CONTRIBUTIONS TO ECONOMIC ANALYSIS 179 Honorary Editor: J.TINBERGEN Editors: D. W. JORGENSON J.WAELBROECK NH q>^C NORTH-HOLLAND AMSTERDAM · NEW YORK · OXFORD TOKYO THE POLITICAL ECONOMY OF SOCIAL SECURITY Edited by: B.A. GUSTAFSSON University of Göteborg Sweden and N. Anders KLEVMARKEN University of Göteborg Sweden til 1989 NORTH-HOLLAND AMSTERDAM · NEW YORK · OXFORD TOKYO ©ELSEVIER SCIENCE PUBLISHERS B.V., 1989 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical photocopying, recording or otherwise, without the prior written permission of the publishers, Elsevier Science Publishers B.V. (Physical Sciences and Engineering Division), P.O. Box 1991,1000 BZ Amsterdam,The Netherlands. Special regulations for readers in the USA. -This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside of the USA, should be referred to the copyright owner, Elsevier Science Publishers B.V, unless otherwise specified. 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. ISBN: 0 444 87141 1 Publishers: ELSEVIER SCIENCE PUBLISHERS B.V. P.O. Box 1991 1000 BZ Amsterdam The Netherlands Sole distributors for the U.S.A. and Canada: ELSEVIER SCIENCE PUBLISHING COMPANY, INC. 655 Avenue of the Americas New York, N.Y. 10010 U.S.A. Library of Congress Catalog1ng-1n-Publicatlon Data Gustafsson, Björn. The political economy of social security / Björn Gustafsson and N. Anders Klevmarken. p. cm. — (Contributions to economic analysis ; 179) Includes bibliographies and index. ISBN 0-444-87141-1 (U.S.) : fl 175.00 (Netherlands) 1. Social security—Germany (West) 2. Social security—Sweden. 3. Social security—Great Britain. 4. Social security. I. Klevmarken, N. Anders. II. Title. III. Series. HD7179.G87 1989 368.4'0094—dc 19 88-38127 CIP PRINTED IN THE NETHERLANDS v Introduction to the series This series consists of a number of hitherto unpublished studies, which are introduced by the editors in the belief that they represent fresh contributions to economic science.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 methods 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 The Political Economy of Social Security 1 B.A. Gustafsson and N. Anders Klevmarken (eds.) © Elsevier Science Publishers B.V. (North-Holland), 1989 /. The Political Economy of Social Security Editor's Introduction and Summary Björn Gustaf sson and N. Anders Klevmarken Department of Economics, Gothenburg University In most Western countries the modern social security systems were gradually developed in the decades of economic growth following World War II. Expendi­ tures on social security have increased rather dramatically since the beginning of the 1960's. A study from the OECD (1985) shows, for instance, that in the United Kingdom they increased from 6.7 per cent in relation to GDP in 1960 to 12.4 percent in 1981, in Sweden from 7.4 to 17.9, in the Federal Republic of Germany from 15.0 to 19.8 and in the United States from 6.0 to 11.1. Although these figures may not be fully comparable, they all show an impressive increase 1}. In the 1980's we have experienced a shift in focus of the political debate from the previous interest in expanding and improving the social security safety net, to concern about the future stability of the social security system. Economic growth has declined and views about the future have become more pessimistic. Expendi­ tures on social security and, in particular, on old age pensions are, however, even without reforms expected to grow, principally because of the aging population. The share of the population above 65 is expected to increase rather dramatically until year 2025 in a number of countries and the support ratio, i.e. the number of workers per pensioner, will decline, see Table 1. Since the social security system in most countries is of the pay-as-you-go type it is obvious that the burden of adjustment to adverse demographic developments falls primarily on the working population. If benefits are not decreased social security taxes will have to be increased. Projections show that for some countries, 1} These figures include the following expenditures: - Pension (Expenditures on old-age, disability or survivors' benefits, other than government employees, and government employee pensions.) - Unemployment compensation (Expenditure on social insurance and other government schemes to compensate individuals for loss of income due to unemployment.) - Other social expenditure (Expenditure on sickness, maternity or temporary disablement benefits, family and child allowances, other social assistance and expenditure on welfare affairs and services.) 