This paper is motivated by a simple idea that has received lamentably little attention in the literature on unemployment policy: different unemployment policies are generally based on different theories of unemployment, and confidence in a policy should depend—at least in part—on the ability of the underlying theory to account for some prominent empirical regularities in unemployment behavior.
This paper is motivated by a simple idea that has received lamentably little attention in the literature on unemployment policy: different unemployment policies are generally based on different theories of unemployment, and confidence in a policy should depend—at least in part—on the ability of the underlying theory to account for some prominent empirical regularities in unemployment behavior.
Some theories depict unemployment as the efficient outcome of market activity. These usually serve to rationalize a laissez-faire policy stance. Other theories depict unemployment as the product of market failures. Here, unemployment must be seen as the symptom of many possible diseases, as many different market failures can produce the same problem of joblessness. And just as different diseases require different treatments, so different market failures may call for different government policies.2 It is because different theories of unemployment focus on different market failures that different policies are generally based on different theories.
It is difficult to evaluate the various unemployment policies through a direct empirical assessment of the market failures identified by the underlying theories. After all, market failures arise when people are not fully compensated for the costs and benefits that they impose on one another, and uncompensated costs and benefits are inherently difficult to measure. For this reason, it is natural to evaluate unemployment policies by investigating the predictive power of the underlying theories. A first step in this direction is to examine the degree to which these theories are able to account for some generally recognized regularities in the movement of unemployment rates in Organization for Economic Cooperation and Development (OECD) countries over the postwar period. This approach would perhaps be too obvious, were it not so frequently at variance with the standard rationalizations of unemployment policies.
Admittedly, the suggested criterion is highly simplistic. In practice, unemployment may generally be expected to arise from several different causes operating simultaneously, and it would not be reasonable to expect any single theory to explain all the salient empirical features of unemployment behavior in the OECD. All that this paper claims is that confronting unemployment policies with these empirical features can give a useful preliminary indication of how significant these policies are likely to be. It would surely be unwise to have a heavy stake in a policy whose underlying theory explains little of how unemployment has evolved in the postwar period.
The paper is organized as follows. The first section deals with the laissez-faire policy stance, based on theories of voluntary unemployment. The next section deals with demand-management policies, which rest on Keynesian theory. The following three sections consider, respectively, supply-side policies aimed at raising workers’ productivity, the interaction between demand- and supply-side policies, and institutional policies designed to change labor market institutions. Concluding remarks are then presented.
The laissez-faire policy stance—in which the government does little or nothing to influence unemployment—is based primarily on models in which swings in unemployment are viewed as the outcome of the optimizing decisions by job seekers and job providers in efficient markets. In this context, active unemployment policy is generally undesirable as it only disturbs the workings of well-functioning markets and interferes with people’s free choices to accept or reject employment.
There are two main types of laissez-faire stances. One argues that government interventions aimed at influencing the long-run equilibrium unemployment rate would be ineffective or undesirable but acknowledges the possible effectiveness and desirability of policies to deal with cyclical swings in unemployment. In particular, it advocates the implementation of predictable policies, whose effects can be readily foreseen by economic agents. This view receives its most forceful expression in the market-clearing variant of the natural rate theory. The other laissez-faire stance discourages intervention not only with the long-run equilibrium unemployment rate, but also with cyclical unemployment swings. This stance rests primarily on intertemporal substitution theory and real business cycle theory.
The market-clearing variant of the natural rate theory is an obvious vehicle for rationalizing the importance of policy predictability. In this theory, unemployment is at its “natural rate” when people’s expectations about wages and prices are correct. Under conditions of perfect competition and perfect information, this natural rate depends only on people’s tastes, technologies, and resource endowments.3 When wage-price expectations are out of line with actual wages and prices, unemployment deviates from its natural rate.
Provided that tastes, technologies, and endowments do not fluctuate cyclically, fluctuations in unemployment—according to this theory—must be explained by fluctuations in expected wages and prices around their actual values.4 In order for this theory to have predictive power, it needs to be combined with a theory of how expectations are formed. The dominant one is the rational expectations theory, which asserts—quite plausibly—that people are not fooled in ways that they themselves could have predicted. To test this hypothesis, a description of people’s “information sets” is required, from which expected wages and prices could be inferred. In practice, this is, of course, an impossible task, so empirical models generally assume that everyone has the same information as the authors of the models—except that the authors have more recent data.
The combination of the natural rate theory with this variant of rational expectations theory has a well-known implication: as people make no systematic expectational errors (errors that they could have predicted), unemployment cannot diverge systematically from its natural rate. Just as expected wages and prices will fluctuate randomly around their actual values, so unemployment will fluctuate randomly around the natural rate.
It is not hard to see why policy predictability is advisable in this context. With well-functioning markets, there is clearly no efficiency case to be made for interfering with the natural rate of unemployment. Policies that have no influence on this natural rate—monetary policies, for example—can affect unemployment only by driving a wedge between actual and expected wages and prices through unexpected policy variations, such as unexpected changes in the money supply. To put it simply, demand-management policies are effective only when they are deceptive. But deceptive policies are generally not in the public interest: if people were initially pursuing their own interests in well-functioning markets—and thereby promoting the public interest as well—unexpected changes in policy would only reduce individual and social welfare. Thus, stabilization policy is reduced to the limited task of being predictable.
The problem with this theory is that it fails to address many of the features of European unemployment over the past decade. In many OECD countries over the 1980s, union density remained constant or even declined, the expansion of unemployment benefits and other forms of welfare state support was arrested or even reversed, and deregulation, privatization, and liberalization of labor markets were not uncommon. On this account, it is difficult to argue that the natural rate of unemployment could have risen significantly. Some economists have maintained that the expansion of the welfare state in Europe over the 1950s, 1960s, and the first part of the 1970s may have been responsible for the growing unemployment in the 1980s, as people adjusted their behavior only gradually to the more generous social provision.5 But it is hard to imagine that these lagged responses should have been so powerful and so delayed as to be responsible for the large, decade-long increases in European unemployment after most welfare states had ceased to expand.
Furthermore, given the stable rates of inflation over much of the decade, it cannot be argued that people’s wage-price expectations were getting further and further out of line with actual wages and prices; nor is it plausible to suppose that these misperceptions could have persisted for so long.
In short, there is nothing in the market-clearing variant of the natural rate theory that provides a clue to the massive rise in European unemployment through the 1980s, Nor does this theory shed useful light on why unemployment has been so much more persistent in Europe than in the United States, or why European unemployment rose with each major recession of the 1970s, 1980s, and early 1990s, while U.S. unemployment has always tended to return to its prerecession level. Can it honestly be believed that Europeans are much slower than Americans to adjust their expectations, so that expectational errors are more persistent in Europe than the United States? Beyond that, theories based on expectational errors reveal little, if anything, about why unemployment spells tend to be longer in Europe than the United States (for given unemployment rates), why U.S. unemployment rates are more variable than most European ones, why unemployment falls unequally among different population groups, and why labor and product markets move so much more closely in tandem in the United States than Europe.
Noninterference with Business Cycles
The case against using stabilization policies in the labor market is made quite explicit in the intertemporal substitution and real business cycle theories.
As the name implies, the intertemporal substitution theory6 is concerned with workers’ desire to engage in intertemporal substitution of work for leisure in response to changing economic incentives. For example, if workers believe that real wages are temporarily depressed and will rise in the future, they may wish to take more leisure now and work harder later. The same may be true if they perceive real interest rates to be temporarily low, as that means that their current wage income cannot be transferred into the future at an advantageous rate.
