unemployment, labor market, social protection, employment, United States, Europe, Rebecca M. Blank, social protection programs, labor market flexibility, The Connection, social programs, economic growth, United Kingdom, employment protection, Richard B. Freeman, National Bureau of Economic Research, unemployment benefits, Brookings Institution, welfare state, employment growth, flexibility, R. Layard, Washington, D.C., Basil Blackwell, OECD, West Germany, protection programs, Harvard University Press, workers, Social assistance programs, Organization for Economic Cooperation and Development, ed, Journal of Human Resources, MIT Press, R. Lawrence, unemployment insurance, income support programs, employment opportunities, employment problems, OECD countries, Economic Flexibility, unemployment rate, European countries, distortionary, labour market flexibility, term flexibility, C. Schultze, unemployment benefit, Oxford, long-term unemployed, social welfare spending, empirical studies, social welfare, european growth, employment adjustment
This PDF is a selection from an out-of-print volume from the National Bureau of Economic Research
Volume Title: Social Protection versus Economic Flexibility: Is There a Trade-Off? Volume Author/Editor: Rebecca M. Blank Volume Publisher: University of Chicago Press Volume ISBN: 0-226-05678-3 Volume URL: http://www.nber.org/books/blan94-1 Conference Date: December 14-15, 1992 Publication Date: January 1994 Chapter Title: Evaluating the Connection between Social Protection and Economic Flexibility Chapter Author: Rebecca M. Blank, Richard B. Freeman Chapter URL: http://www.nber.org/chapters/c11253 Chapter pages in book: (p. 21 - 42)
Evaluating the Connection Between Social Protection and Economic Flexibility Rebecca M. Blank and Richard B. Freeman From the mid-1970s through the early 1980s, most western industrialized countries suffered from slow economic growth and high or rising unemployment. While the United States was no exception, it had markedly better employment growth than western Europe. The United States added some 20 million jobs from 1975 to 1984, whereas employment was stagnant in Europe. European rates of unemployment, which had historically been below those in the United States, came to exceed the U.S. level. Many analysts and policymakers attributed the U.S. "job creation miracle" to that country's flexible and unregulated labor market. A consensus developed that the cure for Europe's employment problems required greater labor market flexibility, specifically including a reduction in social protection programs that impeded flexibility (OECD 1990). To speed up market adjustments and increase employment, many European countries tightened provisions for unemployment insurance benefits, loosened dismissal regulations, decentralized collective bargaining, and sought to increase worker mobility through government subsidies or training. Under Prime Minister Margaret Thatcher, the United Kingdom tried to revamp its entire set of labor market institutions to create a more flexible and market-driven economic system. Developments in the late 1980s and early 1990s have called into question the 1980s analysis that welfare state programs and labor market rigidities were Rebecca M. Blank is professor of economics at Northwestern University
, faculty affiliate at Northwestern's Center for Urban Affairs and Policy Research, and research associate of the National Bureau of Economic Research. Richard B. Freeman is professor of economics at Harvard University
and at the London School of Economics
and directs the Labor Studies Program at the National Bureau of Economic Research. The authors thank the Ford Foundation, the National Bureau of Economic Research, and the Center for Urban Affairs and Policy Research, Northwestern University, for their support for this project. Participants in the conference for this volume provided excellent suggestions and comments, and Rebecca London provided top-quality research assistance. 21
22 Rebecca M. Blank and Richard B. Freeman major causes of the decade's economic problems
. The curative powers of labor market flexibility seem far more limited at this writing than they did a decade ago. The expansion in U.S. employment after the 1983 recession ended in several years of stagnant growth in the 1990s and was accompanied by rising inequality and falling real earnings for many workers. In the United Kingdom, the increase in unemployment in the early 1990s to over three million workers, rising inequality, and financial woes in many sectors raised serious doubts as to the efficacy of the Thatcher reforms. In both countries, the government began to back away from relying solely on market-oriented reforms. High U.S. unemployment in summer 1992 forced the Bush administration to extend unemployment benefits; in fall 1992, U.S. voters chose a new president who promised a more activist government economic policy. Devaluation of the pound forced the British government
from a zero-inflation, noninterventionist economic policy to one more attuned to growth. How valid was the 1980s belief that an extensive social protection system-- defined broadly to include all governmentally sponsored programs that protect individuals or families from serious income declines or job loss1--contributes to economic problems? In what ways can social protection advance or retard economic progress? How strong is the empirical evidence on the effect of welfare state or social safety net programs on the functioning of economies? We begin to examine these questions by reviewing the economic record that gave rise to the belief that social protection harms economic progress. Then we consider the argument that such protection has deleterious effects and the counterargument that it has positive economic effects and, finally, evaluate the empirical evidence on this debate. 1.1 Employment and Growth Woes: The 1980s Analysis The belief that the extensive social welfare programs in the Organization for Economic Cooperation and Development countries in Europe (OECD-Europe) limited labor market flexibility is rooted in the economic development
s of the late 1970s and 1980s, when virtually all major OECD countries suffered from high unemployment and reduced economic growth. Table 1.1 indicates the extent of these problems. Part A shows unemployment rates across a sample of countries. Although rates showed no secular change in the 1960s, during the 1970s they rose everywhere, in many cases to levels once considered highly 1. This includes income transfer programs that protect families against poverty, and in-kind programs such as housing or health insurance that ensure access to particular goods or services, as well as social insurance programs that cushion workers against unemployment, disability, or old age. It also includes employment regulation programs that provide greater job security and programs that mandate employer payments to support particular fringe benefits, such as social security or health care. We limit our discussion to government programs, though we recognize the Rein and Friedman (1992) point that social protection involves other sectors of the society as well.
