Rethinking the Welfare Society
Designing Policy Reforms and Incentive Mechanisms

The second in a series of three CEPR workshops entitled ‘Rethinking the Welfare Society’ was held jointly with the Instituto de Estudios Económicos de Galicia, Pedro Barrié de la Maza (IIEG/PBM) at La Coruña on 27/28 September 1998. The workshop was organized by Kai Konrad (Freie Universität Berlin and CEPR), Martin Rein (Massachusetts Institute of Technology) and Dennis Snower (Birkbeck College, London, and CEPR).

CEPR’s welfare state network has been examining alternative strategies for reforming welfare activities, namely social insurance, life-cycle transfers, redistribution and the provision of welfare services such as healthcare, education and training. The focus has been on the design of policy reforms and incentive mechanisms in the advanced market economies, especially those of the EU and the United States. The intention of this workshop was to explore how the various welfare activities were to be divided amoung the government, business enterprises, households and other economic and political institutions. With government being only one of several sources of welfare activities, the papers covered the whole of ‘welfare society’ rather than just the ‘welfare state’.

Pedro Pita Barros (Universidade Nova de Lisboa and CEPR) opened the first session on ‘Efficiency and Redistribution’ with a presentation on ‘Efficient Capitation Transfer Systems’. The author addressed the issue of how to set incentives in payment schemes for health-care purchasers, such as insurance funds, without allowing them to select the best risks. Reform proposals in several countries had advocated some type of capitation system, in which fixed amounts would be paid ex ante for each insured individual. These proposals faced the problem that incentives to risk-selection were prevalent in the system. A considerable literature has been devoted to ways of mitigating, if not eliminating, the problem without destroying incentives to efficiency. Pita Barros proposed a transfer system that, under some circumstances, would attain efficiency without risk-selection. The system would extend typical linear capitation formulas and could be interpreted as a fixed transfer at the beginning of the period plus an ex-post fund at the end of the period. The ex-post adjustment fund was defined to be a financially balanced scheme inducing a socially optimal level of cost-reducing effort. The novelty lay in the way contributions to the fund were defined. The risks could be categorized in principle by exogenous characteristics such as age, sex or other variables.

Dennis Snower suggested that if this risk grouping were not sufficiently detailed there might still be adverse selection within each category across different insurances. Jean-Charles Rochet (GREQAM, IDEI, Université des Sciences Sociales de Toulouse, and CEPR) pointed out that the model did not control for the effect of different quality in health-care provision. 

Graziella Bertocchi (Università di Modena and CEPR) presented ‘The Politics of Cooptation and the Origin of the Welfare State’ which she had co-authored with Jody Overland and Michael Spagat. The authors considered the historical origins and determinants of the current welfare state. Acknowledging the diverse nature of the social and political contexts in which welfare regimes came into existence in different countries, they constructed a theoretical framework illustrating how a self-interested elite could make strategic use of cooptation to control the threat of revolution by the poor. Assuming non-overlapping generations, such an elite could reduce the probability of overthrow by coopting some poor individuals into the middle class. In the authors’ view, the model captured the essentials of the early evolution of the welfare state, class structure, and political stability.

the result that co-optation might represent the buyout of votes. Hans-Peter Grüner (Universität Bonn and CEPR) remarked that a dictator might choose to pay either for a military force to suppress the poor or for transfers to coopt. The point to be explained is then why the transition from military to transfer expenditures might occur.

The paper given by Jeff Frank (Royal Holloway College, University of London), and entitled ‘How to Ration the Public Provision of Private Goods’, looked at the provision of private goods as a redistributive mechanism and posed the question of how such redistribution could be achieved at minimum cost. The question was high on the agenda of public policy discussion: it was now accepted that, in order to cut welfare-state costs, benefits needed to be better targeted, and that targeting efficiency could be better reached by self-selection as opposed to reliance on the intervention of bureaucrats. Frank discussed alternative targeting provisions, such as low-quality public provision, queuing or waiting lists, by means of which the authority could deter better-off households. His paper concluded that the optimal policy was to combine low-quality provision with waiting lists. Self-selection by quality was preferred to rationing by queues and by other forms of ‘ordeals’. In contrast, rationing via a waiting list, which led to the household obtaining the good on the private market while waiting, could be desirable. Qualifying restrictions on the waiting list were found to be optimal only if the public quality of provision was at the minimum possible level.