The implication is that cyclical swings in employment may be the optimal response—by individual agents and society at large—-to temporary shocks to tastes, technologies, and endowments.7 Whereas most economists used to see business cycles as undesirable and needing to be damped through stabilization policies, the intertemporal substitution theory suggests that this is not so. It is not in the public interest to implement countercyclical monetary and fiscal policies, as these would prevent people from making their optimal dynamic responses to external shocks.
Although this theory can be used to generate an empirical account of much of the unemployment persistence and variability observed in the United States and Other OECD countries,8 it is hard to see how it could provide a reasonable explanation of European unemployment over the past 25 years. Is it plausible to believe that the many millions of Europeans who joined the unemployment register in the mid-1970s, early 1980s, and early 1990s were merely trying to take advantage of the high real wages or high real interest rates expected to occur in the future? Regarding the upward trend in European unemployment rates since the mid-1970s, is it plausible to believe that a very long-term intertemporal substitution is taking place, whereby workers have decided to enjoy considerable free time for two decades, perhaps with the intention of working very long hours for the next two decades? Furthermore, the available empirical evidence indicates that hours of work are unresponsive to real wage and real interest rate variations,9 and that many of these variations tend to be permanent rather than temporary.
The real business cycle theory10 builds on the intertemporal substitution theory and identifies technological shocks as the main source of macroeconomic fluctuations. Much effort has been expended attempting to establish that these fluctuations arise when perfectly informed individuals, all maximizing their utility subject to technological and resource constraints, respond to technological shocks by intertemporally substituting labor, leisure, and consumption. But beyond the predictive problems of the intertemporal substitution theory, it is hard to get an intuitive interpretation of the technological shocks. Whereas technological advances (which the real business cycle theory associates with the booms) are relatively easy to identify, the technological setbacks (which allegedly give rise to the recessions) are not.11 It is hard to see how knowledge and expertise gets lost, particularly on the large scale that is necessary to account for the deep recessions that have been witnessed over the past two decades. Some economists argue that the negative technological shocks reflect such adverse macroeconomic events as oil price hikes or inappropriate investment (such as machinery that turns out not to work or that produces goods for which demand did not materialize). But the downturns in European labor market activity over the past two decades have lasted much longer than the price hikes for oil and other resources, and it would be strange—in the real business cycle world of rational expectations in clearing, perfectly functioning markets—for these shocks to generate investment fluctuations that are large enough to pull the massive OECD recessions in their wake. Beyond that, the long-term increase in European unemployment since the mid-1970s cannot plausibly be explained as the market-clearing outcome of technological shocks.
Demand-management policies to reduce unemployment fall into two broad categories: government employment policies, whereby the government stimulates employment directly by hiring people into the public sector; and product demand policies, which stimulate employment by raising aggregate product demand (for example, through tax reductions, increases in government spending on goods and services, or increases in the money supply).
For the “short run,” in which wages and prices respond sluggishly to demand fluctuations, the main underpinning for both types of policies is Keynesian theory.12 Here, recessions are characterized by deficient labor and product demand reinforcing one another: workers are unemployed because firms are not producing enough goods and services; firms are not doing so because there is too little demand; and demand is deficient because people are unemployed. In short, deficient demand in the labor market originates in the product market, and deficient demand in the product market originates in the labor market. Activity in these two markets thereby goes up and down together. The mechanism that couples these two markets is wage-price sluggishness, A fall in product demand will reduce labor demand if wages do not fall sufficiently; a fall in labor demand will reduce product demand if prices are sluggish downward.
This interaction between product and labor markets gives demand-management policy considerable leverage in the Keynesian theory. A rise in government employment will raise the purchasing power of the newly employed, who, in turn, will demand more goods and services, thereby inducing firms to employ yet more people, and so on. In the same vein, a stimulus to product demand (resulting, say, from a tax reduction) gives firms the incentive to raise employment, which creates more purchasing power, which raises product demand even further, and so on. The more sluggish wages and prices are, the greater these multiplier effects become.
Of course, in practice wages and prices are not sluggish indefinitely, and thus the critical question is how short the Keynesian “short run” really is. Clearly, if it were shorter than the time it takes most firms to make and implement their employment and production decisions, the Keynesian employment repercussions of demand-management policies could not be expected to be significant. Wage-price sluggishness in excess of the relevant production and employment lags is required before Keynesian policies come into their own.
The Keynesian quantity-rationing theory13 provides no guidance in this respect, as it merely assumes wages and prices to be indefinitely rigid. The New Keynesian theories of nominal sluggishness move beyond this primitive assumption. They seek to explain why wages and prices do not change sufficiently to obviate the need for substantial adjustments of output and employment in response to changes in demand. In shedding light on the degree of wage-price sluggishness, the ultimate aim is to help determine the length of time over which Keynesian policy effects are operative. The three dominant New Keynesian theories in this area are the “menu cost” theory, the theory of “near rationality”, and the wage-price staggering theory.14
According to the menu cost theory, small fixed costs of changing prices induce firms to adjust quantities, instead of prices, in response to a sufficiently small change in demand. The same holds even in the absence of price adjustment costs if firms are “nearly rational,” that is, changing their prices only when that has a substantial effect on profits. There are, however, a number of obstacles to using these theories to derive the degree of wage-price sluggishness. First, the existing menu cost models show how product demand variations affect employment when the costs of changing prices are the only adjustment costs. In practice, however, as employment adjustment costs (such as hiring, training, and firing costs) generally exceed the price adjustment costs by a large margin, it is not clear why product demand changes should have Keynesian effects on employment. Second, the menu cost theory implies that prices are either rigid or completely responsive to demand shocks, for the cost of small price changes is generally no different from the cost of large ones. The theory is thus unable to explain an important feature of wage-price sluggishness in practice, namely, that many firms change their prices frequently, but not by sufficiently large amounts to make significant quantity adjustments unnecessary. These two difficulties make it difficult for the menu cost theory to predict the degree of wage-price sluggishness and the short-run effectiveness of Keynesian demand-management policy.
The theory of near rationality is subject to the first of these two difficulties: to explain the effectiveness of Keynesian demand-management policy, the deviation from complete rationality must be sufficiently large to outweigh the costs of adjusting employment and production. Moreover, as it is hard to see how this deviation could be measured empirically, this theory does not yield firm quantitative predictions on the degree of wage-price sluggishness.
The wage-price staggering theory demonstrates that, if wages and prices, once set, are fixed over substantial contract periods and if different wage-price decisions are staggered (rather than set simultaneously), a current change in aggregate product demand will affect production, employment, and unemployment well beyond the expiry of the current contract period. Several important lacunae in this theory, however, keep it from providing a firm basis to predict the degree of wage-price sluggishness. First, the theory does not identify the wage-price adjustment costs that keep wages and prices fixed over substantial intervals. Without a handle on these costs, the length of the contract periods—which plays such an important part in determining the degree of wage-price sluggishness in this context—cannot be derived. Second, the theory rests on the assumption that wages and prices are set in advance in nominal terms; it does not explain why rules for setting wages and prices generally do not involve indexing. If people have no “money illusion”—and therefore understand the difference between changes in value expressed in nominal and real terms—and if simple indexation schemes (such as making the wage depend on an aggregate price index) are easy to formulate and monitor, it remains an open question why so many wages and prices are set in nominal terms.15 Third, the theory does not explain what determines the degree to which rules for setting wages and prices are time dependent (changing as a function of time) versus state dependent (changing as a function of external contingencies).16 This is an important issue because time- and state-dependent rules have very different implications for the degree of wage-price sluggishness following a change in product demand.17 Fourth, little attention has been given to the question of why wage-price decisions are staggered rather than synchronized. Ball and Romer (1989) attribute it to firm-specific shocks, whereas Ball and Cecchetti (1988) suggest that staggering can arise from firms’ incentives to set their prices after they have gained information about their rivals’ price changes. As these examples show, the different sources of staggering imply radically different staggering structures and, presumably, radically different degrees of wage-price inertia. Finally, different sectors of the economy are characterized in practice by vastly different periods of nominal adjustment; the resulting patterns of staggering are enormously complex—perhaps too complex, at the requisite level of disaggregation, to be a convenient predictive tool.