23 The Connection between Protection and Flexibility
Unemployment Rates and Rates of Growth of GDP, 1960-1990
A. Rates of Unemployment
United States OECD-Europe West Germany France United Kingdom Italy Smaller European Country
5.4% 2.9 1.0 1.4 1.3 5.5 3.9
3.5% 3.4 1.2 2.7 2.1 5.6 4.5
5.8% 5.7 3.3 5.9 4.6 7.6 6.4
B. Annual Rates of Real GDP growth
5.4% 7.8 6.2 9.0 5.5 10.8 7.9 1979-90
United States OECD-Europe West Germany France United Kingdom Italy Smaller European
4.5% 4.7 4.1 5.4 3.0 5.7 5.4
3.2% 4.9 4.9 5.4 3.4 4.5 5.6
2.4% 2.6 2.3 2.8 1.5 3.7 2.6
2.6% 2.3 2.0 2.1 2.1 2.4 2.6
Sources: Organization for Economic Cooperation and Development 1992b, tables 2.15 and 3.1; Organization for Economic Cooperation and Development 1991, table 2.15.
recessionary. In the 1980s, unemployment in most European countries increased further. In contrast, after reaching a post-World War II peak in 1983, unemployment fell substantially in the United States. Japan was an exception throughout the period, with a low and stable unemployment rate. Part B of table 1.1 shows rates of Gross Domestic Product
(GDP) growth over the same years. Economic growth slowed between the 1960s and the 1970s and in most countries decelerated further between the 1970s and 1980s. The United States is again an exception. U.S. economic growth was as strong (or weak) in the 1980s as in the 1970s. Much of the growth of U.S. GDP, however, took the form of growing employment per capita rather than growing output per worker (table 1.2). GDP per capita grew at the same rate in the United States as in Europe, whereas growth in GDP per worker in the United States was poor by OECD standards and by past U.S. standards as well. While the aggregate data thus give a mixed message about the U.S. experience--better employment growth but worse productivity growth--analysts of European problems focused largely on the output and employment success of the United States in their assessment of the virtue of deregulated labor markets.2 Table 1.3 explores one reason for this: differential U.S.-European performance in creating full-time jobs and in the duration of joblessness. Part A 2. This point is documented in Freeman (1988) for the first half of the 1980s.
24 Rebecca M. Blank and Richard B. Freeman
Rates of Growth of Employment per Capita, GDP per Capita, and GDP per Employee, 1979-90
Growth of GDP
Growth of employment
Growth of GDP
1.5% 0.3 1.2
Source: Organization for Economic Cooperation and Development 1992b, tables 3.2 and 3.7, with employment per capita obtained by subtracting GDP per employee from GDP per capita.
Part-Time Employment and Duration of Unemployment
A. Part-Time Employment as Share of Total
United States OECD-Europe West Germany France United Kingdom Italy Smaller European Belgium Netherlands
16.4% 11.4 8.2 16.4 5.3 6.0 16.6
18.4% 12.6 9.7 19.4 4.6 8.1 21.4
B. Percentage Unemployed Twelve Months or More
16.9% 13.2* 12.0 21.8* 5.7* 10.2* 33.2 1990
United States OECD-Europe West Germany France United Kingdom Italy Smaller European Belgium Netherlands
4.3%t 28.7 32.6| 29.5 51.2 61.5 35.9
13.3% 39.3 42.2 47.0 57.7 66.3 50.5
5.6% 46.3 38.3 36.0 71.1 69.9 48.4
* 1989 statistic. 11980 statistic. Source: Organization for Economic Cooperation and Development 1991, tables 2.7, 2.9; 1992, table N.
25 The Connection between Protection and Flexibility shows that the share of part-time employees in the labor force rose in most European economies, so that employment growth consisted largely of parttime rather than full-time jobs. Indeed, in many countries, full-time employment fell while part-time employment rose. In the United States, by contrast, while the share of part-time workers increased in the early 1980s recession, it then fell so that it stood at nearly the same level in 1990 as in 1979.3 Part B of table 1.3 shows the large differential in duration of unemployment between Europe and the United States. Jobless spells in Europe were long and rising, whereas those in North America were relatively short, as might be expected in a flexible and adaptable market system. More than any other single fact, the lengthy duration of European joblessness led many to believe that European social policies supporting the unemployed or protecting the employed might be part of the problem of high joblessness rather than part of the solution. At first, economists analyzing this problem stressed the potential adverse effect of social protection and labor institutions on short-run labor market flexibility, limiting wage adjustments in the face of macroeconomic shocks. As high unemployment persisted, however, and estimates of the nonaccelerating inflation rate of unemployment (NAIRU) indicated that the unemployment rate consistent with stable inflation had increased,4 economists came to view labor market flexibility as a tool to reduce the NAIRU and long-term unemployment.5 The notion grew that flexibility was necessary if not sufficient for any major reduction of unemployment and restoration of growth. But what exactly is meant by labor market flexibility? The precise meaning of the term is often unclear, because of the many margins along which economic agents can be flexible.6 The most common usage measures flexibility by the speed of price and quantity adjustment in a changing economic environment. The more flexible market has wages adjusting rapidly when unemployment or prices change; has employment or hours adjusting rapidly when labor demand changes; has greater mobility of labor between different sectors, firms, and geographic areas
as demands shift; has more Rapid Transit
ions from unemployment to employment; and so on. For analysts addicted to Phillips curve regressions, flexibility means larger coefficients on unemployment variables in wage change regressions. For analysts concerned with mobility of labor, 3. Although more difficult to measure, temporary work increased in both the U.S. and the European economies. 4. For instance, the European Unemployment Program conferences reached this conclusion, as summarized in the papers in Bean, Layard, and Nickell (1987) and Dreze and Bean (1990). For a good summary of the changing discussion over the nature of the unemployment program in Europe, see Krugman (1987). 5. For an example of such a policy discussion on how to reduce the NAIRU, see Layard (1986). 6. Boyer (1988) has an excellent discussion of alternative definitions of flexibility and of some ways in which flexibility may or may not be to the long-run advantage of the economy. Metcalf (1987) also presents an extended discussion of various types of labor market flexibility.
