systems: Ageing and Migration

Figure 6.5: Comparison: France, Germany and Italy. Source: CESifo pension model, Eurostat (2000), OECD (2001b), and Bundesausla¨nderbeauftragte (2002), own calculations.

Political Feasibility of Pension Reforms

153

Figure 6.6: Participation in the elections of the European Parliament in Germany. Source: Statistisches Bundesamt (1999b).

resulting from the long phase-in period of the new system. In France and Germany the stylised reform ﬁnds supporters among an increasing number of age groups. 6.3.4. Results A comparison of Figures 6.3 and 6.4 as displayed in Figure 6.5 shows for how long the stylised reform will ﬁnd a majority in a democratic voting process in France, Germany and Italy. In Germany, the indifference age exceeds the median age up to the middle of the next decade. Thus, the young outweigh the old, and there is a majority for the stylised reform towards a partially funded pension system. The majority, however, is slight. In France, the indifference age exceeds the median age until 2007 and stays close to it until 2014. In Italy, the indifference age and the median age coincide until 2006. The majorities for the stylised reform thus will vanish in the next two decades in Germany and France and even earlier in Italy. In fact, the countries will have reached a situation which can be called a gerontocracy, at which point they will no longer be able to initiate further reforms towards more funding. The transition to gerontocracy must be expected even earlier if one takes into account the fact that participation in elections is clearly biased towards older age groups. Figure 6.6 shows for Germany the share of each age group voting in the last two elections of the European Parliament. Among the 21 –24 year old, for example, 50% participated in 1994 and 33% in 1999, while the share of the 50– 59 year olds was 65% in 1994 and 50% in 1999. The decreasing political inﬂuence of

154

Unfunded Pension Systems: Ageing and Migration

the younger age groups due to their falling share in the population will be even further diminished by their lower participation in the electoral process.31

6.4. CONCLUSION For Germany and France, the beginning of the next decade will be very decisive for reforms which initiate a partial transition to a funded pension system. For Italy, the crucial point in time will even be in the present decade. Afterwards, all three countries will be characterised by a gerontocratic system where the old decide over the young. There is an argument that many people do not understand what is going on and especially do not know whether they belong to the winners or losers of a pension reform. A survey conducted in France, Germany, Italy and Spain in 2000 sheds some light on this position (Boeri et al., 2001). Individuals were asked whether they would accept a proposal that implied a reduction of contribution rates by 50% and a reduction of pension beneﬁts as if worked for only 50% of the wages actually earned, starting from now on, without affecting already acquired pension claims. Forty-seven percent would accept this proposal in Germany and Italy, while 4% in Germany and 7% in Italy did not answer. If we interpret the latter individuals as being indifferent and voting for the proposal, we have a majority in favour of the proposal in both countries. In France, however, only 24% would accept the proposal while in Spain, only 19% gave a positive answer.32 As this reform is qualitatively very similar to our stylised reform where we consider a one percentage point reduction in contribution rates and pension beneﬁts—without, however, excepting already acquired pension claims—we can compare the responses with our results. For Italy and Germany, our results are conﬁrmed. We have seen that there is a small majority for the reform right now in Germany and Italy if individuals vote according to the impact of the reform on the present value of their cash ﬂow of contributions and beneﬁts. This majority will vanish in the next decade. Our results are, however, different for France. We ﬁnd 31

Of course, it is possible that this bias will be reversed when the younger individuals feel the necessity to participate more numerously in the elections in order to defend their interests against an increasingly gerontocratic power structure. This reaction will, however, at best delay but not stop the path towards gerontocracy. 32 It is not clear whether individuals have realised that pension beneﬁts which result from 50% of the wages actually earned imply a reduction of pension beneﬁts by less than 50% given minimum pensions and other redistributive instruments. If so, their approval is not surprising—and in fact one would expect even higher rates—as their burden in form of contributions decreases without affecting their beneﬁts to the same extent.

Political Feasibility of Pension Reforms

155

that more individuals in France are positively affected by the reform (than negatively) and should therefore vote in favour of the reform. But this is not reﬂected in the answers. The French seem to be less informed about the mechanisms of the pension system than Germans and Italians. According to Boeri et al. (2001) this can be attributed to two factors: the intensity of the public debate and the fragmentation of the national pension systems. A low coverage in the media and a high degree of fragmentation make it difﬁcult for private-sector employees to know about the conditions that apply for them and the prospect for the future. The French system is very fragmented consisting of different systems with different contribution rates for every sector. In contrast to this, the system is not at all fragmented in Germany and only to some degree in Italy. In addition, the higher reform activities in Germany and Italy in the last decade have led to a larger coverage of pension topics in the media compared to France. What is even more interesting in our context is the question about the characteristics of those who are in favour of the proposal. The rate of acceptance clearly decreases with age, which is what one expects on the basis of our cash ﬂow calculations. In addition, being informed about the pension system and being aware of the crisis increase the probability to accept the proposal as well as being male and rich. We have derived the results on the basis of the assumptions underlying the models of Browning (1975) and Sjoblom (1985). Let us conclude by brieﬂy commenting on some of them. Voters have regarded the voting result as being binding once and for all. But after the point in time when society has become a gerontocracy, reforms which aim at further reducing the size of the unfunded pension system no longer ﬁnd a majority and even reforms which have been passed earlier may be undone by the then-prevailing majority of the old. Constitutional safeguards can be useful to prevent this from happening. In the absence of such safeguards it can only be hoped that the earlier reforms will imply a strong implicit commitment for politicians beyond the time where the majority switches. The result, however, relies on two further assumptions. First, the young do not have the possibility to organise as an interest group. Decisions are taken by simple majority rule with one vote for each individual. Even if we think that this assumption might reﬂect reality reasonably well at the moment, it is possible to imagine that the young might participate qualitatively differently in the political process once their majority is about to vanish. The smaller a group, the easier it is for its members to organise and to exert pressure on the political decision maker despite having relatively few votes (Olson, 1965). Bigger groups, which dispose of more votes, also have more problems to speak with one voice and avoid free riding. The question is then which group—the young or the (retired) old—will better manage to form an effective interest group. There are arguments in favour of the old, given that lobbying is time-intensive, because the old have a lower labour

156

Unfunded Pension Systems: Ageing and Migration

productivity and thus lower opportunity costs to actively support their interests (Mulligan and Sala-i-Martin, 1999). Besides, the old are more homogeneous as for all of them, pension beneﬁts are the dominant feature of the pension system. The young, in contrast, are more heterogeneous as for them, contributions and beneﬁts are important but to a different extent depending on their distance from retirement age. There are, however, also arguments which run the other way. Contributions when levied proportionally on wages create distortions. The deadweight costs rise at an increasing rate with (implicit) taxes (Becker, 1983). This discourages pressure for more transfers from favoured groups because a given increase in revenues yields fewer additional beneﬁts. At the same time, this encourages pressure from contributors because a reduction of (implicit) taxes then has a smaller effect on the amount available for beneﬁts. The young thus have more incentives to defend their interests. The second assumption is that the young only have the voice option and no form of exit option; the young can only vote to express their interests, but they cannot react by reducing their domestic labour supply, i.e. by working less or working in the shadow economy, or by emigrating. That means that we have taken labour supply to be inelastic. This has enabled us to determine the latest point in time before society becomes a gerontocracy. Allowing the young an exit option will add a further component to the analysis beyond the voting mechanism and will alter the results. The mobility of the young can then serve to exert pressure on unfunded pension systems towards a signiﬁcant reduction of intergenerational activities which might counterbalance the gerontocratic power of the old or might even lead to an erosion of old-age security. In total, we must expect that the young will put pressure on unfunded pension systems in one way or another even after they have lost the majority. Nevertheless, it is useful for all voters to be aware of the path to gerontocracy in France, Germany and Italy, in order to be immune against attempts from politicians to manipulate election outcomes. Moreover, the power of the young after the transition to a gerontocracy depends, in particular, on their speciﬁc outside options and/or their ability to organise quickly and effectively. In the following chapters, we will allow the young to emigrate, thus focusing especially on the second possibility to react to a rising pension burden. The assumption of a majoritarian voting system will be maintained whenever voting decisions are analysed.

APPENDIX A6. CALCULATING THE MEDIAN AGE For the calculation of the median age, we have to make some assumptions concerning the participation of foreign-born individuals in the election.

Political Feasibility of Pension Reforms

157

(1) Stock of foreigners. The stock in 1999 was 3.3 million in France, 7.3 million in Germany and 1.3 million in Italy (OECD, 2001b, Table A.1.5). (2) Net migration ﬂow. According to the baseline scenario of Eurostat (2000), the projected net migration ﬂows are as follows

Germany France Italy

2000

2010

2020

2050

300 50 50

250 50 65

200 50 80

200 50 80

Values are given in thousands. Source: Eurostat (2000)—baseline scenario.

(3) Naturalisation. † For Germany, we assume an annual rate of naturalisation of the foreign population of 4%, which corresponds to the average rate of naturalisation in the years 1994 –1999 (Bundesausla¨nderbeauftragte, 2002). † For France, we assume an annual rate of naturalisation of 4% on the basis of the 1999 value of 4.5% (OECD, 2001b, Table A.1.6). † For Italy, we assume an annual rate of naturalisation of 1% on the basis of the 1999 value of 0.9% (OECD, 2001b, Table A.1.6). (4) Share of non-naturalised foreigners of total population. Based on the projections for the total population of the Eurostat (2000) baseline scenario, we calculate the share of non-naturalised foreigners of the total population. (5) Age distribution. We take for all three countries the age distribution of German immigrants in 2000 and assume this distribution to remain constant (Bundesausla¨nderbeauftragte, 2002). Age distributions which change over time to take a potential ageing of the immigrant population into account do not affect the path of the median age signiﬁcantly and are without any impact on the result.

APPENDIX B6. CALCULATING THE INDIFFERENCE AGE For a ﬁctitious example, we illustrate with the help of Figure B6.1 the procedure for calculation of the indifference age:

158

Unfunded Pension Systems: Ageing and Migration

Figure B6.1: Procedure for calculation of the indifference age.

(1) Horizontal direction. We calculate the present value of the change to the cash ﬂow for different birth cohorts starting at different calendar years, i.e. different ages. For younger individuals, the change to the present value is positive (þ ) whereas for older individuals, the change to the present value is negative (2 ). We note the age of each birth cohort at which the present value of the change is closest to zero. This is the indifference age of each birth cohort. (2) Vertical direction. At the same time, we have determined for every calendar year the age of the indifferent birth cohort, i.e. the indifference age. Younger individuals who belong to later birth cohorts are in favour of the reform (þ ) while older individuals are against it (2 ).

CHAPTER 7

Mobility as a Counterforce to Gerontocracy The perspective of political economy when the young have the exit option With the age distribution we are approaching, there is no reason why the majority over forty should not soon attempt to make those of a lower age toil for them. It may be only at that point that the physically stronger will rebel and deprive the old of both their political rights and their legal claims to be maintained. Friedrich A. von Hayek (1960, p. 297) Abstract If the young have the exit option in addition to the voice option, the results change compared to the case where the young can only defend their interests by voting. Even after the young have lost the majority, they are not powerless. The exit option can serve to counterbalance the power of the old. We analyse a model where the mobility of the young can be set strategically by the old. We show that investment in human capital which increases the mobility of the young can be interpreted as such a commitment device of the old to overcome the hold-up problem of human capital investment and to guarantee a certain level of redistribution. By balancing exit option and gerontocratic power, an intergenerationally redistributive system need neither explode nor erode. For Hayek (1960), the physical force of the young must be expected to mitigate the power of the old in a gerontocracy. Thus, even when the old have the absolute majority in the electorate, they cannot exploit this political power in a democratic

160

Unfunded Pension Systems: Ageing and Migration

voting process as their decisions must guarantee that the interests of the young are also respected. Otherwise, there is the threat of a rebellion. Although we abstract from seeing violence by the young as the consequence of an overburdened unfunded pension system, we nevertheless want to enlarge our discussion of Chapter 6 by taking into account outside options of the young. So far, we have assumed that the young cannot act in any way except by voting to defend their interests. In a majoritarian voting process, this gives the young the power to shape political decisions as long as they form the majority, but this also renders them completely powerless as soon as they have lost this majority. We have thus restricted their possibilities to the voice option and implied that in a gerontocracy, the young are at the mercy of the old. The analysis in this chapter focuses on modelling in detail whether our results from Chapter 6 still hold if we abolish the assumption of complete immobility— also in a wider sense—by explicitly giving the young the exit option. We analyse in particular how the possibility of the young to react can serve as a counterforce to a gerontocratic power structure. In section 7.1, we build on the simple voting model from Chapter 6 by now allowing for the possibility of the young to emigrate. Section 7.2 then presents a model where the mobility of the young can be set strategically by the old in a gerontocracy in order to commit to a certain level of taxation. By balancing the exit option and gerontocratic power, intergenerational redistribution towards the old is feasible, and the system need neither explode nor erode.

7.1. VOTING MODEL WITH MOBILITY In Chapter 6 we derived the result that the preferred contribution rate increases with age—in particular, the contribution rate chosen in a democratic voting process increases with the age of the median voter. The preferences of older and younger generations are not taken into account. Let us now assume that the younger generations are no longer homogeneous. a is the mobile part and ð1 2 aÞ is the immobile part with aðuÞ [ ½0; 1 and au ðuÞ . 0 for the contribution rate u [ ½0; 1: The mobile part is thus a positive function of the contribution rate chosen by the median voter meaning that the outside option is the more attractive, the higher the contribution rate. The dependency ratio D for the case of a threeperiod overlapping generation model with population growth rate n; which is the ratio of old individuals to young and middle-aged, becomes DMo ¼

1 : ½ð1 þ nÞð1 2 aðuÞÞ þ ð1 þ nÞð1 2 aðuÞÞ 2

ð7:1Þ

Mobility as a Counterforce to Gerontocracy

161

Differentiating Equation 7.1 with respect to u yields

›DMo 22½ð1 þ nÞð1 2 aðuÞÞð1 þ nÞau 2 ð1 þ nÞau ¼2 . 0: ›u {½ð1 þ nÞð1 2 aðuÞÞ2 þ ð1 þ nÞð1 2 aðuÞÞ}2

ð7:2Þ

In this context, mobility has to be understood in the broadest sense. More precisely, the young have an outside option with respect to the unfunded pension system by simply reducing the labour supply in reaction to the pension burden in those sectors of the domestic labour market, which are covered by the unfunded pension system. a can then be interpreted as those members of the young generation who choose not to work in these sectors. The higher the tax part of the contributions, the stronger the reaction of the labour supply of the contributors is. Breyer (1994b) and Breyer and Stolte (2001) study how this limits the contribution rate preferred by the median voter. Mobility in the form of an elastic labour supply can imply several exit options. Individuals can choose home production or self-employment, respectively, if this is a way—as it is in Germany—to opt out of the unfunded pension system. Mobility can also refer to working in the unofﬁcial sector of the economy or to emigrate (von Hagen and Walz, 1995, and Haupt and Peters, 2003), thus seizing the opportunity provided by the integrated European labour market. a then stands for those young individuals who are self-employed or who supply their labour in the shadow economy or abroad.1 This opting-out requires, however, access to other possibilities to transfer consumption across the life cycle from the working to the retirement periods. One such prerequisite is a functioning capital market for investments in real capital or sufﬁcient investment in human capital in the form of own children. a then denotes those members of the young generation who have access to the capital market or who have enough children to guarantee old-age consumption. We now repeat the calculations of Chapter 6. Individuals maximise utility by 2 choosing the optimal level of consumption in all periods cX t ; and the optimal contribution rate u so that M O max Ut ðcY ð7:3Þ t ; ctþ1 ; ctþ2 Þ X c ;u

subject to the budget constraints Y cY t þ st ¼ ð1 2 uÞWt ; M Y cM tþ1 þ stþ1 ¼ ð1 2 uÞWtþ1 þ st ð1 þ rÞ;

cO tþ2 1

¼ ptþ2 þ ð1 þ

ð7:4Þ

rÞsM tþ1 :

a can also refer to the part of total time the individuals work in self-employment, in the shadow economy or devote to leisure activities. 2 X takes the value X ¼ Y, M, O for the young, middle-aged and old generation.

162

Unfunded Pension Systems: Ageing and Migration

Lifetime consumption must equal labour income, i.e. gross wages W net of contribution rate u in the two working periods, plus pension beneﬁts ptþ2 ¼ Wtþ2 u=DMo in the retirement period. Savings s cancel out in present-value terms—with r as the interest rate—but are necessary for consumption smoothing. We only report the result for the middle-aged individual as we assume that this is the median voter. The respective ﬁrst-order conditions are UcXi þ li ¼ 0;

ltþ1

for i ¼ t þ 1; t þ 2;

Wtþ1 W DMo 2 Wtþ2 uð›DMo =›uÞ 2 ltþ2 tþ2 ¼0 ð1 þ rÞ ½ð1 þ rÞDMo 2

ð7:5Þ ð7:6Þ

with ltþ1 ; ltþ2 as the Lagrange multipliers. Using Equation 7.5, we can rewrite Equation 7.6 as follows: UcMtþ1

Wtþ1 W DMo 2 Wtþ2 uð›DMo =›uÞ ¼ UcOtþ2 tþ2 : ð1 þ rÞ ½ð1 þ rÞDMo 2

ð7:7Þ

Let us compare this result with the corresponding ﬁrst-order condition we have derived in Chapter 6 for a dependency ratio D independent from u : UcMtþ1

Wtþ1 Wtþ2 ¼ UcOtþ2 : ð1 þ rÞ ð1 þ rÞ2 D

ð7:8Þ

Again, we assume utility to increase with consumption at a decreasing rate and normal goods. The left-hand side of Equations 7.7 and 7.8 are identical. With ð›DMo =›uÞ . 0 from Equation 7.2, it can easily be seen that the fraction of the right-hand side is smaller in Equation 7.7, where we allow for mobility of the young in reaction to an increase in the contribution rate, than in Equation 7.8. Thus in this case, old-age consumption must decrease to satisfy Equation 7.7, implying that utility maximising pension beneﬁts and contribution rates must also fall. The contribution rate preferred by the median voter is smaller when the young have the exit option compared to the case where the young only have the voice option. Mobility of the young constitutes a counterforce to the power of the median voter. This result clearly shows the two diametrical problems in the context of a democratic voting process with mobility of the young and redistribution towards the old. With perfect and complete mobility, a redistributive system is not sustainable due to ﬁscal competition (Sinn, 1997) if one abstracts from altruistic motives. Without any mobility and thus without any outside option of the young generation, the redistributive system is sustainable but the preferences of the median voter may lead to partial or total exploitation of the young. An outside

Mobility as a Counterforce to Gerontocracy

163

option for members of the young generation is necessary in order to counterbalance the political power of the median voter—but a perfect outside option endangers the survival of the redistributive system. So far, we have discussed a simple model and several examples of how the exit option of the young from the unfunded pension system counterbalances the power of the old. Even after the transition to a gerontocratic power structure, the pension burden is shared between all generations and not only concentrated on the young. In all these models, the common feature is that the mobility of the young is due to certain institutional features of the international or domestic labour market. In section 7.2, we will assume that the old can determine how mobile the young are. The mobility of the young can then serve as a commitment device for the old towards a lower level of contribution rates, thus guaranteeing a certain level of intergenerational redistribution beneﬁting the old.

7.2. MOBILITY AS A COMMITMENT DEVICE In a closed economy, investment in human capital faces a hold-up problem with respect to taxation. Once the investment is made, human capital is a ﬁxed factor. Consequently, the optimal tax on human capital is high. Boadway et al. (1996) analyse the time consistency problem for a benevolent government and discuss its consequences for the incentives to invest in human capital. Increased international labour mobility, however, changes the constraints which affect optimal education and tax policy.3 As Sinn (1997) shows, the only stable equilibrium with ﬁscal competition in an economy with wage taxation and labour mobility is one without taxation and consequently without redistribution. However, there are some arguments in favour of a less pessimistic view. Mobility costs allow a tax up to these costs without inducing emigration. As these costs are not necessarily exogenous, the government might have an interest in increasing these mobility costs by offering an educational programme with a clear focus on domestically valuable skills. This would increase the scope for redistributive taxation if, and only if, the mobile generation was not able to react. In our model, however, there are possibilities for the mobile (young) generation to react to the strategies of the immobile (old) generation by choosing the amount of human capital investment. Consequently, a necessary condition for the survival of a moderate redistributive system in favour of the old generation is that there is a credible commitment that prevents excessive redistribution. We analyse 3

This section is based on Thum and Uebelmesser (2003). See also Thum and Uebelmesser (2000).

164

Unfunded Pension Systems: Ageing and Migration

the strategies of the young and the old generation and show that in a non-altruistic, gerontocratic world, providing education which increases the mobility of the young generation can be interpreted as a commitment device.4 Transfers to the young and to the old are thus linked in a strategic way.5 Each generation has an incentive to invest in the human capital of the subsequent generation in order to increase its mobility as this solves the problem of time-inconsistency concerning redistributive taxation. This result is similar to Kehoe (1989), which shows that capital mobility—by acting as a commitment device—can solve the hold-up problem of time-consistent taxation for a benevolent government. Factor mobility—in the form of mobile capital or mobile labour—reduces the incentive of the government to tax the returns of this factor ex post in an excessive way.6 The idea is modelled in the following way. Suppose that the population can be divided into two generations: one young generation and one old generation. The young generation is ﬁrst educated and then starts to work. The old generation has the power to levy a tax on the young generation.7 The tax rate is set to equalise the domestic net wage and the foreign net wage, which is the outside option for the young generation. To avoid a situation where the young individuals do not invest in human capital, given the high tax burden they anticipate once they have acquired country-speciﬁc human capital, the old generation would like to commit itself credibly to a low level of taxation in the future. A possible commitment device is to reduce mobility costs by providing the young individuals with language skills and information about foreign customs, laws and regulations. This increases the productivity abroad and correspondingly the wage abroad, thus making the outside option more attractive. A necessary condition for the feasibility of this strategy, however, is the power of the old to control the skill composition. Our model relies on the productivity effect of human capital investment as a link between the young and the old generation and allows for migration of the young. It is thus related to Konrad (1995a,b), who addresses the provision of education and infrastructure. Konrad (1995a) focuses on how the investment incentives are affected by increased mobility with ﬁscal federalism; Konrad (1995b) then discusses how these incentives change with population growth. 4

For public education as commitment device in an altruistic model see Gradstein (2000). Intergenerational transfers are quite signiﬁcant in developed countries: roughly 5% of GDP are transferred to the young in the form of public education and about 10% of GDP are received by the old via public pension systems (OECD, 1996, Economic Policy Committee, 2001, and Thum, 2000). 6 See Andersson and Konrad (2003), who analyse how labour mobility affects the government’s incentive to subsidise human capital and the individuals’ incentives to invest in human capital. 7 The tax can also be interpreted as contributions to an unfunded pension system with a weak contribution–beneﬁt link. 5

Mobility as a Counterforce to Gerontocracy

165

The model is also related to the literature on investment in general versus speciﬁc human capital. Contrary to the conclusion reached by Becker (1964) that ﬁrms do not invest in the general skills of their employees in competitive markets where they cannot collect the returns from these investments, some ﬁrms do in fact provide their workers with general human capital. Kessler and Lu¨lfesmann (2000) show that there is an incentive complementarity between employer-sponsored general and speciﬁc training, which means that employers invest in both types of training. General investment thereby reduces the hold-up problem concerning the provision of ﬁrm-speciﬁc investment. Our model is, however, different in one respect: “general” human capital, i.e. human capital that increases mobility, does not raise productivity equally in all countries but is only productive abroad. The decision for more “general” human capital is, therefore, always a decision against less (country-) “speciﬁc” human capital—something which is not present in the training decision. Section 7.2.1 presents the model for two and for n countries and compares the results with the ﬁrst best. In section 7.2.2, the monetary costs of education are added to the analysis. Section 7.2.3 extends the basic setting and discusses two variants of the basic model and their consequences for the welfare and the income of the young and the old generation. Section 7.2.4 draws some conclusions. 7.2.1. The role of education We are interested in how public education is affected by increased mobility. To answer this question, we analyse as a ﬁrst step the interaction between two countries, each of which takes the behaviour of the other country as given. In a second step, we apply this framework to a federation of n countries where neither country is big enough to exert some market power. In both cases, we assume that there are no informational problems. 7.2.1.1. The model In each country, there is an old generation and a mobile young generation.8 Members of each generation are homogeneous. Both countries compete for the young individuals by setting a proportional wage tax rate. These tax rates can be different for the domestic individuals a country wants to keep and the foreign individuals a country wants to attract. Separating the competition for the mobile 8

We abstract from a possible second group of young individuals in each country with sufﬁciently high mobility costs to render them immobile.

166

Unfunded Pension Systems: Ageing and Migration Education Quantity: Z

Young Generation Old Generation

Education Structure: γ

1

Migration: N Tax: t

2

3

4

Stage

Figure 7.1: Decision structure.

individuals from country i and country j reﬂects the fact that domestic individuals and immigrants are very often treated differently with respect to ﬁscal matters. Even though one does not ﬁnd different wage tax rates for domestic and foreign workers, the total net tax charge is very likely to be different for locals and foreigners when everything—including transfers and taxes and taking into account partially incompatible or incompletely harmonised social security systems—is taken into account. The old generation raises the taxes that have to be paid by the young generation. It is thus assumed that the old generation has the power to tax the young generation, i.e. there is a gerontocracy.9 Additionally, it is assumed that production in country i is determined by a function Yi ¼ Fi ðLi Þ;

ð7:9Þ

which depends on domestically productive labour provided by domestic and immigrated individuals expressed in efﬁciency units Li : Let Hi denote the domestically valuable human capital of the Ni domestic workers and Hij the human capital of the Nij immigrated workers from country j which is productive in country i. We then have for the total amount of efﬁciency units available for production in country i Li ¼ Hi Ni þ Hij Nij :

ð7:10Þ

The function Fi ðLi Þ is assumed to display constant marginal productivity with respect to labour ð›Fi =›Li ¼ Wi . 0Þ: The model has the following decision structure from the point of view of country i (cf. Figure 7.1). In the ﬁrst stage, the old generation decides on the composition of the educational programme such that they determine which skills are taught in school. This can be justiﬁed by considering secondary education, where the curriculum is normally ﬁxed by the government or some school board. 9

See, for example, Konrad (1995a,b) and the discussion in section 7.2.4.

Mobility as a Counterforce to Gerontocracy

167

It is assumed that education can be divided into one part that increases the domestic productivity (for example, law of the domestic country) and one part that increases the productivity in the foreign country (for example, foreign languages). In reality, this separation can hardly be made, since foreign languages, for example, increase both the domestic productivity and the foreign productivity— though to relatively different extents. To know some French is certainly useful in Italy; for working in France, however, very good knowledge of this language is indispensable. The strict separation of human capital into skills that are exclusively productive either at home or abroad is motivated by putting the emphasis on these relative increases. For simplicity, the part of education that increases productivity equally in the home and in the foreign country (such as mathematical skills) is neglected in the analysis. Let us deﬁne gi as the fraction of education that increases only the domestic productivity in country i: In the second stage, each member of the young generation chooses the amount of education. In many countries, the young individuals have some choice concerning the number of years they want to spend in school and the diploma they want to acquire. The amount of education Zi multiplied by the part of the skills that increases domestic productivity gi can be interpreted as an investment in domestically valuable human capital Hi Hi ¼ gi Zi :

ð7:11Þ

In the third stage, the old generation of country i and the old generation of country j set the proportional wage tax ti and tji which applies to the young individuals of country i: Given these tax rates, the members of the young generation decide at the last stage whether to emigrate or to stay in the home country. In the one-shot game here, the old generation thus chooses the structure of education before the tax rate, which corresponds to the observation that the educational structure normally changes less frequently than the tax policy, although in reality, the educational structure and the tax rate are revised on a regular basis so that reputation-building plays an important role. The solution of the model is then obtained as a sub-game perfect equilibrium by solving the decision structure backwards.10 7.2.1.2. Stage 4: migration decision Each member of the young generation compares the wage in the home country i to the wage in the foreign country j; i – j: The domestic net wage in country i for a young 10

For a similar set-up in a related analysis see Andersson and Konrad (2003).

168

Unfunded Pension Systems: Ageing and Migration

individual from country i is given by the marginal product of labour after taxes: wi ¼ Wi gi Zi ð1 2 ti Þ

ð7:12Þ

with Wi as the domestic wage per efﬁciency unit of domestically valuable human capital, Hi ¼ gi Zi ; and ti as the proportional wage tax of country i: The foreign net wage in country j for a young individual from country i is given by wij ¼ Wj ð1 2 gi ÞZi ð1 2 tji Þ

ð7:13Þ

with Wj deﬁned analogously. Hji ¼ ð1 2 gi ÞZi is the human capital of an individual of country i valuable in country j and tji denotes the wage tax of country j which applies to an immigrant from country i: In equilibrium, the net return to human capital has to be equal. In order to keep the young or to attract the young from the other country, respectively, the old generations in both countries engage in tax competition. With regard to the young generation of country i; the old generation has an advantage by disposing of two instruments—the tax rate ti and the structure of the human capital gi : The foreign old generation, in contrast, can only try to attract the young individuals of country i by setting a low tax rate tji : Given a sufﬁciently high gi ; the domestic old generation can always force the foreign tax rate to zero and keep the young generation in the home country.11 In equilibrium, we have thus tji ¼ 0 for the foreign wage tax rate and ti . 0 for the domestic one with all young individuals from country i staying in their home country. This leads to Wi gi Zi ð1 2 ti Þ ¼ Wj ð1 2 gi ÞZi :

ð7:14Þ

7.2.1.3. Stage 3: tax decision The old generation maximises its income by raising a proportional wage tax ti in such a way that the young individuals of country i stay in the home country. From Equation 7.14 we get for the optimal wage tax rate tip tip ¼ 1 2

11

Wj ð1 2 gi Þ : Wi gi

ð7:15Þ

This equilibrium condition abstracts from the possibility that the young generation might have the opportunity to tax the subsequent generation in the future. This setting can be justiﬁed by assuming a singular demographic shock, i.e. that the gerontocratic system is temporary.