2 B. Gustafsson and N.A. Klevmarken Table 1 Demographic changes Percentage of Worker per pensioner total population aged 65 and over 1980 2025 1980 2025 Fed. Rep. of Germany 15.5 23.6 2.75 1.56 Japan 9.0 21.2 5.61 2.38 Sweden 16.4 21.5 2.69 2.10 United Kingdom 14.8 18.5 2.73 2.18 United States 11.2 19.4 4.49 2.53 Source: Halter & Hemming (1987) and own calculations based on data from Statistics Sweden. Table 2 Social Secutiry Tax Rates, 1980 and 2025, Under 1980 Status Quo and with Recent Reforms (In percent) Country 1980 2025 1980 status with recent quo reforms Fed. Rep. of Germany 14.6 25.7 23.1 Japan 6.4 19.5 13.6 United Kingdom 10.5 17.2 14.4 United States 9.4 16.7 15.1 Source: Table 10 in Halter & Hemming (1987) and under unfavourable conditions, social security taxes may have to be close to 30 percent of the wage bill around year 2025. Halter and Hemming (1987) have estimated the future tax rates needed to cover the pension systems of four OECD countries. Also if already decided reforms are taken into consideration a substan­ tial tax increase can be expected, in particular for the Federal Republic of Germany and Japan. Their results are reproduced in Table 2. The recently published projections by the National Swedish Insurance Board show that the increase needed to maintain the present pension obHgations in real terms depends very much on future economic growth 2). If there is no growth at all the tax rate would have to increase from the present 24 per cent to 51.2 per cent in the year 2025, on the other hand an annual growth rate of 2 per cent would lead to a tax decrease by a few percentage points. Social security taxes combined with other taxes may well produce a total tax burden which could compel the voting majority not to honor previous pension obHgations but rather to decrease the benefits of the retired. In Sweden pensions are consumer price and not wage indexed. There is also a ceiling for pension benefits. These two features combined imply that the higher real economic growth the less becomes the future burden of the pension system. The political economy of social security 3 Given this perspective, economic research has turned to the analysis of the properties of existing social security systems, their dependence and impact on national economies, political decision making on social security, reforms which could prevent a possible conflict between retired and working generations, as well as alternatives to existing pay-as-you-go systems. The present volume is based on papers originally presented at a symposium at Fiskebäckskil outside Gothenburg, Sweden in the spring of 1987. They were subsequently refereed and revised. Various aspects of the social security system are covered and the papers represent different research approaches and draw on experiences from several countries. 1. Unemployment and demographic change In his discussion of social security in the Federal Republic of Germany Hans-Jürgen Krupp emphasizes the importance of reducing unemployment. As shown by John Creedy and Richard Disney for Britain a decrease in the unemployment rate will not only reduce the need for unemployment benefits but also for social security benefits generally. Low unemployment also improves the dependency ratio. However, a policy to fight unemployment should not be based on the removal of people from the labour force. For instance, it has been suggested that a lower retirement age could reduce unemployment, but it is not likely to be good policy. On the contrary, its effectiveness is doubtful and, as shown by Krupp, to ease the financial burden of future social security payments one would rather increase the pension age 3). The Federal Republic of Germany is a country of particular interest because its dependency ratio is projected to become very low, with the taxburden of the present social security system projected to be higher than in most other Western countries. Thus, the need for reform is accentuated. However, an unexploited reserve of female labour might contribute to a somewhat brighter future for the German social security system. Female labour force participation rate is low, not only when compared to the Scandinavan countries, but also compared to, for instance, the United States. Robin Boadway and David Wildasin also analyse the consequences of changes in the population growth rate on the social security program. In a theoretical approach they use an overlapping generations model with capital markets fully specified to model the political decision making. A decrease in population growth will increase social security taxes and thus decrease the desired level of social security for any voter below pension age. However, it also increases the median age of the voters. Because older voters will favour higher benefits, the net impact of the two effects cannot be determined theoretically. They also discuss how the length of the voting periods would influence voting 3) It is difficult to gain political support, not only in countries with high unemployment, for a policy which would increase the retirement age. Survey results from Sweden indicate that very few are willing to work additional years even if they were compensated by a higher pension, see Klevmarken (1986). 