26 Rebecca M. Blank and Richard B. Freeman flexibility means greater exit rates from unemployment (and presumably from employment as well). More heterodox analysts, noting the success of Japan in adjusting to the shocks of the period, used flexibility to refer to a broader set of responses, such as adaptive corporate strategies, changeable production technologies, or multiskilled and adaptable workers. When flexibility is restricted to short-term, market-driven wage and employment responses, Japan's commitment to lifetime employment contracts raises the possibility that "flexible rigidities," to use Ronald Dore's (1986) phrase, rather than unconstrained markets are the requisite for economic success. As arguments over labor market flexibility focus largely on the trade-off between European-style social protection and U.S.style short-term flexibility, however, we will use the term flexibility to refer to the speed of short-term quantity and price adjustments in the labor market in the ensuing discussion. What was Europe doing wrong in terms of labor market practices and social protection that the flexible United States was doing right? There is a diverse set of policies and institutions that, arguably, adversely affected labor market flexibility and thus economic progress:7 1. Labor relations practices that strengthen the position of workers within firms can limit employers' ability to adjust to the changing economic environment by creating sticky wages and slow reallocation of labor (within and between firms). Legislation regulating the hiring and firing of workers will reduce employer ability to reduce employment in the face of short-run shocks and can make them reluctant to hire workers in a boom if they fear it is temporary.8 Centralized wage settlements, viewed as helpful in an era of inflationary pressures (Bruno and Sachs 1985) can similarly impair industry or firm wage adjustments to changing market conditions. Worker organizations inside firms can prevent wage adjustments that might increase New hires
(Lindbeck and Snower 1990). 2. The size and nature of the welfare state can slow workers' responses to changing labor market signals. According to this view, generous unemployment benefit programs limit the incentives of workers to reenter the labor market upon becoming unemployed, while extensive income support
programs make long-term unemployment or nonparticipation in the labor market more economically possible and socially acceptable.9 3. Inadequate skills or mobility can limit workers' ability to adjust to changing labor demands. "Skill mismatches" could result in high unemployment 7. A number of the papers in Jessop et al. (1991) summarize and critique these different causal theories. Lawrence and Schultze (1987) and Krugman (1987) also provide good summaries. 8. A variety of different articles discuss this problem, from Piore and Sabel's (1984) call for greater "flexible specialization" to Boyer's (1988) discussion of the failure of the Fordist system of industrial relations. 9. The argument against the welfare state is nicely summarized in Lawrence and Schultze (1987).
27 The Connection between Protection and Flexibility even when demand for new workers remains strong. Attempts to test the skill mismatch hypothesis, however, found little evidence that mismatches were a major factor in the rise of unemployment in the 1980s, leading this explanation to lose its attractiveness.10 But the notion remains that low mobility of labor is one cause of high unemployment, despite the Japanese example of low mobility and low unemployment. 4. Excessively high wages and wage shares can limit long-term job creation. Failure by employers to invest in physical capital during the late 1970s and early 1980s was blamed for permanently lowering employment demand. For instance, Modigliani et al. (1987) estimated that by 1985 available production capacity was 15 percent below that needed for full employment, because of inadequate capital investment. To generate greater investment, wage reductions that raise profitability were viewed as necessary, making the decline in real wages in the United States seem the "right" adjustment and the maintenance of high real wages in Europe the "wrong" strategy. 5. Linking all of these arguments were a series of "hysteresis hypotheses" that tried to explain how a short-term rise in unemployment could permanently raise the NAIRU, so that adverse shocks in unemployment would not be selfcorrecting." Insider/outsider models argued that the unemployed have little say in wage bargaining and thus little effect on wage levels.12 Other models focused on behavioral changes among the long-term unemployed. As the human capital of the unemployed deteriorates and they adjust to unemployment, workers become less productive or stop seeking jobs.13 In either case, a short-run rise in unemployment would lead to higher long-term unemployment in the future. While different analysts stressed the adverse effects of different social protection policies, the broad message was the same: greater flexibility in the labor market and less social protection were the road to reducing unemployment and curing the economic woes of the 1980s. Decreased labor market regulation would permit reductions in wages and related costs of employment when demand fell. Having fewer social benefits would increase incentives for the unemployed and nonemployed to seek work. Targeted job training or related labor market programs might also be necessary to bring the unemployed (especially the long-term unemployed) back into the world of work. 10. For instance, the argument is dismissed by Layard and Nickell, using U.K. data, and by Franz and Konig, using data from West Germany (both articles are in Bean, Layard, and Nickell 1987), but Blanchard (in Dreze and Bean 1990) argues for possible mismatch effects that this empirical work would not have measured. Flanagan (1987) provides a nice summary of the evidence on occupational and spatial mismatch. 11. For instance, the empirical work of Oswald and Blanchflower (1990) indicates that unemployment has a strong effect on wages at low unemployment but little effect at high unemployment. 12. A summary of this approach is in Lindbeck and Snower (1990). 13. Blanchard (in Dreze and Bean 1990) summarizes this argument nicely. Franz (1987) provides empirical support for this hypothesis, using West German data.
28 Rebecca M. Blank and Richard B. Freeman 1.2 Policy Reactions and Outcomes Given the wisespread belief that flexible labor markets were part of the cure, if not the cure, to high unemployment and slow growth, many OECD countries sought to increase labor market flexibility over the 1980s (see table 1.4).14 France tried increased work-sharing. It decentralized collective bargaining, as did the Netherlands and Spain. Italy got rid of automatic wage indexing, the Scala Mobile. Virtually all European countries expanded training measures targeted at unemployed workers. West Germany and Spain introduced shortterm employment contracts. The United Kingdom changed its labor relations laws to weaken trade unions and sought to ensure that the unemployed were really seeking jobs. A wave of privatization reduced the governmental share of employment throughout Europe. Some countries cut unemployment benefits and other social protection programs, aiming to emulate the lower level of benefits in the United States. By the end of the 1980s, most European countries had less state involvement in labor market outcomes, less centralized labor relations, and more limited transfer programs than they had a decade earlier. But the effort to increase labor market flexibility did not reduce aggregate unemployment. In the United Kingdom, policies for greater flexibility appeared to do little on the unemployment front.15 Spain introduced new fixedterm contracts to increase employer flexibility. While unemployment dropped from 1985 to 1990, it never dropped below 15 percent, although nearly all new hires in the 1980s were on these temporary contracts. In some other countries, such as the Netherlands, there were only limited reductions in social programs, suggesting that policy changes large enough to create a U.S.-style flexible market were politically difficult to implement. At the same time, the U.S. economy began to look less attractive. The benefits of growth within a flexible and decentralized labor market turned out to be uneven in the 1980s: the earnings and the income distributions widened, growth had less of a "trickle-down" effect than it had in previous decades, and poverty rates remained high throughout the decade.16 Visitors to the country saw homeless people in the streets and the increasingly Third World appearance of many parts of the Inner Cities
. Some European economies also experienced growing earnings inequality, but only the relatively decentralized and flexible United Kingdom had large rises in inequality.17 In most European countries, moreover, increased earnings inequality did not translate into increased family income inequality, seemingly because these countries had more 14. Boyer (1988) and OECD (1990) discuss different European countries' policy responses to the economic problems of the 1980s. 15. They presumably did help in other ways. For instance the U.K. legislation that weakened unions seems to have spurred unionized firms to higher rates of productivity. 16. For a summary of U.S. inequality trends, see various articles in the Quarterly Journal of Economics 107 (February 1992), or Danziger and Gottschalk (1993). For a discussion of the changes in the effect of the macroeconomy on poverty, see Blank (1993). 17. For instance, see Davis (1992) or Katz, Loveman, and Blanchflower (1992).