Mobility as a Counterforce to Gerontocracy

169

7.2.1.4. Stage 2: education quantity decision At the second stage, each member of the young generation in country i decides on the amount of education by maximising the wage net of tax wi in country i; taking into account the disutility from education. For tractability, the disutility is assumed to be a quadratic function of the amount of education Zi :12 The maximisation problem is thus given by max Wi gi Zi ð1 2 ti Þ 2 Zi2 : Zi

ð7:16Þ

This yields with tip for the optimal amount of education Zip Zip ¼

1 2

Wj ð1 2 gi Þ:

ð7:17Þ

Equation 7.17 shows that there is a negative relationship between the educational level Zi and the skill composition parameter gi : A higher value for gi ; i.e. a more domestically oriented educational structure, decreases the outside option of the young generation. This allows the old generation to extract a larger part of the productivity gain without having to fear emigration. The prospect of higher future taxes, however, decreases the incentives for the young generation to invest in education Zi : The old generation cannot inﬂuence the chosen amount of education Zi directly; all it can do is to choose a certain structure gi : 7.2.1.5. Stage 1: education structure decision In the ﬁrst stage, the old generation maximises its income from the tax revenue by choosing the educational structure gi 13 max Wi gi Zip tip : gi

ð7:18Þ

The ﬁrst order condition is given by Wi Wj ð 12 2 gi Þ ¼ 2Wj2 ð1 2 gi Þ: 12

ð7:19Þ

The increasing marginal disutility of education can also be interpreted in a broader sense. Let’s take w per year as the forgone wage income during the educational period, i.e. the opportunity costs of education per year. If we interpret Z as the level of education (and not as the years of education) with Z ¼ f ðYearsÞ and ZYears . 0; ZYearsYears , 0; then the marginal costs of a higher educational level are increasing in the years of education. 13 Note that Zip and tip are functions of gi :

170

Unfunded Pension Systems: Ageing and Migration

The left-hand side shows two opposing effects on the income of the old generation from a marginal increase in gi ; which can be seen by rewriting this expression as Wi Zip þ Wi gi

›Zip 1 2 gi : ¼ Wi Wj 2 ›gi

ð7:20Þ

First, for a given amount of education Zip the output increases with the educational structure gi : This effect is equal to Wi Zip : Second, the young generation decreases the chosen quantity of education Zi : This effect is equal to Wi gi ›Zip =›gi : The sum of both effects is the term on the left-hand side of Equation 7.19. The right-hand side of Equation 7.19 shows how a marginal increase of gi ; i.e. a more domestically oriented educational structure, inﬂuences the income of the young generation in the form of the foreign wage rate, which is the outside option of the young generation. Differentiation of the foreign income of the young generation wij with respect to gi yields 2Wj2 ð1 2 gi Þ: Increasing gi implies that the migration option becomes less attractive, which increases the taxation possibility for the old generation. The optimum is reached when an increase in gi equalises the effects on the income of the old and of the young generation. For the optimal educational structure we get

gpi ¼

1 Wi þ 2Wj : 2 Wi þ Wj

ð7:21Þ

7.2.1.6. Results For Wi ; Wj positive and ﬁnite, we thus have

1 2

, gpi , 1; which leads to

Proposition 1. The old generation has an incentive to set up an educational system that increases foreign skills ðgpi , 1Þ in order to restrict future taxation. Education is used as a commitment device by the old generation. The lower gi is, the more attractive it is for the young individuals to migrate. Consequently, the old generation cannot set a high tax rate without inducing emigration. This solves the hold-up problem of time-consistent taxation. Reinserting gpi into Zip and tip ; we get as equilibrial values for the amount of education and the tax rate Zip ¼

1 Wi Wj ; 4 Wi þ Wj

ð7:22Þ

Mobility as a Counterforce to Gerontocracy

171

and tip ¼

Wi þ Wj : Wi þ 2Wj

ð7:23Þ

It is worth noting that the tax rate tip is smaller than 1 as long as Wi ; Wj are positive and ﬁnite, i.e. it is never optimal for the old generation of country i to levy a wage tax of 100%. If the tax rate was set above tip ; emigration would lead to zero income for the old generation. Hence, it is optimal for the old generation to choose a ti ; gi combination that does not induce emigration: all members of the young generation stay in the home country generating a positive tax income for the old generation. For country j; we get analogous results. For the special case where the countries are identical we have gpi ¼ 34 and tip ¼ 23 : 7.2.1.7. The n-countries case Extending the two-country setting to n countries is straightforward as the basic arguments do not change. So far, we had two countries i and j competing on two separate markets for the mobile young of each country. Now, we have n separate markets with competition over the mobile young of these n countries. This does not imply, however, that in each of these markets all n countries are involved. Who competes with the respective home country can be inﬂuenced by the home country through the choice of its educational structure. Concerning the skills that are only productive abroad ð1 2 gi Þ; we have to distinguish two cases. First, we can think of these skills as being equally productive in all foreign countries j – i: Thus, we again have gi denoting the domestically productive part of human capital and 1 2 gi denoting the part which is productive abroad, i.e. in the n 2 1 foreign countries. Then, country i has to set the tax rate sufﬁciently low to win the competition against the foreign country or countries which present the best outside option for the young generation. This will always be possible given the fact that the domestic country i has two instruments, gi and ti ; with which to compete, whereas the foreign countries can only vary the tax rate tji : Second, it is also possible that the old generation P can determine the skills for every foreign country separately so that 1 2 gi ¼ j–i gji : However, as the only purpose of an outside option for the young generation ðgi , 1Þ is to serve as a commitment device for the old generation, it is not efﬁcient for the old generation to spread the foreign skills over several countries. It is enough that gji . 0 for one country j – i which brings us back to the two-country model. In fact, a home country always faces only one competitor—or several, but identical, competitors with respect to the marginal productivity of labour—in its

172

Unfunded Pension Systems: Ageing and Migration

home market, even in a setting with n countries. The results for the asymmetric and symmetric case which we have derived for two countries, therefore, carry over to the n country case. 7.2.1.8. Social planner The social planner, who wants to obtain the social optimum for the domestic individuals, chooses the g; Z combination that maximises the domestic tax and labour income for the old and the young generation minus the disutility from education. As he or she does not need to commit him- or herself to a certain policy, the social planner, who only takes the domestic country into account, invests exclusively in domestically valuable skills ðg ¼ 1Þ: Therefore, Equation 7.11 simpliﬁes to Hi ¼ Zi : Assuming a linear utility function with respect to consumption (so far this assumption was not necessary), the maximisation problem can be written as max Ni Wi Hi 2 Ni Hi2 : H

ð7:24Þ

Rewriting the ﬁrst order condition determines the socially optimal investment in human capital Hipp Hipp ¼

1 2

Wi :

ð7:25Þ

Without the feasibility of an intergenerational contract, however, there is a negative relationship between g and Z (see Equation 7.17). The higher the value for g, i.e. the closer g gets to the socially optimal level, the less attractive the investment in human capital becomes for the young generation. Hence, the old generation has to choose a skill composition which at least partly increases productivity in the foreign country (i.e. g , 1) in order to induce the young generation to invest in human capital. From a social planner’s point of view, the members of the young generation learn too many skills which only increase foreign productivity. The socially optimal value for the human capital investment Hipp cannot be reached. This can easily be seen by comparing the amount of domestically valuable human capital Zip gpi with the socially optimal level Hipp :14 In the ﬁrst best scenario, all education is used to increase domestic productivity ðg ¼ 1Þ; whereas in the second-best scenario, g is below 1. Due to the distorted skill composition and the distorted education decision of the young generation, there is too little investment in human capital. These results can be summarised in Proposition 2. 14

See Appendix A7.

Mobility as a Counterforce to Gerontocracy

173

Proposition 2. Without the feasibility of intergenerational contracts, human capital investment is below the social optimum. There are two sources of inefﬁciencies. (1) The structure of education g is distorted towards too many skills that increase only the foreign productivity. (2) The level of education Z is below the social optimum.

7.2.2. Monetary costs of education To single out the effects of the intergenerational setting, we have so far neglected any monetary costs of education. This section analyses the case where monetary costs C per unit of education Z are completely borne by the old generation.15 Only is stage 1 of the analysis above is affected. The optimal skill composition gCp i given by16

gCp i ¼12

1 ðWi 2 CÞ : 2 ðWi þ Wj Þ

ð7:26Þ

Taking the costs into account does not change the qualitative results of the analysis. Quantitatively, however, the costs lead to a higher optimal value for gCp i ; p gCp i . gi :

ð7:27Þ

This is intuitively clear when thinking about the following mechanism: a higher g induces the young generation to choose less education Z because the outside option is less attractive. This reduces the costs of providing for Z for the old generation. The disadvantage of a higher g in the form of a lower Z is then compensated by the advantage of lower costs. It can be shown that the total Cp Cpp investment in human capital gCp i Zi is again below the socially optimal level Hi from the point of view of a social planner who is only interested in the domestic country: HiCpp ¼

15 16

1 2

ðWi 2 CÞ:

The young generation is assumed to bear costs of education only in the form of disutility. See Appendix B7 for all derivations of this section.

ð7:28Þ

174

Unfunded Pension Systems: Ageing and Migration

Proposition 3. Each generation has an incentive to invest in the human capital of the subsequent generation, even if the monetary costs are fully borne by the old generation. The human capital investment, however, is again below the social optimum.

7.2.3. Extensions We have seen that with and without the monetary costs of education, the old generation allows for a positive degree of mobility of the young generation which serves as a commitment device to overcome the hold-up problem. Redistributive taxation is thus feasible. The domestically valuable human capital investment is positive but lower than socially optimal. It is, therefore, worthwhile analysing two variants of the model and to compare these results to the results of the basic model with respect to the allocative and distributive implications. So far, we have allowed for continuous choices of the educational quantity Zi $ 0 by the young generation and have abstracted from introducing other possible instruments of the old generation to inﬂuence this choice. In the ﬁrst variant, we now analyse the possibility of restricting the values that Zi can take. In the second variant, we examine whether it might be advantageous to consider a subsidy to compensate the young generation for (part of) the disutility. Whether these two variants bring us closer to the social optimum than the basic model will then be analysed. 7.2.3.1. Discrete choice of Z From the point of view of the old generation, education that is only productive in the home country is preferred. But setting gi ¼ 1 induces the young generation to choose no education at all (cf. Equation 7.17), resulting in no income for the old generation. Very often, however, the possible choices of the amount of education are directly or indirectly restricted, for example, via regulations which introduce some minimum amount of education required for obtaining a degree. Suppose that this minimum amount of education Z i . 0 is chosen such that the income of the young generation is zero, making the young generation indifferent between Zi ¼ 0 and Z i . 0 but leading to a positive output.17 This output is then completely redistributed to the old generation. 17 This can be motivated by assuming that the signalling function of a degree is essential for the wage determination. Years of school Zi ; 0 , Zi , Z i ; will not be chosen although this would be possible.

Mobility as a Counterforce to Gerontocracy

175

The income of the young generation Wi gi Zi ð1 2 tip Þ 2 Zi2 ¼ Wj ð1 2 gi ÞZi 2 Zi2

ð7:29Þ

is zero for Zi ¼ 0 and Z i ¼ Wj ð1 2 gi Þ: Thus the old generation might wish to make Z i a requirement for a degree, thus indirectly restricting the choice of the young generation to 0 and Z i : If we assume that in this case the young individuals choose Z i ; we can derive the optimal choice of the educational structure from an optimisation problem analogous to Equation 7.18: ð1 2 gi ÞWj max Wi gi Z i tip ¼ Wi gi ½Wj ð1 2 gi Þ 1 2 ð7:30Þ gi g i Wi yielding for the optimal structure

giDp ¼

1 Wi þ 2Wj : 2 Wi þ Wj

ð7:31Þ

With giDp ¼ gpi and Z i ¼ 2Zip and no change in the wage tax rate, it can easily be seen that the tax base will be larger, thus leading to a higher income for the old generation, when the choice of the amount of education is (indirectly) restricted. The old generation increases its income at the expense of the young generation whose income reduces to zero. 7.2.3.2. Direct subsidies If the members of the old generation are able to inﬂuence the costs of the young generation to acquire education, i.e. the disutility, via some kind of subsidy, they can also use this instrument to increase their income. The income of the young generation is then Wi gi Zi ð1 2 tip Þ 2 Zi2 fðBi Þ;

ð7:32Þ

where fðBi Þ is a function that is decreasing in subsidies Bi at an increasing rate and fð0Þ ¼ 1 and limBi !1 fðBi Þ ¼ a $ 0:18 The young generation chooses ZiBp ¼

Wj ð1 2 gi Þ ; 2fðBi Þ

which exceeds Zip for Bi . 0 in order to maximise the income when educational subsidies are available. The structure of education, which maximises the income 18

See Andersson and Konrad (2003) for an identical formulation of educational subsidies.

176

Unfunded Pension Systems: Ageing and Migration

of the old generation, max Ni Wi gi ZiBp tip 2 Ni Bi ;

ð7:33Þ

g

is

giBp ¼

1 Wi þ 2Wj ; 2 Wi þ Wj

which equals gpi : Given giBp and ZiBp ; the value for Bi that maximises the income of the old generation has to fulﬁl Að2f0 ðBpi ÞÞ ¼ fðBpi Þ2 ; where A ¼ Wi gi Zi ti : As the old generation can always choose Bi ¼ 0; the income of the old generation with an educational subsidy is at least equal to the income without subsidy ! 1 2 1 2 Bpi $ 0: Wi gi Zi ti fðBpi Þ The income of the young generation is always at least as high as the disutility is not larger, the amount of education is not smaller, but the tax rate and the educational structure remain unchanged. 7.2.3.3. Welfare comparison So far, we have analysed the extent to which these two additional instruments affect the income of the young and the old generation. Now, we turn to the welfare analysis. Allowing for a restriction of the quantity of education to two discrete values does not change the socially optimal amount of human capital investment from the point of view of one country—taking only those individuals into account who have not emigrated. As we have seen, the social optimum is H pp ¼ 12 Wi ; which exceeds the domestically productive human capital investment of the basic model H p ¼ gpi Zip : Restricting the quantity of education to two discrete values leads to investment in human capital of the amount19 Dp p p p p p H D ¼ gDp i Zi ¼ 2gi Zi . H ¼ gi Zi ;

but H D , H pp :

ð7:34Þ

Thus, the social optimum will not be reached although the human capital with a restricted choice of education exceeds the human capital with an unrestricted choice.

19

See Appendix A7 for the derivation.

Mobility as a Counterforce to Gerontocracy

177

Next, we have to look at the domestically productive human capital when subsidising is possible. In this case, the social optimum can be derived from max Ni Wi Hi 2 Ni Bi 2 Ni fðBi ÞHi2 H

ð7:35Þ

leading to H Bpp ¼

Wi : 2fðBi Þ

It turns out that human capital investment with the possibility to subsidise is never smaller than in the basic model Bp H B ¼ gBp i Zi ¼

gpi Zip $ H p ¼ gpi Zip ; as fðBi Þ # 1; fðBi Þ

ð7:36Þ

but does not reach the social optimum H B , H Bpp : It is worth noting that if we allow for these variants of the basic model the human capital investment gets closer to the social optimum, but the fundamental structure of the model does not change. In particular, the structure of education is unaffected and still serves as a commitment device ðgp , 1Þ in order to counterbalance the gerontocratic power of the old generation. 7.2.4. Conclusion We have seen that in a mobile, non-altruistic world, increasing the mobility of the young generation enables the old generation to commit itself credibly to a lower tax level in the future. To say it differently, a more attractive outside option of the young generation, which serves as the commitment device prevents emigration and preserves a redistributive system towards the old. Although there is an efﬁciency loss due to human capital investment that increases the foreign productivity but not the domestic one, this system will not erode. This result is important as it allows us to draw some conclusions about the possible development of old-age security on the European level. To derive this result we have taken two stylised facts into account: higher (potential) mobility of the young and gerontocratic power. As we have seen in Chapter 3, both are or will be relevant for Europe in the next few decades. Do we already live in a gerontocratic system or will we do so in the near future? The fact that the median voter is younger than the retirement age does not have to be a criticism of the assumed gerontocracy. Individuals close to the retirement age

178

Unfunded Pension Systems: Ageing and Migration

have similar interests to the retired and the median voter will get older in the years to come.20 But what about intra-European migration? The levels have never been very high. One might expect that a higher degree of European integration will lead to more intra-European mobility and consequently to a more efﬁcient allocation of labour in the coming years. When efﬁcient allocation is reached, migration might settle at some steady-state level. This would be in line with our model. Making it easier for the young to emigrate does not necessarily mean that everybody leaves his or her country. On the contrary, it is exactly this better outside option that may make migration (partially) redundant—at least for reasons other than an efﬁcient allocation. We are now able to derive how systems of intergenerational redistribution towards the old evolve for different degrees of mobility of the young. In Chapter 6, we have analysed the case where the young are completely immobile; the voice option via voting is the only possibility for them to defend their interests. Once the old have the majority, an explosion of the system can no longer be excluded. In this chapter, we have enlarged the analysis to include an exit option for the young. More precisely, we have considered an intermediate case between no mobility and total mobility where (potential) mobility of the young can be controlled by the old in such a strategic way as to counterbalance gerontocracy. This enables the old to commit to a certain level of taxation. Thus, an intergenerationally redistributive system needs neither explode nor erode. We can also draw some conclusions for the case with total mobility of the young where the old cannot inﬂuence mobility in any way. An erosion of the system might then be the result. In the Chapter 9, we will look in detail at scenarios of restricted and unrestricted mobility within the European Union. The analysis will focus on the potential danger of a “race to the bottom” due to competition between the European countries. The question is then where to allocate the responsibilities for social security issues optimally—in particular concerning old-age security—in order to ensure an efﬁcient allocation of labour and a positive level of intergenerational redistribution.

APPENDIX A7. SOCIAL PLANNER PROBLEM The socially optimal education level is given by (cf. Equation 7.25) H pp ¼ 20

See Chapter 6.

1 W: 2 i

ðA7:1Þ

Mobility as a Counterforce to Gerontocracy

179

Comparing the socially optimal amount of human capital to the amount realised without a social planner, we get for 0:5 , gp # 1

gpi Zip , H pp , gpi

ð1 2 gpi ÞWj W , i: 2 2

ðA7:2Þ

By using Equation 7.21, this can be easily veriﬁed: ðWi þ 2Wj Þ½ð2Wi þ 2Wj Þ 2 ðWi þ 2Wj ÞWj , Wi ð2Wi þ 2Wj Þ2 ; ðA7:3Þ

Wi ðWi þ 2Wj ÞWj , Wi ð2Wi þ 2Wj Þ2 ; 0,

4Wi2

þ 7Wi Wj þ

2Wj2 :

The human capital investment with (indirectly) restricted choice of the educational quantity is also smaller than the relevant social optimum

giDp ZiDp , H pp , gpi ð1 2 gpi ÞWj ,

Wi : 2

ðA7:4Þ

This can be shown by ð2Wi þ 2Wj Þ2 ; 2 0 , 2Wi2 þ 3Wi Wj : ðWi þ 2Wj ÞWj ,

ðA7:5Þ

APPENDIX B7. MONETARY COSTS OF EDUCATION If the monetary costs of C per unit of education Z are completely borne by the old generation, only stage 1 of the analysis above is affected. Equation 7.18 thus becomes ðB7:1Þ max Ni Wi gi Zi ti 2 Ni Zi C; g

with the ﬁrst order condition 1 C 2 gi ¼ 2Wj2 ð1 2 gi Þ 2 Wj : Wi Wj 2 2

ðB7:2Þ

The interpretation of this equation is identical to the interpretation in section 7.2.1 with the exception that there is an additional disadvantageous term on the righthand side caused by the costs of education. Rearranging this equation yields the optimal educational structure gCp i

gCp i ¼12

1 ðWi 2 CÞ : 2 ðWi þ Wj Þ

ðB7:3Þ

180

Unfunded Pension Systems: Ageing and Migration

It can again be seen that 0:5 , gCp i , 1 (for Wi 2 C . 0). Thus, taking the costs into account does not change the qualitative results of the analysis. Quantitatively, however, the monetary costs C lead to a higher optimal value for g: In order to obtain the social optimum for this case, a social planner has to take the production and the costs of education, i.e. the disutility of education as well as the monetary costs, into account max Ni Wi Hi 2 CNi Hi 2 Ni Hi2 : H

ðB7:4Þ

Rewriting the ﬁrst order condition yields the socially optimal investment in human capital H Cpp ¼

1 1 W 2 C: 2 i 2

ðB7:5Þ

Cpp In order to show that ZiCp gCp ; one has to take into account that C [ i , Hi p 1 ½0; Wi as g [ ð 2 ; 1Þ and that

ZiCp ¼

1 ðWi 2 CÞ Wj : 4 ðWi þ Wj Þ

ðB7:6Þ

Cpp 21 For C ¼ 0; it has already been shown that ZiCp gCp : And for C ¼ Wi ; it is i # Hi Cp Cp Cpp easy to show that Zi gi ¼ Hi ¼ 0: Therefore, it remains to check 8 9 8 9 .> .> > > > > < > < > = = Cp Cp Cpp Cp Cp ¼ Hi , Wj ð1 2 gi Þgi ¼ Wi 2 C: Zi gi ðB7:7Þ > > > > > > : > : > ; ; , , Cp Cp We deﬁne Q ; Wj 1 2 gi gi and differentiate Q with respect to C

›Q ›gCp ¼ Wj i 1 2 2gCp i ; ›C ›C

ðB7:8Þ

›gCp 1 i ¼ : 2ðWj þ Wi Þ ›C

ðB7:9Þ

where

With Equations B7.8 and B7.9, one gets Wj ›Q ¼ 1 2 2gCp i : ›C 2ðWi þ Wj Þ 21

See the argumentation in section 7.2.1.

ðB7:10Þ

Mobility as a Counterforce to Gerontocracy

181

In order to sign the derivative ›Q=›C; we use the following two relationships: Wj , 1; Wi þ Wj 21 , 1 2 2gCp , 0: i

ðB7:11Þ ðB7:12Þ

The latter follows from the fact that gCi [ ð 12 ; 1Þ: Hence, the derivative ›Q=›C lies within the following interval: 2

1 ›Q , , 0: 2 ›C

ðB7:13Þ

This has to be compared with the derivative of the socially optimal solution

›HiCpp 1 ¼2 : 2 ›C

ðB7:14Þ

As a comparison of Equations B7.13 and B7.14 shows, a marginal increase in the costs C means that the marginal decrease of the socially optimal education level HiCpp is higher in absolute terms than the marginal decrease of the education level Cp gCp within the interval C [ ½0; Wi : We know that at the right border of the i Zi Cp Cpp interval (i.e. C ¼ Wi ) HiCpp ¼ gCp i Zi ¼ 0; and so the socially optimal value Hi Cp Cp exceeds gi Zi for all C within the interval.

This Page Intentionally Left Blank

CHAPTER 8

Qualitative Aspects of Migration Who is most likely to emigrate Comes over one an absolute necessity to move. And what is more, to move in some particular direction. A double necessity then: to get on the move, and to know whither. David H. Lawrence (1885 –1930) Abstract Migration is low for most European countries. But migration puts nevertheless pressure on unfunded pension systems if those who pay contributions to an unfunded pension system leave the country. This will lead to competition for contributors and set a “race to the bottom” in motion. Growing integration of the countries of the European Union will intensify this development. In order to shed some light on the aspect of emigration from an industrialised country by natives, we analyse who is most likely to react to these incentives. As statistical information is thin with respect to emigration, we present estimations about intended emigration from Germany. We ﬁnd that young individuals with an above-average school degree are most likely to think about emigrating. But these individuals are exactly those who are needed as contributors to the unfunded pension systems. Reactions of national pension systems must therefore be expected. As we have seen in Chapter 3, population growth is primarily dominated by fertility rates and life expectancies. Migration does not play a very important role for most European countries and will be very probably insufﬁcient to provide a solution to the ageing problem. It is, however, possible that despite the low general level of migration, those few who migrate are enough to put pressure on

184

Unfunded Pension Systems: Ageing and Migration

the unfunded pension systems. If the “wrong” individuals, i.e. young individuals with high incomes who pay contributions to the pension system, start migrating in order to escape the rising pension burden, the ﬁnancial stability of the system deteriorates. Even worse, the danger of an unbalanced pension budget might induce the national pension systems to react and to engage in a competition over contributors which might result in a “race to the bottom”. It is not sufﬁcient to observe that labour mobility is still limited—even though the majority of Germans, for example, is not yet mobile internationally,1 which hints at a high degree of general inertia or insufﬁcient incentives. As Sinn (1990, 1998) states, all that is needed for governments to feel the political pressure of factor mobility is marginal mobility.2 As soon as some people react to different situations at home and abroad by migrating, the unfunded pension systems are affected because they rely on a well-balanced ratio of contributors to retirees. In this chapter, we, therefore, analyse in detail who is most likely to react to these incentives. We present estimations about the quality of migration from Germany as an example of an industrialised country. By knowing which characteristics make it more likely for an individual to think about emigrating and by considering the projected frequency of these characteristics—notably with respect to the demographic development—we can formulate hypotheses about the possible evolution of emigration in the coming decades. Various studies have looked into the social and economic integration of immigrants in countries like the US, Canada, Australia and Israel beneﬁting from an exhaustive collection of data. These analyses focus on who immigrates (for example, Borjas, 1987, 1994) and on how immigrants coming from different countries of origin and arriving at different points in time adapt to the new environment (for example, Chiswick, 1978, and Borjas, 1994, for a survey).3 Although emigration and immigration are only two different sides of the same coin, the migration literature is mostly about immigration. Emigration, on the contrary, has not been much examined with the exception of emigration from industrialised countries in the form of return migration (see DaVanzo (1983) and Dustmann (1996) for a survey) and emigration from developing countries linked 1

See Chapter 3. According to Sinn (1998), the concept of marginal mobility is similar to the marginal concept predominant on private goods markets where it is also marginal demand that drives the competitive behaviour of ﬁrms and not intramarginal demand. 3 Examples of empirical analyses can be found in Beggs and Chapman (1988, 1990, 1991), Chiswick and Miller (1985), Dustmann (1993), Greenwood and McDowell (1991), Mayer and Riphahn (2000), and Schmidt (1997). 2

Qualitative Aspects of Migration

185

to the brain drain problem (for example, Hamada, 1996). So far, there are not many studies which analyse emigration—in particular by natives—from industrialised countries. One reason for this observation might be that in general, information about emigration—in contrast to immigration—is hard to ﬁnd. The US Census Bureau has recently developed some techniques to estimate the number of emigrants, which underlines the difﬁculty of obtaining reliable emigration data.4 In Germany, emigrants are legally obliged to give notice when leaving the country. These data collected by the German Federal Statistical Ofﬁce thus give an initial idea about the quantity and quality of emigration from Germany. However, the number of emigrants is very probably underreported due to registration problems. This must be kept in mind when evaluating the plausibility of the data we use for our estimations. In both cases, statistical information is thin and comprises at most an approximation of the volume without much additional information about the characteristics of emigrants, let alone information about the destination country. We, therefore, want to analyse emigration from Germany in detail with a special interest in the qualitative—not quantitative—aspects and estimate the effects of different characteristics on the probability of thinking about emigrating. We consider detailed information at the individual, household, and regional level which helps us to determine the characteristics of those who intend to emigrate relative to those who want to stay in Germany. These ﬁndings allow us to evaluate who is most likely to emigrate in the next few decades. In section 8.1, we describe the data. The intention to migrate is estimated in sections 8.2 and 8.3 concludes.

8.1. DESCRIPTION OF THE DATA The data for this analysis stem from the German Socio-Economic Panel (GSOEP) which consists of about 6000 households.5 We use the waves 10 from 1993, 13 from 1996, 14 from 1997, and 15 from 1998 because these waves are the only ones where individuals are asked about their intentions to emigrate. Wave 13 in addition comprises information about the reasons for moving. We exclude foreigners as they are self-selected in the sense that they have already displayed 4

See Bashir and Robinson (1994) for the foreign-born population and Fernandez (1995) for the US born population. 5 See the Appendix A5 for some additional information about the GSOEP.