Nevertheless, many economists agree that the Keynesian view sheds some light on unemployment behavior during deep recessions. When economies suffer from high unemployment and low capital utilization, increases in aggregate demand generally lead to declines in unemployment, and demand reductions usually lead to increases in unemployment. But the 1980s have exposed an important shortcoming of the Keynesian theory: for most of that decade, European labor and product markets did not move together at all. Product demand started to pick up toward the end of 1982, but unemployment did not start to fall until 1986 in the United Kingdom and even later in most other European countries. This gap is simply too large to be explained by inventory dynamics or lags between inputs and outputs in production processes; the Keynesian vision of tightly linked labor and product demand is called into question. The link was much stronger in the United States than in most European countries over the 1980s. This disparity is too large to be rationalized simply in terms of greater wage-price sluggishness in the United States than in Europe. Consequently, it becomes difficult to account for the serial correlation in OECD unemployment rates—and the greater degree of serial correlation in the European countries than in the United States—through the serial correlation in aggregate demand. Nor does the Keynesian theory explain readily why unemployment durations over the past two decades have been much longer in the European Community (EC) than in the United States or Japan, even after normalizing for differences in unemployment rates.
Lastly, the various New Keynesian theories of wage-price sluggishness can do no more than explain why unemployment can respond to variations in aggregate demand for a limited period of time. The longrun rise of European unemployment over the past two decades clearly cannot be rationalized on this basis. This does not mean that demand-management policies cannot conceivably influence unemployment over the longer run, but merely that such longer-term effectiveness must rest on something other than wage-price sluggishness. Several possibilities will be considered below in the discussion on the interaction between demand- and supply-side policies.
Job Search Support and Information Dissemination
This general policy approach covers such measures as counseling the unemployed, assisting them with personal problems, such as alcoholism and drug addiction, and alerting them to available training opportunities.18 This approach also involves disseminating information about available labor services to firms and about available vacancies to workers.
Many of the market failures addressed by these policies can be analyzed effectively through the theory of search and matching.19 In this theory, workers are not perfectly informed about the available jobs, and firms are not perfectly informed about the available workers. Thus, both sides of the market engage in search. Each agent acquires information up to the point at which the cost of searching for an additional job (or worker) is equal to the discounted stream of expected future returns from that job (or worker). Unemployment arises because jobless workers know that, although there are vacant jobs with wages sufficiently high to make the returns from the search exceed the costs, they do not know precisely where these jobs are and may not find them right away. The result is “frictional unemployment,” which is never eliminated in aggregate since there are always some workers getting fired, some quitting, some entering the labor force, and some retiring from it. At center stage in all search models lies a “matching function,” which specifies how the expected number of matches is related to the number of unemployed workers and the number of vacant jobs.
It is not possible, of course, to attribute the rise in European unemployment to a deterioration of this matching technology, because the dissemination of labor market information has, if anything, improved with the passage of time. Nor are the recent periods of high unemployment related to comparatively high degrees of labor market “turbulence,” that is, sectoral imbalances responsible for job creation and job destruction.20
Thus, if imperfect information about vacant jobs and unemployed workers were the only problem for job search support and information dissemination to overcome, its potential would probably be quite limited. However, this policy may also be useful in overcoming the discouragement and demoralization that prevent many long-term unemployed from seeking jobs effectively. The search and matching theory views this problem as the consequence of a decline in people’s returns from job searches as their unemployment spells lengthen. The declining returns could be due to the depreciation or obsolescence of their skills and to a resulting fall in Firms’ efforts to attract these workers. Another reason why workers’ search intensity may decline is that their preferences gradually change. In particular, the long-term unemployed can become accustomed and reconciled to remaining jobless, adopt it as a way of life, and stop searching seriously at all.21 Counseling and personal assistance may help to mitigate these problems by restoring the attitudes and expectations necessary for successful job search strategies.
The potential importance of this policy approach may be highlighted by the recognition that the decline of search intensity with unemployment duration undoubtedly plays a significant role in explaining unemployment persistence (namely, the dependence of current unemployment rates on past unemployment rates).22 This approach also helps explain why the burden of unemployment is distributed unequally. If the search intensity of workers falls the longer that they remain unemployed, and if the corresponding search intensity of potential employers falls as well, the expected future length of these workers’ unemployment spells will depend positively on how long they have already been unemployed.
Aside from the search and matching theory, another rationale for supporting job searches and improving information dissemination—as well as for implementing various other policies to be discussed below—comes from the efficiency wage theory. Here, firms are assumed to have imperfect information about individual employees’ productivities and are thus unable to make their wage offer contingent on their employees’ performance. The firms, as wage setters, observe that by raising their wage offers they are able to increase the average productivity of their workforce, because higher wage offers enable a firm to recruit more highly qualified employees or motivate employees to work harder.23 In other variants of the theory, higher wages discourage workers from quitting the firm, thereby reducing the firm’s labor turnover costs.24 Consequently, firms may have an incentive to keep wages above levels that would be necessary to ensure full employment. The unemployed are unable to get jobs by offering to work for less than the prevailing wage because it is not in the firms’ interests to allow wages to fall.
In this context, policies that improve the dissemination of information about workers’ ability, motivation, and quit behavior would enable firms to base their wage offers more closely on workers’ individual productivities and potential labor turnover costs, thereby reducing the role of wages as an incentive mechanism and bringing down the associated level of unemployment.
The great strength of the efficiency wage theory is that it provides one conceivable explanation for why, even under perfectly flexible wages, people may be unemployed even though they would prefer to do the jobs of the current job holders at less than the prevailing wage.25 Beyond that, however, it is not clear that the theory can shed much light on why EC unemployment has risen over the past two decades. It might be argued that with the decline in assembly line production, which is fairly easy to monitor, firms have placed increasing reliance on wages as an incentive mechanism. But if that were an important consideration, it would apply to both the EC and the United States, leaving the open question of why unemployment in the EC has risen so much relative to that in the United States. Moreover, the advance of computer technology has improved firms’ monitoring capabilities in many sectors of the economy, making them less reliant on wages to motivate and attract employees.
It could also be argued that the use of wages as an incentive mechanism is more important in countries with more stringent job security legislation. After all, the more costly it is for firms to fire their employees, the lower are the chances a shirking or incompetent worker will be dismissed, and thus the greater the wages that firms have to pay (relative to unemployment benefits) in order to stimulate productivity. However, if this interaction between labor turnover costs and efficiency wages were significant in practice, it would be hard to explain why the EC unemployment rate averaged less than the U.S. unemployment rate over the 1950s and 1960s, given that job security legislation tended to be more stringent in the EC than in the United States over the entire postwar period.
Furthermore, the efficiency wage theory does not explain why the average duration of unemployment in Europe has significantly exceeded that in the United States and Japan since the mid-1970s, why labor and product market activities tend to move together in the United States but not in Europe, or why unemployment in many countries varies less within a business cycle than from one cycle to the next. These phenomena clearly cannot be ascribed to differences in monitoring technologies through time and across countries.