29 The Connection between Protection and Flexibility
Sample Policy Changes Implemented over the 1980s to Increase Labor Market Flexibility in Europe
Established short-time unemployment benefits Created programs to assist temporary work placements Weakened dismissal laws
Increased decentralization in bargaining Weakened dismissal laws Increased training for long-term unemployed Decreased workweek
Weakened dismissal laws Increased incentives for early retirement Increased limits on unemployment benefit receipt Decreased workweek
Eliminated automatic wage indexation
Decentralized wage agreements Lowered relative minimum wage Created programs to assist temporary work placements Increased limits on unemployment benefit receipt Increased training for long-term unemployed
Decentralized wage agreements Decreased workweek Increased training and job creation for long-term unemployed Increased availability of part-time and short-term work
Increased training and job search requirements for those receiving unemployment benefits
Implemented privatization of major government-owned industries Decentralized wage agreements Weakened dismissal laws Increased limits on unemployment benefit receipt Increased training for long-term unemployed
Source: Organization for Economic Cooperation and Development 1990.
extensive transfer systems. The rise of inequality in the United States, accompanied by a very visible urban homelessness problem, reminded analysts of the benefits of social protection and transfer programs.18 In addition, the cyclical downturn in the U.S. economy in the early 1990s showed that flexibility did not guarantee permanently lower unemployment or an ongoing strong economy. By the early 1990s the claim that weakening social protection systems 18. Another possible factor affecting thinking was the political and economic change
s in eastern Europe. These changes made it clear that market capitalism is not easily or quickly transplanted to economies with a history of control and planning and that historical economic, political, and institutional differences shape the possibilities for economic reform.
30 Rebecca M. Blank and Richard B. Freeman would increase labor market flexibility and cure economic stagnation could not be made as strongly or convincingly as it had been at the height of Reaganomics and Thatcherism. Reforms to increase flexibility by weakening welfare state programs did not deliver what they had promised. The conventional wisdom of the 1980s--that social protection impairs flexibility in ways that harm economic performance--deserves a serious rethink. 1.3 The Case against Social Protection Every program that protects people from the consequences of unemployment provides an incentive not to work and is a drag on the economy. --Archetypal opponent of social protection The argument that social protection harms economic outcomes is familiar to economists because it is derived from first principles about the effect of interventions in perfectly functioning markets. Most institutional interventions create incentives for market participants to behave differently than they otherwise would, implying some distortionary loss of social welfare from the competitive ideal. In terms of labor supply, transfer programs drive a wedge between individual utility-maximizing outcomes and the socially efficient outcome. Unemployment assistance or welfare benefits may lead a worker to choose leisure instead of work until such benefits end. This maximizes the worker's well-being but lowers output and lengthens the spell of unemployment. Similar distortionary side effects are likely to accompany other social protection measures. Lowcost public housing in a particular city may limit workers' mobility, leading a worker to reject jobs in other areas of the country. Programs that mandate fringe benefit packages may constrain worker choices, producing nonoptimal and inefficient outcomes as well as high labor costs. Generous sick benefits may produce an epidemic of headaches, backaches, and related ills until all benefit days are used up. In terms of labor demand, social assistance
may also change employer decisions in socially inefficient ways. Laws that constrain employers' hiring or firing behavior, for instance, may distort market wage signals, reduce profitability due to "excess labor," or lead employers to hire fewer workers because of the additional expected future cost. Laws that levy taxes on employers to fund retirement or unemployment payments may induce firms to hide income or to hire off the books or may make it unprofitable to hire low-wage labor, depending on the incidence of those taxes. Socially determined minimum wages, such as the French SMIC or Italy's collectively bargained Scala Mobile adjustments, can lower employment in low-wage activities. The case against social programs also stresses that there can be sizable in-
31 The Connection between Protection and Flexibility vestment opportunity costs and deadweight losses to raising the taxes necessary to support program
s. Social assistance programs may divert capital from more fruitful alternative investments. If those opportunities have higher multipliers or greater job creation potential, money spent to support low-income families could permanently lower employment opportunities, leaving both the unemployed and the rest of society worse off in the long run. Most social protection policies create some "excess burden" through the tax system and expand public sector employment, which may further add to labor market rigidities. Furthest from neoclassical analysis but of great importance in popular discussion is the possibility that income support programs may change individual preferences by creating "dependency." If labor-leisure preferences are malleable over the short run, programs that induce people to remain out of the labor market may permanently shift their preferences toward greater leisure and away from work, resulting in lower labor market involvement even after the transfer program comes to an end. Some of the discussion over hysteresis effects in unemployment evokes this argument, as the long-term unemployed become less and less attuned to work as their spells lengthen. Finally, there is an additional potential negative impact in an open economy if social protection programs have any or all of the above negative effects. By burdening business, social programs may reduce trade competitiveness. This leads to falling exports, rising imports, currency devaluation, and deleterious effects on a country's long-term employment and economic growth. Whether an extensive welfare state can survive in an open economy that trades with countries that have less-extensive programs is a question that the Common Market and the North American Free Trade Agreement will put to the text. 1.4 The Case against the Case against Social Protection Social programs cost resources and may have some undesirable effects on market efficiency
, but they generate benefits that are greater than their costs and may be viewed as investments that pay off in higher long-run productivity. --Archetypal supporter of social programs There are two ways to criticize the conventional criticism of social protection programs. The first "case against the case against"19 admits that social programs have distortionary costs that impact labor markets but argues that the benefits of programs, which critics rarely measure, exceed the costs. The second argument denies that social programs have deleterious effects on markets 19. Making cases against cases is a tradition in economics fathered by Robert Solow, to whom we give full credit for this mode of argumentation about where the burden of proof lies in discussing a policy issue.