186

Unfunded Pension Systems: Ageing and Migration

Table 8.1: Description of the dependent variable and sample characteristics. Womena

Move

Menb

Cases

Percent

Cases

Percent

Never Probably not Yes, it depends Yes, very much

1842 3643 3399 1111

18 36 34 11

1300 3419 3735 1158

14 36 39 12

Sum

9995

100

9612

100

Mean and standard error are calculated for the values 1 ¼ “never”,… and 4 ¼ “yes, very much”. Percentages may not add to 100% due to rounding. Source: Waves of 1993, 1996, 1997 and 1998 (GSOEP). a Mean ¼ 2.378; Std. error ¼ 0.909. b Mean ¼ 2.494; Std. error ¼ 0.873.

their mobility in at least one instance.6 As the gender can be expected to inﬂuence the propensity to move in different ways—for example, through stronger family ties for women (Naskoteen and Zimmer, 1980) and through different educational and professional careers—the sample will be subdivided into a female and a male sub-sample. After excluding individuals with missing values for relevant variables—mostly concerning the propensity to migrate and the schooling and work history, the male sample population reduces to 9612 observations and the female sample population to 9995 observations. Potential correlations in the error terms due to the fact that the sample includes repeated answers from given individuals in subsequent years are taken into account. Table A8.1 describes sample characteristics for the variables used in the empirical analysis. 8.1.1. Intention to migrate We use the stated propensity to move to another country (“move”) as the dependent variable. Table 8.1 presents sample characteristics for this variable with the response categories “never”, “probably not”, “yes, it depends”, and “yes, very much”. Men and women show a very high willingness to emigrate. We see that 51% of the men and 45% of the women are thinking about emigrating strongly (“yes, very much”) or at least to some extent (“yes, it depends”). We will comment below on the order of magnitude of these responses given the quite low numbers of individuals who actually emigrate. 6

For an analysis of the migration propensity based on a sample of Germans and foreigners see Uebelmesser (2004).

Qualitative Aspects of Migration

187

Table 8.2: Reasons of those with a positive propensity to migrate. Reasons to migrate

Better job Retirement Training/education Family Other reason No answer Sum

Women

Men

Cases

Percent

Cases

Percent

269 176 83 140 367 70

24 16 8 13 33 6

522 190 115 34 314 67

42 15 9 3 25 5

1105

100

1242

100

“move” ¼ “yes, it depends” or “move” ¼ “yes, very much”. Percentages may not add to 100% due to rounding. Source: Waves of 1993 (GSOEP).

Table 8.2 gives an overview of the reasons for migration stated by those who show a positive propensity to migrate.7 Better professional opportunities play an important role for 42% of the men and for 24% of the women. This hints at relevant differences in wages and/or employment probabilities in the destination country relative to Germany. However, reasons which are not directly linked to economic differences prevail. Sixteen percent of the women and 15% of the men want to spend their retirement period abroad. The motivation to migrate is thus not related to wage or employment differentials, although the general economic situation in the destination country is important to judge the purchasing power of the pension beneﬁts abroad.8 Better institutions for training and education are the reason given by 8% of the women and 9% of the men with a positive propensity to migrate. Again, the economic situation only plays a role in an indirect way when assuming that these individuals hope to give themselves better job opportunities abroad—or at home—after having completed their studies abroad.9 For women, family reasons are of importance in 13% of the cases compared to 3% of the cases for men. Here, the economic situation indirectly inﬂuences the decision to migrate when thinking that those individuals may follow their partner or other family members who have

7 A more extensive discussion of the economic and non-economic reasons of the potential migrants can be found in Uebelmesser (2004). 8 Note, however, that migration after retirement does not affect an unfunded pension system. 9 We do not have information about the intention of individuals to return after having completed their education. See Dustman (1995, 1997) for an analysis of the long-run effects of return migrants on the welfare state.

188

Unfunded Pension Systems: Ageing and Migration

migrated earlier—possibly for economic reasons. Thus, in general, the economic situation in the destination country compared to Germany plays some role for the propensity to migrate—either directly or indirectly. Tables 8.3 and 8.4 show the breakdown of the responses for women and men according to a number of personal, household and regional characteristics. Various patterns are immediately apparent. Table 8.3: Intention to emigrate (women). Variable

Yes, very much Yes, it depends Probably not

Never

Total sample

Individual data Age ,20 Age 20–29 Age 30–39 Age 40–49 Age 50–59 Age 60 þ Married Foreign partner German partner Not married Children 0– 6 Children 7– 16 No children under 16 West Germany East Germany

95 (24) 372 (15) 305 (10) 175 (9) 127 (10) 37 (4) 508 (9) 10 (11) 498 (8) 603 (15) 327 (10) 499 (10) 697 (12) 908 (13) 203 (6)

152 (38) 940 (37) 1040 (36) 728 (36) 384 (31) 155 (17) 2985 (33) 53 (59) 1932 (33) 1414 (35) 1129 (35) 1770 (36) 1940 (34) 2558 (38) 841 (26)

108 (27) 942 (37) 1149 (40) 792 (40) 417 (34) 235 (26) 2333 (39) 22 (24) 2311 (39) 1310 (32) 1216 (38) 1952 (39) 1901 (33) 2273 (34) 1370 (42)

41 (10) 297 (12) 414 (14) 301 (15) 298 (24) 491 (53) 1134 (19) 5 (6) 1129 (19) 708 (18) 513 (16) 731 (15) 1146 (20) 1030 (15) 812 (25)

396 2551 2908 1996 1226 918 5960 90 5870 4035 3185 4952 5684 6769 3226

Education Elementary Secondary Higher secondary University degree Occupational training

304 (9) 460 (10) 347 (15) 154 (14) 695 (10)

898 (28) 1534 (34) 967 (45) 490 (44) 2301 (33)

1063 (33) 1813 (40) 767 (34) 383 (35) 2716 (38)

960 (30) 711 (16) 171 (8) 81 (7) 1350 (19)

3225 4518 2252 1108 7062

Occupation Worker Self-employed Trainee Employee Civil servant Unemployed

104 (11) 44 (13) 77 (17) 457 (11) 35 (11) 101 (11)

271 (29) 145 (43) 186 (41) 1472 (37) 140 (45) 257 (27)

353 (38) 106 (32) 141 (31) 1593 (40) 113 (37) 353 (38)

203 (22) 41 (12) 48 (11) 459 (12) 22 (7) 226 (24)

931 336 452 3981 313 937

Income Yes 1053 (12) Low net income 275 (13) Middle net income 463 (10) High net income 315 (13) No (retired) 58 (5)

3167 (36) 646 (31) 1524 (34) 997 (43) 232 (20)

3352 (38) 776 (38) 1731 (39) 846 (36) 291 (25)

1276 (14) 357 (17) 741 (17) 178 (8) 566 (49)

8848 2054 4459 2335 1147

Values in parenthesis are percentages. Source: Waves of 1993, 1996, 1997 and 1998 (GSOEP).

Qualitative Aspects of Migration

189

Table 8.4: Intention to emigrate (men). Variable

Yes, very much Yes, it depends Probably not

Never

Total sample

Individual data Age ,20 Age 20–29 Age 30–39 Age 40–49 Age 50–59 Age 60 þ Married Foreign partner German partner Not married Children 0– 6 Children 7– 16 No children under 16 West Germany East Germany

70 (17) 395 (16) 348 (12) 170 (9) 131 (10) 44 (6) 528 (9) 16 (26) 512 (9) 630 (16) 327 (10) 489 (12) 748 (13) 927 (14) 231 (7)

132 (33) 1074 (44) 1166 (41) 731 (39) 465 (35) 167 (23) 2024 (36) 33 (53) 1991 (36) 1711 (42) 1129 (35) 1567 (39) 2325 (40) 2715 (42) 1020 (33)

154 (38) 782 (32) 1027 (36) 750 (40) 477 (36) 229 (32) 2110 (38) 11 (18) 2099 (38) 1309 (32) 1216 (38) 1511 (37) 1994 (34) 2127 (33) 1292 (42)

49 (12) 184 (8) 316 (11) 231 (12) 240 (18) 280 (39) 899 (16) 2 (3) 897 (16) 401 (10) 513 (16) 499 (12) 819 (14) 737 (11) 563 (18)

405 2435 2857 1882 1313 720 5561 62 5499 4051 3185 4952 5886 6506 3106

Education Elementary Secondary Higher secondary University degree Occupational training

381 (11) 371 (11) 406 (14) 221 (12) 758 (11)

1082 (33) 1248 (37) 1405 (47) 859 (47) 2507 (36)

1146 (35) 1312 (39) 961 (32) 601 (33) 2589 (38)

712 (21) 400 (12) 188 (6) 132 (7) 1046 (15)

3321 3331 2960 1813 6900

Occupation Worker Self-employed Trainee Employee Civil servant Unemployed

269 (10) 99 (16) 75 (15) 344 (12) 85 (12) 85 (11)

886 (34) 259 (43) 195 (38) 1331 (45) 287 (42) 253 (33)

1021 (39) 196 (33) 188 (37) 1082 (36) 241 (35) 272 (36)

458 (17) 49 (8) 55 (11) 213 (7) 71 (10) 152 (20)

2634 603 513 2970 684 762

Income Yes 1078 (12) Low net income 225 (10) Middle net income 563 (13) High net income 290 (13) No (retired) 80 (9)

3503 (40) 782 (35) 1672 (40) 1049 (46) 232 (26)

3135 (36) 885 (40) 1484 (35) 766 (32) 284 (31)

990 (11) 322 (15) 469 (11) 199 (9) 310 (34)

8706 2214 4188 2304 906

Values in parenthesis are percentages. Source: Waves of 1993, 1996, 1997 and 1998 (GSOEP).

Men in general are more likely to consider emigrating than women and younger individuals think more often about leaving Germany than older individuals—with the exception of men under 20. A higher school degree makes thinking about emigrating more likely for men and women. Occupational training and a university degree, however, do not

190

Unfunded Pension Systems: Ageing and Migration

further increase the probability compared to a secondary degree or a higher secondary degree, respectively. Concerning occupation as an important factor, we ﬁnd for men and for women that employed people in general are more likely to consider emigrating than the unemployed and retired. For the speciﬁc forms of occupations considered explicitly, probabilities are very high for most of the occupations—including self-employed and civil servants. In addition, the probability to think about emigrating increases with income levels.10 Single men and women are more likely to consider emigrating than those who are married to a German partner, while those with a foreign partner show a higher propensity to emigrate. Children, however, do not play an important role in the decision. Concerning differences between East and West Germany, individuals living in West Germany are more likely to emigrate than individuals living in the Eastern part of Germany. 8.1.2. Discussion of the intention variable Given the few actual emigrants in the GSOEP, and given the fact that in general information about emigration—in contrast to immigration—is hard to ﬁnd, the variable on the intention to move allows an approximation of the underlying data. As Manski (1990, p. 935) states, “intentions data do potentially convey information about behaviour”. To get a feeling for the reliability of the data of the GSOEP, we compare them with similar data from the study on “Performance of the European Union Labour Market” by the European Commission (1995). In this study, individuals are asked whether they would be willing to work in an EC member state different from the one of which they are a national (Table 8.5). Thirty-four percent of the men and 21% of the women gave a positive answer. This allows us to compare the results of the two data sets at least indirectly. If we only consider data from the 10th wave (1993) of the GSOEP with information about the reasons to move, we ﬁnd that 22% of the men and 11% of the women name better job opportunities abroad as a possible reason to think about emigrating.11 In both data sets, men are more willing to migrate for professional reasons than women. It is not surprising that the numbers from the GSOEP are smaller than those in the study of the European Commission (1995), given the fact

10 For those individuals without any (information about) net wage income, we have simulated net wage income using the Heckman procedure (1979). 11 Ratio of those who name “better job opportunities” as the reason to move (Table 8.2) to all individuals (2489 women and 2420 men) of wave 1993.

Qualitative Aspects of Migration

191

Table 8.5: Willingness to work in a “foreign” EC State. Variable

Yes (%)

No (%)

Total Of which Men Women

28

72

34 21

66 79

Source: European Commission (1995).

that in the GSOEP individuals have to choose the most likely reason among several reasons so that other reasons might crowd out work-related reasons.12 Taking this into account, one can state that the order of magnitude is comparable. It is, however, still necessary to give an explanation as to how one can reconcile the different orders of magnitude of the statistical information of the German Federal Statistical Ofﬁce (see Chapter 3) and the responses in the GSOEP and the survey of the European Commission. For this, we break down the responses into age groups. From the GSOEP data (Tables 8.3 and 8.4), we see that the propensity to emigrate (“move” ¼ “yes, very much” or “move” ¼ “yes, it depends”) decreases with age for men and women, with few exceptions. For women, it is 62% for those under 20 and decreases for the 20 to 59-year-old from 52 to 41%. Women aged 60 and over are less likely to consider emigrating (21%). The pattern is very similar for men although the level of those who consider emigrating is higher compared to women except for men under 20 (50%). For the 20 to 59-year-old, the level decreases from 60 to 45% and drops to 29% for those aged 60 and over.13 The observations from the German Federal Statistical Ofﬁce (Statistisches Bundesamt, 2000) are quantitatively different but not qualitatively. The absolute number of emigrants is much lower than what one would expect from the answers to the willingness-to-migrate question in both surveys. In 1993, only 86,619 Germans emigrated and in 1996, 1997 and 1998, the volumes were only slightly higher with 118,430, 109,903 and 116,403 emigrants, respectively. If we breakdown the data by age groups (Table 8.6), we ﬁnd, however, a similar proﬁle for intended and real emigration. The data from the German Federal Statistical Ofﬁce show a relative increase in emigration up to the age of 25– 30 12

See Table 8.2 for the other reasons. The study of the European Commission (1995) conﬁrms this general trend. The willingness to work abroad is highest for those below 31 years with 39% and decreases to 27% for the 31–49-year-old and to 15% for the 50-year-old and older individuals.

13

192

Age groups

1993

1996

1997

1998

Absolute number

Per 1000 of the age group

Absolute number

Per 1000 of the age group

Absolute number

Per 1000 of the age group

Absolute number

Per 1000 of the age group

, 18 18–25 25–30 31–50 51 þ

20,260 9900 12,984 30,835 12,640

1.3 1.4 1.8 1.3 0.5

25,312 12,545 15,521 46,359 18,693

1.6 2.0 2.4 1.8 0.7

22,426 11,493 12,124 44,472 17,388

1.4 1.8 2.3 1.7 0.6

22,443 12,435 14,845 48,600 18,080

1.4 2.0 2.7 1.9 0.6

Total

86,619

1.1

118,430

1.4

109,903

1.3

116,403

1.4

Source: Statistisches Bundesamt (2000).

Unfunded Pension Systems: Ageing and Migration

Table 8.6: Emigration from Germany.

Qualitative Aspects of Migration

193

and a decline thereafter with a sharp drop for those aged 51 years and older. This is exactly reﬂected by the answers of men whose propensity to emigrate also peaks for the age group 20– 29 and falls markedly for those 60 years and older, while it is closely mirrored by the answers of women, whose willingness declines over all age groups. For the following analysis, we thus follow Burda et al. (1998) in assuming that intentions are a monotonic function of the underlying driving variables which motivate migration. We will, therefore, concentrate on identifying those characteristics which make it more likely for an individual to think about emigrating. We will interpret the results of the estimation accordingly, namely that individuals with these characteristics will in fact be over-proportionally represented among the future emigrants.

8.2. ESTIMATION OF THE INTENTION TO MIGRATE According to the standard human capital model,14 the mobility decision of an individual is guided by the comparison of the present value of lifetime earnings— labour income and pension beneﬁts—in the home country and in the foreign country net of migration costs for migration at a certain age. As with all other decisions, the individual chooses the alternative that maximises the utility: in this case the utility of lifetime earnings. Thus, within this framework, migration occurs when the utility with migration exceeds the utility without migration. The human capital model thus suggests comparing the economic situation in the source and in the destination country by considering the monetary and non-monetary migration costs. This modelling has, however, two shortcomings in our context. First, it neglects any reasons which are not earnings-related, but which play an important role when thinking about migration as illustrated in Table 8.2. Second, it requires that the destination country with its speciﬁc characteristics is known. In general, however, information about the volume of emigration in general and about characteristics of the destination country in particular is mostly lacking.15 Our approach alleviates both problems. We assume that the emigration decision is a function of individual characteristics, characteristics of the household, and characteristics of the (home) region. We thus include economic but also 14

See Sjaastad (1962) for an early version of this model. In Germany, emigrants are legally obliged to give notice when leaving the country. However, the number of emigrants is probably underreported due to registration problems and information about the destination country is very limited. 15

194

Unfunded Pension Systems: Ageing and Migration

non-economic factors which can be important for the (potential) migration decision and aim at identifying their effects. In addition, we abstract from variables concerning the destination countries for the analysis which can be justiﬁed as we are only interested in the attitude towards migration and not in the probability of migrating to a speciﬁc country. It is reasonable to assume that there is at least one country for individuals with a positive propensity to migrate for which the utility exceeds the utility without migration. Following this approach, we now focus on a systematic analysis of the effect of each independent variable on the dependent variable. We implicitly assume that the decision to emigrate can be approximated by these variables.16 As the dependent variable, we use the reported propensity to move to another country which can be viewed as an ordered response with four categories: “never”, “probably not”, “yes, it depends” and “yes, very much”. The statistical model for categorical data is an ordered model, for example, an ordered Probit model.17 The interpretation of the estimated coefﬁcients in an ordered Probit model is not straightforward as the value and the sign can differ from those of the marginal effects. Therefore, the interpretation focuses on the statistical signiﬁcance of the coefﬁcients and on the marginal effects derived from simulations. The estimation results of the ordered Probit model for the male and female sub-sample are given in Tables B8.1 and B8.2, with column 1 presenting results of the basic conﬁguration and columns 2 and 3 adding measures of the professional situation and the environment, respectively. Simulations are carried out for the whole sample such that ﬁrst the baseline probabilities for each outcome are predicted on the basis of the estimated coefﬁcients (Tables B8.1 and B8.2). Then single variables are ﬁxed at two different values and the probabilities are predicted for each of the two values. The continuous variables “age” and “unemployment rate” are ﬁrst set at the average value and then at the average value plus one standard deviation. Dummy variables, for example, residence in West Germany, are set at 0 and at 1, respectively. The difference in the predicted probability for each variable set at two different values is then divided by the baseline probabilities. Tables 8.7 and 8.8 display changes of the propensity to migrate measured in percent of the baseline probabilities, thus making the relative impact of a variable comparable across different outcomes and across sub-samples. The results of the simulation conﬁrm mostly what one would expect. If we have a human capital theory a` la Sjaastad (1962) in mind, we would expect 16

Burda et al. (1998) follow a similar approach in their analysis of the intention to migrate from East to West Germany. 17 See Appendix C8 for a short description of the probit model.

Qualitative Aspects of Migration

195

Table 8.7: Simulation results (women). Variable Baseline probability

Yes, very much Yes, it depends Probably not

Never

0.122

0.340

0.359

0.178

23.180 0.779

20.486 0.302

0.057 20.140

1.705 20.755

20.116

20.042

0.019

0.082

Individual data Age: 38 vs. 52* Residence: West German vs. East German*** Married to German partner vs. not married** Married to foreign partner vs. not married*** Children 0– 6 vs. none*** Children 7– 16 vs. none

0.507

0.119

20.086

20.214

20.412 20.064

20.172 20.024

0.082 0.013

0.429 0.056

Education Secondary vs. elementary*** Higher secondary vs. elementary*** University degree vs. none** Occupational training vs. none

0.339 0.771 0.044 20.039

0.140 0.263 0.010 20.014

20.065 20.150 20.020 0.008

20.343 20.612 20.052 0.033

Occupation Unemployment rate: 12% vs. 16% Worker vs. not Self-employed vs. not** Trainee vs. not Employee vs. not** Civil servant vs. not Unemployed vs. not

20.061 20.026 0.386 0.072 0.143 20.119 0.012

20.023 20.010 0.125 0.027 0.052 20.050 0.004

20.013 0.005 20.079 20.015 20.029 0.024 20.002

0.054 0.025 20.279 20.063 20.120 0.121 20.010

20.059

20.020

0.011

0.042

0.006

0.002

20.001

20.004

20.596

20.266

0.103

0.652

Income Middle net income vs. low net income High net income vs. low net income None (retired) vs. low net income***

Reference categories are elementary degree, East German residence, not married and low net income. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level, respectively (see Table B8.1— most extensive estimation).

the propensity to move to decrease with age as the shorter payoff period of the human capital investment decreases the net gains of migration—at least if migration is considered for economic reasons.18 The coefﬁcients for the age variables are signiﬁcant at the 10% level in the female sub-sample but not in 18

But the human capital theory cannot explain migration after retirement. For those individuals who think about emigrating in order to spend their years as retirees abroad, the willingness to migrate is supposed to increase with age.

196

Unfunded Pension Systems: Ageing and Migration

Table 8.8: Simulation results (men). Variable Baseline probability Individual Data Age: 38 vs. 52 Residence: West German vs. East German*** Married to German partner vs. not married** Married to foreign partner vs. not married*** Children 0– 6 vs. none*** Children 7– 16 vs. none

Yes, very much

Yes, it depends

Probably not

Never

0.130

0.388

0.353

0.129

20.433 0.670

20.153 0.227

0.116 20.195

0.519 20.759

20.113

20.037

0.025

0.088

0.966

0.140

20.183

20.328

20.226 20.074

20.078 20.024

0.070 0.023

0.247 0.074

Education Secondary vs. elementary*** Higher secondary vs. elementary*** University degree vs. none Occupational training vs. none

0.255 0.467

0.088 0.144

20.078 20.141

20.285 20.455

0.016 20.054

0.003 20.017

20.008 0.016

20.020 0.051

Occupation Unemployment rate: 12 vs. 16% Worker vs. not*** Self-employed vs. not Trainee vs. not*** Employee vs. not Civil servant vs. not* Unemployed vs. not**

0.046 20.336 0.249 20.334 20.050 20.239 20.244

0.014 20.108 0.056 20.107 20.013 20.072 20.087

20.014 0.103 20.071 0.103 0.015 0.073 0.076

20.043 0.347 20.172 0.344 0.041 0.228 0.278

0.072

0.024

20.022

20.073

0.183

0.056

20.056

20.172

20.053

20.019

0.016

0.059

Income Middle net income vs. low net income High net income vs. low net income None (retired) vs. low net income

Reference categories are elementary degree, East German residence, not married and low net income. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level, respectively (see Table B8.1—most extensive estimation).

the male sub-sample. As to the marginal effect, an increase in the age of the individual increases the probability of staying in Germany and decreases the probability of migrating. Compared to a 38-year-old, the alternative “never” becomes 171% more likely for a 52-year-old woman and 52% more likely for a 52-year-old man whereas the probability for the alternative “yes, very much”

Qualitative Aspects of Migration

197

decreases by 318% for women and 43% for men.19 The rather low signiﬁcance of the effects hint at other potential reasons for thinking about migrating which are not captured by the human capital theory, e.g. joining friends and family members or emigrating in order to spend the years as retirees abroad. As we focus on emigration from an industrialised country, the propensity to migrate should increase with the years of education and training (Borjas, 1996). First of all, the geographic region which makes up the relevant labour market is larger for highly educated individuals than for less educated ones. Second, highly educated individuals might be more efﬁcient at learning about employment opportunities in alternative labour markets which reduces migration costs. Last but not the least, higher education implies better knowledge of foreign languages which is an essential prerequisite for economic and social integration.20 The extent to which human capital is transferable from the home country to the destination country depends in many cases on general communication skills. The simulation shows that the signiﬁcant school and university variables have effects in the expected direction. Holding a “higher secondary degree” increases the probability for the “yes, very much” alternative by 77% (47%) in the female (male) sub-sample and decreases the probability for the “never” alternative by 61% (45%) compared to an elementary degree. Holding a university degree has a small signiﬁcant effect on the propensity to migrate for women, but does not have a signiﬁcant effect for men. Occupational training, however, is without any signiﬁcant inﬂuence in both sub-samples. The occupational situation plays a signiﬁcant role for a “self-employed” and for an “employee” in the female sub-sample and for a “worker”, a “trainee”, a “civil servant”, an “unemployed” and a “self-employed”—albeit at the 11% signiﬁcance level—in the male sub-sample. Being self-employed positively inﬂuences the probability to migrate for women and men. It seems, therefore, that the entrepreneurial spirit of the self-employed outweighs the counter-arguments brought forward by Naskoteen and Zimmer (1980).21 For civil servants and unemployed individuals in the male sub-sample, the effect is signiﬁcant and negative, which implies that both groups are less likely to migrate in order not to

19 This huge effect is partly caused by the fact that a change in age by 14 years (one standard deviation) is considered. 20 See, for example, for the relevance of language skills for social integration Chiswick and Miller (1995) and for economic integration Dustmann (1994). 21 Naskoteen and Zimmer (1980) argue that self-employment should lead to a smaller propensity to move as the self-employed are less susceptible to promotion opportunities.

198

Unfunded Pension Systems: Ageing and Migration

lose the safe job at the government or the claims of the unemployment insurance, respectively. Apparently, unemployed individuals are afraid to forego their insurance claims, although it might be worthwhile to consider migrating if the probability of ﬁnding a suitable job abroad, i.e. the expected income abroad, is sufﬁciently high. The level of the wage income does not signiﬁcantly inﬂuence the propensity to emigrate, the only exception being no income (retired) in the female sub-sample, which has a negative effect. As to the private environment, the partner variable for those who are married to a German should have a negative impact on the propensity to move abroad as it is both partners together or the family as a whole who must gain by migrating.22 Moving with the partner or the family—especially when there are children—induces higher migration costs as all members of the family incur monetary and non-monetary costs when trying to adapt to a foreign environment. In contrast, those with a foreign partner should show a higher propensity to emigrate, which implies that for those couples the migration costs are lower. We ﬁnd that being married signiﬁcantly inﬂuences the propensity to migrate in both sub-samples. With a German partner, women (men) are 12% (11%) less likely to think about migration “very much” compared to being not married. A foreign partner, however, increases the probability to consider migrating “very much” by 51% (97%) for women (men). This shows that moving with a partner makes it more difﬁcult and, therefore, less likely than moving alone—at least when the partner is German; but this conclusion is reversed when the partner is foreign. Children in the household have a signiﬁcant effect in the expected direction in both sub-samples—though this effect is almost twice as large for small children of 0 –6 years in the female sub-sample. This shows that the mobility of women is more affected by family ties. What is quite surprising at ﬁrst sight is the signiﬁcance of living in the western part of Germany in both sub-samples and the impact this variable has on the propensity to migrate. If an individual lives in West Germany, the probability for the alternative “yes, very much” will increase by 78% for women and by 67% for men, whereas the probability for the alternative “never” decreases by 76% for women and for men. One explanation for this phenomenon could be that more mobile individuals from East Germany have already migrated either to the West or to a foreign country or—to put it differently—that there is a selection bias regarding the mobility of individuals

22

See Mincer (1978) for an analysis of migration decisions of families.

Qualitative Aspects of Migration

199

who still live in East Germany. The state unemployment rate, however, has no signiﬁcant effect.

8.3. CONCLUSION In order to shed some light on a so far rather neglected aspect of migration, namely emigration from an industrialised country by natives, we determine the characteristics of an individual and his or her environment which positively or negatively inﬂuence the propensity to migrate. It is important to know more about the qualitative aspects of emigration, especially with respect to the growing integration of the countries of the European Union and the consequences of migration for unfunded pension systems. We have seen in Chapter 3 that net migration is not an important phenomenon for countries of the European Union and is not expected to increase over the next 50 years. There might be more intra-EU migration at the expense of immigration from non-EU countries but without a signiﬁcant increase in total migration volumes. It is, however, possible that the low level of migration is sufﬁcient to put pressure on the unfunded pension systems in the EU countries if those individuals who are mostly needed to carry the pension burden start emigrating. We have, therefore, focused the analysis on providing some indications about the qualitative aspects of emigration from Germany, i.e. who is most likely to emigrate. What the analysis has shown is that older men and women are less likely to migrate. In the light of an older society, this effect would reduce the total mobility of Germans in the future. However, another result of the analysis is that children in the household reduce the propensity to migrate. Fewer children in the future would thus lead to a more mobile society. Besides, a rise in the educational levels for more recent cohorts—something which we have not explicitly dealt with in our estimations—would also increase mobility. Which effect dominates is a priori difﬁcult to say. What can, however, be derived from the analysis concerns the characteristics of the probable emigrants: young and with an above-average school level. But these individuals are exactly those who are needed as contributors for the unfunded pension systems. There is the argument that if the young and well educated individuals in other European countries show comparable migration behaviour we will merely observe an exchange of mobile individuals with similar characteristics. This argument, however, is only true if there is no competition for migrants between countries. As the young and well-educated are particularly mobile and are

200

Unfunded Pension Systems: Ageing and Migration

particularly needed in all countries as contributors, there will be much competition over precisely these individuals. Thus, even though the effect of ageing on total mobility is ambiguous, it is quite clear that emigration of the young and well educated will exert pressure on the German pension system and in a similar way on pension systems in other EU countries if it is not possible—as legally prohibited for migration within the European Union—to discriminate between immigrants and the local population. As Sinn (1998, p. 117) has stated, “marginal mobility and non-discrimination are sufﬁcient to produce strong competitive pressure on the institutions concerned”. There is always the illusion of being able to attract more young and well-educated individuals by scaling down the pension system and reducing thus in a twofold way the burden for contributors, i.e. through a reduced system and more contributors, without hurting the pensioners too much, thanks to the increased number of contributors. And again, immigration and emigration are just different sides of the same coin as the attempt to attract young and highly skilled immigrants is identical to the ﬁght to hold back the potential emigrants—both leading to a likely “race to the bottom” of inter- and intragenerationally redistributive activities.23 In Chapter 9, we will focus on the possible competition for mobile individuals among European countries which might be intensiﬁed by the growing integration within the European Union. The effect of migration on the ﬁnancial sustainability of unfunded pension systems depends on the distribution of competencies at the national and European level. We will, therefore, analyse in detail intra-European migration with respect to efﬁciency and distribution against the background of European regulations.