Of course, many efficiency wage models explain how unemployment may rise in response to a drop in labor productivity, a rise in the real interest rate, or a rise in the unemployment benefit. However, as with the search models, the efficiency wage models cannot lay unique claim to these predictions. The efficiency wage models do not add much to what other theories have to say in this respect. Similarly, the inclusion of labor turnover costs in an efficiency wage setting can explain why unemployment rates tend to be serially correlated, and differences in the magnitude of these costs can help account for intercountry differences in such serial correlation, as well as intercountry differences in duration of unemployment. But labor turnover costs are not an intrinsic building block of efficiency wage models. These models can rationalize the existence of unemployment even in the absence of labor turnover costs, and the addition of these costs to a wide variety of other theories would yield equivalent insights into unemployment dynamics.
Policies to Stimulate Worker Mobility
Some policies designed to reduce the burden of housing costs to the poor, such as rent control or low-cost public housing, reduce worker mobility by inhibiting workers from moving to available jobs and thereby create unemployment. This is a potentially significant problem in a number of OECD countries containing booming and slumping regions, as well as large differentials in house prices and rents across these regions.
These differentials may be incorporated in models of search and matching to provide an explanation for regional differences in unemployment rates. Although, as noted, there is little if any evidence that the rising European unemployment rates over the past two decades can be attributed to greater volatility of sectoral labor market shocks or to reduced availability of information about unemployed people or vacancies, differentials in house prices and rents can become an especially serious source of mismatch in the labor market, as they often expand when the degree of sectoral imbalance rises.26 In particular, the greater the discrepancy between the excess of vacancies over unemployment in the booming regions and the excess of unemployment over vacancies in the slumping ones, the greater the differentials in house prices and rents are likely to be, owing to the greater discrepancy in excess demand across regions. Thus, as the degree of mismatch rises, the impediments to matching may rise in tandem.
Rent control and housing subsidies tied to the current place of residence give further leverage to these impediments to matching. Replacing these policy interventions by more efficient ways of redistributing income (such as conditional negative income taxes, which are discussed in the following section) could therefore help reduce unemployment. A similar argument can be made for implementing policies that increase the portability of health insurance and pensions between firms.
Policies Centering on Human Capital Formation
Policies that focus on human capital formation include government training programs and training subsidies to firms or workers,27 as well as policies that reduce the rate of interest and thereby reduce the rate at which future returns to human capital formation are discounted.28
These policies may be analyzed through a special set of search and matching models that explain how unemployment can arise on account of market failures in the demand for, and supply of, training.29 First, as unemployed people have few firm-specific skills, training them may involve a large “poaching externality.” Specifically, if unemployed people were given training, a large share of the benefits from that training, in imperfectly competitive labor markets, would fall neither on the firms supplying the training nor on the workers receiving it, but on third parties, namely, firms that may poach the workers after they have been trained. In that event, the social benefit from training would exceed the private benefit, regardless of how the costs of training were distributed between the trainer and trainee. The market would then generate too few matches between firms and currently unemployed workers; with relatively few workers becoming productive and profitable through training, an inefficiently large number of them would remain jobless (see Snower (1994b)), This problem may become magnified considerably through the “low-skill, bad-job trap”: a deficient supply of trained job seekers induces firms to create an excessive number of unskilled vacancies; these, in turn, further reduce workers’ incentives to acquire training, leading to even more unskilled vacancies, and so on (see Snower (1994a)). These market failures are likely to be especially pronounced with regard to the long-term unemployed, who tend to be particularly poorly endowed with firm-specific skills and thus particularly prone to the poaching externality and the low-skill, bad-job trap.
Some of the rise in European unemployment over the past two decades might arguably be due to the interaction between these market failures, on the one hand, and the joint pull of skill-biased technological change and international trade, on the other. Technological developments that raise the productivity of skilled workers relative to unskilled workers, as well as rising trade with countries that have a comparative advantage in producing goods relatively intensive in unskilled labor, reduce the demand for unskilled labor relative to the demand for skilled labor. If the market failures discussed above are responsible for a deficiency in the acquisition of skills and an excessive number of unskilled workers without jobs, these two factors could lead to a rise in unemployment.
In addition, an expansion of trade or an increased rate of technological change could generate unemployment by raising the amount of labor market turbulence, particularly by increasing the rate of job creation and destruction.30 This, of course, is not an argument for the implementation of policies limiting the degree of technological change or international trade, for—as is well known—these two factors generally permit a given amount of goods and services to be produced with less labor input and thereby could improve the overall material standard of living, provided that the appropriate redistributions from the winners to the losers can be made without substantial loss of efficiency. Rather, the above diagnosis is an argument for job search support in order to improve the effectiveness of the matching process.
Thus, government training programs or training subsidies to the unemployed—particularly the long-term unemployed—may have a role to play in combating unemployment. Many government training programs, however, are ill-suited to firms’ needs. This is scarcely surprising, as these needs are extremely diverse while government training programs are inevitably standardized and limited in variety. Training subsidies granted to firms thus appear preferable, for the firms then have the incentive to make the resulting training maximally appropriate to their available jobs. To keep firms from illicitly diverting the training funds to other purposes, it may be necessary to provide training subsidies only for programs leading to nationally recognized qualifications granted by institutions independent of the firms receiving the subsidies.31
The Interaction Between Demand- and Supply-Side Policies
Policies Centering on Physical Capital Formation
These policies range from government infrastructure investment to policies that raise the rate of capital utilization, stimulate the entry of firms, or promote physical capital formation by reducing the user cost of capital. What these policies all have in common is that they raise the level of capital services provided in the economy; consequently, if labor and capital are complementary in the production process, they increase the marginal product of labor, thereby raising employment and reducing unemployment. By simultaneously increasing investment demand and capital supply, these policies work on both the demand and supply sides. In so doing, they illustrate how demand management policies can be effective in the “longer run,” that is, over a time span long enough to permit full adjustment of wages and prices.
A growing number of economists have come to suspect that the effectiveness of demand-management policy is undersold by the short-run Keynesian mechanisms described above, which rest on wage-price sluggishness. Many believe that aggregate demand had a role to play in sustaining periods of prolonged low unemployment in the 1960s and prolonged high unemployment in the 1980s in Europe. However, for that to have been the case, the influence of aggregate demand on employment must extend well beyond the span over which wages and prices can be presumed sluggish.
The long-run effectiveness of demand-management policy may be analyzed through theories of imperfect labor market competition. In this vein, it is useful to picture labor market equilibrium in terms of the intersection of a downward-sloping aggregate labor demand curve and an upward-sloping wage-setting curve in real wage-employment space. The aggregate labor demand curve depicts the horizontal sum of firms’ profit-maximizing relations between labor demand and the real wage under imperfect competition, and the wage-setting curve represents the real wage that emerges, at any given level of employment, from wage bargaining or efficiency wage minimization. Furthermore, the labor supply curve lies to the right of the wage-setting curve. The difference between labor supply and equilibrium employment is the equilibrium level of unemployment.
In this context, an increase in product demand can reduce unemployment by shifting the wage-setting curve or the labor demand curve outward. If only the wage-setting function shifted (along an unchanged labor demand curve), the real wage would move countercyclically. However, as real wage movements are often acyclic or even procyclical (particularly in the United States), it is important to explore how demand management policy can shift the labor demand curve, thereby allowing for the possibility of procyclical real wage movements.32 The most likely channels through which demand management can affect unemployment in the long run involve significant supply-side effects. Moreover, the interaction between demand- and supply-side policies becomes crucial in this regard.