4 B. Gustafsson and N.A. Klevmarken behaviour. In the absence of altruistic motives for transfers from young to old, shorter voting periods combined with uncertainly about the willingness of future generations to vote for a maintained benefit level will induce the median voter to opt for relativly low benefits and taxes. 2. Distributional issues A pay-as-you-go system works as a chain letter. The generation which intro­ duces it gets a windfall gain from their children. As long as one generation is able to pass this implicit debt on to the next generation, the chain is not broken and no generation experiences a windfall loss. If, however, some generation decides to break the chain, for instance, because they find the financial burden too severe, their parents will experience a loss of benefits. Also a mature and functioning pay-as-you-go system may result in a (minor) redistribution between generations depending upon the precise rules of the system, the demographics of its population and the long-run changes of the national economy. In addition, most social security systems redistribute resources from affluent and healthy groups to more disadvantaged groups. Ann-Charlotte Stählberg has analysed the redistributional properties of the Swedish social security system using micro data. Although the first generations covered by the earnings-related pension system did not get full coverage, she finds that they got a sizable windfall gain, with 75-90 percent of their benefits representing an intergenerational transfer payment. More interestingly, however, Stählberg also finds that the working generations of today (in her study the birth cohorts 1944-1950) on average are not likely to get an actuarially fair compensa­ tion for what was paid into the system. The pensions may thus become the subject of an intergenerational controversy. Whether this will become an im­ portant political issue is to a large extent dependent on the rate of growth of the Swedish economy, (cf footnote 2 on page 2). Stählberg also shows that there is a redistribution from men to women and between socioeconomic groups. The major channels are the disability pension and the sickness benefits. Females use the latter benefits more than men and blue-collar workers use both types of benefits more frequently than white-collar workers. The social security system does not only influence the distribution of income but also the distribution of wealth. When the pay-as-you-go systems were introduced their consequences for savings behaviour were much discussed. There is no paper in this volume which follows this line of research. Michael Hurd, however, analyses how the relative shares of annuity wealth and bequeathable wealth are influenced by social security in the United States. Because it is not possible to borrow against future social security benefits, an exogenously given social security annuity might force people to follow a consumption path which yields less utility than one would have obtained from an annuity with the same expected present value, but in which the consumption path could be chosen independently of the annuity path. The political economy of social security 5 Based on a sample of single people Hurd estimates the distribution of the marginal rate of substitution between annuity and bequeathable wealth. His estimates are sensitive in levels to the method of estimation, but he concludes that while a majority in the sample have a marginal rate of substitution exceeding one, there is a substantial fraction with a marginal rate less than one. They are thus overannuitized. Hurd also shows that the marginal rate of substitution varies positively with bequeathable wealth, which means that at the margin the wealthy gain more utility from an increase in social security benefits than do the poor. 3. The feasability of transition and stability As discussed above there are strong forces seeking to reform the present pay-as-you-go systems. When the rates of population and economic growth are low in relation to the real rate of interest - a situation experienced in many countries in the 1980's - it is natural to investigate whether a transition to a capital reserve system is feasible without loss of welfare. For a conversion policy to be successful, the retired generations at the time of conversion must not loose their pension. Under the pay-as-you-go system they would have had no-incen­ tives to accumulate funds and must be provided for by the government. Part of the future gains of the working generations deriving from conversion to a capital reserve system might be transferred to the retired generations if the government takes up a loan and uses it for pension payments. The government of a small open economy might be able to borrow abroad without affecting the interest