APPENDIX A8. DESCRIPTIVE STATISTICS See Table A8.1.

APPENDIX B8. PROBIT ESTIMATION See Tables B8.1 and B8.2.

23

Immigrants from non-European countries may react even more strongly to differences in the economic conditions in individual European countries. Their (differential) mobility is very high as they need to choose only the destination country given that they have already decided to move (Sinn, 1998).

Qualitative Aspects of Migration

201

Table A8.1: Descriptive statistics. Variable

Dummy

Women

Men

Mean

SD

Mean

SD

Individual data Age Residence: West Germany Married Married to German partner Married to foreign partner Children 0– 6 Children 7– 16

£ £ £ £ £ £

38.842 0.677 0.596 0.587 0.009 0.170 0.267

14.099 0.468 0.491 0.492 0.094 0.375 0.442

38.323 0.677 0.579 0.572 0.006 0.155 0.238

13.408 0.468 0.494 0.495 0.080 0.362 0.426

Educationa Elementary degreeb Secondary degree Higher secondary degree University degree Occupational training

£ £ £ £ £

0.323 0.452 0.225 0.111 0.707

0.468 0.498 0.418 0.314 0.455

0.346 0.347 0.308 0.189 0.718

0.476 0.476 0.462 0.391 0.450

Occupation Worker Self-employed Trainee Employee Civil servant Unemployed

£ £ £ £ £ £

0.093 0.034 0.045 0.398 0.031 0.094

0.291 0.180 0.208 0.490 0.174 0.291

0.274 0.063 0.053 0.309 0.071 0.079

0.446 0.242 0.225 0.462 0.257 0.270

Incomec Low net incomeb Middle net income High net income None (retired)

£ £ £ £

0.206 0.446 0.234 0.115

0.404 0.497 0.423 0.319

0.230 0.436 0.240 0.094

0.421 0.496 0.427 0.292

12.461

4.223

12.475

4.235

Unemployment rate (State level) No. of observations

9995

9612

Source: Waves of 1993, 1996, 1997 and 1998 of the GSOEP—except for Statistisches Bundesamt (1994, 1999a) for the unemployment rate. a Variables indicating the highest degree obtained by the individuals—corresponding in the German system to “Hauptschule”, “Realschule” and “Fachoberschule/Gymnasium”, respectively. b Omitted in the estimation to avoid multicollinearity. c Low net income referring to the ﬁrst quartile of the wage distribution and high net income to the fourth quartile.

202

Variable

Coefﬁcient

Age Age2 Age3

20.080*** 0.001 20.000***

0.026 0.001 0.000

20.096** 0.002** 20.000**

0.027 0.001 0.000

20.051* 0.001* 0.000**

0.027 0.001 0.000

Education Secondary degree Higher secondary degree University degree Occupational training

0.207*** 0.427*** 0.141** 20.009

0.040 0.052 0.062 0.041

0.199** 0.418** 0.126* 20.028

0.040 0.053 0.063 0.042

0.197*** 0.387*** 0.128** 20.021

0.041 0.061 0.064 0.042

20.051***

0.004

20.008 0.106* 0.152* 0.106* 0.319** 0.075 0.215**

0.008 0.048 0.069 0.053 0.077 0.079 0.036

20.009 0.006 0.040 20.015 0.194** 20.073 0.078**

0.008 0.049 0.070 0.055 0.080 0.083 0.038

Occupation Unemployment rate (state level) Unemployed Trainee Worker Self-employed Civil servant Employee

Std. error

Coefﬁcient

Std. error

Coefﬁcient

Std. error

Unfunded Pension Systems: Ageing and Migration

Table B8.1: Parameter estimates (women).

Individual data West German residency Married to foreign partner Married to German partner Children 0– 6 Children 7– 16

0.443**

0.074

Income Retired Middle net income High net income

Pseudo R2 Log-likelihood

22.788 21.673 20.514

0.222 0.221 0.221

0.051

22.077 20.952 0.216

0.255 0.255 0.255

0.057 212,235.948

0.076 0.122 0.038 0.041 0.033

20.362*** 20.028 0.003

0.085 0.048 0.071

21.046 20.563 0.612

0.283 0.283 0.283

0.062 212,164.948

212,101.354

***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level, respectively. Reference categories are elementary degree, East German residence, not married and low net income.

Qualitative Aspects of Migration

l1 : threshold for probit l2 : threshold for probit l3 : threshold for probit

0.430*** 0.321*** 20.098** 20.256*** 20.036

203

204

Variable

Coefﬁcient

Std. error

Coefﬁcient

Std. error

Coefﬁcient

Std. error

Age Age2 Age3

20.037 0.001 0.000*

0.025 0.001 0.000

20.053* 0.001 20.000*

0.026 0.001 0.000

20.019 0.000 20.000

0.027 0.001 0.000

Education Secondary degree Higher secondary degree University degree Occupational training

0.168*** 0.354*** 0.054 20.050

0.040 0.048 0.052 0.044

0.159** 0.296** 0.048 20.041

0.041 0.050 0.054 0.045

0.153** 0.261** 0.035 20.031

0.041 0.052 0.057 0.045

20.039***

0.004

0.007 20.111 20.210** 20.169** 0.146* 20.093 0.003

0.008 0.063 0.071 0.054 0.074 0.072 0.053

0.006 20.153* 20.196** 20.197** 0.120 20.135 20.026

0.008 0.064 0.075 0.055 0.075 0.077 0.055

Occupation Unemployment rate (state level) Unemployed Trainee Worker Self-employed Civil servant Employee

Unfunded Pension Systems: Ageing and Migration

Table B8.2: Parameter estimates (men).

Individual data West German residency Married to foreign partner Married to German partner Children 0– 6 Children 7– 16

0.435**

0.074

Income Retired Middle net income High net income

Pseudo R2 Log-likelihood

22.175 21.023 0.218

0.236 0.236 0.236

21.755 20.594 0.659

0.275 0.275 0.275

0.082 0.150 0.041 0.041 0.033

20.032 0.042 0.103

0.098 0.054 0.081

21.380 20.214 1.045

0.280 0.280 0.280

0.037

0.042

0.045

211,678.270

211,607.723

211,572.045

***, ** and * denote statistical signiﬁcance at the 1, 5 and 10% level, respectively. Reference categories are elementary degree, East German residence, not married and low net income.

Qualitative Aspects of Migration

l1 : threshold for probit l2 : threshold for probit l3 : threshold for probit

0.393** 0.559** 20.099* 20.139** 20.044

205

206

Unfunded Pension Systems: Ageing and Migration

APPENDIX C8. PROBIT MODEL Let ypij be a continuous, latent variable which represents the attitude towards emigration of individual i in the region j: Assume that ypij is a linear function of Xij and Zj ; parameters b and g; and a stochastic term uij such that ypij ¼ X 0ij b þ Z 0j g þ uij

ðC8:1Þ

The unobserved variable ypij is represented by a variable yij which is related to the four categories in the following way 8 1; iff ypij # l1 > > > > > < 2; iff l1 , ypij # l2 yij ¼ ðC8:2Þ > > 3; iff l2 , ypij # l3 > > > : 4; iff ypij . l3 ; where ll ; l ¼ 1; 2; 3; are unobservable thresholds to be estimated. With the standard normal distribution F; the probabilities for an individual to be part of the four categories are given by Prðy ¼ 1Þ ¼ Fðl1 2 X 0 b 2 Z 0 gÞ; Prðy ¼ 2Þ ¼ Fðl2 2 X 0 b 2 Z 0 gÞ 2 Fðl1 2 X 0 b 2 Z 0 gÞ; Prðy ¼ 3Þ ¼ Fðl3 2 X 0 b 2 Z 0 gÞ 2 Fðl2 2 X 0 b 2 Z 0 gÞ;

ðC8:3Þ

Prðy ¼ 4Þ ¼ 1 2 Fðl3 2 X 0 b 2 Z 0 gÞ: For all probabilities to be positive, we need

l1 , l2 , l3 :

ðC8:4Þ

CHAPTER 9

Sustainability of Pension Systems with Systems Competition National and European responsibilities for old-age security with (partially) mobile labour

Article 8a of the Treaty gives every citizen of the Union the right to reside freely within the territory of the Member States. […] The challenge now is therefore to create a real European mobility area […]. European Commission (1994, p. 26) A Europe with competing tax systems and unrestricted migration would be like an insurance market where the customers can select their company and pay the premium after they know whether or not a loss has occurred. Hans-Werner Sinn (1990, p. 502) Abstract If national pension systems are not sufﬁciently harmonised within the European Union, the principle of free movement of labour endangers their ﬁnancial sustainability while the low degree of harmonisation distorts the allocation of labour. A systems competition between the Member States over contributors, which would intensify the problems of sustainability and efﬁciency, must be avoided. First, we abstract from the possibility of compensating transfers. We use the results known from the theoretical literature based on the employment principle to evaluate the realised institutional distribution of competence between the European Union and the Member States and the consequences for the allocation of labour and the level of redistribution. Second, we discuss alternative options.

208

Unfunded Pension Systems: Ageing and Migration

So far, the degree of harmonisation is sub-optimally low. Rules the binding nature of which the European Union would have to guarantee are needed. Although the European Union has gained competence in the area of social policy in recent decades, Member States are still far away from conceding more fundamental responsibilities to the European level. So far, the total volume of intra-European migration is rather low. With further European integration and thus lower barriers to migration, this is expected to change. The projections of Eurostat make one think that the increase in intra-European migration will be at the expense of immigration from non-EU countries.1 But even if the total volume of migration will not change by much, mobility of those most needed as contributors to the unfunded pension systems, i.e. the young with an above-average school level, might be enough for countries to engage in competitive activities in order to attract these individuals. We have seen in Chapter 7 how (potential) mobility of the young can secure the survival of intergenerational redistribution by acting as a counterforce against gerontocratic structures. A redistributive system can be maintained when the level of mobility is controlled by the old, i.e. the young are mobile conditional on the old granting them mobility. The old use mobility as a strategic instrument to commit credibly to a certain level of redistributive taxation. But when the level of mobility is determined by factors out of reach for the old, the results are substantially different. Then, mobility is associated with fears that it endangers redistribution. An unhealthy competition among countries over the mobile individuals might initiate a “race to the bottom” of old-age security systems.2 The realisation of one of the most fundamental goals of the European Union, i.e. free labour mobility as part of the common internal market, has consequences for efﬁciency and redistribution. If mobility is not distorted due to legal or real impediments, a higher degree of mobility allows a more efﬁcient allocation of labour across Member States which increases the national incomes and thus the resources available for redistribution. However, this is only the case if redistributive activities are sufﬁciently harmonised. Otherwise efﬁciency is not reached while, in addition, mobility reduces the scope for redistribution by putting pressure on the national ﬁscal systems in general and on the national pension systems in particular (Sinn, 1998). These two arguments make 1

See Chapter 3. For an analysis of minimum standards as the central instrument of co-ordination in the area of social policy in reaction to the danger of “social dumping”, see Kolmar (2000). 2

Sustainability of Pension Systems with Systems Competition

209

it clear that economically reasonable levels of harmonisation of the national pension systems are needed to guarantee allocative efﬁciency and to enable redistributive activities. It is especially important to avoid competition among countries in areas where the state should step in on the national level according to the “selection principle” because a market failure has been identiﬁed (Sinn, 1997, 2003a). Introducing competition on the European level cannot lead to an efﬁcient outcome if old-age security fulﬁls the criteria for an intervention of the state on the national level. A (rather) centralised solution is then the best solution. In fact, old-age security organised as unfunded pension systems involves a social contract between generations – alive and not yet born. This contract, which has been concluded on behalf of the individual by its parents or grandparents, would not have any binding effect in the private insurance market. An intervention of the state is thus necessary to guarantee the fulﬁlment of the social contract.3 This intervention can take place in different ways. In this chapter, we analyse the institutional distribution of competence between the European Union and the Member States in order to evaluate the realised level of harmonisation and the consequences for the allocation of labour and the level of redistribution. Harmonisation is understood as a rather general concept. On the one hand, it comprises measures which link the national pension systems closer to one another. These measures are thus particularly suitable for systems which are similar in important respects. On the other hand, it includes measures which regulate the reassignment of migrants without modifying the national pension systems. In addition to similar systems, these measures are also well applicable for systems, which show essential differences. Both types of measures will be discussed in the following. So far, harmonisation is sub-optimally low. Free labour mobility is ensured by the mutual recognition of pension claims by the Member States. This, however, does not remove the allocative distortions and does not prevent the erosion of redistributive activities. As will be shown, the main competence for pension systems could stay with the Member States. But the European Union must have the competence to guarantee the binding nature of the respective harmonisation rules. Although the European Union has gained responsibilities in the area of social policy in the last decades, the Member States are reluctant to concede further, more fundamental responsibilities. A more courageous approach leading to a more pronounced involvement of the European level is, however, needed.

3

We abstract here from discussing the optimality of the initial conclusion of this contract.

210

Unfunded Pension Systems: Ageing and Migration

In what follows we assume that the European Union is a closed economy and abstract from interactions with non-EU countries via, e.g. external migration ﬂows. We focus exclusively on the ﬁrst pillar of pension systems, i.e. on mandatory, publicly organised unfunded systems. We start from a status quo where social security is based on the employment principle in section 9.1. Increasing labour mobility endangers the ﬁnancial sustainability of nationally organised unfunded pension systems if no form of harmonisation is realised. For pension systems which are similar in important aspects, section 9.2 analyses the optimal level of harmonisation while section 9.3 looks at the effective level of harmonisation. In section 9.4 we evaluate these empirical ﬁndings on the basis of the theoretical results. Section 9.5 discusses alternative options for harmonisation which are also suitable for systems which differ in important ways. Section 9.6 concludes.4

9.1. STATUS QUO It is a well-known fact from the traditional literature on ﬁscal federalism that for efﬁciency reasons, redistribution policy should be best allocated at a governmental level which has a geographical scope that coincides with the relevant factor markets. With increasing labour mobility, redistributive activities should consequently be shifted more and more to European institutions in order to guarantee an efﬁcient allocation of labour within the borders of the European Union (Krieger, 2001). The reason for this conclusion is that in determining their pension systems, countries exert externalities on other countries which they do not take into account.5 If country f reduces the net wage by extending the pension system, this leads to a reallocation of labour from country f to other countries of the Union until the net wages are again equalised across countries—even though at a lower level. But country f only considers the outﬂow of labour, i.e. contributors to the pension system, and neglects the inﬂow of labour in all other countries. This inﬂow, however, generates welfare gains in the recipient countries as it increases the sum of the contributors and thus the sum of the contributions paid to the pension systems. This reduces the costs for extending the pension system in the other countries or alternatively allows these countries to maintain their present systems at lower costs—both corresponding to lowering the implicit taxes. The resulting level of redistribution, which is too low because of the positive

4 5

This chapter is an extensive and enlarged version of Uebelmesser (2003b). See Zodrow and Mieszkowski (1986) for a discussion of similar externalities with mobile capital.

Sustainability of Pension Systems with Systems Competition

211

ﬁscal externalities, can only be avoided by some form of harmonisation which results in an internalisation of these externalities. One solution would be to shift the responsibilities for redistribution to a European institution (Sinn, 1998).6 This chapter analyses in detail whether a centralised pension system is really necessary for allocative efﬁciency and redistributive activities or whether more decentralised forms of harmonisation are sufﬁcient. In general, efﬁciency and redistribution are not as different as they might seem at ﬁrst sight. Apart from implying more social peace and less crime which allows individuals to pursue their activities in a more efﬁcient way, the welfare state has two more important implications for efﬁciency (Sinn, 2000b). First, the welfare state via its redistributive elements provides insurance against risks that cannot be insured by the private insurance market. Second, this insurance encourages individuals to take more risks which in total increases the national product. Redistribution and efﬁciency should not be regarded as opposites. In our context, the challenge is, therefore, to determine the level of harmonisation which allows the Member States to best achieve efﬁcient allocation of labour without eroding intergenerational redistribution.

9.1.1. Implicit taxes in an international context A simple diagram is sufﬁcient to illustrate the basic consequences of insufﬁciently harmonised pension systems for efﬁciency and distribution in a static general equilibrium context with labour mobility (Wildasin, 1999). Figure 9.1 shows the allocation of the total population N ¼ Nf þ Ng between the two countries f and g where we assume that all individuals in both countries work, i.e. N ¼ L and Ni ¼ Li for i ¼ f ; g where L denotes the labour force. In the absence of any public intervention, with equal goods prices and without migration costs, individuals will migrate until gross wages are equal in both countries. With competitive labour markets, this implies equality of the marginal productivities of labour Wi ; i ¼ f ; g: The resulting allocation of labour with Nfp individuals in country f and N 2 Nfp individuals in country g is efﬁcient. Now, we add activities of the public sector to our analysis. Wildasin (1999) deﬁnes the difference between the beneﬁts an individual receives and the payments an individual has to make as the net ﬁscal beneﬁt (per year) from residing in one country i, Gi : If beneﬁts are different from payments, the system 6

See also Dre`ze (2000), who promotes the idea of an integrated Union-wide old-age security system. Beside this, he also considers lump-sum or matching grants from the Union to the Member States which we, however, will not discuss in our analysis.

212

Unfunded Pension Systems: Ageing and Migration

Figure 9.1: Distortion of the allocation of labour due to implicit taxes.

creates a wedge between gross wages, i.e. the marginal productivities, and net wages. Depending on the sign of the net ﬁscal beneﬁt, the Wi curves have to be shifted up or down by Gi : As individuals base their migration decision on net wages taking the net ﬁscal beneﬁts into account, the allocation of labour is only efﬁcient if Gf ¼ Gg : If net ﬁscal beneﬁts differ across countries, Gg – Gf ; the allocation of labour is distorted, Nf£ – Nfp leading to an efﬁciency loss (shaded triangle). As we are interested in how differently attractive unfunded pension systems inﬂuence the migration decision—and as migration decisions imply a comparison of the present value of lifetime net income in countries f and g, we reinterpret the static model depicted in Figure 9.1 in present-value terms, thus allowing for a multi-period perspective.7 If we abstract from other taxes and transfers and concentrate only on possible distortions due to the unfunded pension systems, Wi then presents the present value of the lifetime gross wage income an individual can earn in country i and Gi stands for the total implicit taxes of the unfunded pension system in country i. The lifetime net wage income in country i, 7

See Sjaastad (1962) for an early version of the human capital model which this migration model is based on.

Sustainability of Pension Systems with Systems Competition

213

Wi 2 Gi ; thus depends on the country-speciﬁc productivity as well as on the country-speciﬁc taxes implicit in the unfunded pension system. For this, we assume that the membership in the national pension systems follows the employment principle. Positive or negative implicit taxes distort the migration decision while at the same time—as will become clear later—migration inﬂuences the size of the implicit taxes. This interpretation of the net ﬁscal beneﬁt allows us to argue again within the concept of implicit taxes. First, however, we look at calculations of the implicit taxes for some European countries to get an idea about the order of magnitude of this distortion. 9.1.2. Calculations for European countries Wildasin (1999) has calculated implicit taxes for the pension systems of seven European countries. Individuals are assumed to earn the mean wage in each country, with their lifetime earning growing over time to take account of the effect of experience on wages. All individuals start working at age 20 and retire when they are entitled to receive full retirement beneﬁts. Individuals are either single or married with a non-working spouse and no children. To simplify the analysis, spouses are assumed to die at the same age as the individual. Calculations are presented for individuals of age 20 and 40. Table 9.1 displays implicit taxes over the life cycle of individuals in selected European countries depending on their age and their marital status. Given the many assumptions needed to arrive at these results, the absolute values might be less informative than the relative differences across countries. What can, however, be said when looking at individual countries is that the present value of

Table 9.1: Implicit taxes by age and marital status (in euros and as a percentage of lifetime wealth). Country Belgium Denmark Germany France Italy Luxembourg Netherlands

Single age 20

Married age 20

Single age 40

Married age 40

230,152 (213) 211,438 (23) 253,059 (216) 213,634 (26) 228,698 (213) 233,543 (211) 291,018 (231)

228,224 (212) 27289 (22) 253,059 (216) 29652 (24) 228,698 (213) 233,543 (211) 287,810 (230)

220,240 (28) 23707 (21) 238,758 (211) 15,558 (8) 18,173 (9) 215,939 (25) 284,439 (228)

215,503 (26) 6701 (2) 238,758 (211) 24,786 (12) 18,173 (9) 215,939 (25) 276,772 (225)

Calculations for contribution rates and pension formulae operative in 1986 for countries with comparable data of contributions and beneﬁts. Source: Wildasin (1999, pp. 265).

214

Unfunded Pension Systems: Ageing and Migration

Table 9.2: Change of implicit taxes by age and marital status through migration (in euros and as a percentage of lifetime wealth). Migration from Germany to Belgium Denmark France Italy Luxembourg Netherlands

Single age 20

Married age 20

Single age 40

Married age 40

9134 (3) 42,906 (12) 33,394 (10) 12,631 (4) 14,115 (4) 252,672 (215)

11,942 (3) 47,055 (14) 37,375 (11) 12,631 (4) 14,115 (4) 249,464 (214)

8371 (2) 39,957 (11) 53,756 (15) 40,835 (12) 14,770 (4) 254,874 (216)

12,205 (3) 45,740 (13) 58,883 (17) 40,835 (12) 14,770 (4) 250,615 (215)

Calculations for contribution rates and pension formulae operative in 1986 for countries with comparable data of contributions and beneﬁts. Source: Wildasin (1999, pp. 268).

contributions falls short of the present value of pension beneﬁts for young individuals in all countries, ranging from a loss of 3 –31% of lifetime wealth for singles and 2 –30% for married individuals.8 The middle-aged have to carry fewer implicit taxes over the remainder of their life cycle in all countries. This effect can be explained by a combination of life-cycle and cohort effects. As we have seen in Chapters 5 and 6, lifetime implicit taxes are lower for early-born generations— and even negative for the introductory generations—and higher for later-born ones. At the same time, implicit taxes decrease over the life cycle, which implies higher taxes for younger individuals than for older ones. Both effects together are responsible for the differences in implicit taxes we observe for the two age-groups. It is now interesting to calculate the changes in implicit taxes which result from migration to another country. Table 9.2 shows how the implicit taxes change for a 20- and 40-year-old who migrates from Germany to one of the other six countries. It is assumed that the individuals retain the mean wage from their country of origin.9 With the exception of the Netherlands, Germans who migrate to any one of the other countries reduce their implicit tax burden quite substantially. Again, it must be noted that these results should not be used to derive the exact value of the implicit tax wedge and of the gains from migrating since the calculations are based on contribution rates and pension formulae operative in 1986. The exercise is, however, useful to get an idea of the order of magnitude of 8

The implicit tax of 16% of lifetime wealth for young individuals in Germany is higher than the implicit tax calculated with the pension model developed by CESifo for the Council of Advisors of the German Ministry of Economics (Wissenschaftlicher Beirat, 1998). For the birth cohort of 1966, the implicit tax rate amounts to approximately 40% of total contributions (see Chapter 6). If we assume for simplicity an average contribution rate of 20%, we get an implicit tax of 8% of total lifetime income. 9 This explains why the results in Table 9.2 for the 20-year-old cannot be obtained exactly by subtracting the respective values in Table 9.1.

Sustainability of Pension Systems with Systems Competition

215

the pension burden relative to lifetime wealth. It can easily be seen that implicit taxes of this magnitude—together with the explicit taxes from which we have so far abstracted—might affect the decision to migrate. Depending on how sensitive migrants are with respect to these tax differences, an efﬁcient allocation of labour is then very improbable.

9.1.3. Discussion We now use the theoretical and empirical insights of the last sections to illustrate the incentives of countries to attract immigrants which may result in the “race to the bottom” of intergenerational redistribution. We assume a ﬁxed-beneﬁt pension system. Thus, demographic changes only affect the working generations, but not the retirees. If due to ageing in the form of higher life-expectancy the implicit taxes increase in country f, this induces individuals to emigrate; but fewer people in country f lead to lower contributions to the pension system. To balance the pension budget, country f must either raise the contribution rate or lower the pension beneﬁts or both. Both actions might induce even more migration by the age-groups who are negatively affected. This shows that a country with a pension system with a high implicit tax experiences emigration, while a country with a low implicit tax is characterised by immigration. Without any form of harmonisation, these migration trends reinforce themselves. Following this argumentation, countries face incentives to reduce contribution rates in the ﬁrst place in order to initiate this migration process in their favour. Countries which want to avoid the occurrence of migration at their expense follow suit. In the end, we can observe a “race to the bottom” with respect to oldage security.10 It is important to note that from the perspective of allocative efﬁciency, the development might be no reason for intervention. Complete elimination of national pension systems also eliminates the distortions of the migration decision due to implicit taxes. The partial or complete abolition of an existing unfunded pension system as a reaction or equivalently the transition to a funded pension system is, however, not Pareto-improving if there are no idiosyncratic and static inefﬁciencies which originate from the unfunded pension system.11 It is not

10

See also Begg et al. (1993). Compare the analysis in Chapter 5 and the discussion of the implications of the welfare state for efﬁciency above. 11

216

Unfunded Pension Systems: Ageing and Migration

possible to ﬁnd an abolition or transition path which makes all generations better off. Either the old generations lose their pension beneﬁts or the young have to contribute without acquiring own claims or both. The mix of funded and unfunded elements in the pension system which results from ﬁscal competition can, therefore, not be judged on the basis of efﬁciency, but is essentially a question of intergenerational fairness. This criterion can be operationalised for the question of interest in this analysis by looking at the documents which constitute the European Union and its preceding organisations.12 The objectives mentioned reﬂect the will of the Member States and their citizens. Although there is no explicit statement concerning the protection of national old-age security systems, it is possible to ﬁnd indirect indications against an abolition of unfunded pension systems at the expense of a few present (or future) generations. Concentrating the transitional burden on some generations is certainly in conﬂict with “[t]he raising of the standard of living and quality of life” for all generations which has been a priority for the Member States since the very beginning (Art. 2 ToR13).14 The development of social policy issues has, however, been pursued with varying intensity over the last decades.15 The ﬁrst concrete initiatives in the 1960s and 1970s to foster social integration of the Member States were mainly motivated by the objective to reduce barriers to migration. The focus was thus on the (technical) efﬁciency of the allocation of labour. This started to change in the 1980s. The necessity to give equal priority to economic and social objectives ﬁnally let to the promulgation of the Charter of the Fundamental Social Rights of Workers, the so-called Social Charter in 1989 and was reafﬁrmed by the inclusion of the Social Policy Agreement as part of the Social Policy Protocol in the Treaty of Maastricht (1993). This shows the growing conviction of the Member States concerning the importance of social policy for ensuring a high and equal standard of living. The economic objective of technical efﬁciency has thus been complemented by the social objective of a reasonable level of protection for employees.16

12

As preceding organisations, we understand the European Coal and Steel Community (ECSC) from 1951 and the European Economic Community (EEC) and the European Atomic Energy Community (EURATOM) both from 1958 as well as the European Community (EC) founded in 1967 by uniting the institutions of ECSC, EEC, and EURATOM. If we speak about the European Union in general, we also refer implicitly to these organisations where appropriate. 13 Treaty of Rome (1958). The dates refer to the year where the Treaty was ratiﬁed. 14 See also the Preamble to the Treaty of Rome. 15 For a more extensive discussion see section 9.3.1. 16 For further references see Atkinson (1995) and Kolmar (1999).

Sustainability of Pension Systems with Systems Competition

217

In the following, this evolution of social policy issues and the underlying will of the Member States will serve as the normative foundation of the analysis.

9.2. THEORETICAL RESULTS WHEN PENSION SYSTEMS ARE SIMILAR Despite some convergence of old-age security originating from the Beveridgean or Bismarckian tradition, public pension systems still display some differences with respect to their aim and scope. While in Anglo-Saxon countries, pension beneﬁts are mainly supposed to guarantee a uniform, basic level of social protection for old age, continental countries take the view that the beneﬁts should enable the recipients to keep their standard of living during retirement. It is evident that linking pension systems by harmonising their contribution rates and/or beneﬁt levels is very difﬁcult if this means to reconcile very different levels of size and generosity. The level of redistribution, which results, need not equal the level of redistribution most preferred by the majority of the Member States. Options, which are particularly suitable for old-age security systems which differ signiﬁcantly, are, therefore, analysed in section 9.5. But before turning to the case of different systems, we assume that countries have sufﬁciently similar public pension systems and that membership in the pension systems is according to the employment principle. We further abstract from the possibility of interjurisdictional transfers and individual payments. In order to avoid an inefﬁcient allocation of labour on the one hand without initiating a “race to the bottom” on the other hand, the national pension systems need to be harmonised. For this restricted policy space, we determine the minimum degree of harmonisation sufﬁcient for an efﬁcient allocation of labour and intergenerational redistribution given different degrees of mobility. We then use this as a benchmark to evaluate the realised level of harmonisation. Of course, complete centralisation on the European level eliminates all incentives to migrate stemming from the national pension systems thus allowing an efﬁcient allocation of labour with respect to marginal productivities. Starting from complete centralisation, we then analyse how much competence could be left with the Member States without inducing inefﬁcient migration and without initiating a collapse of the national pension systems. To put it differently, the question is which minimum level of harmonisation would be sufﬁcient for allocative efﬁciency and redistributive activities. This refers to a level of harmonisation that reduces the barriers to mobility and at the same time avoids a “race to the bottom”. The logical challenge is that a higher degree of mobility in accordance with the priority of the European

218

Unfunded Pension Systems: Ageing and Migration

Figure 9.2: Levels of harmonisation. Source: following Breyer and Kolmar (2002).