Because the labor demand curve is the set of real wage-employment combinations at which the real marginal value product of labor is equal to the real wage, a change in product demand can shift the labor demand curve only if it affects the real marginal value product of labor at any given level of employment. It is easy to show that this shift occurs whenever the product demand change affects (i) the price elasticity of product demand, (ii) the imperfectly competitive interactions among firms, (iii) the degree of capital utilization, (iv) the user cost of capital, (v) the number of firms in operation, and (vi) the marginal product of labor.33
Of the channels through which product demand changes can be transmitted to employment, the first two do not appear to provide a firm foundation for the effectiveness of demand-management policies.
With respect to the price elasticity of product demand, some authors have suggested that changes in government spending can affect employment by changing the composition of product demand, which, in turn, changes the associated price elasticity of aggregate demand (see, for example, the survey in Dixon and Rankin (1994)). There are, however, good reasons to believe that this hypothesis would be a tenuous basis for government policy. First, an increase in government spending would shift the labor demand curve outward through this channel only when the public sector price elasticity of demand exceeds that of the private sector, and there is no evidence that this is consistently the case across sectors and through time. Second, this transmission mechanism has the implausible implication that, if an increase in government expenditures shifts the labor demand curve outward, a tax reduction must shift that curve inward, as the former policy would raise public sector spending relative to private sector spending (thereby raising the aggregate price elasticity), while the latter policy would have the opposite effect. Moreover, affecting the price elasticity through changes in the composition of domestic versus foreign expenditures does not provide a firmer foundation for policy. In fact, if—as appears plausible—the foreign price elasticity exceeds the domestic one, an increase in domestic demand would reduce the aggregate elasticity and thereby move the labor demand curve inward!
As for imperfectly competitive interactions among firms, other economists have suggested that oligopolists might behave more competitively in a boom, so that a rise in product demand could shift the labor demand curve outward via its influence on competition.34 However, Rotemberg and Saloner (1986) show that this effect holds only when firms are implicitly colluding oligopolists, and that this induced-competition channel is a weak foundation for demand-management policy.
That leaves the other four channels, which appear to be more promising avenues for transmitting demand-management policies to employment. All four of these channels make the employment impact of demand-management policies depend on their supply-side effects, which therefore ensures a special role for supply-side policies in enhancing the effectiveness of demand management.
Lindbeck and Snower (1994) have shown that, when there is excess capital capacity, demand-management policy can affect the marginal product of labor by influencing the degree of capital utilization. To fix this idea, consider the following sequence of labor market decisions. First, each firm sets its supply of physical capital and determines from the range of its available technologies those that are to become accessible through its capital stock. Next, the nominal wage is determined (for example, through bargaining between the firm and its employees). Then, the firm observes the position of its product demand curve and makes its employment decisions. Under these circumstances, an unanticipated adverse product demand shock could make it unprofitable for a firm to operate at full capacity.35 A subsequent favorable demand shock would induce a firm not only to hire more labor at the existing level of capital services, but also to raise the degree of capital utilization. When economies emerge from recessions in this way, with workers recalled to operate vacant machines and restart idle assembly lines, the capital brought back into use is often highly complementary to labor. Through this channel, expansionary demand-management policy may raise the marginal value product of labor, leading to procyclical movements of the real wage.
An increase in product demand that reduces the real interest rate will thereby also reduce the user cost of capital, increasing the size of the capital stock and shifting the labor demand curve outward, provided that labor and capital are Edge-worth complements in production (that is, the marginal product of labor depends positively on the capital stock). This could occur either through expansionary monetary policy, or, as Greenwald and Stiglitz (1988) indicate, through a decline in the risk premium on investment brought about by the expansion of demand. Naturally, if the rise in demand takes the form of an increase in government spending, the real interest rate may rise (rather than fall), shifting the labor demand curve inward through the mechanism described above. Moreover, even if the real interest rate falls, the labor demand curve still will shift inward when labor and capital are Edgeworth substitutes.
With respect to the number of firms in operation, increases in product demand can (as demonstrated, for example, by Pagano (1990) and Snower (1983b)) induce entry of new firms, which shifts the labor demand curve outward, both directly and indirectly by increasing the degree of product market competition. Specifically, if nominal wages are temporarily rigid, a rise in product demand can reduce the real wage by raising prices, leading to the entry of new firms. Once nominal wages adjust, this entry ceases, but the recently entered firms remain operative. In this way, a temporary nominal wage rigidity can give product demand policy an influence on employment in the longer run (see Lindbeck and Snower (1987c)).
If the increase in government spending takes the form of industrial infrastructure investment, there may obviously be a direct stimulus to the marginal product of labor. In this case, expansionary demand-management policy shifts the labor demand curve outward through its effect on the capital stock.
The policy implications regarding these four channels are potentially of considerable significance: the longer-term influence on employment of product demand policy depends on the availability of a limited number of supply-side channels of transmission. Supply-side policies—such as those that reduce barriers to the entry of new firms,36 or those that augment industrial infrastructure—can help open these supply-side channels and thereby improve the long-term effectiveness of demand management. In the long run, therefore, demand- and supply-side policies are interdependent.
Although it is always difficult to assess the empirical importance of long-term policy effects, the above analysis does offer potentially interesting explanations of some long-run stylized facts about OECD unemployment. First, the relatively low unemployment rates experienced in Europe and the United States during the 1950s and 1960s came at a time of a significant buildup of industrial infrastructure, whereas the higher unemployment since the mid-1970s has been associated with a general rundown of industrial infrastructure in these areas. The analysis suggests that these two developments may bear some relation to one another. Second, the analysis also implies that the greater ease of entry and exit of firms in the United States, compared with European firms, may help in part to explain why the rise in aggregate demand in the aftermath of the recent recessions has had a greater impact in reducing unemployment in the United States than in Europe. Finally, the theory helps explain why the influence of demand management on unemployment appears to be particularly pronounced when this policy is associated with a decline in real interest rates and the availability of ample excess capital capacity.
Low-Wage Subsidies and Payroll Tax Reductions
This set of policies is designed to address the worsening of the relative position of workers at the bottom of the earnings distribution in many OECD countries over the past two decades.37 This deterioration has taken the form of lower relative real wages in the United States (and, to a lesser degree, in the United Kingdom) and higher relative unemployment rates in many continental European countries. Providing subsidies or payroll tax reductions to low-wage workers is meant to raise firms’ demand for these workers, thereby reducing their unemployment rates and raising their take-home pay.38 It has been suggested that these policy measures be financed through a rise in value-added tax (VAT) or the “CO2” tax. Econometric simulations, such as those reported in Dreze, Malinvaud, and others (1994), suggest that the expansionary employment effect of a drop in the payroll tax on low-wage earners may substantially outweigh the contractionary effect of a corresponding rise in the VAT.
As these policies reduce unemployment by reducing employers’ labor costs at the bottom of the wage spectrum, their effectiveness does not appear to be very sensitive to the precise underlying cause of the unemployment (in contrast to profit-sharing subsidies). Regardless of whether unemployment is generated by union pressures, efficiency wage considerations, or insider-outsider conflict, a drop in labor costs is bound to raise employment, as it permits firms to substitute labor for capital and enables them to reduce product prices and thereby create more demand.
However, three major factors limit the effectiveness of these policies: (i) “deadweight” (subsidies or tax reductions received by workers who would have become employed anyway); (ii) “displacement” (through which incumbent employees are displaced by the subsidized new recruits); and (iii) “substitution” (through which firms that benefit from these policies drive out of business firms that do not benefit). Clearly, the more closely the subsidies and the payroll tax reductions are targeted at the low-wage workers, the smaller are the effects of deadweight and substitution, and the larger that of displacement.