rate. However, Harrie Verbon finds that such a transition is not possible without loss to at least one generation. He also analyses the case of a temporary transition to a capital reserve system. Also in this case it is impossible to completely cover the government debt by taxing the gains of the (future) working generations. Verbon shows, however, that after a reconversion to a pay-as-you-go system, when the population growth rate exceeds the real rate of interest, the debt per capita decreases. Two modes of financing a pay-as-you-go system are compared in the theoreti­ cal paper by Bernard van Praag and Remco Oostendorp. More specifically they analyse how an employer financed unemployment insurance versus a worker financed insurance affects the labour market. They use a stylized model of a representative profit maximizing firm which demands labour of differing quality. The demand for labour has to satisfy the balance equation of the pay-as-you-go system, i.e. total contributions must equal the sum of all benefits. They find that there may exist multiple equilibria, all of which need not be stable. In a numerical example, using a CES-production function, they show that an em­ ployer financed system is more likely to be unstable than a worker financed one. In a system with high benefits, increased employer contributions to meet a potential deficit may increase unemployment dramatically because the wage bill becomes too high to make employment profitable. 6 B. Gustafsson and N.A. Klevmarken 4. Economic incentives and the impact of changes Any major reform of the social security system will clearly have to address macro economic issues. Edward Palmer and Märten Palme employ a small econometric n aero model of the traditional type to simulate the effects of a fully financed increase in pension benefits in Sweden. When financed by higher employer contributions, the increase in labour costs initially affects prices, wages as well as profits, but the costs are gradually passed on to wages. For instance, the partial effect on the money wage rate of a percentage point increase in the employer contribution is an immediate decrease of 0.2 per cent and after six years a decrease of 0.7 per cent. The effects of a higher contribution rate are counterbalanced by the demand stimulus from increased pension payments. While the combined effect on GDP is small, the trade balance is worsened. Wages, profits and to some extent gross business investments decrease. There is also a shift in employment from industry to the public sector. These results thus demonstrate the interdependence between social security policy and general macro economic policy. Changes in the social security system do not only concern the distribution of welfare but also short-run economic balance and economic growth. Another approach to incorporate macro economic effects in a study of changes in the social security system is taken by Peter ten Hacken, Arie Kapteyn and Isolde Woittiez. To analyse the effects of changes in the unemployment benefits on labour supply and income distribution in the Netherlands they use an integrated micro-macro model. The aggregate labour supply equation in a con­ ventional macro model is replaced by a micro simulation model based on household data. Household labour supply is explained in a neoclassical frame­ work taking nonconvex budget sets into account. A specific feature of their approach is an attempt to model preference interdependence. An advantage of the micro simulation approach is that the effects on labour supply can be investigated for various household types and that other distributional issues can be investigated as well. Although their work is still of an experimental character, some of the results are suggestive. For instance, they find that only singles and, in particular, single females adjust their labour supply to changes in the unemploy­ ment benefits. Married men and women are not responsive. Unemployment benefits and labour supply is also the topic of the contribution by Anders Björklund and Bertil Holmlund. Their problem is to determine how much, the successively more generous unemployment compensation in Sweden, has contributed to the marked increase in unemployment duration. They use an approach different from that of ten Hacken et al. In a search theoretical framework they exploit about twenty years of time-series information from the Swedish labour force surveys. Their results suffer the usual shortcomings of time-series analysis, i.e. uncertainty in model specification and imprecise esti­ mates. The evidence to support the hypothesis that increased unemployment benefits contributed to the increase in unemployment duration is rather weak. They did not, like ten Hacken et al, split their sample by marital status but by age, type of coverage and duration. Only for medium-term unemployed above the