Union endangers the national pension systems in the case of insufﬁciently harmonised pension systems. In what follows, we will, therefore, analyse how pension systems should be harmonised before evaluating how they are harmonised at the moment. Figure 9.2 gives an overview of the different levels of harmonisation ranging from complete decentralisation to complete centralisation. National pension systems without any form of harmonisation are called decentralised. The weakest form of harmonisation is the mutual recognition of pension beneﬁts.17 Coordination of contribution rates leads to a stronger link between national pension systems. In such a system, an absolute level which contribution rates may neither exceed nor undercut might be deﬁned or a relative level, in relation to other countries’ contribution rates, which determines the national contribution rate. For the special case where contribution rates are equal across countries, the national pension systems are called equalised. Further harmonisation would then be a single centralised system on the European level. As has been shown graphically (Figure 9.1), implicit taxes of national pension systems distort the allocation of labour as they drive a wedge between gross wages, i.e. the marginal productivity of labour, and net wages. In the following, the condition for an internationally efﬁcient allocation of labour will be derived

17

Examples of this are the exportability of beneﬁts and the conversion of insurances periods. See EEC regulations 1408/71 and 574/72 for the details.

Sustainability of Pension Systems with Systems Competition

219

and contrasted with the migration equilibrium within the framework of a dynamic model. This will enable us to determine what level of harmonisation of the national pension systems is sufﬁcient for an efﬁcient solution. For this, we will discuss the case of unrestricted mobility as the benchmark. The ﬁnal goal of the European Union is the realisation of the four liberties—among them the free movement of labour. Barriers to mobility are consequently dismantled, which leads to a reduction of mobility costs. This development will allow efﬁcient allocation of labour but endangers the ﬁnancial sustainability of the social security systems, thus making redistribution difﬁcult. This result will then be contrasted with the more realistic case of restricted mobility. Due to some mobility costs, allocation of labour might no longer lead to equalisation of marginal productivities while redistributive activities can more easily be carried out. The challenge is then to determine the optimal level of harmonisation to allow free movement of labour and redistribution among generations.18 9.2.1. Unrestricted mobility—the benchmark As a benchmark, we ﬁrst look at the case of unrestricted mobility in all periods (Breyer and Kolmar, 2002). Let a denote the mobile part of the population. Unrestricted mobility then implies a ¼ 1: To analyse possible distortions of labour mobility due to nationally organised pension systems, we again make use of the concept of implicit taxes in order to identify and quantify these distortions. We consider a two-period overlapping-generation (OLG) model of a union with two countries f and g identical to the model developed in Chapter 2 for the onecountry case. Individuals live for two periods—a working period and a period where they are retired. They are born in either of the two countries of the union and may migrate to the other country before starting to work. The union is characterised by unrestricted mobility of the young individuals—an assumption which will be modiﬁed later. Goods and factor markets are fully integrated. O In period t; the union is inhabited by N Y t young individuals and Nt old individuals. The distribution of the young individuals over the two countries is Y;g described by ðN Y;f t ; Nt Þ; whereas the distribution of the young individuals after Y;g migration is denoted by ðN Y;f t ; Nt Þ: There is a national unfunded pension system in both countries. In the working period t; individuals in country i ¼ f ; g inelastically supply one unit of labour and receive a gross wage income Wti according to their marginal productivity and a net 18

The terms “unrestricted” and “restricted” refer to the part of the population for which mobility costs are low enough to migrate. They do not imply that there are any restrictions due to interventions from the government.

220

Unfunded Pension Systems: Ageing and Migration

wage income of Wti ð1 2 ui Þ: ui denotes contributions to the national pension system of country i which we assume to stay constant over time but which we allow to differ between the two countries. They spend their net income to consume cY;i t and to save st : In period t þ 1; individuals retire. They ﬁnance their secondO;i with their pension beneﬁts pitþ1 and their savings ð1 þ period consumption ctþ1 rtþ1 Þst ; where rtþ1 is the market rate of return. A balanced budget of the national pension system requires that in all periods t total contributions equal total beneﬁts of those who have acquired pension claims in country i during their working period t 2 1 such that NtY;i ui Wti ¼ NtO;i pit :

ð9:1Þ

We already know the intertemporal budget constraint for an individual living in country i, which equates in present-value terms lifetime consumption with net labour income and pension beneﬁts. In our case, we can write with Equation 9.1 cY t þ

cO pitþ1 tþ1 ¼ Wti ð1 2 ui Þ þ 1 þ rtþ1 1 þ rtþ1 ¼ Wti ð1 2 ui Þ þ

Rewriting, we get for country i cit

¼

Wti

2u

i

Wti

Y;i Ntþ1 ui i Wtþ1 : O;i 1 þ rtþ1 Ntþ1

! rtþ1 2 iitþ1 ; 1 þ rtþ1

ð9:2Þ

ð9:3Þ

where cit is lifetime consumption of an individual born in period t and iitþ1 is the internal rate of return of an unfunded pension system. iitþ1 can be approximated Y;i =NtY;i 2 1; where by the sum of the growth rate of the population nitþ1 ¼ Ntþ1 Y;i O;i i i Nt ¼ Ntþ1 ; and the growth rate of productivity gtþ1 ¼ Wtþ1 =Wti 2 1: In what follows, we assume the growth rate of productivity to be zero. The term in brackets is once again the implicit tax rate tit :19 If the internal rate of return falls short of the interest rate, which is empirically the case in all industrialised countries, the part tit of the contributions is lost when compared to an investment in the capital market. Output Yti in country i at period t is produced with the production function

Yti ¼ F i Kti ; Lit ð9:4Þ 19

See Chapter 2.

Sustainability of Pension Systems with Systems Competition

221

with capital Kti and labour Lit : The production function has constant returns to scale and fulﬁls the Inada conditions. It is country-speciﬁc but time-invariant. In what follows, we assume that all individuals of the young generation work, i.e. Lit ¼ NtY;i : Proﬁt maximisation leads to the equalisation of factor prices, i.e. gross wage Wti and interest factor 1 þ rt ; and the respective marginal productivities Wti ¼ FNi ðKti ; NtY;i Þ;

and

1 þ rt ¼ FKi ðKti ; NtY;i Þ;

ð9:5Þ

where FAi ðKti ; NtY;i Þ ¼ ›F i =›A; A ¼ Kti ; NtY;i : An internationally efﬁcient allocation of labour results in the largest sum of output in the union for the given amount of labour. The Lagrangean equation is L ¼ F f ðKtf ; NtY;f Þ þ F g ðKtg ; NtY;g Þ þ lt ðN 2 NtY;f 2 NtY;g Þ

ð9:6Þ

with lt as the Lagrange parameter. Maximising Equation 9.6 with respect to labour NtY;f and NtY;g yields FNf ¼ FNg , Wtf ¼ Wtg :

ð9:7Þ

Efﬁciency thus requires that the marginal productivities of labour and—according to Equation 9.5—the gross wages are equal in all countries. With unrestricted mobility, individuals migrate to the country where lifetime income is highest. While efﬁciency demands equalisation of marginal productivities, i.e. gross wages, the migration equilibrium implies equalisation of net wages: ! ! g f g g g rtþ1 2 itþ1 f f f rtþ1 2 itþ1 Wt 2 Wt u ¼ Wt 2 Wt u : ð9:8Þ 1 þ rtþ1 1 þ rtþ1 If the migration equilibrium is efﬁcient, i.e. if marginal productivities are equalised across countries, we must get ! ! g f g g rtþ1 2 itþ1 f f rtþ1 2 itþ1 ¼ Wt u : ð9:9Þ Wt u 1 þ rtþ1 1 þ rtþ1 Efﬁciency requires that implicit taxes are equal in both countries. The problem to be solved is then to determine for all relevant migration scenarios the level of harmonisation that leads to equalised implicit taxes. All distortions of the migration decision due to implicit taxes are then neutralised and an efﬁcient allocation of labour can be realised. For this to be achieved, the policy variables are the country-speciﬁc contribution rates—if we abstract from the option of a centralised pension system. Contribution rates must be set to balance differences in the population growth due to fertility differences after migration.

222

Unfunded Pension Systems: Ageing and Migration

According to Homburg and Richter (1993), a time path of distributions of people is internationally efﬁcient if either of the following two conditions is met: (1) ui ¼ 0 or (2) uf ¼ ug and nitþ1 ¼ gitþ1 ¼ 0 for i ¼ f ; g:20 This means that it is not sufﬁcient to equalise contribution rates. Along a perfect foresight path, the population growth rates—they argue—may well diverge, thus leading to different implicit taxes of the national pension systems in different countries. As Homburg and Richter (1993) consider a stationary state with nitþ1 ¼ gitþ1 ¼ 0 as “more an analytical ﬁction than a positive approach” (p. 59) and as they regard efﬁcient harmonisation outside a stationary state as rather impossible, their policy recommendations concerning the optimal level of harmonisation are twofold: one option would be a transition to a funded system (condition 1) which does not inﬂuence locational choices of individuals as its rate of return equals the interest rate r in all countries.21 Another option would be to replace the national pension systems with a European pension system (consequence from condition 2). As the internal rate of return of such a European system would equal the European growth rate of the wage sum, individuals could not change this rate of return by migrating from one European country to another. Their locational choice would thus not be inﬂuenced. Breyer and Kolmar (2002) show, however, for the restricted policy space considered here22 that the interpretation by Homburg and Richter (1993) is not correct with unrestricted mobility. The equalisation of contribution rates is sufﬁcient for an efﬁcient allocation of labour even with differences in fertility rates. In a steady-state equilibrium with perfect foresight, the differences in fertility rates are eliminated by (unrestricted) migration. It is, however, important that the pension system does not lead to additional incentives to migrate due to differences in contribution rates. Consequently, an efﬁcient migration equilibrium necessitates the equalisation of contribution rates—either at a positive level (condition 2) or at the level of zero (condition 1). The latter case, in fact, implies that there is no pay-as-you-go pension system. As Breyer and Kolmar (2002) point out, equalisation of the contribution rates is sufﬁcient and necessary for an efﬁcient allocation of labour; centralisation is not required.

20

In contrast to Homburg and Richter (1993), who consider lump-sum contributions, we have to add the condition gitþ1 ¼ 0: 21 Aspects of efﬁciency and feasibility of a transition of an unfunded pension system to a funded one are discussed in Chapters 5 and 6. 22 Note that we abstract here from the possibility of interjurisdictional transfers and individual payments.

Sustainability of Pension Systems with Systems Competition

223

9.2.2. Restricted mobility—the relevant scenarios The case of unrestricted mobility with a ¼ 1 is a helpful benchmark—though in sharp contrast to European reality.23 In order to draw conclusions for the optimal level of harmonisation, it is, therefore, necessary to analyse more realistic scenarios with restricted mobility today, i.e. a , 1: Today, the European Union is characterised by a very low level of intraEuropean migration which hints at signiﬁcant monetary and non-monetary costs of migration. Whether these impediments to migration will vanish in the near or not so near future is difﬁcult to say. It is, therefore, useful to focus on two cases which reﬂect the most probable scenarios for the European Union: restricted mobility ða , 1Þ today and either unrestricted mobility ða ¼ 1Þ or restricted mobility ða , 1Þ in the future. As we will see, the degree of mobility in the future determines the minimum requirements of harmonisation for an efﬁcient allocation of labour. 9.2.2.1. Case I: Restricted mobility today and unrestricted mobility in the future For unrestricted mobility from period t þ 1 onwards, we know from the analysis above that an efﬁcient allocation of labour requires equal contribution rates. It must only be veriﬁed whether equal contribution rates from period t þ 1 onwards are compatible with efﬁciency in period t. Breyer and Kolmar (2002) show that this is in fact the case. As a result, an equalisation of contribution rates is the minimum level of harmonisation which leads to an efﬁcient allocation of labour. 9.2.2.2. Case II: Restricted mobility today and in all future periods With restricted mobility in all periods, we have to distinguish three cases concerning the degree of mobility from periods t þ 1 onwards (Breyer and Kolmar, 2002). For the minimum level of harmonisation, the size of the mobile group relative to the differences in the fertility rate is essential. The reason for this is that the internal rates of return of the unfunded pension systems and thus the implicit tax rates in both countries depend on the growth rates of the population after migration has taken place. If the mobile group is big enough to compensate for the fertility differences ða $ aÞ; the population distribution can converge over time (Case IIa). There exists 23

See Chapters 3 and 8 for a discussion of the qualitative and quantitative aspects of intra-European migration.

224

Unfunded Pension Systems: Ageing and Migration

some period tp from which on the population growth rate in both countries is identical. This brings us back to the results for unrestricted mobility. Equal contribution rates are then necessary and sufﬁcient for an efﬁcient allocation of labour in all periods. If the mobile group is, however, not big enough to balance the fertility difference ða , aÞ; two further cases have to be distinguished. First, mobility is large enough so that the fertility difference can be compensated and the population distribution between the two countries will be stabilised on the initial level (Case IIb). Second, mobility is too small. The population distribution will diverge further (Case IIc). As in these two cases the growth rate of the population cannot be equalised by migration, equal contribution rates in both countries would not lead to a neutralisation of the country-speciﬁc implicit taxes (Equation 9.9). Breyer and Kolmar (2002) show that for these cases co-ordinated contribution rates are necessary for efﬁciency, i.e. the contribution rates must be set in a speciﬁc ratio to one another ðui ¼ gðu j Þ; i – jÞ which depends on the countryspeciﬁc population growth rates. 9.2.2.3. Summary Figure 9.3 summarises the minimum requirements of harmonisation for the cases of restricted mobility in the present period and differing degrees of mobility in

Figure 9.3: Minimum requirements for an efﬁcient allocation of labour. (Restricted mobility ða , 1Þ in the present period and different degrees of mobility in the following periods.)

Sustainability of Pension Systems with Systems Competition

225

the following periods that lead to an efﬁcient allocation of labour. The ﬁgure also includes the case of no mobility at all in the following periods ða ¼ 0Þ: Depending on the productivity and fertility differences between the two countries compared to the interest rate, this scenario can require centralisation, co-ordination of contribution rates or no harmonisation at all.24 As this scenario is not very probable, we only report the results as a point of reference without discussing them in detail. In comparing the harmonisation requirements for different degrees of mobility from period t þ 1 onwards, one should note that a growing mobile part a [ ½0; 1 of the population does not induce a continuous change of the required level of harmonisation. In contrast, equal contribution rates are necessary and sufﬁcient for an efﬁcient allocation of labour for the case of completely mobile labour. Intermediate degrees of mobility demand equal contribution rates if mobility is large enough to balance fertility differences and co-ordinated contribution rates if mobility compensates fertility differences only partially. For the case of completely immobile labour, harmonisation can require everything from centralised pension systems to completely decentralised pension systems. Different degrees of mobility require different levels of harmonisation to guarantee efﬁcient allocation of labour. If we characterise the present situation in the European Union with respect to migration as one with restricted mobility, there are two relevant scenarios to distinguish. On the one hand, mobility could be equally restricted in the following periods. On the other hand, mobility could increase and become (close to) unrestricted. For both cases, a co-ordination of contribution rates across countries—either at the same level or at a level which reﬂects the differences in the population growth rates—is sufﬁcient for efﬁciency. With these migration scenarios in mind, centralisation of national pension systems in the form of a uniﬁed European pension system is not required. No matter how long the time horizon, Wildasin (2000) expects mobility to increase but does not consider perfect mobility to be a realistic scenario. The projections of Eurostat (2000) conﬁrm this view. The total volume of net migration to countries of the European Union will even slightly decrease in the next 50 years from a total of 661,000 in 2000 to 622,000 in 2050. Even though it can be assumed that intra-EU migration will be more important in the next few decades at the cost of immigration from non-EU countries, the numbers indicate that mobility will be far from 100%. The focus so far has been on the migration decision of an individual and not of a dynasty. With a . 0; each descendant of this individual has the same option to migrate. But dynasties become relevant for the case with no mobility from the 24

For the derivation of the results, see Breyer and Kolmar (2002).

226

Unfunded Pension Systems: Ageing and Migration

second period onwards, which we have, however, regarded as rather unrealistic in the context of the European Union. Migration would then mean that the immigrant starts a new dynasty with children and grandchildren in the immigration country. As Sinn (2001) has shown, the value of an immigrant in an unfunded pension system equals his or her gross contributions as his pension beneﬁts will be ﬁnanced by the contributions of his descendants. These contributions then constitute a large positive ﬁscal externality which needs to be taken into account.

9.3. INSTITUTIONAL DISTRIBUTION OF COMPETENCE BETWEEN THE NATIONAL AND THE EUROPEAN LEVEL Even though the institutional arrangements are subject to change, it is useful to have a closer look at them. Today’s institutions and today’s distribution of competence make a certain level of harmonisation more likely to emerge. This level can then be compared to the optimal level of harmonisation derived in section 9.6 in order to determine the need for reform.25 9.3.1. The evolution of social policy issues on the European level This historical evolution of the European Union as brieﬂy discussed above shows that social policy as a by-product of economic policy or as an independent objective to complement economic integration are both anchored in the understanding of the European Union. To get a better understanding we look at the last decades in more detail. We distinguish three periods with respect to the development of social policy in the European Union and its preceding organisations. The ﬁrst period (1958 –1973) extended from the Treaty of Rome to the ﬁrst oil crisis and was characterised by high growth rates and low unemployment rates. Social considerations were of no importance. The second period (1974 –1999) was marked by the end of the postwar boom and rising unemployment rates, which increased the necessity of social policy. This period also saw the realisation of the common internal market and the 25

The theoretical results have been derived on the basis of the benevolent planner approach which neglects the fact that policies have to be supported by the majority of the people of all Member States (Krieger, 2001). We abstract from a detailed analysis of the political economy aspect of harmonising national pension systems—see Vaubel (1994) for a survey—and concentrate instead on a discussion of the institutional distribution of competence between the national and the European level. As this distribution is the outcome of a democratic process, it reﬂects the preferences of the (majority of the) populations of the Member States.

Sustainability of Pension Systems with Systems Competition

227

common monetary policy with further implications for social policy. In the third period (1999 – today), social policy in most areas plays an equally important role as economic policy.26 At the beginning, the objective of “raising of the standard of living and quality of life” was regarded as a by-product of increased efﬁciency of the economic activities within the European Union. Mobility of goods, services and factors leads to allocative efﬁciency and thus contributes to a higher standard of living. A “common market and an economic and monetary union” (Art. 2 ToR) is thus the precondition and should comprise “an area without internal frontiers in which the free movement of goods, persons, services and capital is ensured” (Art. 7a ToR). The focus of the process of European integration was on the creation of an Economic Union, and not of a Social Union. The competence for social policy stayed with the Member States (Breyer and Kolmar, 1996). In the second and third periods, it was realised that economic prosperity does not always imply social prosperity as well. The European Community has, therefore, attached increasing importance to social policy issues—thus addressing the social dimension of the economic integration more directly. Consequently, the Economic Union has been complemented step by step by a Social Union—a process that took several decades in total and led to a shift of competence from the Member States to the European level. We consider the second period as the time from 1974 up to before the Treaty of Amsterdam (1999) where social policy issues gained importance on the European level, but mostly in the form of non-binding or only partially binding programmes. The third period, from the Treaty of Amsterdam (1999) up to today, has seen a shift to more fundamental responsibilities of the European Community for social policy issues.27 One can think about this development in the following way (Figure 9.4): at the beginning of the process of the European integration, the Economic Union was the centre of attention. Social considerations were reduced to instances where the effective mobility of production factors, especially of labour, was endangered by a lack of harmonisation. The efforts to ensure the mutual recognition of pension claims of migrant workers continue until today. From about 1974 onwards, the awareness for social policy issues independent from economic considerations has increased – ﬁrst in the form of declarations of intent and since 1999 in the form of more speciﬁc competence.27 In what follows, we will look at the evolution of social considerations within the Economic Union and the Social Union in order to see how the importance—and 26

For a similar overview see Mosley (1990). See in particular Treaty of Rome (1958) and Treaty of the European Community (1993): Art. 117, 118 ToR and Art. 136, 137 ECT (Appendices A9 and B9).

27

228

Unfunded Pension Systems: Ageing and Migration

Figure 9.4: Evolution of social considerations in the European Union.

consequently the competence—of the social dimension has changed over the last few decades. We focus on social policy issues as part of the Economic Union and as part of the Social Union before and after the Treaty of Amsterdam.

9.3.1.1. Economic Union The ﬁrst initiatives to foster social integration of the Member States were motivated by the objective to reduce barriers to migration. This becomes clear when looking at the respective Articles in the Treaty of Rome where a homogeneous concept of social policy is missing (Breyer and Kolmar, 1996). European Community measures, which concern in particular old-age security, are limited to the regulation of the mutual recognition of pension claims. Art. 48 ToR (39 ECT28) states explicitly that the freedom of movement for workers has to be secured by “abolition of any discrimination based on nationality”, while Art. 8a ToR (18 ECT) extends the right of free movement to every citizen of the European Union. Art. 49 ToR (40 ECT) regulates the abolition of administrative obstacles in general. Art. 51 ToR (42 ECT) mentions measures necessary to provide freedom of movement in the ﬁeld of social security. In particular, it states that

28

Treaty of the European Community (1993).

Sustainability of Pension Systems with Systems Competition

229

The Council shall […] make arrangements to secure for migrant workers and their dependants: (a) aggregation, for the purpose of acquiring and retaining the right to beneﬁt and of calculating the amount of beneﬁt, of all periods taken into account under the laws of the several countries; (b) payment of beneﬁts to persons resident in the territories of Member States. Old-age security is thus a relevant issue for the European Union in the sense that a lack of harmonisation endangers the free movement of workers. In particular, the acquisition of pension claims and the payment of pension beneﬁts must be secured for migrant workers. The details can be found in EEC regulation 1408/71 and (implementing) EEC regulation 574/72. Both deﬁne procedures concerning the recognition of pension claims of migrant workers such that migrants and nonmigrants have to be treated equally. The interpretation of the principles stated in the two regulations lies in the responsibility of the European Court of Justice. With regard to questions of oldage security, the Court pursues the position that migrant workers must not experience disadvantages with respect to their pension claims. This implies that acquired pension claims must not be lost and that rules that restrict the accumulation of pension claims are incompatible with the freedom of movement.29 The Court itself has already stated in 1967 that “the […] regulations, regarded as a whole, are intended, in certain circumstances, to beneﬁt the migrant worker as compared with the situation which would result for him from the exclusive application of national law”.30 This position has been reafﬁrmed by the Court in 1988 by emphasising that “the rules contained in Regulation 1408/71 must guarantee to workers who move within the Community all the beneﬁts which have accrued to them in the various Member States whilst limiting them ‘to the greatest amount’ of such beneﬁts”.31 The European Court of Justice positively discriminates against migrant workers compared to non-migrant workers as for most migrant workers pension beneﬁts are higher by application of the EEC regulations 1408/71 and 574/72 than on the basis of national law (Keller, 2000). In addition to this pro-migration ruling, the European Court of Justice has extended the scope of the regulations and thus the competence of the European Union continuously in the area of social security. This concerns in particular the 29

See Judgement of the Court of 21 October 1975, case 24/75 (Petroni), European Court reports, 1975. See Judgement of the Court of 5 July 1967, case 1/67, (Ciechelski), European Court reports, 1967, section 2. 31 See Judgement of the Court of 14 March 1989, case 1/88 (Baldi), European Court reports, 1989, section 22. 30

230

Unfunded Pension Systems: Ageing and Migration

wider interpretation of the term “employee” and of the types of beneﬁts. In total, the number and the types of beneﬁts subject to European regulations rise. As these rulings are intended to stimulate intra-European migration and economic integration, von Maydell (1992) talks about the European Court of Justice as “Motor der Integration” (p. 322).32 For more than 40 years, the pension systems of the Member States have thus been linked in a way to avoid the loss of pension claims for migrant workers. This is seen as a precondition for free movement of labour. 9.3.1.2. Social Union—before the Treaty of Amsterdam Apart from these regulations, social policy at the European level—especially in the early years—was characterised by many non-binding programmes and resolutions. In 1961, the European Social Charter was signed emphasising the social dimension—but on a merely voluntary basis. Social policy remained primarily a national responsibility. In 1972, the heads of state or government emphasised for the ﬁrst time in the “Declaration of Paris” that the same importance had to be attached to social policy as to an economic and monetary union.33 In 1974, the content of this declaration was speciﬁed by the ﬁrst social action programme of the Council and complemented by the succeeding social action programme in 1984.34 Together, these programmes deﬁne the framework of the social policy of the European Community (Onur, 1993). With the adoption of these programmes, the authority of the European Community to issue regulations in the ﬁeld of social policy was accepted. The Single European Act (1987) expanded the competence of the European Community emphasising the aim “to improve the economic and social situation by extending common policies and pursuing new objectives” (Preamble). In 1988, the European Commission published the action programme “The Social Dimension of the Internal Market”, which emphasised the necessity to give equal priority to economic and social objectives. In 1989, this was agreed upon as a resolution by the European Parliament. By stating explicitly that “the adoption at Community level of the fundamental social rights” was necessary and “should not be jeopardised because of the pressure of competition or the search for increased competitiveness”, the strong position of the social component compared to the economic component was emphasised.35 This process led to the Charter of 32

Motor of integration (own translation). See General Report on Activities of the European Community (1972). 34 See OJ C 013/1974, p. 0001 and OJ C 067/1984, p. 0002. 35 OJ C 096/1989, p. 0061. 33

Sustainability of Pension Systems with Systems Competition

231

the Fundamental Social Rights of Workers, the so-called Social Charter, in 1989, which advocates minimum standards—or level co-ordination—in major areas of labour law.36 Although not binding due to the resistance of the UK, it nevertheless reafﬁrmed the conviction of the other Member States concerning the importance of social policy for ensuring a high and equal standard of living. This was once more highlighted with the promulgation of the Social Policy Agreement as part of the Social Policy Protocol of the Treaty of Maastricht (1993)—due to the persisting resistance of the UK, however, relegated to the annex. Concerning the stand the European Union has taken on a common social policy, the Treaty has again made clear the interdependence between the creation of a common internal market and the development of a social equilibrium. The Preamble to the Treaty (EUT37) stated that the Member States are determined to “promote economic and social progress” while conﬁrming at the same time “their attachment to fundamental social rights” and their resolution “to facilitate the free movement of persons”. However, the competence of the European Union in the area of social policy was mostly limited to supporting and complementing the activities of the Member States. 9.3.1.3. Social Union—after the Treaty of Amsterdam In 1999, the Social Policy Agreement was included in a slightly stronger version in the Social Chapter of the Treaty of Amsterdam (Art. 136– 145 ECTa38). This has fundamentally changed the competence of the European Union with respect to social policy issues (Eichenhofer, 2002). Until the Treaty of Amsterdam, the competence of the European Union were restricted to measures supporting a cooperation in social areas between the Member States as laid down in Art. 118 ToR: Without prejudice to the other provisions of this Treaty and in conformity with its general objectives, the Commission shall have the task of promoting close co-operation between Member States in the social ﬁeld […] To this end, the Commission shall act in close contact with Member States by making studies, delivering opinions and arranging consultations both on problems arising at national level and on those of concern to international organisations […] 36 This was also stressed as the best method to avoid unfair competition in the White Paper on Social Protection in 1994 (COM (95) 333). 37 Treaty of the European Union (1993). 38 Treaty of the European Community as amended by the Treaty of Amsterdam (1999).

232

Unfunded Pension Systems: Ageing and Migration

Since the Treaty of Amsterdam, the European Union for the ﬁrst time has legal authority, as detailed in Art. 137 ECTa: 1. With a view to achieving the objectives of Article 136, the Community shall support and complement the activities of the Member States in the [social] ﬁelds […] 2. To this end, the Council may adopt, by means of directives, minimum requirements for gradual implementation, having regard to the conditions and technical rules obtaining in each of the Member States. Such directives shall avoid imposing administrative, ﬁnancial and legal constraints in a way which would hold back the creation and development of small and medium-sized undertakings […] The Treaty includes general permission to act with respect to social policy issues if the functioning of the internal market is endangered, as well as the special permission laid down, for example, in the Art. 39 – 42 ECT concerning the freedom of movement. The incorporation of the Social Policy Agreement has substantially extended the powers of the European Union and granted legislative authority in the area of social policy. The concrete majority requirements determine the distribution of competence (Art. 137 ECTa). In some areas (equality between men and women, working conditions, etc.), the Council may adopt directives by a qualiﬁed majority under the co-decision procedure after consulting the Economic and Social Committee and the Committee of the Regions (Art. 251 ECTa). In other areas, however, the Council can only act unanimously on a proposal from the Commission after consulting the European Parliament and the Economic and Social Committee. Social security issues require unanimity of the Council, thus implying a veto power for each country. But it is very probable that unanimity will soon be replaced by the majority principle to guarantee the capacity to act within an enlarged European Union. The Treaty of Nice (not yet ratiﬁed) will further extend the European Union competence in the area of social security. The Council, acting unanimously on a proposal from the Commission after consulting the European Parliament, may decide to render the procedure referred to in Art. 251 (qualiﬁed majority) applicable to those areas of social policy, which are currently subject to the rule of unanimity. This “bridge”, however, cannot be used for social security (Art. 137 ECTn39).

39

Treaty of the European Community as amended by the Treaty of Nice (not yet ratiﬁed).