Another potential drawback of these policies is that, by raising the take-home pay of unskilled workers relative to skilled workers, they reduce the returns to training. Insofar as labor and capital are complementary in production, the resulting fall in human capital acquisition may also lead to a fall in physical capital formation. For this reason, it appears desirable that these policies be supplemented by subsidies to education and training. This additional element, however, would substantially increase the cost of the intervention. Another drawback is that these policies may encourage the excessive creation of unsatisfying, dead-end jobs providing little potential for advancement. In that event, the unemployment trap would be replaced by the “trap of the working poor.” But even so, workers would experience a rise in their living standards: as the take-up is voluntary, workers and firms would avail themselves of these policy measures only if it were to their advantage.
The case for implementing recruitment subsidies is similar to that for implementing low-wage subsidies and payroll tax reductions: they bring down labor costs and thereby promote employment and reduce unemployment.39 In fact, they are better targeted because they are granted only to new recruits.
Once again, the effects of deadweight, displacement, and substitution limit the employment effect of recruitment subsidies. Obviously, the deadweight effect is generally smaller for recruitment subsidies than for low-wage subsidies or payroll tax reductions, but the displacement and substitution effects are likely to be larger. In any event, the aggregate employment impact of recruitment subsidies is invariably less than the number of jobs subsidized. Beyond that, their effectiveness is likely to be further reduced by the ways in which they are financed. Employer-based taxes will directly discourage employment, and income taxes will reduce product demand and thus indirectly discourage employment. In either case, however, the positive effect of the recruitment subsidies on employment will generally outweigh the negative effect of the taxes.
It is sometimes alleged that another deficiency of recruitment subsidies—one that is shared by low-wage subsidies and payroll tax reductions—is that they distort firms’ decisions concerning factor composition, for example, by encouraging labor at the expense of capital. This matter is quite unlikely to have macroeconomic significance. The inefficiencies resulting from a distorted labor-capital mix are generally insignificant in comparison with the inefficiencies associated with long-term unemployment. Besides, as the efficiency wage, insider-outsider, and union theories suggest, free market activity may often be associated with market failures that give rise to excessive wages and deficient unemployment. In this context, recruitment subsidies may correct for existing distortions rather than create distortions themselves.
“Benefit transfers” involve giving long-term unemployed the opportunity to use part of their unemployment benefits to provide vouchers for firms that hire them (see Snower (1995a)). The longer a person is unemployed, the larger is the voucher. Larger vouchers are also granted to firms that use them entirely on training. Workers provide vouchers on a regular basis to the firms that hire them, and the amount of the voucher gradually falls during the employment period. In this way, benefit transfers are a combination of several different structural policies: the vouchers are equivalent to a special type of recruitment subsidy; the voucher supplements for training are a special type of training subsidy; and the transfer of unemployment benefits amounts to a reform of the unemployment benefit system.
The reasons for using benefit transfers are various. First, they permit people to transfer funds out of a system that discourages employment in order to give firms an incentive to create employment. Second, benefit transfers extend the choice sets of workers and firms. Workers offer the vouchers to potential employers when their expected wage offers are sufficiently high; the employers accept the vouchers when the resulting labor costs are sufficiently low. Thus, benefit transfers are used only when both parties are made better off. Third, benefit transfer schemes are costless to the government because the vouchers are financed through the forgone unemployment benefits. Fourth, benefit transfers are not inflationary, as the long-term unemployed have no significant effect on wage inflation, and as the vouchers reduce labor costs and thereby exert downward pressure on prices. Fifth, benefit transfer schemes function as automatic stabilizers, as a fall in unemployment reduces the amount spent on unemployment benefits, which, in turn, reduces the funds available for the employment vouchers. Sixth, by providing generous vouchers to firms that use them for training, such schemes give these firms an incentive to maximize the productivity-enhancing effect of this training. Finally, the schemes could help to overcome regional unemployment problems. Regions with high unemployment would command a disproportionate share of training subsidies, which might give firms an incentive to relocate there and provide the unemployed with the requisite skills, even in the face of existing barriers to mobility.
In view of these advantages, benefit transfers appear desirable as a first line of attack against long-term unemployment. Once the employment-creating potential of unemployment benefits has been exploited in this way, further measures may well be necessary to bring European unemployment down to socially acceptable levels.
Institutional policies, as their name suggests, aim to change labor market institutions to reduce unemployment. These policies come in many guises, of which only the most prominent will be considered here.
Policies to Reduce the Power of Labor Unions
Policies to reduce the power of labor unions include, interalia, restrictions on secondary picketing, laws prohibiting closed-shop agreements, and regulations restricting the coverage of union wage agreements. These policies may be analyzed straightforwardly through the theory of labor unions. In the traditional variants of this theory,40 all union members are assumed to have identical preferences and an equal share in the available work. The union then represents the interest of its members by exerting its monopoly power in wage setting, much like sellers of goods or services exert their monopoly power in price setting. The resulting wages will be higher—and employment lower—than would otherwise be the case. If all workers in the economy belong to unions, aggregate employment will be less than it would be under full employment. The difference is unemployment (or underemployment).
More recent theories recognize that unions take greater account of the interests of their employed members than of the unemployed and that the employed workers have greater access to work than do the unemployed. The unemployment arising in this setting may be voluntary from the vantage point of the employed union members, but it is generally involuntary from the vantage point of the unemployed, as the latter could be made better off by a wage reduction associated with a rise in employment.
The main theoretical weakness of this theory lies not in what it explains, but in what it does not explain, namely, why the unemployed do not leave unions that do not represent their interests and start new unions making lower wage claims. Nor does this theory identify the basis of the unions’ “clout.” As union coverage in most market economies is far below 100 percent, it is not clear why employers do not simply throw out high-wage union members and hire low-wage nonmembers instead.41
On the empirical front, there is some evidence of an inverse relation between intercountry differences in unemployment rates, on the one hand, and intercountry differences in indexes of union power and union coverage, on the other, over the postwar period. Yet the union theories have not performed well in predicting movements of unemployment over the past decade. In the first part of the 1980s, for example, union membership in the United Kingdom and several Other European countries fell while unemployment rose. For this reason, it is premature to say that unemployment policies designed to reduce union power are on a firm predictive foundation.
Reforming the Wage-Bargaining System
In recent years, there has been a growing call to strengthen firm-level and national-level bargaining at the expense of bargaining at the sectoral level.42 This policy strategy is based on the analysis of Calmfors and Driffill (1988), who explore how the economic efficiency of wage bargaining depends on the number of independent agents engaged in bargaining. They argue that, when there is a high degree of centralization in bargaining-—with few unions confronting few employers’ confederations, such as in Austria and Sweden—the negotiating partners internalize most of the effects of their claims; in particular, the unions take account of the price increases associated with their wage claims, and the employers take account of the wage increases associated with their employment and pricing decisions. The resulting wage-employment outcome is therefore reasonably efficient. However, when there are large numbers of negotiating workers and firms, with each occupying a small portion of the market, the resulting activity is efficient for the standard competitive reasons. The United States approximates this setup. Calmfors and Driffill claim that it is only in the intermediate range, where the independent negotiators are sufficiently few in number to have market power but sufficiently numerous to ignore the external effects of their decisions, that gross inefficiencies arise. Calmfors and Driffill adduce some empirical evidence in favor of this thesis, and Layard, Nickell, and Jackman (1991, p. 55) provide cross-section evidence that the unemployment rates in 20 OECD countries tend to be inversely related to the degree of union and employer coordination in each country.