32 Rebecca M. Blank and Richard B. Freeman at all and claims instead that they are investments in a more productive work force that promise substantial economic returns. The first defense of social programs begins with the observation that the case against them focuses almost exclusively on one side of the benefit-cost calculus: the losses in social well-being due to distortionary incentives. Haveman (1985, 17) may be right that "the gains [of welfare state programs] are relatively familiar and directly experienced," but the criticism of these programs rarely acknowledges these benefits. The standard 1980s economists' study of unemployment insurance or income support, for instance, is an exercise in estimating distortionary labor supply or related costs, not in assessing how well or poorly the program fulfills its purpose of reducing economic uncertainty through income and/or employment guarantees. A complete costbenefit analysis of any program requires, of course, that the increase in economic well-being among citizens that is due to the program be set against its costs, direct and indirect, through lost growth or productivity. Simply showing that programs have distortionary effects or inefficiency costs does not make the case against them. Many social insurance programs provide "goods" that competitive markets intrinsically cannot provide because of the presence of moral hazards, externalities, or other forms of market failure, which means these programs cannot be criticized for crowding out market alternatives. For instance, the competitive labor market is unlikely to offer socially optimal levels of workplace-related benefits such as sickness benefits or unemployment insurance, because of the adverse selection of workers into insurance scheme
s. If a firm offers particularly good sickness benefits, it will attract a disproportionate number of workers prone to illness, which will be costly. In contrast, if all firms offer identical (mandated) sickness packages, no single firm will bear the cost of attracting these more costly workers. The abortive effort by one major British bank to offer private unemployment insurance in the 1970s and the difficulties of unions in particular industries to provide private schemes support this point. The claim that social programs reduce economic growth is, the case against the case against argues, weak even in terms of standard theory. Existing analyses of welfare triangle losses have no compelling predictions for investment or growth. They show that social protection creates a static efficiency loss-- lower GDP--but not that it reduces growth rates
. Programs that waste resources reduce the pie available for investment, but they may not alter the share of savings and investment in national output. Standard growth theory
does not give a clear prediction about how static distortions alter growth, leading Mancur Olson (1982) and others who believe that those distortions have longterm Growth effects
to make more heterodox arguments for their position. In a similar vein, conclusions about distortionary costs are typically derived from first-best models of economic equilibrium. But in a second-best (or thirdbest) world where other distortions already exist due to taxation, regulation, or institutional structures, it is unclear that social protection programs produce
33 The Connection between Protection and Flexibility less-efficient outcomes. They may offset inefficiencies and distortions caused by other political or economic constraints. For instance, if union collective bargaining contracts require firms to lay off younger workers before older workers
, raising unemployment among younger workers, employment protection laws that mandate all workers be given four months notification before layoff may produce an employment outcome for younger workers that is closer to the first-best equilibrium. The implication is that the effects of any program must be analyzed in the context of the entire economic system in which it fits rather than as an isolated change in an otherwise ideal competitive world. In short, without denying that social protection programs cause some inefficiencies, one can still reject the claim that such programs are socially deleterious. Looking only at efficiency losses provides an incomplete analysis that fails to evaluate the benefits of the programs and ignores the broad context in which these programs work. The second attack on the case against social protection stresses the role of programs in enhancing human capital and productivity in the labor market. This is the type of argument put forth by the Clinton administration in 1993 for expanding various government expenditure programs. From a human capital perspective, social protection programs could create long-term incentives for employers and workers to invest in training. For instance, laws that limit dismissal might induce employers to invest more in worker training, since these laws create the long-term attachment that makes investment in specific skills profitable. Income protection programs might allow greater opportunities for returning to school or investing in retraining. For instance, maternity leave programs, that allow women to leave the labor market for several years when they have a child, such as Sweden's, provide a chance for women to acquire additional job training. Greater employment and income security may increase workers' productivity in other ways as well. Japanese workers with lifetime contracts may be more productive because they concentrate better on the job and feel more committed to their firm. A worker whose job is protected may be more willing to look for ways to improve productivity and to accept new machines or technological changes. A generous income transfer program may make workers more likely to take risks, change jobs, or move locations. Social assistance programs aimed at children and teenagers, such as health care programs, educational assistance, child allowances, or teen apprenticeships, are a particular favorite of those who look upon the welfare state as a productive investment
. Programs that shift resources to children are a "capital investment" in a country's future labor resources. Assistance to families with small children may improve the health or emotional well-being of those children, resulting in lower future social expenditures and higher future productivity. Assistance to teens may expand their labor market knowledge and improve their skills. If social protection and economic growth and flexibility complement each
34 Rebecca M. Blank and Richard B. Freeman other in these ways, emphasis on short-term "speed of adjustment" trade-offs would be shortsighted and inaccurate. But a case can even be made that social programs are not harmful in the short run. First, there are potential countercyclical social benefits in programs that reduce cyclical fluctuations in consumption spending and provide employment protection. For instance, worker uncertainty about possible job losses arguably can reduce consumer confidence and expenditures on durables, delaying economic recovery, as in Britain in the early 1990s. Second, short-run flexibility can have negative social effects, so that a bit of inflexibility may be virtuous. Bubbles in financial markets
and speculative swings or panics make it clear that rapid responses to economic changes are not always good; overshooting equilibria result from too much rather than too little flexibility. 1.5 Evaluating the Evidence In reviewing the arguments about the relationship between social protection programs, labor market flexibility, and aggregate economic welfare, we are struck by how little evidence is available on many issues. The economics literature contains many studies that measure the static efficiency impacts of some social protection policies but few that deal with labor market adjustment per se or with effects on economic growth. Probably the most-extensive literature on the efficiency costs of social programs attempts to measure the impact of income transfers on labor supply and of unemployment benefits on the duration of unemployment. The evidence shows behavioral responses to the incentives in these programs but has not yielded a definitive consensus over whether the magnitude of this response is large or small. If, as some studies suggest, two months of additional unemployment benefits increase the time a person is unemployed by one to two weeks (Katz and Meyer 1990), one analyst may regard this with horror while another may find it an acceptable cost of assisting the unemployed.20 Moffitt (1992, 16) has summarized the literature on the effect of welfare support on the labor supply of recipients in the United States by stressing the "considerable uncertainty regarding the magnitude of the effects" even after twenty years of research. Research on other routes by which programs may reduce efficiency is limited. The distortionary effects of social regulation on employers has been studied for some programs in some countries, but the results do not yield a simple generalization. When Sweden provided a highly generous sick-leave policy, this seemed to produce excessive use of sick leave (OECD 1991). But, to take another example, Houseman (1991) finds that European employment protection preserved jobs in steel compared to the loss of employment in the United 20. For a review of the literature on the effects of unemployment insurance, see Atkinson and Micklewright(1991).