Sustainability of Pension Systems with Systems Competition

233

Both the Treaty of the European Community and the Treaty of the European Union recapitulate the evolution of the idea of a Social Union which complements the Economic Union. In Art. 136 ECTa and in the Preamble (EUTa)40 “fundamental social rights such as those set out in the European Social Charter signed at Turin on 18 October 1961 and in the 1989 Community Charter of the Fundamental Social Rights of Workers” are mentioned. Thus, these two documents have also been awarded an ofﬁcial character. This development towards more European Community competence in social policy issues, which can be observed in the Treaties of Amsterdam and Nice, is accompanied by a discussion of fundamental rights. The Charter of Fundamental Rights was proclaimed in 2000—at the moment it is still non-binding, but will probably be part of a future European constitution. Art. 34 of the Charter states explicitly the position of the European Union concerning social security and social assistance: 1. The Union recognises and respects the entitlement to social security beneﬁts and social services providing protection in cases such as […] old age, […] in accordance with the procedures laid down by Community law and national laws and practices. 2. Everyone residing and moving legally within the European Union is entitled to social security beneﬁts and social advantages in accordance with Community law and national laws and practices. […] Social security including old-age security will become an important fundamental right. Although the concrete interpretation and application is still open, the responsibility of the European Union for the realisation of this right is clearly visible (Eichenhofer, 2002).

9.3.2. Consequences for the distribution of competence and the level of harmonisation This historical evolution of social policy shows that the two approaches—social policy as a by-product of economic policy or as an independent objective to complement economic integration—are both anchored in the understanding of the European Union. Over the last few decades, however, a shift has been

40

Treaty of the European Union as amended by the Treaty of Amsterdam (1999).

234

Unfunded Pension Systems: Ageing and Migration

observable. The awareness for social policy issues at the European level has increased—though not always continuously. Social policy issues are no longer reduced to supporting the evolution of the common market, but are more and more regarded as objectives in themselves. Correspondingly, the competence of the European Community has been extended to include legislative authority in the area of social policy since the Treaty of Amsterdam (1999). The increasing importance of social policy issues at the European level as well as the pro-European position of the European Court of Justice gives us grounds to think that the competence of the European Community has been unambiguously extended to social policy issues which were previously the responsibility of the Member States, and that this process is going to continue with consequences for the level of harmonisation. In order to be able to evaluate whether this impression is correct, we look in detail at the distribution of competence. The question is whether enough competence has already been transferred to the European level to establish the level of harmonisation – equalisation or co-ordination of contribution rates—needed to guarantee an efﬁcient allocation of labour and an optimal level of redistribution. 9.3.2.1. Distribution of competence The realisation of European integration has been concerned with the creation of the Economic Union on the basis of the four freedoms of movement. It has not been a primary goal to appropriate social policy competence from the Member States (Breyer and Kolmar, 1996). The competence for social policy is thus principally located at the national level. The European institutions lack the legal basis to intervene, for example, in the nationally organised old-age security systems. From the legal perspective, the territorial principle, which corresponds to the principle of employment if economically interpreted, applies to all areas of social policy. The territorial principle, however, is limited if free mobility of labour (Art. 39 ECTa) is endangered. Legislative activities at the European level can then be justiﬁed. But Art. 136 ECTa emphasises once more the two different views regarding the role of the European Community, one believing in the market and the other seeing the danger of “social dumping”. By pointing out that “the functioning of the common market […] will favour the harmonisation of social systems”, there is a reference to the neo-liberal view. But this process also requires “procedures provided for in this Treaty and from the approximation of provisions laid down by law, regulation or administrative

Sustainability of Pension Systems with Systems Competition

235

action”. The Member States thus believe that such a development will not ensue from the common market alone, but also needs concrete actions on the European level. The principle of subsidiarity as understood by the European Community is intended to prevent the competence in the area of social policy from becoming one-sidedly concentrated at the European level. Art. 5 ECTa states that […] In areas which do not fall within its exclusive competence, the Community shall take action, in accordance with the principle of subsidiarity, only if and insofar as the objectives of the proposed action cannot be sufﬁciently achieved by the Member States and can therefore, by reason of the scale or effects of the proposed action, be better achieved by the Community. […]41 On the one hand, Art. 5 regulates the areas in which the European Community may act if this is deemed necessary—including areas not speciﬁcally mentioned in the Treaty. On the other hand, this is clearly restricted to instances where the subsidiarity principle is not violated and where the actions of the European Community can be justiﬁed by the objectives of the Treaty. The additional competence which can be derived for the European level is thus limited by the subsidiarity principle. The application of the principle of subsidiarity was speciﬁed in 1997 in the Protocol (No. 30 ECTa) on the application of the principles of subsidiarity and proportionality.42 Actions of the European Community need to fulﬁl the requirements of Art. 5 ECTa for which the following guidelines have been developed: the respective area is of transnational interest; measures taken by the European level dominate measures on the national level due to scope and impact; and measures of the Member States or no measures at all violate the requirements of the Treaty, for example, with respect to economic and social integration. Concerning the harmonisation of old-age security across Member States, the last guideline allows the European Union to act if the free movement of labour is otherwise prevented or distorted. Consequently, Eichenhofer (2002) considers the principle of subsidiarity as not very effective in protecting the competence of the Member States. De jure, the economic importance of old-age security alone sufﬁces to justify the competence of the European Community in this area.

41 42

See also Art. 2 EUTa which explicitly refers to Art. 5 ECTa. See also declaration 43 of the Final Communique´ to the Treaty of Amsterdam (1997).

236

Unfunded Pension Systems: Ageing and Migration

De facto, the distribution of competence is, however, not very clear.43 Given the high degree of interaction between economic integration and social policy issues, the ﬁelds of social policy which are related to the freedom of movement and other aspects of the common internal market are numerous. This implies that the competence for social policy is shared between the European Community and the Member States with the centre of gravity shifting more and more to the European level as the integration process advances. However, with regard to pension issues, the main competence still remains with the Member States—either because of their speciﬁc responsibilities in this area or because of their veto power. This is of course not without consequence for the level of harmonisation. 9.3.2.2. Impact on harmonisation In section 9.2, we analysed the level of harmonisation which guarantees an efﬁcient allocation of labour without a complete erosion of intergenerational redistribution. We derived that for the realistic cases of restricted mobility today and restricted or unrestricted mobility in the future, co-ordination or equalisation of the contribution rates leads to optimal results. We now want to see what impact the distribution of competence between the Member States and the European Community has on the level of harmonisation and compare the outcome with the optimal level. Again, it is useful to look at the competence and the resulting level of harmonisation separately for social activities within the Economic Union and the Social Union. We begin with actions of the European Community within the framework of the Economic Union. The measures which are intended to reduce the obstacles to free movement of labour support the mutual recognition of pension claims according to our classiﬁcation in Figure 9.2. It is, therefore, possible for the European Community to demand that national pension systems recognise pension claims acquired in other Member States in order to ensure that workers—and members of their families—who exercise their right of free movement do not suffer negative consequences with regard to their pension beneﬁts.44 And which level of harmonisation results from the increase in competence within the Social Union? The legislative authority of the European Community in the social area is described in Art. 136 and 137 ECTa. Although the objective of supporting the social development in various areas is reafﬁrmed, it is emphasised that “[…] the Community and the Member States shall implement measures 43

The so-called Post-Nice Process is supposed to clarify this by 2004. A more concrete deﬁnition of European and national competence as well as the role of national parliaments within the European Union is part of the agenda. 44 See Art. 42 ECTa and EEC regulations 1408/71 and 574/72.

Sustainability of Pension Systems with Systems Competition

237

which take account of the diverse forms of national practices […]” (Art. 136 (2)), and it is required that “the Council may adopt […] minimum requirements for gradual implementation, having regard to the conditions and technical rules obtaining in each of the Member States” (Art. 137 (2)). This is concretised in the Treaty of Nice as “excluding any harmonisation of the laws and regulations of the Member States” (Art. 137 (4)). With regard to the national social security systems, Art. 137 (4) ECTn more speciﬁcally states that [t]he provisions adopted […] shall not affect the right of Member States to deﬁne the fundamental principles of their social security systems and must not signiﬁcantly affect the ﬁnancial equilibrium thereof […]. The last point in particular underlines that the European Community is not entitled to appropriate in any way the responsibility of the Member States for the national pension systems. With regard to the harmonisation efforts, the legislative authority of the European Community seems to be restricted to those measures, which reduce the obstacles to free movement of labour. As we have already observed, the level of harmonisation—limited to the mutual recognition of pension claims—is low. Although the general competence of the European Community has been extended to instances where “measures of the Member States or no measures at all violate the requirements of the Treaty, for example, with respect to economic and social integration” (Protocol No. 30, 1997), the concrete provisions are limited to the adoption of “minimum requirements for gradual implementation” (Art. 137 (2) ECTa). The Economic Union is thus the driving force for the harmonisation efforts and not the Social Union. It has been realised that a higher level of co-ordination is needed in various areas of social policy including old-age security; and that closer co-operation would be useful given the similar challenges which the national pension systems have to face. At the same time, it is evident that different views on how to organise a pension system coexist in Europe as a result of the distinct histories of social security systems in each Member State. Art. 137 (2) ECTa points at a way to deal with this problem by stating that […] the Council […] may adopt measures designed to encourage cooperation between Member States through initiatives aimed at improving knowledge, developing exchanges of information and best practices, promoting innovative approaches and evaluating experiences […].

238

Unfunded Pension Systems: Ageing and Migration

The Council has, therefore, decided to initiate the so-called “open method of co-ordination”. The expression was coined at the Lisbon Summit in 2000 to indicate a process “where explicit, clear and mutually agreed objectives are deﬁned, after which peer review enables Member States to learn from the best practices in Europe. This method respects local diversity, is ﬂexible, and simultaneously wants to ensure progress in the social sphere” (Vandenbroucke, 2002, p. 35). To say it differently: the open method aims at exchanging information and learning from one another based on a common process of planning, evaluating, reviewing, comparing and adapting national social policies on the basis of common goals (European Council, 2000). This is intended to lead to greater convergence towards the main goals of the European Community. Given the demographic development all Member States have to face, these goals include the ﬁnancial sustainability of pension systems in the Member States. The European Community has the responsibility to organise this process, but as Riester (2002) puts it “The idea […] is not for the European Union to lay down laws or to promote harmonisation” (p. 27). The objective of the open method is to make the national pension systems more uniform without implying uniform social security laws based on the idea of initiating a process of convergence in the medium-term via the identiﬁcation and realisation of best practices (Eichenhofer, 2002). The motivation for this approach can be found in the necessity of all Member States to prepare their national pension systems for the forthcoming demographic crisis. Learning from one another should help when designing reforms. Reforms of the national pension systems will very probably include measures, which aim at redistributing the pension burden among generations, thus affecting the implicit taxes and, therefore, the distortions of the efﬁcient allocation of labour. But it is evident that the open method does not and cannot deal with the impact of unfunded pension systems on the efﬁcient allocation of labour, or with the danger of a “race to the bottom” of old-age security if the national pension systems are not sufﬁciently harmonised.

9.4. COMPARISON OF THE THEORETICAL AND INSTITUTIONAL RESULTS After having analysed the theoretically optimal and the institutionally realised level of harmonisation, we now turn to compare both results. As Breyer and Kolmar (1996) point out, “[i]t is not the level of social transfers per se, but the organisational structure that might be responsible for the non-achievement of maximum gross national products” (p. 148f). The focus will, therefore, be on

Sustainability of Pension Systems with Systems Competition

239

the impact of the distribution of competence and the resulting level of centralisation on the efﬁcient allocation of labour and the level of distribution. 9.4.1. Allocation of labour Two aspects have to be distinguished concerning the efﬁcient allocation of labour with respect to pension systems. The ﬁrst one concerns the question of whether all obstacles have been removed so that free movement is possible within the European Union. The second one deals with efﬁciency, asking whether mobility is distorted. As we have seen, countries of the European Union are obliged to mutually recognise pension claims acquired in other Member States (Art. 42 ECTa). Migrating from one country to another country is not impeded by losing pension beneﬁts. Individuals will thus migrate to countries where their labour income over the life cycle net of explicit and implicit taxes is maximal.45 If we abstract from differences in explicit taxes and restrict our attention to old-age security, the choice of the destination country will then be determined by differences in marginal productivities, i.e. in gross wages if we assume complete labour markets, and differences in the taxes implicit in the national pension systems. While migration that equates gross wages results in an optimal allocation of labour, this is not the case for migration that leads to equal wages net of implicit taxes—if implicit taxes differ across countries.46 Analysis of the institutional regulations of the European Union has shown that an equation and thus neutralisation of implicit taxes is not an explicit goal of the process of convergence, which is mainly based on the open method. A reduction of the distortions due to different tax burdens imposed by the national pension systems can only emerge as an accidental by-product. The European Court of Justice more explicitly inﬂuences migration when interpreting the EEC regulations 1408/71 and 574/72 in a pro-migration way. In fact, the European Court of Justice has to rule in accordance with the European Treaties, which explicitly put forward the objective of supporting the process of European integration. The Court assumes implicitly that intra-European migration promotes integration. The question is, however, whether the decisions of the Court with respect to migration stimulate a sub-optimal level of migration in the sense of 45

See section 9.1. The consequences are identical for the allocation of labour if the pension systems are transformed to notional deﬁned contribution systems ﬁnanced by pay-as-you-go contributions, as put forward by Feldstein (2001). Abstracting from transitional aspects, individual accounts which can be easily transferred from one country to another country do not eliminate the distortion of the allocation of labour as the migration decision still depends on gross wage net of taxes implicit in the national pension systems.

46

240

Unfunded Pension Systems: Ageing and Migration

too much migration. Positively discriminating migrants compared to nonmigrants might lead to distortions of the optimal allocation of labour. The economic objective and the objective of the Court differ concerning migration; potentially inefﬁcient migration might be encouraged or even intensiﬁed. Summarising, national pension systems do not present important obstacles to migration. The principle of mutual recognition ensures that the free movement of labour is not hindered. Measures which guarantee that migration is undistorted are, however, absent. Neither the national countries nor the European institutions seem to be aware of the distortions of the allocation of labour which stem from taxes implicit in the national pension systems—as well as from explicit taxes. The regulations concerning the free movement ignore the presence of distortions and the Court does not use its authority to create the preconditions for an efﬁcient allocation of labour. 9.4.2. Level of redistribution Despite the increase in competence of the European Community for social policy issues, responsibility for the organisation of the national pension systems stays with the Member States. Only when the functioning of the common market in general and the free movement of labour in particular are endangered, does the European Union have the right to intervene. To guarantee that mobility is not impeded by incompatibilities of the national pension systems, Member States must mutually recognise pension claims. All initiatives to harmonise the national pension systems further—for example, in the sense of co-ordinating or equalising contribution rates—depend on the voluntary agreements of the Member States. The open method allows an exchange of experiences and a step-by-step convergence of the different national pension systems. It might, however, lack the binding character necessary to avoid a “race to the bottom” of intergenerationally redistributive activities, even though the European Community has the legislative authority to transform the results of the open method into binding laws (Eichenhofer, 2002). But it must be doubted whether the Member States will agree on the harmonisation of fundamental aspects of the national pension systems, such as the contribution rate. If they do not, and pension systems remain decentralised, a merely voluntary commitment not to engage in a “race to the bottom” is not credible.47 47

Breyer and Kolmar (2002) show that the co-ordination requirements are incompatible with the voluntary co-ordination of Member States. If every country maximises national welfare without taking into account the external effects on other countries, a unilateral deviation is proﬁtable.

Sustainability of Pension Systems with Systems Competition

241

9.4.3. Evaluation The analysis has shown that the chosen approach towards the harmonisation of national pension systems neither leads to an efﬁcient allocation of labour nor prevents a “race to the bottom” of intergenerational redistribution. So, how should the competence be optimally distributed in order to achieve these objectives given the restricted policy space? To answer this question, let us turn again to the principle of subsidiarity. In the context of the distribution of competence between the Member States and the European Community, subsidiarity indicates that the responsibility should be with the lowest level, which can fulﬁl the task in the best way. For the economic application of the subsidiarity principle with respect to old-age security, the existence and the range of (ﬁscal) externalities need to be determined. The political decision-making should then be allocated to the level at which an internalisation can be best guaranteed and an efﬁcient allocation of labour and an optimal level of redistribution can be reached. The main features of pension systems are that they redistribute incomes across generations and within generations from individuals with a lower life expectancy to individuals with a higher one (Kolmar, 1997). The ex post enforcement of the contract is, however, not possible in the context of the employment principle. The net contributors, i.e. the young or those who have information indicating that their life expectancy is below average, can escape by migrating. This makes it clear that a more pronounced involvement of the European level is needed. This result is nothing else than an application of the “selection principle” (Sinn, 1997, 2003a). If old-age security fulﬁls the criteria for an intervention of the state at the national level according to the “selection principle”, which we have seen to be the case, introducing competition on the European level cannot lead to an efﬁcient outcome. Even though the main competence for old-age security can stay with the Member States, the necessary responsibilities must be conceded to the European Union to allow an effective monitoring of the harmonisation rules. If this appears to be a too demanding requirement given the still persisting fundamental differences between the national pension systems, there are solutions which might be less controversial.

9.5. ALTERNATIVE OPTIONS WHEN PENSION SYSTEMS ARE DIFFERENT It has been realised that a higher level of harmonisation is needed in various areas of social policy including old-age security; and that a closer cooperation is useful

242

Unfunded Pension Systems: Ageing and Migration

given the similar challenges which the national pension systems have to face. So far, we have analysed the impact of increased labour mobility on unfunded pensions systems, which are very similar with respect to important aspects. But it is evident that different views on how to organise a pension system coexist in Europe given the distinct histories of social security systems in each Member State. Especially the harmonisation of pension systems in the Anglo-Saxon tradition a` la Beveridge, where beneﬁts are mainly supposed to guarantee a uniform, basic level of social protection for old age, and continental pension systems a` la Bismarck, where beneﬁts are set such that the recipients can keep their standard of living during retirement, is very demanding if not impossible. It is, therefore, necessary to discuss alternative options which can be applied to unfunded pension systems, which differ with respect to the aim and scope of oldage security. For this, we distinguish the case where migration implies the immediate change from the pension system of the country of origin to the pension system of the destination country according to the employment principle and the case where migration does not at all affect the membership in a pension system (home-country principle) or only with a delay of several years (principle of delayed integration). We now also include transfers and payments among the available measures thus enlarging the policy space. 9.5.1. Transfers and individual payments For the moment, we maintain the assumption that membership in pension systems follows the employment principle and discuss other options which lead to an efﬁcient allocation of labour without a complete erosion of redistributive activities. A direct way to attack the problem of externalities of migration on national pension systems is to internalise these effects via interjurisdictional transfers (Kolmar, 1997). If changes in the pension system of one country cause external effects in other countries, the responsible country has to pay transfers to compensate the other countries. As can be shown (Breyer and Kolmar, 2002) only compulsory transfers work. In addition, it must be assumed that the information problem can be solved and that the interjurisdictional contracts are renegotiation proof. It is also possible to shift the responsibility for the transfers from the interjurisdictional level, i.e. the country, to the individual. Externalities on the pension systems of the source country and/or the destination country due to migration can then be internalised via payments from the individual. Depending on the direction of the external effects of the migration decision, an exit or an entry fee could be constructed.

Sustainability of Pension Systems with Systems Competition

243

For interjurisdictional transfers and individual payments, the requirements are much less demanding than for the case discussed above with the harmonisation of national pension systems in the form of equalisation or co-ordination of contribution rates. As these transfers and payments do not affect the general structure of the national pension systems, old-age security systems remain more independent from one another. It is, therefore, no longer necessary that they are similar in important aspects. In contrast, externalities due to migration can be internalised no matter how different the pension systems are. It is evident that with 25 Member States, bilateral bargaining would prove to be a very complicated and complex issue. A central authority would, therefore, be needed for setting the transfers and payments, coordinating the redistribution of income across countries and controlling the compliance (Wildasin, 1991, 1994). So even with this very much decentralised option, the Member States would be required to concede the necessary—albeit smaller—responsibilities to the European level. 9.5.2. Home-country principle and delayed integration The inefﬁciencies with respect to the allocation of labour and the danger of a complete erosion of redistributive activities stem from the opportunity of individuals to opt out of an existing social contract after the individual risk is realised. Abolishing this option would remove the incentives to do so. To avoid distortions of the migration decision due to taxes implicit in social security systems, Sinn (1990) proposes to replace the employment principle by the home-country principle. Individuals choose a social security system of one Member State once for their entire life. Then, migrating does no longer help to escape the system; especially the implicit taxes of the chosen pension system cannot be avoided by changing the place of employment. The externalities of migration on the national pension systems are completely internalised. The migration decision is thus not affected by differences of the pension burden between countries, but depends on the difference in gross wages. An efﬁcient allocation of labour results. Replacing the employment principle by the home-country principle eliminates any form of ﬁscal competition. In this respect the home-country principle is comparable to a close linking of pension systems as discussed above. Whether less competition should be preferred to more competition or vice versa depends on what one thinks about the general operation of unfunded public pension systems (Oates, 2001, 2002). We thus enlarge the analysis by taking political economy arguments into account. If we assume that those responsible for the national

244

Unfunded Pension Systems: Ageing and Migration

pension systems seek to promote social welfare, the home-country principle is the best solution. If, however, we adhere to the Leviathan belief and assume that national pension systems are dominated by the interplay of special interest groups, the evaluation of the home-country principle is less positive. In the speciﬁc context of unfunded pension systems, those who determine the fundamental parameters of old-age security are for the main part the voters. The home-country principle, therefore, bears the risk that the young—once they have opted for a certain national pension system—can be more easily exploited by the old in a gerontocracy. As we have seen in Chapter 7, the young can counterbalance the power of the old—even when they no longer have the political majority—if they have the option to decrease their domestic labour supply, e.g. by emigrating. Depriving them of the option to escape the system reduces their possibilities to react to even higher pension burdens shifted to them from the old. There is thus a trade-off between too much migration motivated by the employment principle, which endangers the ﬁnancial sustainability of the national pension systems, and too few outside options due to the home-country principle. Although this problem concerns intergenerational redistribution and not efﬁciency, it is necessary to take this into account when thinking about putting the home-country principle into practice. If ﬁscal competition is welcome in order to counter-balance gerontocratic tendencies, the employment principle should be advocated. If ﬁscal competition is, however, seen as endangering the efﬁcient allocation of labour, the homecountry principle should be chosen. But both alternative principles present rather extreme ways of how to assign migrants to jurisdictions. It has, therefore, been proposed to opt for a middle course with the principle of delayed integration (Sinn, 2002, and Richter, 2002).48 This principle means that migrating from one country to another country results in the assignment to the ﬁscal and social systems of the destination country with some delay. The principle is closer to the employment principle for a shorter delay and more similar to the home-country principle for a longer delay. Thus, the two essential aspects—distortions of labour allocation and containment of gerontocratic tendencies—are both taken into account to a certain extent. Compared to the employment principle differences in national pension systems, which result in different implicit taxes do not distort the migration decision for short-term migration and play a smaller role for long-term migration. At the same time, the reassignment of migrants to the pension system of the destination country—even though with some delay—works as a counterbalance to potential gerontocratic tendencies. 48 See also the report of the Council of Advisors of the German Ministry of Finance (Wissenschaftlicher Beirat, 2001).

Sustainability of Pension Systems with Systems Competition

245

Once more, a central authority would be needed for setting the exact rules and controlling the compliance. The Member States would again have to concede the necessary responsibilities to the European level.

9.6. CONCLUSION Although for all three options discussed—harmonisation, transfers and payments and delayed integration—competence has to be shifted from the Member States to the European Union, the extent to which the Member States restrict their responsibility for old-age security is different. The requirements for equalising or co-ordinating contribution rates are by far the most demanding. First, national pension systems must display similarities in the most important aspects and second, the Member States must be willing to forgo fundamental decision-making powers. Even though, the national pension systems have already converged to a certain extent during the last decades as a reaction to exogenous factors as we have seen in Chapter 4, fundamental differences still exist especially with respect to systems organised as “ﬂat-rate beneﬁt” systems in the tradition of Beveridge and “social insurance type” systems in the tradition of Bismarck. Beside, the analysis of the institutional development has shown that the Member States are still far away from approving the necessary shift of competence to the European level where the national pension systems are concerned. Transfers and payments as well as a change from the employment principle to the principle of delayed integration have, therefore, some merits. First, as the national pension systems remain rather independent from one another, differences with respect to size and generosity do not present important obstacles. Second, although these options also require that the Member States concede some responsibilities to the European level, the scope is nevertheless much more reduced. To sum up, the relation between migration and national pension systems must be considered when thinking about a reform of old-age security in the European Union. The principle of free movement of labour might endanger the ﬁnancial sustainability of non-coordinated national pension systems while at the same time the lack of co-ordination distorts the allocation of labour. As we have seen, it is not necessary to insist that national pension systems are replaced by a uniform European pension system in order to cure both problems. The main responsibility can stay with the Member States; but the European Union must be involved sufﬁciently to guarantee the binding nature of the harmonisation rules.

246

Unfunded Pension Systems: Ageing and Migration

If the Member States fail to agree on the role of the European Union, it is highly possible that the demographic pressure on national pension systems induces the Member States to fall back on either or both of two strategies: ﬁrst, to set a “race to the bottom” in motion to attract contributors and second, to erect obstacles to prevent contributors from emigrating. As this development would be the exact opposite of the objectives of the European Union, it is very important to start with reforms as soon as possible. The European Council has realised this and has initiated the process of the open method. We have, however, seen that the intended convergence is not able to avoid an unhealthy competition among the pension systems of the Member States. It must be hoped that the open method is only the ﬁrst step towards a more courageous approach leading to a more pronounced involvement at the European level in the area of old-age security. APPENDIX A9. ART. 117 ToR AND ART. 136 ECT The evolution of the understanding of the role of the European Union with respect to social policy issues is reﬂected in Art. 117 ToR and in the modiﬁcations by Art. 136 ECT (Treaty of Amsterdam): Art. 117 ToR—Treaty of Rome (1958) Member States agree upon the need to promote improved working conditions and an improved standard of living for workers, so as to make possible their harmonisation while the improvement is being maintained. They believe that such a development will ensue not only from the functioning of the common market, which will favour the harmonisation of social systems, but also from the procedures provided for in this Treaty and from the approximation of provisions laid down by law, regulation or administrative action. Art. 136 ECT (ex Art. 117)—Treaty of Amsterdam (1999) The Community and the Member States, having in mind fundamental social rights such as those set out in the European Social Charter signed at Turin on 18 October 1961 and in the 1989 Community Charter of the Fundamental Social Rights of Workers, shall have as their objectives the promotion of employment, improved living and working conditions, so as to make possible their harmonisation while the improvement is being maintained,

Sustainability of Pension Systems with Systems Competition

247

proper social protection, dialogue between management and labour, the development of human resources with a view to lasting high employment and the combating of exclusion. To this end the Community and the Member States shall implement measures which take account of the diverse forms of national practices, in particular in the ﬁeld of contractual relations, and the need to maintain the competitiveness of the Community economy. They believe that such a development will ensue not only from the functioning of the common market, which will favour the harmonisation of social systems, but also from the procedures provided for in this Treaty and from the approximation of provisions laid down by law, regulation or administrative action. APPENDIX B9. ART. 118 ToR AND ART. 137 ECT The change of competence at the European level can be seen in Art. 118 ToR and in the modiﬁcations by Art. 137 ECT (Treaty of Amsterdam and Treaty of Nice): Art. 118 ToR—Treaty of Rome (1958) Without prejudice to the other provisions of this Treaty and in conformity with its general objectives, the Commission shall have the task of promoting close co-operation between Member States in the social ﬁeld, particularly in matters relating to: † † † † † †

employment; labour law and working conditions; basic and advanced vocational training; social security; prevention of occupational accidents and diseases; occupational hygiene; the right of association, and collective bargaining between employers and workers.

To this end, the Commission shall act in close contact with Member States by making studies, delivering opinions and arranging consultations both on problems arising at national level and on those of concern to international organisations. Before delivering the opinions provided for in this Article, the Commission shall consult the Economic and Social Committee.