On this account, it has been argued, wage-bargaining systems need to be either highly centralized or highly decentralized.43 Policies that reduce the power of labor unions, reduce labor turnover costs, and promote international trade are all likely to strengthen decentralized, firm-level bargaining. Government sponsorship of “social pacts”—whereby unions accept targets for nominal wage growth (based on productivity growth and price inflation), firms accept targets for price increases (based on wage inflation), the central bank sets the growth of the money supply with a view to noninflationary growth, and the fiscal authority aims to control unemployment—encourages centralized, national-level bargaining. As a practical matter, however, wage-bargaining systems are very difficult to reform; thus, this should be seen more as a long-term structural policy desideratum than as a shortterm policy instrument.
Reforming the Unemployment Benefit System
The main deficiency of all unemployment benefit systems is that, in helping to cushion the blow of unemployment, they worsen the underlying problem in two ways. First, unemployment benefits discourage job search because, when an unemployed person finds a job, the unemployment benefits are withdrawn and taxes are imposed. Second, unemployment benefits put upward pressure on wages by improving incumbent workers’ negotiating positions. The first effect lies in the domain of search and matching theory, the second in the province of bargaining theory. Together, these effects make unemployment benefit systems inherently inefficient and inequitable.
In reforming unemployment benefit systems, it is important to distinguish carefully between the equity and efficiency objectives of these systems. The equity goal is simply to redistribute income from the rich to the poor. The efficiency goal is to respond to market failures in the provision of unemployment insurance.44 However, unemployment benefits are generally a very poor tool to accomplish these objectives.
With regard to equity, it is worth keeping in mind that, for most poor people, employment is the best—and often the only—way to overcome poverty. Thus, it is particularly unfortunate that, by making the distribution of employment opportunities more unequal, unemployment benefits discourage employment. Clearly, a more effective way to redistribute income from rich to poor is to use income as the criterion of redistribution; the employment criterion is obviously a blunt instrument for this purpose because some employed are poor while some unemployed are well-off.
With regard to efficiency, the gains from provision of unemployment insurance must be set against the efficiency losses that arise when unemployment benefits discourage employment and encourage unemployment. It is by no means a foregone conclusion that the efficiency gains will exceed the associated losses. In any case, the unemployment benefit schemes that predominate in Europe—characterized by either flat-rate components or ceilings on benefits that depend on past wages—have much less in common with optimal unemployment insurance schedules than with standard redistributive schemes. In short, the unemployment benefits encountered in practice are not designed to yield major efficiency gains by correcting for failures in the unemployment insurance market.
These deficiencies, however, are not the only problems arising from unemployment benefits. The efficiency wage, labor union, and insider-outsider theories identify failures associated with free market activity that tend to yield excessively high wages and excessively low employment. Unemployment benefit systems exacerbate these market failures by further driving up wages and discouraging employment. Furthermore, these market failures are perpetuated through various dynamic effects. As noted above, the longer people are unemployed, the more their skills depreciate and become obsolete, the more discouraged and ineffective they become in the job search process, and the more wary firms become of hiring them. When governments reward unemployment through unemployment benefits and penalize employment through income taxes, they amplify these dynamic effects by discouraging unemployed from competing for jobs and becoming “enfranchised” in the wage determination process. As a result, unemployment becomes less effective in moderating wages or raising firms’ returns from searching for new recruits. In this way, unemployment benefit systems make unemployment more persistent and put the long-term unemployed at a greater disadvantage in competing for jobs.
For all these reasons, unemployment benefit reform has become a topic of growing policy interest throughout Europe. But while it is relatively easy to recognize the need for reform, it is very difficult to agree on its content. The critical question is how to provide a safety net for the disadvantaged and the unfortunate without dramatically reducing people’s incentives to fend for themselves and thereby creating more disadvantaged and unfortunate.
A growing number of European economists argue that unemployment benefits should be generous but made available for only a limited period of time (see, for example, Layard, Nickell, and Jackman (1991)). The generosity is allegedly required to give people the opportunity to make judicious job matches, which credit constraints might otherwise keep them from doing. Limited benefit duration, it is claimed, is necessary to induce people to find work quickly, before they become discouraged and stigmatized, and lose their skills. This advice sounds eminently sensible, but little attempt has been made thus far to explore whether the theory behind it captures empirically important determinants of unemployment. It seems doubtful that workers’ credit constraints are an important aspect of the European unemployment problem. If they were, the problem would be that unemployment duration was too short, resulting in overemployment. This, it appears, is the least of Europe’s worries.
Beyond that, the prescription to shorten benefit duration characteristically becomes vague on the subject of how to deal with people who remain jobless after their unemployment benefits have expired. Some economists recommend that these people be given training, while others put more emphasis on job counseling. However, that still leaves the question of how to treat those left unemployed even after completion of the training and counseling. Many European economists still hold the popular European opinion that the social safety net—in the form of income support and a range of welfare state benefits—is required to keep these hapless individuals from destitution. However, a nominally short benefit duration may cease to give unemployed people an effective incentive to find jobs promptly.
This is, in fact, the problem that the current, unreformed European benefit systems face. Many European countries—such as Germany, France, Greece, Ireland, and the Netherlands—grant some form of unemployment insurance of limited duration, followed by unemployment assistance that is frequently unlimited. It is hard to see how the disincentive effects generated by these systems could be overcome simply by shortening the time span for unemployment insurance and inserting a period of training and counseling prior to the receipt of open-ended unemployment assistance.
Overall, it is safe to say that unemployment benefit reform should be guided by the objective to overcome its two biggest deficiencies, namely, the disincentive effects and the imperfections in targeting the poor. It is arguable that both deficiencies could be mitigated simply by replacing unemployment benefit systems by a conditional negative income tax program,45 whereby receipt of negative income taxes is made to depend on the applicants’ ability to pass stringent tests on their willingness and readiness to work.46
Work Sharing and Early Retirement
Work sharing and early retirement have begun to look attractive to an increasing number of European policymakers, particularly in Germany. It is based on the view that, as there is a fixed amount of work to be done in an economy in any given period of time, it is the job of the policymakers to decide how this work is to be distributed across the available workforce. If it is currently distributed unequally, with most people in the workforce working full time and some remaining unemployed for prolonged periods, work sharing and early retirement could spread the job opportunities more equitably.
But to call this a “theory” is an overstatement. Most economists would rather call it the “lump-of-labor fallacy,” as it is well understood that the amount of work to be done in an economy is not a fixed number of hours that is beyond the influence of policymakers.47 Keynesian theory drives this point home particularly forcefully: the more people are employed, the more they earn, the greater their purchasing power, the more they spend, and—completing the cycle—the more people firms will seek to employ.
In addition to their nonexistent theoretical foundation, job sharing and early retirement schemes suffer from a number of serious problems. First, they tend in practice to increase nonwage labor costs, particularly those associated with hiring, screening, training, and administration. Thus, these schemes may be expected to discourage employment and create more unemployment. Second, insofar as they are successful in reducing the pain from unemployment by distributing it among more people, they lessen the political pressure on governments to address the unemployment problem through more promising means. Third, in reducing the number of unemployed people competing for jobs, job sharing and early retirement schemes may well drive up wages and stimulate price inflation. These results may, in turn, induce governments to implement restrictive macroeconomic policies, which would raise unemployment, possibly creating a further perceived need to redistribute job opportunities through yet more work sharing and early retirement. The main advantage of work sharing and early retirement schemes is that they may enfranchise a larger number of people in the wage determination process and thereby moderate the insiders’ wage demands. It appears unlikely, however, that this advantage would dominate the disadvantages discussed above.