35 The Connection between Protection and Flexibility Kingdom, and Abraham and Houseman (1993) find that West Germany's employment protection law did not reduce the long-run adjustment of labor to changes in shipments. Lazear (1990), by contrast, finds that such provisions reduce employment across countries but reports that this result was sensitive to specification. Because it is difficult to design empirical studies that separate the effect of social programs on export competitiveness, investment spending, and dynamic labor market behavior from other factors, there is relatively little work on these topics. Even if estimated efficiency costs are viewed as substantial, there is limited evidence linking these costs directly to the difference in labor market outcomes between the United States and Europe that sparked belief in the virtues of flexibility. Burtless (1987) indicates that the more extensive unemployment benefits in European countries did not cause the relative rise in European versus U.S. unemployment rates over the 1970s and 1980s, because the estimated effects are far too small to generate the huge observed unemployment changes. The introduction of temporary contracts in Spain seemed to increase employment, but Spanish unemployment still remained the highest in Europe. Similarly, introduction of temporary contracts in West Germany appears to have had no great effect on employment practices. The papers in this volume add to the research linking social protection programs to labor market developments within and across countries in the last decade. In most cases, the studies reject the existence of a substantial tradeoff between these programs and market flexibility. Abraham and Houseman (chap. 3) find that loosening of dismissal law in West Germany, France, and Belgium had little effect on the speed of employment adjustment. BorschSupan (chap. 5) finds that implementing or loosening tenant protection laws in West Germany did not change mobility or housing construction rates. Blank (chap. 8) finds that employment and wages in the public sector are as variable as in the private sector
in the United States and only slightly less variable in the United Kingdom. Holtz-Eakin (chap. 6) finds little evidence that inequitable health benefits between jobs limit worker mobility in the United States or West Germany. Rebick (chap. 7) finds little evidence that the implementation of early retirement programs in Japan induced elderly workers to leave the labor market. What is striking in all of these papers is the lack of evidence that the government programs had substantial effects on labor market adjustment. Why has research failed to turn up the large trade-off that the 1980s conventional wisdom posited? One possible reason is that there is in fact no sizable trade-off between specific programs, flexibility, and efficiency. Programinduced inefficiency losses may have much more limited effects on labor market dynamics than economists would like to believe. This may be because the programs are embedded in a larger system of employment and family relations, so that changes in one program do not change incentives as much as might at first appear to be the case and thus do not induce large changes in behavior. Another possibility is that the program changes are too modest and the time
36 Rebecca M. Blank and Richard B. Freeman period covered by the studies too short to capture the "big" adverse effect of these welfare state programs on the overall operation of society. Behavior may change slowly as experience and information are acquired over time. The claim that social protection limits economic adjustment may not be well tested by looking at marginal changes in program parameters. Perhaps European nations have not achieved a substantial increase in labor market flexibility because they have not gone far enough in cutting back their degree of social protection. Only major changes, abolishing whole sets of programs, may produce the degree of market flexibility needed to bring down long-term unemployment. Whatever the reason for the findings, our reading of the evidence in this volume and elsewhere is that there is little empirical evidence for large tradeoffs between labor market flexibility and social protection programs in general. At the present state of knowledge, the best attitude toward the trade-off hypothesis is one of open-minded skepticism. If there is little evidence of substantial trade-offs between social protection programs and labor market adjustment, however, estimates of the magnitude of purported benefits of such programs are almost nonexistent. Researchers have made only a few attempts to estimate the benefits of programs to individuals and thus to provide a fuller cost-benefit analysis.21 Lampman's (1984) analysis of the full effects of changes in social welfare spending in the United States between 1950 and 1978 on economic and social well-being and Haveman's (1985) comparison of U.S. and Netherlands social welfare spending took a broad benefit-cost view but lacked the necessary microstudies of how citizens value benefits that would provide a definitive welfare accounting. Hansen and imrohoroglu's (1992) evaluation of the behavioral and social welfare effects of an unemployment insurance scheme relied on simulation rather than on detailed empirical analysis of these effects. Their finding that, with some plausible parameters, the social benefits of unemployment insurance can outweigh the efficiency costs supports the notion that a full accounting of this program may yield a different welfare assessment than is implicit in studies focused on its undesirable effects in lengthening spells of joblessness. The claim that social protection should be viewed as an investment in future productivity is based upon even less solid evidence. We know of no studies that persuasively link labor market protection programs to worker productivity nor that definitively connect child and family assistance programs with labor market productivity in later life. Compensatory preschool programs have been linked with greater school achievement and more years of schooling, both of which are correlated with economic success in the labor market (Barnett 1992), but there is no evidence on whether income assistance for poor families 21. For instance, Haveman and Wolfe (1984) attempt to measure a wide variety of nonmarket effects from public education
. Weisbrod (1983) tries to measure the net social benefits of alternative programs for the mentally ill. Kemper, Long, and Thornton (1984) estimate the net social benefit of a job-training program. Haveman (1985) discusses the full range of social benefits and costs that should be considered in evaluating income protection programs.