248

Unfunded Pension Systems: Ageing and Migration

Article 118a (1) Member States shall pay particular attention to encouraging improvements, especially in the working environment, as regards the health and safety of workers, and shall set as their objective the harmonisation of conditions in this area, while maintaining the improvements made. (2) In order to help achieve the objective laid down in the ﬁrst paragraph, the Council, acting in accordance with the procedure referred to in Article 189c and after consulting the Economic and Social Committee, shall adopt, by means of directives, minimum requirements for gradual implementation, having regard to the conditions and technical rules obtaining in each of the Member States. Such directives shall avoid imposing administrative, ﬁnancial and legal constraints in a way which would hold back the creation and development of small and medium-sized undertakings. (3) The provisions adopted pursuant to this Article shall not prevent any Member State from maintaining or introducing more stringent measures for the protection of working conditions compatible with this Treaty. Art. 137 ECT (ex Art. 118)—Treaty of Amsterdam (1999) (1) With a view to achieving the objectives of Article 136, the Community shall support and complement the activities of the Member States in the following ﬁelds: † improvement in particular of the working environment to protect workers’ health and safety; † working conditions; † the information and consultation of workers; † the integration of persons excluded from the labour market, without prejudice to Article 150; † equality between men and women with regard to labour market opportunities and treatment at work. (2) To this end, the Council may adopt, by means of directives, minimum requirements for gradual implementation, having regard to the conditions and technical rules obtaining in each of the Member States. Such directives shall avoid imposing administrative, ﬁnancial

Sustainability of Pension Systems with Systems Competition and legal constraints in a way which would hold back the creation and development of small and medium-sized undertakings. The Council shall act in accordance with the procedure referred to in Article 251 after consulting the Economic and Social Committee and the Committee of the Regions. The Council, acting in accordance with the same procedure, may adopt measures designed to encourage cooperation between Member States through initiatives aimed at improving knowledge, developing exchanges of information and best practices, promoting innovative approaches and evaluating experiences in order to combat social exclusion. (3) However, the Council shall act unanimously on a proposal from the Commission, after consulting the European Parliament, the Economic and Social Committee and the Committee of the Regions in the following areas: † social security and social protection of workers; † protection of workers where their employment contract is terminated; † representation and collective defence of the interests of workers and employers, including co-determination, subject to paragraph 6; † conditions of employment for third-country nationals legally residing in Community territory; † ﬁnancial contributions for promotion of employment and jobcreation, without prejudice to the provisions relating to the Social Fund. […] (5) The provisions adopted pursuant to this Article shall not prevent any Member State from maintaining or introducing more stringent protective measures compatible with this Treaty. […] Art. 137 ECT (ex Art. 118)—Treaty of Nice (not yet ratiﬁed) (1) With a view to achieving the objectives of Article 136, the Community shall support and complement the activities of the Member States in the following ﬁelds:

249

250

Unfunded Pension Systems: Ageing and Migration (a) improvement in particular of the working environment to protect workers’ health and safety; (b) working conditions; (c) social security and social protection of workers; (d) protection of workers where their employment contract is terminated; (e) the information and consultation of workers; (f) representation and collective defence of the interests of workers and employers, including co-determination, subject to paragraph 5; (g) conditions of employment for third-country nationals legally residing in Community territory; (h) the integration of persons excluded from the labour market, without prejudice to Article 150; (i) equality between men and women with regard to labour market opportunities and treatment at work; (j) the combating of social exclusion; (k) the modernisation of social protection systems without prejudice to point (c). (2) To this end, the Council: (a) may adopt measures designed to encourage cooperation between Member States through initiatives aimed at improving knowledge, developing exchanges of information and best practices, promoting innovative approaches and evaluating experiences, excluding any harmonisation of the laws and regulations of the Member States; (b) may adopt, in the ﬁelds referred to in paragraph 1(a) to (i), by means of directives, minimum requirements for gradual implementation, having regard to the conditions and technical rules obtaining in each of the Member States. Such directives shall avoid imposing administrative, ﬁnancial and legal constraints in a way which would hold back the creation and development of small and medium-sized undertakings. The Council shall act in accordance with the procedure referred to in Article 251 after consulting the Economic and Social Committee and the Committee of the Regions, except in the ﬁelds referred to in paragraph 1(c), (d), (f) and (g) of this Article, where the Council

Sustainability of Pension Systems with Systems Competition shall act unanimously on a proposal from the Commission, after consulting the European Parliament and the said Committees.The Council, acting unanimously on a proposal from the Commission, after consulting the European Parliament, may decide to render the procedure referred to in Article 251 applicable to paragraph 1(d), (f) and (g) of this Article. (3) A Member State may entrust management and labour, at their joint request, with the implementation of directives adopted pursuant to paragraph 2. In this case, it shall ensure that, no later than the date on which a directive must be transposed in accordance with Article 249, management and labour have introduced the necessary measures by agreement, the Member State concerned being required to take any necessary measure enabling it at any time to be in a position to guarantee the results imposed by that directive. (4) The provisions adopted pursuant to this Article: † shall not affect the right of Member States to deﬁne the fundamental principles of their social security systems and must not signiﬁcantly affect the ﬁnancial equilibrium thereof; † shall not prevent any Member State from maintaining or introducing more stringent protective measures compatible with this Treaty. […]

251

This Page Intentionally Left Blank

CHAPTER 10

Conclusion Twenty years from now you will be more disappointed by the things you didn’t do than by the ones you did. Mark Twain (1835 –1910) A more globalised world with more integrated factor and goods markets increases the demands placed on social security systems for several reasons. On one hand, national economies are exposed to more intensive competition, especially from low-wage countries, while they are at the same time affected to a larger extent by global economic disturbances. On the other hand, individuals have more fragmented employment histories, very often with jobs in several countries. The national systems, therefore, need to be adjusted to provide security for individuals against global economic uncertainties in a way that is appropriate given the individual work biographies. These general considerations also apply to old-age security. But in addition to the changes in the basic conditions due to global changes, industrialised countries experience another challenge which—if not responded to appropriately—might result in a serious crisis: the ageing of the population. Pension systems in most countries rely, for the main part, on pay-as-you-go ﬁnancing and thus depend on a well-balanced ratio of old to young, i.e. of recipients to contributors. This so-called dependency ratio will, however, worsen signiﬁcantly in the next few decades and endanger the sustainability of old-age security. The analysis has shown that reforms are needed to prepare national pension systems for this demographic development and the increase in (selective) mobility. Reforms, however, need the support of the majority of the electorate in a majoritarian voting process. This does not present any problem for reforms which make all generations better off whereas the question of feasibility is important for reforms which aim at a more fair distribution of the pension burden across

254

Unfunded Pension Systems: Ageing and Migration

generations. In order to evaluate the chance of success of a reform, it is then necessary to take the path of society towards gerontocracy into account as this determines the time frame within which reforms have to be realised. After the young have lost their majority, it is more difﬁcult or even impossible to reform unfunded pension systems in a fundamental way. At a minimum this requires that the young are able to defend their interests by means other than voting. To analyse how the results change if the young have an exit option, we enlarge the approach in this direction by assuming that the young can emigrate. If the old have means to inﬂuence the degree of mobility, old-age security can be sustained. If mobility, however, is independent from actions of the old, this threatens the survival of intergenerational redistribution. Free movement of labour is an important goal, especially in the context of the integrated market of the European Union. It is thus necessary that reforms are designed such that they take into account the effects of ageing on the sustainability of unfunded pension systems as well as the consequences of the increasing mobility of the young. Reforms, therefore, should not be limited to considering exclusively the national perspective. On the contrary, it is particularly important to include a European perspective as this is also the relevant migration area. If obstacles to free mobility are further and further dismantled, national pension systems must be sufﬁciently harmonised to keep up with this development. Otherwise, unfunded pension systems with different implicit taxes distort allocative efﬁciency and endanger redistributive activities. In this book, we have shown for different settings of political power distribution and for different degrees of mobility what would happen if the status quo were to be maintained and what could and should be done to guarantee the survival of oldage security based on a fair sharing of the pension burden. Neither explosion nor erosion is the inevitable fate of unfunded pension systems, but to avoid either happening, reforms are necessary as soon as possible which take into account the demographic changes and the increasing mobility of some groups of society, which will otherwise intensify the pressure on unfunded pension systems in the next few decades. Only if we opt for fundamental reforms now which reduce the amount of intergenerational transfers and thus decrease the dependencies across generations, will we be able to look at old-age security 20 years from now without too much disappointment.

References Aaron, H. (1966). The social insurance paradox. Canadian Journal of Economics and Political Science, 32, 371– 374. Andersson, F. and Konrad, K. A. (2003). Globalization and risky human-capital investment. International Tax and Public Finance, 10, 211 – 228. Atkinson, A. B. (1995). Incomes and the Welfare State, Cambridge University Press, Cambridge. Atkinson, A. B. and Sandmo, A. (1980). Welfare implications of the taxation of savings. Economic Journal, 90, 529– 549. Atkinson, A. B. and Stiglitz, J. E. (1980). Lectures on Public Economics, Chapter 12, McGraw-Hill, London. Auerbach, A. J., Gokhale, J. and Kotlikoff, L. J. (1991). “Generational accounting: a meaningful alternative to deﬁcit accounting,” in Tax Policy and the Economy, D. Bradford (ed.), vol. 5, MIT Press, Cambridge, MA, pp. 55– 110. Auerbach, A. J., Gokhale, J. and Kotlikoff, L. J. (1994). Generational accounting: a meaningful way to evaluate ﬁscal policy. Journal of Economic Perspectives, 8, 73– 94. Barro, R. J. (1979). On the determination of public debt. Journal of Political Economy, 87, 940 – 971. Bashir, A., Robinson, G. J. (1994). Estimates of emigration of the foreign-born population: 1980– 1990, Population Division Working Paper 9, Washington, DC. Becker, G. S. (1964). Human Capital, The University of Chicago Press, Chicago. Becker, G. S. (1983). A theory of competition among pressure groups for political inﬂuence. The Quarterly Journal of Economics, 98, 371 – 400. Beckmann, K. (2000). A note on the tax rate implicit in contributions to pay-as-you-go public pension systems. FinanzArchiv, 57, 63– 76. Begg, D., Cre´mer, J., Danthine, J. -P., Edwards, J., Grilli, V., Neven D. J., Seabright, P., Sinn, H. -W., Venables, A. J., Wyplosz, Ch. (1993). Making Sense of Subsidiarity: How Much Centralization for Europe? CEPR, London. Beggs, J. J. and Chapman, B. J. (1988). Immigrant wage adjustment in Australia: cross section and time-series estimates. The Economic Record, 64, 161 – 167. Beggs, J. J. and Chapman, B. J. (1990). Search efﬁciency, skill transferability and immigrant relative unemployment rate in Australia. Applied Economics, 22, 249 – 260.

256

References

Beggs, J. J. and Chapman, B. J. (1991). “Male immigrant wage and unemployment experience in Australia,” in Immigration, Trade, and Labor Market, J. M. Abowd and R. B. Freeman (eds.), The University of Chicago Press, London, pp. 369 – 384. Besendorfer, D., Borgman, C. and Raffelhu¨schen, B. (1998). Ein Pla¨doyer fu¨r intergenerative Ausgewogenheit: Rentenreformvorschla¨ge auf dem Pru¨fstand. ifo-Studien, 44, 209 – 231. Blanchard, O. J. and Fischer, S. (1993). Lecture on Macroeconomics, 6th edn. MIT Press, Cambridge, MA, Chicago and London. Blundell, R. and MaCurdy, T. (1999). “Labor supply: a review of alternative approaches,” in Handbook of Labour Economics, O. C. Ashenfelter and D. Card (eds.), vol. 3A, NorthHolland, Amsterdam, pp. 1559– 1695. Boadway, R., Marceau, N. and Marchand, M. (1996). Investment in education and the time inconsistency of redistributive tax policy. Economica, 63, 171 – 189. Boeri, T., Boersch-Supan, A. and Tabellini, G. (2001). Would you like to shrink the welfare state? The opinions of European citizens. Economic Policy, 32, 7 – 44. Boeri, T., Boersch-Supan, A. and Tabellini, G. (2002). Pension reforms and the opinions of European citizens. American Economic Review (Papers & Proceedings), 92, 396 – 401. Boldrin, M., Dolado, J. J., Jimeno, J. F. and Peracchi, F. (1999). The future of pensions in Europe. Economic Policy, 29, 289–320. Bonin, H. and Feist, K. (1999). Pension reform in Germany and redistribution between living generations, Discussion Paper 78/99, Institut fu¨r Finanzwissenschaft, Universita¨t Freiburg, Freiburg. Boockmann, B. and Steiner, V. (2000). Cohort effects and the returns to education in West Germany, ZEW Discussion Paper 00–05, Mannheim. Borgmann, C., Krimmer, P. and Raffelhu¨schen, B. (2001). Rentenreformen 1998– 2001: Eine (vorla¨uﬁge) Bestandsaufnahme. Perspektiven der Wirtschaftspolitik, 2, 319 – 334. Borjas, G. J. (1987). Self-selection and the earnings of immigrants. The American Economic Review, 77, 531– 533. Borjas, G. J. (1994). The economics of immigration. Journal of Economic Literature, 32, 1667– 1717. Borjas, G. J. (1996). Labor Economics, McGraw-Hill, New York. Breyer, F. (1989). On the intergenerational pareto efﬁciency of pay-as-you-go ﬁnanced pension systems. Journal of Institution and Theoretical Economics, 145, 643 – 658. Breyer, F. (1994a). The political economy of intergenerational redistribution. European Journal of Political Economy, 13, 705– 724. Breyer, F. (1994b). Voting on social security when labor supply is endogenous. Economics and Politics, 6, 119 – 130. Breyer, F. (2000). Kapitaldeckungs-versus Umlageverfahren. Perspektiven der Wirtschaftspolitik, 1, 383 – 405. Breyer, F. (2001). Why funding is not a solution to the social security crisis, IZA Discussion Paper 328, Bonn. Breyer, F. and Craig, B. (1997). Voting on social security: evidence from OECD countries. European Journal of Political Economy, 13, 705–724.

References

257

Breyer, F. and Kolmar, M. (1996). “Social policy in a common market: labour market, social arbitrage, and public ﬁnance,” in Maastricht: Monetary Constitution without a Fiscal Constitution? R. Holzmann (ed.), Nemos, Baden-Baden, pp. 129 – 152. Breyer, F. and Kolmar, M. (2002). Are national pension systems efﬁcient if labor is (im)perfectly mobile? Journal of Public Economics, 83, 347 – 374. Breyer, F. and Stolte, K. (2001). Demographic change, endogenous labor supply and the political feasibility of pension reforms. Journal of Population Economics, 14, 409 – 424. Breyer, F. and Straub, M. (1993). Welfare effects of unfunded pension systems when labour supply is endogenous. Journal of Public Economics, 50, 77 – 91. Brinch, C., Hernæs, E. and Strøm, S. (2001). Labour supply effects of an early retirement programme, CESifo Working Paper 463, Munich. Browning, E. K. (1975). Why the social insurance budget is too large in a democracy. Economic Inquiry, 13, 373– 388. Browning, M., Deaton, A. and Irish, M. (1985). A proﬁtable approach to labour supply and commodity demands over the life-cycle. Econometrica, 53, 503 – 543. Brugiavini, A. (1999). “Social security and retirement in Italy,” in Social Security and Retirement around the World, J. Gruber and D. A. Wise (eds.), The University of Chicago Press, Chicago and London, pp. 181– 238. Brunner, J. K. (1994). Redistribution and the efﬁciency of the pay-as-you-go pension system. Journal of Institution and Theoretical Economics, 150, 511 – 523. Brunner, J. K. (1996). Transition from a pay-as-you-go to a fully funded pension system: the case of differing individuals and intragenerational fairness. Journal of Public Economics, 6, 131– 146. Bundesausla¨nderbeauftragte. (2002). Daten und Fakten zur Ausla¨ndersituation, Berlin. Bundesministerium fu¨r Arbeit und Sozialordnung. (2001). Die neue Rente: Solidarita¨t mit Gewinn, Berlin. Burda, M. C., Ha¨rdle, W., Mu¨ller, M. and Werwatz, A. (1998). Semiparametric analysis of German East-west migration intentions: facts and history. Journal of Applied Econometrics, 13, 525–541. Burkhauser, R. V., Kreyenfeld, M. and Wagner, G. G. (1997). The German socio-economic panel: a representative sample of reunited Germany and its parts. Vierteljahreshefte zur Wirtschaftsforschung, 66, 7 – 16. Buslei, H. and Steiner, V. (1999). Bescha¨ftigungseffekte von Lohnsubventionen im Niedriglohnbereich, Nomos, Baden-Baden. Bu¨tler, M. (2000). The political feasibility of pension reform options: the case of Switzerland. Journal of Public Economics, 75, 389– 416. Casamatta, G., Cremer, H. and Pestieau, P. (2001). Demographic shock and social security: a political economy perspective. International Tax and Public Finance, 8, 417 –431. Chand, S. K. and Jaeger, A. (1996). Aging population and public pension schemes, Occasional Paper 147, IMF, Washington D.C. Chiswick, B. R. (1978). The effect of Americanization on the earnings of foreign-born men. Journal of Political Economy, 86, 897– 921.

258

References

Chiswick, B. R. and Miller, P. W. (1985). Immigrant generation and income in Australia. The Economic Record, 61, 540– 553. Chiswick, B. R. and Miller, P. W. (1995). The endogeneity between language and earnings: international analyses. Journal of Labor Economics, 13, 246 – 288. Cigno, A. and Rosati, F. C. (1996). Jointly determined saving and fertility behaviour: theory and estimates for Germany, Italy, UK and USA. European Economic Review, 40, 1561– 1589. Cigno, A., Casolaro, L. and Rosati, F.C. (2000). The role of social security in household decision: VAR estimates of saving and fertility behaviour in Germany, CESifo Working Paper 394, Munich. DaVanzo, J. (1983). Repeat migration in the United States: who moves back and who moves on? Review of Economics and Statistics, 65, 552– 559. Dre`ze, J. H. (2000). Economic and social security in the twenty-ﬁrst century, with attention to Europe. The Scandinavian Journal of Economics, 102, 327 – 348. Dustmann, C. (1993). Earnings adjustments of temporary migrants. Journal of Population Economics, 6, 153 – 168. Dustmann, C. (1994). Speaking ﬂuency, writing ﬂuency and earnings of immigrants. Journal of Population Economics, 7, 133– 156. Dustmann, C. (1995). Savings behavior of return migrants. Zeitschrift fu¨r Wirtschafts-und Sozialwissenschaften, 115, 511– 533. Dustmann, C. (1996). Return migration: the European experience. Economic Policy, 11, 215 – 250. Dustmann, C. (1997). Return migration, uncertainty and precautionary savings. Journal of Development and Economics, 52, 295– 316. Economic Policy Committee. (2001). Budgetary challenges posed by ageing populations. EPC/ECFIN/655/01-EN ﬁnal. Eichengreen, B. (1991). Is Europe an Optimum Currency Area? NBER Working Paper 3579, Cambridge, MA. Eichenhofer, E. (2002). Der aktuelle Stand der europa¨ischen Sozialpolitik. Deutsche Rentenversicherung, 6, 322– 331. European Commission. (1994). European Social Policy—a Way Forward for the Union (White Paper). KOM(94) 333, Brussels. European Commission Directorate-General for Economic and Finance Affairs. (1995). Performance of the European Union Labour Market. European Economy, 3, Brussels. European Council. (2000). Presidency Conclusions Lisbon, http://ue.eu.int/. Eurostat. (2000). Revised long-term national population scenarios for the European Union, Final report prepared by Statistics Netherlands, Luxembourg and Brussels. Fehr, H. (2000). Pension reform during the demographic transition. Scandinavian Journal of Economics, 102, 419– 443. Feldstein, M. (1974). Social security, induced retirement, and aggregate capital accumulation. The Journal of Political Economy, 82, 905 – 926. Feldstein, M. (1996a). The missing piece in policy analysis: social security reform. American Economic Review, 86, 1 – 14.

References

259

Feldstein, M. (1996b). The missing piece in policy analysis: social security reform. NBER Working Paper 5413, Cambridge, MA. Feldstein, M. (2001). The future of social security pensions in Europe, NBER Working Paper 8487, Cambridge, MA. Feldstein, M. (2002). “Introduction—an American perspective,” in Social Security Pension Reform in Europe, M. Feldstein and H. Siebert (eds.), The University of Chicago Press, Chicago and London, pp. 1 – 17. Feldstein, M. and Liebmann, J. (2002). “Social security” in Handbook of Public Economics, A. A. Auerbach and M. Feldstein (eds.), vol. 4, North-Holland, Amsterdam, pp. 2245– 2324. Feldstein, M. and Siebert, H. (2002). Social Security Pension Reform in Europe, The University of Chicago Press, Chicago and London. Fenge, R. (1995). Pareto-efﬁciency of the pay-as-you-go pension system with intragenerational fairness. FinanzArchiv, 52, 357– 363. Fenge, R. and Schwager, R. (1995). Pareto-improving transition from a pay-as-you-go to a fully funded pension system in a model with differing earning abilities. Zeitschrift fu¨r Wirtschafts-und Sozialwissenschaften, 115, 367– 376. Fenge, R., Uebelmesser, S. and Werding, M. (2002a). Second-best properties of implicit social security taxes: theory and evidence, CESifo Working Paper 743, Munich. Fenge, R., Uebelmesser, S. and Werding, M. (2002b). Old-age provision in ageing societies: equity, efﬁciency, and sustainability, study commissioned by the Economic and Social Research Institute (ESRI), Tokyo. Fenge, R., Meier V., Holzner C. and Werding, M. (2002c). Alterssicherungssysteme im internationalen Vergleich. Finanzierungen, Leistungen, Besteuerung. Ifo Institute for Economic Research, Munich. Fernandez, E.W. (1995). Estimation of the annual emigration of U.S. born persons by using foreign censuses and selected administrative data: Circa 1980. Population Division Working Paper 10, Washington, DC. Franco, D. (2002). “Italy—a never ending pension reform,” in Social Security Pension Reform in Europe, M. Feldstein and H. Siebert (eds.), The University of Chicago Press, Chicago and London, pp. 211– 262. Franz, W. (1985). An econometric analysis of female work participation, education, and fertility: theory and empirical evidence for the Federal Republic of Germany. Journal of Labour Economics, 3, 218– 234. Galasso, V. and Profeta, P. (2002). The Political economy of social security: a survey. European Journal of Political Economy, 18, 1– 29. Galasso, V. and Profeta, P. (2004). Lessons for an aging society: the political sustainability of social security systems. Economic Policy, 38, 63– 115. Gern, K.-J. (2002). “Recent developments in old-age pension systems: an international overview,” in Social Security Pension Reform in Europe, M. Feldstein and H. Siebert (eds.), The University of Chicago Press, Chicago and London, pp. 439 – 478. Gradstein, M. (2000). An economic rationale for public education: the value of commitment. Journal of Monetary Economics, 45, 463 –474.

260

References

Greenwood, M. J. and McDowell, J. M. (1991). Differential economic opportunity, transferability of skills, and immigration to the United States and Canada. The Review of Economics and Statistics, 73, 612– 623. Gros, D. and Hefeker, D. (1999). Factor Mobility, European Integration and Unemployment, EC/DGV—OECD/DEELSA Seminar: Wages and Employment, European Commission, Brussels, pp. 125– 146. Gruber, J. and Wise, D. A. (1999). Social Security and Retirement around the World, The University of Chicago Press, Chicago and London. Hagen, J. von and Walz, U. (1995). “Social security and migration in an ageing europe,” in Politics and Institutions in an Integrated Europe, B. Eichengreen, J. Frieden and J. von Hagen (eds.), Springer, Berlin, New York, pp. 177– 192. Haisken-de New, L.P. and Haisken-de New, J.P. (1997). A primer on the German institutional background for GSOEP users, http://panel.gsoepp.de/soepinfo2001 (Basic Information-Institutional Background). Hamada, K. (1996). Strategic Approaches to the International Economy: Selected Essays of Koichi Hamada, Elgar, Cheltenham. Haupt, A. and Peters, W. (2003). Voting on public pensions with hand and feet: how young migrants try to escape from gerontocracy. Economics of Governance, 4, 57 – 80. Hayek, F. A. von. (1960). The Constitution of Liberty, The University of Chicago Press, Chicago and London. Heckman, J. J. (1979). Sample selection bias as a speciﬁcation error. Econometrica, 47, 153 – 161. Heckmann, J. J. and MaCurdy, T. E. (1980). A life cycle model of female labour supply. Review of Economic Studies, 47, 47– 74. Hirshman, A. O. (1970). Exit, Voice, and Loyality: Responses to Decline in Firms, Organizations, and States, Harvard University Press, Cambridge, MA. Hirte, G. (1999). Raising the retirement age—why should anybody lose? Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 219, 393– 408. Hirte, G. (2000). Privatizing old-age insurance—economic effects and political feasibility, Diskussionsbeitra¨ge der Katholischen Universita¨t Eichsta¨tt, Wirtschaftswissenschaftliche Fakulta¨t Ingolstadt 138, Eichsta¨tt. Holzmann, R. (1998). Financing the transition to multipillar, Social Protection Discussion Paper 9809, The World Bank, Washington, DC. Holzmann, R., Palacios, R. and Zviniene, A. (2001). Implicit Pension Debt: Issues, Measurement and Scope in International Perspective, The World Bank, Washington, DC. Homburg, S. (1988). Theorie der Alterssicherung, Springer, Berlin. Homburg, S. (1990). The efﬁciency of unfunded pension schemes. Journal of Institution and Theoretical Economics, 146, 640– 647. Homburg, S. (1991). Interest and growth in an economy with land. Canadian Journal of Economics, 24, 450 – 459. Homburg, S. (1997). “Old-age pension systems: a theoretical evaluation,” in Reforming the Welfare State, H. Giersch (ed.), Springer, Berlin, Heidelberg, New York, pp. 233 –246.

References

261

Homburg, S. and Richter, W. (1990). “Eine efﬁzienzorientierte reform der GRV,” in Bevo¨lkerung und Wirtschaft, B. Felderer (eds.), Duncker & Humblot, Berlin, pp. 183 – 191. Homburg, S. and Richter, W. (1993). Harmonizing public debt and public pension schemes in the European community. Journal of Economics, 7(Suppl.), 51 – 63. Ja¨ger, T. and Raffelhu¨schen, B. (1999). Generational Accounting in Europe: An Overview, European Commission (eds.), European Economy—Generational Accounting in Europe, Brussels, pp. 1– 16. Kaltenborn, B. (2000). Reformkonzepte fu¨r die Sozialhilfe: Finanzbedarf und Arbeitsmarkteffekte, Nomos, Baden-Baden. Kehoe, P. J. (1989). Policy cooperation among benevolent governments may be undesirable. Review of Economic Studies, 56, 289– 296. Keller, C. (2000). Die Zukunft der gesetzlichen Rentensysteme in der Europa¨ischen Union, Institut fu¨r Wirtschaftspolitik an der Universita¨t zu Ko¨ln, Cologne. Kessler, A.S. and Lu¨lfesmann, C. (2000). The theory of human capital revisited: on the interaction of general versus speciﬁc investment, CEPR Discussion Paper 2533, London. Kifmann, M. and Schindler, D. (2001). Smoothing the implicit tax rate in a pay-as-you-go pension system. FinanzArchiv, 57, 261– 283. Killingsworth, M. R. and Heckman, J. (1986). “Female labour supply: a survey,” in Handbook of Labour Economics, O. C. Ashenfelter and R. Layard (eds.), vol. 1, NorthHolland, Amsterdam, pp. 103– 204. Kolmar, M. (1997). On the efﬁciency of national social security systems in the European Union, EIB Papers 2, Luxembourg. Kolmar, M. (1999). Optimale Ansiedlung sozialpolitischer Entscheidungskompetenzen in der Europa¨ischen Union, Beitra¨ge zur Finanzwissenschaft 7, J.C.B. Mohr (Siebeck, P.), Tu¨bingen. Kolmar, M. (2000). Soziale Mindeststandards in Europa? Journal of Applied Social Science Studies, 120, 207– 237. Konrad, K. A. (1995a). Fiscal federalism and intergenerational redistribution. FinanzArchiv, 52, 166 –181. Konrad, K. A. (1995b). Social security and strategic inter-vivos transfers of social capital. Journal of Population Economics, 8, 315– 326. Krieger, T. (2001). Intergenerational redistribution and labor mobility. A survey. FinanzArchiv, 58, 339– 361. Krugman, P. (2002). Notes on social security. Mimeo, http://www.wws.princeton.edu/ ~pkrugman/socsec.html (10 February 2003). Lindbeck, A. and Persson, M. (2003). The gains from pension reform. Journal of Economic Literature, 41, 74–112. Lu¨deke, R. (1988). “Staatsverschuldung, intergenerative Redistribution und umlageﬁnanzierte gesetzliche Rentenversicherung: Eine andere Sicht der Lasten durch ein negatives Bevo¨lkerungswachstum,” in Wirtschaftliche Strukturprobleme und soziale Fragen, J. Klaus and P. Klemmer (eds.), Duncker & Humblot, Berlin, pp. 167 – 181.

262

References

Mackenroth, G. (1952). Die Reform der Sozialpolitik durch einen deutschen Sozialplan, Schriften des Vereins fu¨r Socialpolitik, N.F.4, Duncker & Humblot, Berlin, pp. 39-76. Manski, C. F. (1990). The use of intentions data to predict behaviour: a best-case analysis. Journal of the American Statistical Association, 85, 934 – 940. Marquardt, M. and Peters, W. (1997). Collective madness: how ageing inﬂuences majority voting on public pensions, Discussion Paper 79, Europa-Universita¨t Viadrina. von Maydell, B. (1992). Die Entwicklung in Europa und ihre Auswirkungen auf die Rentenversicherung. Deutsche Rentenversicherung, 5 – 6, 315 – 326. Mayer, J. and Riphahn, R. (2000). Fertility assimilation of immigrants: evidence from count data models. Journal of Population Economics, 13, 241– 261. Merton, R. C. (1983). “On the role of social security as a means for efﬁcient risk sharing in an economy where human capital in not tradable,” in Financial Aspects of the United States Pension System, Z. Bodie and J. B. Shoven (eds.), The University of Chicago Press, Chicago and London, pp. 325– 358. Mincer, J. (1978). Family migration decision. Journal of Political Economy, 86, 749 – 773. Mosley, H. G. (1990). The social dimension of European integration. International Labour Review, 129, 147 – 164. Mulligan, C. B. (1998). Substitution over time: another look at life-cycle labour supply. NBER Macroeconomics Annual, 1998, 75– 152. Mulligan, C.B. and Sala-i-Martin, X. (1999). Gerontocracy, retirement, and social security, NBER Working Paper 7117, Cambridge, MA. Mu¨nz, R. and Ulrich, R. (1996). Internationale Wanderungen von und nach Deutschland, 1945– 1994. Allgemeines Statistisches Archiv, 80, 5 – 35. Murphy, K. M. and Welch, F. (1998). Perspectives on the social security crisis and proposed solutions. American Economic Review, 88, 142 – 150. Naskoteen, R. A. and Zimmer, M. (1980). Migration and income: the question of selfselection. Southern Economic Journal, 46, 840– 851. van den Noord, P. and Herd, R. (1993). Pension Liabilities in the Seven Major Economies, Economics Department Working Paper, 142, OECD, Paris. Oates, W. E. (2001). Fiscal competition and European Union: contrasting perspectives. Regional Science and Urban Economics, 31, 133– 145. Oates, W. E. (2002). Fiscal and regulatory competition: theory and evidence. Perspektiven der Wirtschaftspolitik, 3, 377– 390. OECD. (1996). Education at a Glance, OECD, Paris. OECD. (1998). Maintaining Prosperity in an Ageing Society, OECD, Paris. OECD. (1999). Trends in International Migration. Continous Reporting System on Migration, Annual Report, SOPEMI, Paris. OECD. (2000). Trends in International Migration. Continous Reporting System on Migration, Annual Report, SOPEMI, Paris. OECD. (2001a). Ageing and Income: Financial Resources and Retirement in 9 OECD Countries, OECD, Paris. OECD. (2001b). Trends in International Migration. Continous Reporting System on Migration, Annual Report, SOPEMI, Paris.