Frank’s conclusions provoked considerable debate, not least because of the suggestion that individuals in need of assistance should be placed on waiting lists without regard for the urgency of their need – a seemingly counter-intuitive proposal that derived from the formulation of the utility function and the particular set-up of the model. Kai Konrad argued that a queuing-only solution might very well be first best if the quality of the private good provision was already at the minimum level. Using the example of public housing, however, Dennis Snower pointed out that letting individuals wait for provision of acceptable dwellings could hardly be optimal if unmet need was not even considered in the allocation of the housing stock. In response, Frank emphasized that the decision of a middle-class individual to consume slum housing might well be voluntary, in which case the queuing solution would be justified.

The ensuing three papers dealt with the theme of ‘Risk Management’. In ‘Rethinking the Welfare Society: The Case for Equal Opportunity Policies’, Hans-Peter Grüner (Universität Bonn and CEPR) compared two social welfare systems with respect to their consequences for risk-taking, entrepreneurship, aggregate output, and individual and social welfare. System 1 was the classical welfare state, relying exclusively on the redistribution of earned income. In system 2, the state provided equal opportunities for all individuals before investment decisions were made. This was achieved by state redistribution of initial financial wealth, with the rest left to the market. Income in Grüner’s model was generated only by risky investments in which the agents were either entrepreneurs or providers of finance. In contrast with previous studies, agent heterogeneity in the form of wealth endowment was observable in this model, while entrepreneurial effort constituted private information.

The author concluded first that ex-post redistribution was detrimental to entrepreneurial incentives, but that ex-ante redistribution could enhance productive efficiency. He demonstrated that the set of equilibrium allocations coincided with the set of constrained Pareto-optimal allocations. This implied that ex-ante redistribution was at least as good as, and did not need to be supplemented by, any ex-post redistributive measure. The paper also compared the roles of the two welfare systems when agents differed in their entrepreneurial ability. Equal opportunity policies then led the market to select entrepreneurs of higher quality by generating an equilibrium in which only the most talented individuals received credit for their projects. Finally, it was shown that there were cases where individual gambling behaviour could completely offset attempts to generate more ex-post equality through ex-ante redistribution.

Kai Konrad commented that it is possible in principle to render an economy less, rather than more, equal by ex- ante redistribution. Jeff Frank noted that Grüner’s results depended on the existence of a minimum firm size, an assumption which Grüner considered reasonable. Responding to a question from Dennis Snower about the implications for welfare-policy reform, Grüner argued that it was important to create equal ex-ante opportunities, and that involuntary insurance should be abandoned as an incentive-reducing ex-post redistributive mechanism.

Jean-Charles Rochet (GREQAM, IDEI, Université des Sciences Sociales de Toulouse, and CEPR) presented ‘The Political Economy of Public Health Insurance’, which was co-authored with Dominique Henriet. The paper examined the determinants of the principle and the scale of public provision of health insurance, which was observed to vary widely across countries. A large positive correlation existed between preferences for redistribution and the share of the public sector in health expenditures, suggesting that health insurance was being used as a redistributive device. The study was based on Mirrlees’s classic 1971 income-tax model, and introduced an illness risk that varied across the population and was not observable a priori by individuals. The government had available the income-tax schedule and public insurance coverage as policy tools. It was assumed that the insurance company had the same information as individuals, thereby obviating any adverse selection problems in the model.

The authors showed that public provision of health insurance occurred only if, on average, the probability of illness (or the morbidity index) was bigger for lower income groups. This was in accordance with theoretical predictions that, in such circumstances, public provision constituted an efficient instrument for income redistribution, complementary to income taxation. Different governments then chose different levels of public coverage according to their redistributive preferences. The theory also predicted a positive correlation between the extent of public health insurance and the progressivity of income taxes. Dennis Snower pointed out that the model would have different implications if the probability of illness was a more direct function of income rather than of productivity. Pedro Pita Barros suggested that there might be alternative explanations for the empirical correlation between health-care provision and the propensity for redistribution.

Fiorella Padoa Schioppa Kostoris (Università degli Studi di Roma, ISPE, and CEPR) presented ‘Family Income and Wealth, Youth Unemployment and Active Labour Market Policies’, written jointly with Claudio Lupi (ISPE). This empirical study investigated the unemployment situation among younger workers, based on cross-section micro data from Italy. The results showed that, together with product-market and labour-market conditions, personal and family characteristics played a crucial role in explaining youth activity and unemployment rates of short and long duration. In particular, the income effect seemed relevant for participation decisions, while family wealth helped reduce youth unemployment. The study simulated the effectiveness of various policy instruments in reducing youth unemployment, and found support for those that were long lasting and were targeted through means-testing of family income and wealth. The empirical analysis was careful to distinguish between first-time job seekers and those who had been previously employed.