37 The Connection between Protection and Flexibility improves children's long-term economic outlook. On the contrary, some evidence suggests that parents' participation in welfare programs in the United States raises the probability of welfare participation among their children when grown (Gottschalk 1992). Given the extensive public discussion of the value and costs of social programs, and the resources spent on different programs, it is unfortunate that we lack the studies that might provide the full benefit-cost assessment of the effects of the programs necessary to make a scientific case for reducing or increasing their scope. 1.6 Improving Analysis of Social Protection Programs Our assessment of the arguments and evidence on the effects of social protection programs has implicitly suggested some directions for future research in this area. We conclude this essay by bringing these suggestions together. First, the greatest immediate need is for detailed studies that measure the benefit side of social programs, in particular assessing the nonincome value of the income protection and job security of social insurance programs. Given the increasing recognition of the importance of risk and risk aversion in economic theory, this is an area in which empirical research has failed to keep pace with theory.22 To evaluate social programs properly, we need to measure the value to individual workers and families of reducing the risk and ensuring incomes or employment. This may require more interaction between economists and other Social Scientist
s, such as social psychologists, who more routinely use nonincome measures of personal well-being. Second, there is also a need to analyze the effects of specific programs within the general equilibrium of social and market institutions. The impact of any particular program depends on the environment of institutions and economic conditions in which it is located. For instance, the differential effect of weakening employment protection in Spain versus West Germany presumably reflects differences in labor market institutions and in the strength of unions or works councils in the two countries. Similarly, multiple and simultaneous programs can cause offsetting behavioral incentives or can reinforce each other in ways that create synergies, so that the effect of several programs together may be different from any individual program alone. For example, Hanratty's paper (chap. 10 in this volume) indicates that time-limited welfare in France moves women into the labor force, but shows that this occurs in part because France's educational system admits children to full-time schooling at exactly the point when women lose their benefits. On the one hand, programs that appear distortionary in isolation may not be distortionary when viewed in a broader systemic context. If a generous minimum-income support system is combined with job creation or skill training programs, the expected adverse 22. Best known, perhaps, is the literature measuring the effect ofjob-related risks on wage rates.
38 Rebecca M. Blank and Richard B. Freeman effects on labor supply may be muted or overwhelmed in importance by the positive effects of job training. On the other hand, it is also possible that a host of social protection programs may coalesce to produce an inflexible system, so that reforming a single program may fail to produce the benefits that reformers expect. Research that measures joint effects from multiple programs would provide insight into these issues and indicate the extent to which effective policy might be better conceptualized as a combination of interactive programs rather than as a list of separate program efforts.23 Third, greater attention should be given to the gap between a program's legal requirements and its implementation. Often research parameterizes programs by their legal definition, with only scant attention to implementation issues. For instance, according to law, welfare programs to single parents in the United States levy extremely high tax rates on earnings, between 67 and 100 percent. Many studies have used these legislated rates in their estimation of the impact of welfare on labor supply, generally concluding that high tax rates seem to have little effect on labor supply decisions of recipients. Yet the actual marginal tax rate
s on earnings faced by most women are much lower, because of a combination of work expense deductions and income underreporting. Perhaps the inference that labor supply is only modestly responsive to the tax rates in income support programs is correct because the effective tax rates are actually quite low. Fourth, we should go beyond determining the immediate effects of a program on labor supply or mobility behavior to the multiple and possible longterm effects on a broader set of variables, including skill formation. This means looking at efficiency, flexibility, and well-being effects of programs over a time horizon greater than one or two years. At this stage we are not even sure whether effects increase or decrease over time. Economists normally expect long-run effects to exceed short-run effects (as the number of constraints on behavior is reduced, own partial derivatives increase by the Le Chatelier principle), but when one program changes, the use and purpose of a whole range of other programs may also change in an offsetting way. Changes in employee dismissal laws may lead to differences in the way unemployment benefits are paid out, with uncertain consequences. To assess more reliably the effects of specific program reforms, we need at least some understanding of how the entire interconnected system of programs adjusts. Finally, the purported long-term investment effects of social programs should be analyzed. Do particular investments in children's education or health programs or transfers to families with children pay off in terms of future adult productivity? Is the claim valid that many social programs are investments in the future, or are these more properly viewed as consumption transfers? The availability of long-term longitudinal data makes studies of long-term effects 23. For an example of multiple-program analysis with respect to the U.S. income support system, see Blank and Ruggles (1992).