References

263

Oksanen, H. (2002). Pension reforms: key issues illustrated with an actuarial model, European Commission Directorate-General for Economic and Financial Affairs Economic Papers 174, Brussels. Olson, M. (1965). The Logic of Collective Action, Harvard University Press, Cambridge, MA. Onur, L. (1993). “Die EG-Sozialpolitik—Zwischenbilanz und Perspektiven: Weg oder Irrweg zur Europa¨ischen Union?” in: Sozialrecht und Sozialpolitik in Europa, Schriftenreihe des Arbeitskreises fu¨r Europarecht an der Universita¨t Osnabru¨ck (ed.), Vol. 4, Osnabru¨ck. Palmer, E. (2002). “Swedish pension reform: How did it evolve, and what does it mean for the future?,” in Social Security Pension Reform in Europe, M. Feldstein and H. Siebert (eds.), The University of Chicago Press, Chicago and London, pp. 171 – 210. Pencavel, J. (1986). “Labour supply of men: a survey,” in Handbook of Labour Economics, O. C. Ashenfelter and R. Layard (eds.), vol. 1, North-Holland, Amsterdam, pp. 3– 102. President’s Commission to Strengthen Social Security. (2001). Final Report, http://www. csss.org (10 Feb. 2003). Prognos (1987). Zur gesamtwirtschaftlichen Entwicklung vor dem Hintergrund einer schrumpfenden Bevo¨lkerung, Basel. Razin, A. and Sadka, E. (1995). Population Economics, MIT Press, Cambridge, MA. Rhee, C. (1991). Dynamic efﬁciency with land. Review of Economic Studies, 58, 791 – 797. Richter, W. (2002). Social security and taxation of labour subject to subsidiarity and freedom of movement. Swedish Economic Policy Review, 9, 47 –74. Riester, W. (2002). “The German pension reform in the European context”, in Open Coordination of Old-age Security in the European Union, Verband Deutscher Rentenversicherungstra¨ger (ed.), DVR Schriften 35, Frankfurt, pp. 26 – 30. Ru¨rup, B. (2002). “The German pension system—status quo and reform options,” in Social Security Pension Reform in Europe, M. Feldstein and H. Siebert (eds.), The University of Chicago Press, Chicago and London, pp. 137– 170. Sachversta¨ndigenrat zur Begutachtung der gesamtwirtschaftlichen Entwicklung. (1988). Jahresgutachten 1988/89, Stuttgart. Sachversta¨ndigenrat zur Begutachtung der gesamtwirtschaftlichen Entwicklung. (1996). Jahresgutachten 1996/97, Stuttgart. Sachversta¨ndigenrat zur Begutachtung der gesamtwirtschaftlichen Entwicklung. (2001). Jahresgutachten 2001/02, Stuttgart. Samuelson, P. A. (1958). An exact consumption-loan model of interest with or without the social contrivance of money. The Journal of Political Economy, 66, 467 – 482. Sandmo, A. (1974). A note on the structure of optimal taxation. American Economic Review, 64, 701 – 706. Sandmo, A. (1987). A reinterpretation of elasticity formulae in optimum tax theory. Economica, 54, 89– 96. Schmidt, C. M. (1997). Immigrant performance in Germany: labor earnings of ethnic German migrants and foreign guest-workers. Quarterly Review of Economics and Finance, 37, 379– 397.

264

References

Schnabel, R. (1998). Rate of return of the German pay-as-you-go pension system. FinanzArchiv, 55, 374– 399. Schwarze, J. (1995). Simulating German income and social security tax payments using the GSOEP, Cross-National Studies in Aging. Program Project Paper 19, Syracuse. Schweizer, U. (1995). Endogenous fertility and the Henry George theorem. Journal of Public Economics, 61, 209– 228. Sinn, H.-W. (1990). Tax harmonization and tax competition in Europe. European Economic Review, 34, 489– 504. Sinn, H.-W. (1997). The selection principle and market failure in systems competition. Journal of Public Economics, 66, 247– 274. Sinn, H.-W. (1998). European integration and the future of the welfare state. Swedish Economic Policy Review, 5, 113–132. Sinn, H.-W. (1999a). Pension reform and demographic crisis: why a funded system is needed and why it is not needed, CESifo Working Paper 195, Munich. Sinn, H.-W. (1999b). Die Krise der Gesetzlichen Rentenversicherung und Wege zu ihrer Lo¨sung, Jahrbuch 1998 der Bayerischen Akademie der Wissenschaften, Beck Verlag, Munich, pp. 95 – 119. Sinn, H.-W. (2000a). Pension reform and demographic crisis. Why a funded system is needed and why it is not needed. International Tax and Public Finance, 7, 389 – 410. Sinn, H.-W. (2000b). The threat to the German welfare state. Atlantic Economic Journal, 28, 279 – 294. Sinn, H.-W. (2001). The value of children and immigrants in a pay-as-you-go pension system: a proposal for a transition to a funded system. ifo Studien, 47, 77 – 94. Sinn, H.-W. (2002). EU enlargement and the future of the welfare state. Scottish Journal of Political Economy, 49, 104– 115. Sinn, H.-W. (2003a). The New Systems Competition, Yrjo¨ Jahnsson Lectures 1999, Basil Blackwell, Oxford and Malden, MA. Sinn, H.-W. (2003b). “The crisis of Germany’s pension insurance system and how it can be resolved,” in Public Finances and Public Policy in the New Century, S. Cnossen and H.-W. Sinn (eds.), MIT Press, Cambridge, MA, pp. 269 – 292. Sinn, H.-W. (2004). The pay-as-you-go pension system as a fertility insurance and enforcement device. Journal of Public Economics, 88, 1335– 1357. Sinn, H.-W. and Thum, M. (1999). Gesetzliche Rentenversicherungen: Prognosen im Vergleich. FinanzArchiv, 56, 104– 135. Sinn, H.-W. and Uebelmesser, S. (2000). Rentenreform, strategische Mehrheit und demographische Entwicklung: Wann kippt Deutschland um? ifo-Schnelldienst, 28–29, 20–25. Sinn, H.-W. and Uebelmesser, S. (2002). Pensions and the path to gerontocracy in Germany. European Journal of Political Economy, 19, 153 – 158. Sinn, H.-W. and Werding, M. (2000). Rentenniveausenkung und Teilkapitaldeckung: ifo-Empfehlungen zur Konsolidierung des Umlageverfahrens. ifo-Schnelldienst, 53, 12–25. Sinn, H.-W., Flaig, G., Werding, M., Munz, S., Du¨ll, N., Hofmann, H. (2001). EUErweiterung und Arbeitskra¨ftemigration. Wege zu einer schrittweisen Anna¨herung der Arbeitsma¨rkte. ifo-Beitra¨ge zur Wirtschaftsforschung, Vol. 2, Munich.

References

265

Sjaastad, L. A. (1962). The costs and returns of human capital migration. Journal of Political Economy, 70(Suppl.), 80– 93. Sjoblom, K. (1985). Voting for social security. Public Choice, 45, 225 – 240. Social Protection Committee. (2001). Adequate and Sustainable Pensions—A Report by the Social Protection Committee on the Future Evolution of Social Protection. http://europa. eu.int/comm/employment_social/soc-prot/social/news/goteborg_en.pdf. Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70, 65– 94. Statistisches Bundesamt. (1994). Statistisches Jahrbuch fu¨r die Bundesrepublik Deutschland, Metzler und Poeschel, Wiesbaden. Statistisches Bundesamt. (1999a). Statistisches Jahrbuch fu¨r die Bundesrepublik Deutschland, Metzler und Poeschel, Wiesbaden. Statistisches Bundesamt. (1999b). Wahl der Abgeordneten des Europa¨ischen Parlaments aus der Bundesrepublik Deutschland am 13. Juni 1999, Wiesbaden. Statistisches Bundesamt. (2000). Statistisches Jahrbuch fu¨r die Bundesrepublik Deutschland, Metzler und Poeschel, Wiesbaden. Statistisches Bundesamt. (2001). Statistisches Jahrbuch fu¨r die Bundesrepublik Deutschland, Metzler und Poeschel, Wiesbaden. Stein, P. (No Year). Fu¨rst Bismarcks Reden, Zwo¨lfter Band: Unter drei Kaisern 1887– 1890, Reclam: Leipzig. Strøm, S. and Wagenhals, G. (1991). Female labour supply in the Federal Republic. Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 208, 575 – 595. Swan, T. W. (1956). Economic growth and capital accumulation. Economic Record, 32, 334 – 361. Thum, C. (2000). Die o¨ffentliche Bereitstellung privater Gu¨ter, Peter Lang, Frankfurt. Thum, C. and Uebelmesser, S. (2000). Brain Gain and the Transferability of Skills, Mimeo, Universityof Munich, Munich. Thum, C. and Uebelmesser, S. (2003). Mobility and the role of education as a commitment device. International Tax and Public Finance, 10, 549 – 564. Thum, M. and von Weizsa¨cker, J. (2000). Implizite Einkommensteuer als Meßlatte fu¨r die aktuellen Rentenreformvorschla¨ge. Perspektiven der Wirtschaftspolitik, 1, 453 – 468. Tobin, J. (1958). Estimation of relationship for limited dependent variables. Econometrica, 26, 24 – 36. Uebelmesser, S. (2003a). Political Feasibility of Pension Reforms, Mimeo, University of Munich, Munich. Uebelmesser, S. (2003b). Harmonisation of old-age security within the European Union, CESifo Working Paper 1108, Munich. Uebelmesser, S. (2004). To Go or not to Go: Emigration from Germany, Mimeo, University of Munich, Munich. United Nations. (2000). Replacement migration: is it a solution to declining and aging populations?, Working Paper (ESA/P) 160, United Nations, New York. United Nations. (2001a). World Population Ageing: 1950 – 2050, United Nations, New York.

266

References

United Nations, (2001b). World Population Prospects: The 2000 Revision, vol. II, United Nations, New York. Untiedt, G. (1992). Das Erwerbsverhalten verheirateter Frauen in der Bundesrepublik Deutschland, Physica, Heidelberg. Vandenbroucke, F. (2002) Pension policy as a matter of common interest, in Open Coordination of Old-age Security in the European Union, Verband Deutscher Rentenversicherungstra¨ger (ed.), DVR Schriften 35, Frankfurt, 31 – 38. Vaubel, R. (1994). The public choice analysis of European integration: a survey. European Journal of Political Economy, 10, 227– 249. Verband Deutscher Rentenversicherungstra¨ger. (2001). Rentenversicherung in Zeitreihen, Verband Deutscher Rentenversicherungstra¨ger, Frankfurt am Main. Wellisch, D. (1999). Finanzwissenschaft III – Staatsverschuldung, Verlag Franz Vahlen, Munich. Werding, M. (2002a). Child expenditure and public pensions. International Studies on Social Security, 8, 141– 165. Werding, M. (2002b). Calculation of the Implicit Debt for Germany. Mimeo, Ifo Institute for Economic Research, Munich. Werding, M. and Blau, H. (2001). Auswirkungen des demographischen Wandels auf die o¨ffentlichen Finanzen: Modellrechnungen fu¨r die staatlichen Alterssicherungssysteme, Ifo Institute for Economic Research, Munich. Wildasin, D. E. (1991). Income redistribution in a common labor market. American Economic Review, 81, 757– 774. Wildasin, D. E. (1994). Income redistribution and migration. Canadian Journal of Economics, 27, 637 – 656. Wildasin, D. E. (1999). “Public pensions in the EU: migration incentives and impacts,” in Environmental and Public Economics: Essays in Honor of Wallace E. Oates, A. Panagariya, P. R. Portney and R. M. Schwab (eds.), Elgar, Cheltenham, pp. 253 – 282. Wildasin, D.E. (2000). Factor mobility and ﬁscal policy in the EU: policy issues and analytical approaches. CESifo Working Paper 344, Munich. Wissenschaftlicher Beirat beim Bundesministerium fu¨r Wirtschaft. (1998). Grundlegende Reform der gesetzlichen Rentenversicherung. Bundesministerium fu¨r Wirtschaft, Studienreihe Nr. 99, Bonn. Wissenschaftlicher Beirat beim Bundesministerium der Finanzen. (2001). Freizu¨gigkeit und soziale Sicherheit. Bundesministerium der Finanzen, Bonn. Wrede, M. (1999). Pareto efﬁcient pay-as-you-go pension systems with multi-period lives. Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 219, 494 – 503. Yahoo. (2002). Historical prices of stock indices, http://ﬁnance.yahoo.com (historical prices). Ziliak, J. P. and Kniesner, T. J. (1999). Estimating life cycle labour supply tax effects. Journal of Political Economy, 107, 326– 359. Zodrow, G. R. and Mieszkowsi, P. (1986). Pigou, Tiebout, property taxation and the underprovision of local public goods. Journal of Urban Economics, 19, 296 – 315.

Symbol Glossary Variable a c B C D d E F; f g g h H i i K k I L l l m N n h v p r r s S

Explanation Mobile part Consumption Subsidies Monetary costs Dependency ratio Adjustment factor for implicit tax rates for women Household income Production function Growth rate of the wages/productivity Quality of education Hours worked Human capital Internal rate of return of pension systems Internal rate of return Capital Capital intensity Investment Labour Labour supply Lagrange multiplier Growth rate of the contribution rate Population Growth rate of the population Uncompensated price elasticity Wage income ðv ¼ weÞ Pension beneﬁts Interest rate Relative price of one pension unit Saving rate Implicit saving

268

s f T t t Gt u Q U w W Y Z B used with g, Z; H C used with g, Z; H d used with i D used with g, Z; H f used with c; l; p; W F used with i M used with c; l; N m used with c; l; p; W Mo used with D nd used with i O used with c; l; N Y used with c; l; N

Symbol Glossary Compensated price elasticity Function of subsidies Implicit tax Tax rate Implicit tax rate Tax to be levied from individuals of generation t Contribution rate Total contributions Utility function Net (marginal) wage per (efﬁciency) unit of labour Gross (marginal) wage per unit of labour Output Quantity of education Index for subsidies Index for monetary costs Index for “dependent on timing” Index for discrete choice Index for women Index for funded pension systems Index for middle generation Index for men Index for mobility Index for “not dependent on timing” Index for old generation Index for young generation

Subject Index Actuarial fairness 5, 19, 20, 34, 54– 57, 67, 69, 71, 74, 76, 78, 82, 111, 112 Ageing 4, 6, 8, 25, 28, 29, 34, 37, 45, 47, 50, 54, 56, 59, 62, 65, 67, 140, 146, 148, 150, 157, 183, 200, 215, 253, 254 CESifo pension model 1, 24, 84, 105, 135, 147, 149 Cohort effects 96, 97, 102– 104, 214 Commitment device 159, 163, 164, 170, 171, 174, 177 Delayed integration 242– 245 Denmark 1, 4, 25, 35– 39, 41, 42, 44, 51, 52, 56 –59, 65– 67, 213, 214 Dependency ratio 47, 49– 51, 54, 56, 59, 67, 134, 135, 137, 139, 141, 144, 150, 160, 162, 253 Distortion 5, 6, 56, 69– 74, 81, 87, 96, 106, 109, 111, 112, 132, 136, 137, 156, 208, 212, 213, 215, 219, 221, 238 – 240, 243, 244 Distribution of competences 207– 208, 217, 226 – 237, 239, 240, 241, 245 Education 163 – 165, 187, 189, 195– 197, 199, 201, 202, 204 discrete choice 174, 175 monetary costs 173, 174, 179– 181 quantity 166, 167, 169, 170, 173, 176, 177 structure 166, 167, 169, 170, 171, 173, 177 subsidies 174 – 176 Economic Union 227, 228, 233, 234, 236, 237

Efﬁciency/Inefﬁciency 4– 8, 19, 69 – 71, 73 –76, 107, 109, 110, 113, 136, 137, 173, 177, 178, 207 – 212, 215 – 219, 221 – 225, 227, 234, 236, 238 – 244, 254 Employment principle 207, 210, 213, 217, 241 – 245 European Court of Justice 229, 230, 234, 239, 240 Eurostat 24, 35 – 37, 44 – 46, 48, 49, 51, 52, 147 – 150, 152, 157, 208, 225 Exit option 7, 131, 132, 156, 159 – 163, 178, 254 Fertility rates 4, 33 – 35, 37, 45 – 47, 49 –51, 54, 135, 137, 148, 183, 221 – 225 France 1, 25 – 27, 35 – 40, 42, 44, 46, 47, 49, 51, 52, 55 – 59, 62, 63, 67, 68, 131, 133, 142, 144 – 148, 150 – 157, 213, 214 Freedom of movement 228, 229, 232, 234, 236 Generational accounting 26 – 28 Gender differences 76, 81 – 83, 89, 108, 109 Gerontocracy 6, 7, 131, 132, 147, 153 – 156, 159, 160, 163, 164, 166, 177, 178, 208, 244, 254 German Socio-Economic Panel (GSOEP) 92, 116, 117, 188, 190, 191 Germany 1 – 4, 6, 7, 16 – 18, 24 – 28, 34 –40, 42 –53, 55, 56, 58 – 63, 67, 68, 70, 76, 79, 83 – 86, 89 – 90, 92,

270

Subject Index

95, 105, 111, 114, 116, 117, 131, 133, 135, 143 – 157, 161, 183– 185, 187 – 191, 197 –200, 213, 214 Riester Reform 60, 62, 63, 142 Growth model 228, 229, 232, 234, 236 dynamical efﬁciency/inefﬁciency 15, 16, 18, 23, 55 Golden Rule 15, 16, 23, 31 Harmonisation 7, 8, 207– 211, 215, 217 – 219, 221 –227, 229, 233 – 238, 240 –243, 245, 254 centralisation 209, 211, 217, 218, 221, 222, 224, 225, 235 co-ordination 8, 218, 224, 225, 234, 236, 237, 240, 243, 245 decentralisation 211, 218, 225, 240, 243 equalisation 8, 218, 221– 224, 234, 236, 240, 243, 245 mutual recognition 209, 218, 227, 228, 236, 237, 239, 240 Heckman 90, 92, 98, 117, 118 Home-country principle 242– 244 Implicit debt 5, 6, 9, 19, 21–29, 33, 69 – 71, 74, 75, 132 Implicit taxes 5, 6, 8, 9, 19– 23, 28, 29, 33, 53, 56, 69 – 71, 74–82, 84– 86, 89, 90, 104 –105, 107– 113, 123, 128, 129, 132 – 137, 146, 148, 150, 151, 156, 210 – 215, 218– 224, 238 – 240, 243, 244, 254 intergenerational context 133 international context 211 intrapersonal context 76 Indifference age 131, 147, 149, 150, 152, 153, 157, 158 Internal rate of return 5, 12– 24, 29, 53, 55 Interest groups 155, 244 Interjurisdictional transfers 217, 242, 243 Intergenerational fairness 57, 136, 137, 140, 216 introductory gains 5, 19, 69, 71, 74

Inverse elasticity rule 69, 88, 89, 100, 105– 107, 109 – 112 Italy 1 – 3, 18, 25 – 27, 35 – 40, 42, 44, 46, 47, 49, 51 – 56, 58, 59, 63 – 65, 67, 68, 131, 133, 142, 144 – 157, 213, 214 Japan 2, 3, 17, 25, 26, 35, 37, 46, 48– 50, 82 Labour allocation 5, 7, 8, 37, 178, 207– 212, 215 – 219, 221 – 225, 234, 238 – 245 Labour supply elasticity 6, 68, 69, 75, 81, 87–91, 96 –98, 100, 110, 112, 113, 123, 128, 129, 132, 136 Leviathan 244 Life cycle 5, 6, 29, 69, 70, 75 – 78, 80, 81, 84, 85, 87, 89, 95 – 97, 102 – 106, 108– 110, 112, 113, 123, 132, 134, 161, 213, 214, 239 Life-expectancy 4, 17, 33, 34, 36, 37, 45, 47, 49, 50, 54, 63 – 65, 82 – 84, 86, 135, 137, 148, 183, 215, 241 Mackenroth 72 Market rate of return 5, 14 – 23, 29, 53, 55 Median age 131, 147 – 149, 152, 153, 156, 157, 160 – 163, 177, 178 Median voter 6, 138, 140, 147, 149 Migration 4, 5, 7, 19, 33, 34, 36 – 50, 157, 163, 164, 166, 167, 170, 171, 177, 178, 184 – 187, 190 – 194, 199, 200, 208, 210, 212, 214 – 218, 221 – 226, 228, 229, 239, 240, 242 – 245, 254 intention 185, 186, 188 – 190, 193 reasons 185, 197, 188, 190, 191 Mobility conditional 7, 208 marginal 184, 200 restricted 178, 219, 223 – 225, 236 unconditional 7 unrestricted 178, 219, 221, 222, 225, 236 Naturalisation 157 Net pension liability 26 – 28

Subject Index Open method of co-ordination 238– 240, 246 Outside option 156, 160– 162, 164, 177, 178, 244 Overlapping generation model (OLG) 10, 29, 76, 77, 133, 138, 160, 219 Pareto improvement 4, 6, 15, 16, 23, 69– 71, 73 – 75, 109, 132, 133, 135, 215 Pension expenditure 24– 26, 57– 59 Pension reform 2, 5, 6, 14, 25, 28, 51, 53, 54, 56–59,67–70,75,131–133,136–138, 144, 226, 238, 245, 253, 254 feasibility 4 – 7, 57, 68, 131– 133, 141, 142, 144, 146, 147 fundamental 5, 24, 53, 59, 67 knowledge 67 – 68, 154 parametric 6, 53, 57, 67 Pension systems Beveridge 1, 54, 217, 242, 245 Bismarck 1, 54, 217, 242, 245 deﬁned beneﬁt (DB) 53– 57, 59, 63– 65, 67, 137, 144 deﬁned contribution (DC) 53– 57, 59, 63 – 67, 144 funded 2, 5, 9, 10, 13– 16, 18, 19, 21, 29, 53 – 57, 59, 67– 74, 136, 137, 142, 144, 147, 153– 155, 215, 216, 254 unfunded/pay-as-you-go 1, 2, 3 – 21, 23, 29, 46, 47, 50, 53– 57, 59, 67, 69 – 78, 84, 89, 108– 112, 131– 133, 135 – 137, 141, 142, 144, 147, 153 – 155, 161, 163, 183, 184, 199, 208 – 210, 212, 213, 215, 216, 238, 242 – 244, 253, 254 Pillars of old-age security 2 – 4, 210 Population growth 5, 33, 34, 36, 45, 54, 183, 221 Age pyramid 48 Age structure 5, 47 Probit 194, 200, 206 Race to the bottom 8, 178, 183, 184, 200, 208, 215, 217, 238, 240, 241, 246

271

Ramsey rule 88 Risk-return relation 18 Second-best optimality 87 Selection principle 209, 241 Simulation 90 – 92, 143, 194 – 197 Social contract 18, 209, 243 Social planner 172 – 174, 176, 177 Social Union 227, 228, 230, 231, 233, 236, 237 Spain 25, 27, 35 – 40, 42, 44, 46, 47, 49, 51, 52, 55, 57, 58, 68, 154 Standardised work biography 113 Subsidiarity principle 235, 241 Sustainability 2, 5, 7, 28 – 29, 33, 46, 47, 50, 53, 54, 56 – 58, 67 – 69, 132, 133, 136, 137, 162, 200, 207, 219, 238, 244, 245, 253, 254 Sweden 2 – 4, 18, 25, 35 – 39, 41, 42, 44, 51 –53, 55, 56, 58, 59, 63 – 65, 67 Time-consistency 163, 164 Tobit 91, 92, 98 – 100, 119, 124 Treaty of Amsterdam 227, 228, 230 – 232, 234, 246 – 248 Treaty of Maastricht 216, 231 Treaty of Nice 232, 237, 247, 249 Treaty of Rome 216, 226, 228, 246, 247 UK/United Kingdom 1– 3, 25 – 27, 35– 40, 42, 44, 46, 47, 49, 51 – 53, 55, 57 –59, 65 –67, 231 US/United States 2, 3, 16, 25, 26, 35, 37, 46, 48 – 50, 82, 135, 184 Voiceoption19,131,156,159,160,162,178 Voting 6, 131, 132, 138, 141, 142, 152, 153, 155, 156, 159, 169, 162, 178, 254 majority 6, 7, 28, 131, 132, 138, 140 – 142, 144, 147, 152 – 156, 159, 160, 178, 244, 253, 254 model 138, 160 participation 153 Zero-sum game 23

This Page Intentionally Left Blank

Cognitive Economics: New Trends, Volume 280 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Fighting Corruption in the Public Sector, Volume 284 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Designing the New European Union, Volume 279 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Unfunded Pension Systems: Ageing and Migration, Volume 264 (Contributions to Economic Analysis)

Cognitive Economics: New Trends, Volume 280 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Fighting Corruption in the Public Sector, Volume 284 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Designing the New European Union, Volume 279 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Tourism, Trade and National Welfare, Volume 265 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Explaining Growth in the Middle East, Volume 278 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

New Directions in Macromodelling, Volume 269 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Hydrocarbon Migration Systems Analysis

Urban Dynamics and Growth: Advances in Urban Economics, Volume 266 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Pension Economic

Ageing And Pension Reform Around The World

Ageing And Pension Reform Around The World

The Single Server Queue (Contributions to Economic Analysis)

Contributions to non-standard analysis

Pension Systems: Sustainability and Distributional Effects in Germany and the United Kingdom (Contributions to Economics)

Modeling Pension Systems

European Union and the Euro Revolution, Volume 283 (Contributions to Economic Analysis)

Introduction to Economic Analysis

Introduction to Economic Analysis

Pension Politics : Consensus and Social Conflict in Ageing Societies

Advances in Psychology Volume 52 - Human Mental Workload (Contributions to Economic Analysis)

The Cultures of Economic Migration (Studies in Migration and Diaspora)

Costs and Benefits of Collective Pension Systems

Families Caring across Borders: Migration, Ageing and Transnational Caregiving

Globalisation, Migration and Socio-Economic Change

Regionalisation, Growth, and Economic Integration (Contributions to Economics)

International Migration: Trends, Policy and Economic Impact (Contemporary Economic Policy)

Sexuality, Sexual Health and Ageing (Rethinking Ageing)

Spatial Economic Analysis - Volume 1, Issue 2

Spatial Economic Analysis - Volume 1, Issue 1

Communications in Mathematical Physics - Volume 264

Unfunded Pension Systems: Ageing and Migration, Volume 264 (Contributions to Economic Analysis)

Cognitive Economics: New Trends, Volume 280 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Fighting Corruption in the Public Sector, Volume 284 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Designing the New European Union, Volume 279 (Contributions to Economic Analysis) (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Tourism, Trade and National Welfare, Volume 265 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Explaining Growth in the Middle East, Volume 278 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

New Directions in Macromodelling, Volume 269 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Hydrocarbon Migration Systems Analysis

Urban Dynamics and Growth: Advances in Urban Economics, Volume 266 (Contributions to Economic Analysis) (Contributions to Economic Analysis)

Pension Economic

Ageing And Pension Reform Around The World

Ageing And Pension Reform Around The World

The Single Server Queue (Contributions to Economic Analysis)

Contributions to non-standard analysis

Pension Systems: Sustainability and Distributional Effects in Germany and the United Kingdom (Contributions to Economics)

Modeling Pension Systems

European Union and the Euro Revolution, Volume 283 (Contributions to Economic Analysis)

Introduction to Economic Analysis

Introduction to Economic Analysis

Pension Politics : Consensus and Social Conflict in Ageing Societies

Advances in Psychology Volume 52 - Human Mental Workload (Contributions to Economic Analysis)

The Cultures of Economic Migration (Studies in Migration and Diaspora)

Costs and Benefits of Collective Pension Systems

Families Caring across Borders: Migration, Ageing and Transnational Caregiving

Globalisation, Migration and Socio-Economic Change

Regionalisation, Growth, and Economic Integration (Contributions to Economics)

International Migration: Trends, Policy and Economic Impact (Contemporary Economic Policy)

Sexuality, Sexual Health and Ageing (Rethinking Ageing)

Spatial Economic Analysis - Volume 1, Issue 2

Spatial Economic Analysis - Volume 1, Issue 1

Communications in Mathematical Physics - Volume 264

Recommend Documents