Four papers were included in the conference session on ‘Social Insurance’. Assaf Razin (Tel Aviv University, and CEPR) presented ‘Migration and Pension’, which was written with Efraim Sadka. Their analysis was motivated by the frequently observed opposition to immigration, based on the assumption that immigrants were net beneficiaries of the welfare state and net consumers at the expense of the native population. The paper evaluated the impact of immigrants in a society with a pay-as-you-go social security system with redistribution. The authors accepted that migration would have implications for the financial soundness of a state pension system, which is an important pillar of any welfare state. Although it was reasonable to expect that young migrants, even if low-skilled, could help society to pay benefits to the current cohort of the elderly, it might also be reasonable to argue that, by virtue of being net beneficiaries of the welfare state, they (the migrants) would adversely affect the current younger generation of natives.

In the static model, migration was found to have adverse effects. When Samuelson’s concept of the economy as an everlasting machine was employed, however, migration was shown to be a Pareto-improving measure. Thus all income groups (low and high) and all age groups (young and old) existing at the time of the migrants’ arrival were rendered better off once allowance was made for the possibility of a surplus in the pension system. In response to Jeff Frank’s observation that pensions in the model were not differentiated by skill, Razin explained that this was representative of the redistributive aspects of the system. Michele Boldrin pointed out that immigration might allow governments temporarily to postpone necessary reforms. Dennis Snower argued that the results hinged on the immigrants’ employment status in the official, as opposed to the shadow, economy. Kai Konrad stressed that the positive immigration effect was determined by the positive productivity and employability assumptions for immigrants.

‘Informal Family Insurance and the Design of the Welfare State’ was presented by Robert MacCulloch (ZEI, Universität Bonn, and University of Oxford). The paper, which was co-authored with Rafael Di Tella, investigated the problem of unemployment benefit provision when the family was also a provider of social insurance. The model sought to explain why extended families seemed to be more prevalent in countries with a relatively undeveloped welfare state, and why political parties appeared to associate a large welfare state with ‘weak’ families.

The authors first presented a model in which risk-sharing motives governed intra-family transfers and in which more generous unemployment benefits, provided by the state, replaced and crowded out family risk-sharing arrangements one-for-one. The model was then extended to capture the idea that the state had an advantage vis-à-vis the family in the provision of insurance because it could tax individuals, whereas the family had to rely on self-enforcing agreements. In this case, the effect of state transfers on intra-family transfers and on total insurance transfers to the unemployed fell more than one-to-one with increases in the state’s generosity. This implied that any increases in state-provided benefits would be followed by a one-for-one reduction in intra-family transfers as families sought to return to the initial level of risk-sharing. The increased generosity of state benefits, however, made defecting from the informal family risk-sharing contract more attractive. Hence family transfers had to be reduced even further to maintain the incentive-compatibility of the informal risk-sharing contract. An interesting implication was that if families could sustain generous informal insurance arrangements, then the state could maximize social welfare either by staying out, or by becoming the sole provider of unemployment benefits. The results still held when families were assumed to be better than the state at monitoring the job-search activities of the unemployed.

Discussion of the paper centred on the effect of welfare provisions on the stability of the family and the endogeneity of family formation. Assaf Razin pointed out that the model presented a simplified depiction of intra-family decision-making which imposed severe constraints on household allocation. In response to Dennis Snower’s question about the paper’s implications for welfare-state reform, the presenter pointed out that the potentially drastic effects of changes in state benefits needed to be considered, where even an increase in benefit provision might cause a net decline in total benefits.

The paper entitled ‘Can and Should a Pay-As-You-Go Pension System Mimic a Funded System?’ was presented by John Hassler (Institute for International Economic Studies, Stockholm, and CEPR) and was written  jointly with Assar  Lindbeck.  The authors set out to answer two questions: first, whether a  pay-as-you-go pension system could be made sufficiently actuarially fair to avoid labour market distortions; and second, which generations would benefit if the pay-as-you-go system set actuarially fair returns. The paper considered the possibility of letting a pay-as-you-go pension system mimic a fully-funded system. Generally, it turned out to be impossible to make a less-than-fully-funded system actuarially fair on average. But a non-funded pay-as-you-go system could provide an actuarially fair implicit return on the margin, thereby increasing economic efficiency. The benefits of this would accrue entirely to current pensioners as a windfall gain, unless compensating transfers were implemented. Such a system could be thought of as a pay-as-you-go system that mimicked a fully-funded system in combination with lump-sum transfers to current pensioners from current and future workers.