39 The Connection between Protection and Flexibility increasingly possible. The argument that social protection programs are investment- and productivity-enhancing policies needs to be seriously evaluated both by proponents and by skeptics. In short, we believe that the research and policy community actually knows quite a bit less about the aggregate effect of social protection programs on individual behavior or on the aggregate economy than is typically claimed. Our reading of the debate over flexibility, social protection, and the labor market suggests that analyses more directly focused on the key points in that debate would enrich our understanding of how the welfare state and labor market interact in needed and useful ways. References Abraham, K., and S. N. Houseman. 1993. Job security in America: Lessons from Germany. Washington, D.C.: Brookings Institution
. Atkinson, A. B., and J. Micklewright. 1991. Unemployment compensation and labor market transitions: A critical review. Journal of Economic Literature 29:1679727. Barnett, W. S. 1992. Benefits of compensatory preschool education. Journal of Human Resources
27:279-312. Bean, C. R., R. Layard, and S. Nickell. 1987. The rise in unemployment. Oxford: Basil Blackwell. Blank, R. M. 1993. Why were poverty rates so high in the 1980s? In Poverty and prosperity in the U.S. in the late twentieth century, ed. D. Papadimitriou and E. Wolf. London: Macmillan. Blank, R. M., and P. Ruggles. 1992. Multiple program use in a dynamic context. Report to the U.S. Bureau of the Census. Washington, D.C. Boyer, R. 1988. The search for labour market flexibility. Oxford: Clarendon. Bruno, M., and J. Sachs. 1985. The economics of worldwide stagflation. Cambridge: Harvard University Press. Burtless, G. 1987. Jobless pay and high European unemployment. In Barriers to Euro- pean growth: A transatlantic view, ed. R. Lawrence and C. Schultze. Washington, D.C: Brookings Institution. Danziger, S., and P. Gottschalk. 1993. Uneven tides: Rising inequality in the 1980s. New York: Russell Sage Foundation. Davis, S. J. 1992. Cross-country patterns of changes in relative wages. NBER Working Paper no. 4085. Cambridge, Mass.: National Bureau of Economic Research. Dore, R. 1986. Flexible rigidities. Stanford, Calif.: Stanford University Press. Dreze, J. H., and C. R. Bean. 1990. Europe's unemployment problem. Cambridge: MIT Press. Flanagan, R. J. 1987. Labor market behavior and European economic growth. In Barriers to european growth: A transatlantic view, ed. R. Lawrence and C. Schultze. Washington, D.C: Brookings Institution. Franz, W 1987. Hysteresis, persistence, and the NAIRU: An empirical analysis for the Federal Republic of Germany. In The fight against unemployment, ed. R. Layard and L. Calmfors. Cambridge: MIT Press.
40 Rebecca M. Blank and Richard B. Freeman Freeman, R. B. 1988. Evaluating the European view that the United States has no unemployment problem. American Economic Review 78:294-99. Gottschalk, P. 1992. Is the correlation in welfare participation across generations spurious? Boston College. Manuscript. Hansen, G., and A. imrohoroglu. 1992. The role of unemployment insurance in an economy with liquidity constraints and moral hazard. Journal of Political Economy 100:118-42. Haveman, R. H. 1985. Does the welfare state increase welfare? Leiden, The Netherlands: H. E. Stenfert Kroese. Haveman, R. H. and B. L. Wolfe. 1984. Schooling and economic well-being: The role of nonmarket effects. Journal of Human Resources 19:377-407. Houseman, S. N. 1991. Industrial restructuring with job security: The case of European steel. Cambridge: Harvard University Press. Jessop, B., H. Kastendiek, K. Nielson, and O. K. Pederson. 1991. The politics of flexibility: Restructuring state and industry in Britain, Germany, and Scandinavia. Worcester, England: Billing and Sons. Katz, L. E, G. W. Loveman, and D. G. Blanchflower. 1992. A comparison of changes in the structure of wages in four OECD countries. Working paper prepared for the NBER Comparative Labor Markets Project Conference on Differences and Changes in Wage Structures. Cambridge, Mass.: National Bureau of Economic Research. In Differences and changes in wage structure, ed. L. E Katz and R. B. Freeman. Chicago: University of Chicago Press.
Forthcoming. Katz, L. F, and B. D. Meyer. 1990. The impact of the potential duration of unemployment benefits on the duration of unemployment. Journal of Public Economics 41:45-72. Kemper, P., D. A. Long, and C. Thornton. 1984. A benefit-cost analysis of the supported work experiment. In The national supported work demonstration, ed. E. G. Hollister Jr., P. Kemper, and R. A. Maynard. Madison: University of Wisconsin Press. Krugman, P. R. 1987. Slow growth in Europe: Conceptual issues. In Barriers to European growth: A transatlantic view, ed. R. Lawrence and C. Schultze. Washington, D.C.: Brookings Institution. Lampman, R. J. 1984. Social welfare spending: Accounting for changes from 1950 to 1978. Orlando: Academic Press. Lawrence, R. Z., and C. L. Schultze. 1987. Overview, in Barriers to European growth: A transatlantic view, ed. R. Lawrence and C. Schultze. Washington, D.C.: Brookings Institution. Layard, R. 1986. How to beat unemployment. Oxford: Oxford University Press. Lazear, E. P. 1990. Job security provisions and employment. Quarterly Journal of Eco- nomics 60:699-726. Lindbeck, A., and D. Snower. 1990. Demand- and supply-side policies and unemploy- ment: Policy implications of the insider-outsider approach. In Unemployment and wage determination in Europe: Labour market policies for the 1990s, ed. B. Holmlund and K-G. Loftgren. Oxford: Basil Blackwell. Metcalf, D. 1987. Labour market flexibility and jobs: A survey of evidence from OECD countries with special reference to Europe. In The fight against unemployment, ed. R. Layard and L. Calmfors. Cambridge: MIT Press. Modigliani, E, M. Monti, J. H. Dreze, H. Giersch, and R. Layard. 1987. Reducing unemployment in Europe: The role of capital formation. In The fight against unemployment, ed. R. Layard and L. Calmfors. Cambridge: MIT Press. Moffitt, R. 1992. Incentive effects of the U.S. welfare system: A review. Journal of Economic Literature 30:1-61.
41 The Connection between Protection and Flexibility Olson, M. 1982. The growth and decline of nations. New Haven, Conn.: Yale University Press. Organization for Economic Cooperation and Development (OECD). 1990. Labor market policies for the 1990s. Paris: OECD. . Employment outlook: 1991. Paris; OECD. . 1992a. Employment outlook: 1992. Paris: OECD. . 1992b. Historical Statistics, 1960-1990. Paris: OECD. Oswald, A. J. and D. G. Blanchflower. 1990. The wage curve. In Unemployment and wage determination in Europe: Labour market policies for the 1990s, ed. B. Holmlund and K-G. Loftgren. Oxford: Basil Blackwell. Piore, M. J., and C. F. Sabel. 1984. The second industrial divide. New York: Basic Books. Rein, M., and B. L. Friedman. 1992. Social protection and economic change. Cambridge: MIT Press. Weisbrod, B. A. 1983. Defining benefits of public programs: Some guidance for policy analysts. In Public expenditure and policy analysis, ed. R. H. Haveman and J. Margolis. Boston: Houghton Mifflin.