Michele Boldrin criticized the failure to consider the changing rate of return on capital, which was empirically important. Kai Konrad questioned the use of a welfare function in an overlapping-generations framework.

David Miles (Imperial College, London, Merrill Lynch and CEPR) presented his paper, written jointly with Andreas Iben, on ‘The Reform of Pension Systems: Winners and Losers Across Generations in the UK and Germany’. The authors used a stylized model of the United Kingdom and Germany to perform simulations to show which generations might be direct gainers, and which losers, from a transition to funded state pensions. The authors estimated the required pre-reform equilibrium structure of intergenerational bequests, if different generations were to be insulated from the effects of a transition to a fully-funded pension system. They showed that it was likely that more than one generation would be direct losers from such a transition, especially in Germany. To prevent those generations from being net losers, the chain of bequests (in the initial equilibrium) needed to satisfy one condition: namely, that the cumulated value of the sum of the losses of all the previous generations of direct losers needed to be less than the pre-reform bequest of each generation to the next generation. Calculating the required chain of bequests given the actual demographic structure and the pension system, it was shown that the critical level of bequests was highly sensitive to the rate of return on assets, the initial generosity of the state pension scheme and the scale of future demographic shifts.

Michele Boldrin pointed out that it was not possible to assume that the growth of the wage bill would be permanently below the return on capital. Either the capital stock would disappear to maintain a given capital-labour ratio, or the labour share would need to go to zero. John Hassler noted that an underlying assumption was that both the return on capital and the growth of the wage bill had the same stochastic characteristics. Fiorella Padoa Schioppa Kostoris commented that the fact that labour union representatives are typically of advanced age, and do not represent the interests of the younger workers, might present a problem for the transition to a fully- or partially-funded pension system.

In the final conference session – devoted to ‘Education’ – Michele Boldrin (Universidad Carlos III, Madrid, and CEPR) presented a paper, written jointly with Ana Montes Alonso (Universidad Carlos III, Madrid), on ‘Intergenerational Transfer Institutions: Public Education and Public Pensions’. The authors noted that credit markets for the financing of human capital investments are rare, with the consequence that competitive equilibrium allocations cannot achieve either static or dynamic efficiency. They suggested, however, that public funding for education of the young could be interpreted as a means for the younger generation to ‘borrow’ from the older generations in order to accumulate human capital. Pay-as-you-go public pension systems could then be seen as a mechanism for the ‘borrowers’ to repay the capitalized value of their educational debt to the ‘lenders’ via their social security contributions from which the pensions are financed. The two welfare-state institutions – public education and public pensions – could thus be viewed as mutually supportive and, in an overlapping-generations model, a properly designed intergenerational pact of this nature would help to achieve a more efficient allocation of resources over time.

The authors tested the main predictions of their model by using micro and macro data from Spain. Among the policy implications of their analysis were that individual contributions and investments in human capital should be compared when considering the rate of return in the pension system. Hans-Peter Grüner pointed out that the model assumed that only market participants, but not the government, were subject to moral-hazard behaviour. Jean-Charles Rochet suggested that there might be business-cycle implications if the rate of return in the pension system changed over time. David Miles thought that it was counter-intuitive to assume that the capital market misfunctioned only for the group of young individuals seeking to invest in human capital.

The final paper was presented by Gianni de Fraja (University of York and CEPR) under the title ‘The Design of Optimal Education Policies’. The paper considered the optimal education policy of a budget-constrained utilitarian government. The key assumptions were that households differ in their income, and in the intellectual ability of their children; income is observable by the government, but ability is private information; and households can choose to use private education, but cannot borrow to finance it. The optimal education policy derived by de Fraja is elitist: it increases the spread between the educational achievements of the bright and the less bright children, compared with both private provision and the first-best policy. It also implies that the education received by less bright children depends positively on parental income. Finally, it is input-regressive, in the sense of Arrow (1971): thus households with higher incomes and brighter children contribute less towards the cost of the education system than do households with lower incomes and less bright children.

Hans-Peter Grüner drew attention to the underlying assumption in the paper that moral hazard did not affect state activity. Jean-Charles Rochet suggested that de Fraja’s model implied that inequality would increase in the long term, a fact confirmed by empirical observations.