Policies Centered on Labor Turnover Costs
Policies to reduce unemployment by mitigating the harmful effects of labor turnover costs are as varied as the turnover costs themselves. Some involve dismantling job security legislation (for example, by passing laws reducing statutory severance pay or simplifying mandated firing procedures); others reduce the ability of incumbent workers to exploit existing labor turnover costs in order to boost their wages (by imposing, for example, legal restrictions on strikes and picketing); yet others help the unemployed surmount the obstacles created by turnover costs (by implementing training subsidies, recruitment subsidies, profit-sharing schemes, policies to reduce the barriers to the entry of new firms, and the reform of wage-bargaining systems). This section focuses attention on the first two groups of policies; policies in the third group have been discussed in the preceding section.
What the first two groups of policies have in common is that they reduce the market power of the “insiders” (incumbent employees whose jobs are protected by significant labor turnover costs) and thereby strengthen the position of the “outsiders” (who are either unemployed or have jobs that are not protected in this way). In the process, insiders become less insulated from the forces of labor demand and supply, and firms find it easier to hire and fire employees. The up-shot is that insiders’ wages face downward pressure, as insiders now face greater competition from outsiders, and employment becomes more responsive to variations in revenue and cost conditions. The first effect stimulates employment48 because insiders’ wages fall and firms raise their demand for new recruits, who eventually turn into insiders. The second effect reduces the degree of employment and unemployment persistence.
This policy approach lies in the domain of the insider-outsider theory (see, for example, Lindbeck and Snower (1986) and (1988a)). According to this theory, labor turnover costs, falling at least in part on firms, give market power to the insiders, who know that their employers would find it costly to replace them. The insiders are assumed to use this power to pursue their own interests in the wage-setting process. Although the resulting insider wages are higher than they otherwise would be, the labor turnover costs discourage firms from firing the insiders. Of course, firms are also discouraged from hiring new entrants by the excessive wages.
Some of the labor turnover costs (such as training costs) are an intrinsic part of the production process; others (such as severance payments) are primarily associated with rent-seeking activities. The rent-related turnover costs give the insiders preferential conditions of employment over the outsiders. Unemployment can then arise on account of the outsiders’ inferior employment opportunities. In this context, policies that reduce labor turnover costs or check insiders’ ability to exploit them in wage setting will generally lead to a reduction in unemployment.
The insider-outsider theory is able to account for a variety of empirical regularities in unemployment behavior. The relatively high labor turnover costs in Europe—both in their own right and through their influence on insiders’ wages—play a role in making European unemployment more persistent (serially correlated) than U.S. unemployment. Because high labor turnover costs make firms reluctant both to hire and to fire employees, they raise the duration of unemployment. In this way, Europe’s high labor turnover costs can lead to its high unemployment durations and low unemployment variability, in comparison with the United States. Furthermore, because labor turnover costs raise insiders’ job retention rates relative to outsiders’ job acquisition rates, they imply that unemployment falls more heavily on population groups whose work patterns are relatively unstable (with high entry and exit rates in the job market), such as young people.
Insofar as many of the full-time unskilled jobs in the traditional industrial sectors are associated with significant labor turnover costs, the insider-outsider theory also explains why wages in these sectors have not fallen with falling demand. It also helps explain why much service sector and temporary employment—associated with relatively low turnover costs—has been buoyant in comparison with industrial employment in OECD countries.
When business cycles are short-lived and mild, most European countries—facing higher labor turnover costs—may be expected to do relatively little hiring or firing, hoarding labor in the slumps and bringing it back into use in the booms. But in the face of deep, prolonged recessions, these countries will stop hoarding and start firing labor. In the subsequent recovery, firms will be comparatively slow to rehire, fearing that they may incur further firing costs should the recovery not materialize. Investment in labor-saving capital equipment may then take the place of new employment. This helps explain why unemployment rates in Europe, which were significantly lower than in the United States in the 1950s and 1960s (when business cycles were short-lived and mild), have been significantly higher since the mid-1970s; why U.S. unemployment has been more variable than European unemployment; and why production and employment move together to a greater degree in the United States than in Europe.
Under profit-sharing contracts, a part of workers’ remuneration is paid as a fraction of the profits earned by their firms (see Weitzman (1983) and (1984)). For a given level of remuneration, it is clear that a firm’s marginal cost of employment is lower under profit sharing than under a fixed wage because (under diminishing returns to labor) the profit share declines as employment rises, whereas a fixed wage, by definition, does not. Consequently, it is alleged, profit-sharing contracts lead to lower unemployment than do wage contracts. Weitzman has suggested that, in a world where wages seldom involve profit sharing, government subsidies for profit sharing are called for because firms have insufficient incentives to offer profit-sharing contracts.
The claim that profit-sharing contracts reduce unemployment is less general than it may appear at first sight. The effectiveness of profit sharing depends crucially on what is generating the unemployment. If, for instance, the unemployment is an efficiency wage phenomenon, the switch from wage contracts to profit-sharing ones will do little, if anything, to reduce unemployment, as workers’ incentives to shirk and quit depend on the total amount of remuneration, not on how this amount is divided between wages and profit shares. Similarly, firms’ ability to attract workers of relatively high productivity would not be affected by a switch from wage to profit-sharing contracts.
However, if the unemployment is predominantly generated by insider-outsider considerations, profit sharing may have an effective role to play. In the insider-outsider theory, the outsiders are unable to “bribe” insiders to forgo the rent-seeking activities that keep the outsiders from getting jobs. The insiders may, for example, boost their wages and protect themselves from competition with outsiders by refusing to cooperate with them in the process of production, thereby creating an insider-outsider productivity differential; or they may harass outsiders who offer to work for less than the prevailing wages and thereby make the available jobs more disagreeable for those outsiders. Alternatively, the insiders may be involved in determining the wages of new entrants and may use their market power to drive entrant wages up, thereby discouraging the employment of entrants, which would drive down the marginal product of insiders.
In this context, profit-sharing contracts may be construed as a device that permits the outsiders to bribe the insiders to stop these activities, so that everyone-—the insiders, the outsiders, and their employers—can be made better off. In particular, if insiders were given a bonus for consenting to profit-sharing contracts for new entrants, the firm’s marginal cost of hiring new entrants would fall, the entrants would receive more than they did when they were unemployed, and the firm’s profits would rise. In the process, of course, unemployment would fall.
Although profit-sharing schemes are promising in this context, it is important to be aware of some potential difficulties. First, it may be impossible to induce the insiders to consent because their rent-seeking activities—like their harassment activities—may not be objectively monitorable. Second, to make profit sharing operational may require implementing costly monitoring procedures that enable workers to gain access to profit information.49 Third, the extra profit generated through the introduction of profit sharing may be insufficient to compensate the insiders for their loss of market power resulting from the inflow of new entrants. Fourth, the extra profit generated may be insufficient to pay the premium that the new entrants would require to induce them to bear the income risk associated with profit sharing. Finally, the insiders may refuse to be bribed because that would create a two-tier remuneration system that would give firms an incentive to lay off the insiders and retain the entrants, once the latter had been fully trained.
It has become a platitude to say that every sensible piece of economic policy advice rests on a reasoned analysis of the underlying policy problem, and that every reasoned analysis is based on a theory of how the economy functions. Politicians may believe that their policy proposals rest simply on “common sense”; but if there is any sense underlying this common sense, it exists in the form of a coherent, self-contained theory. As Keynes (1936) put it:
The ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than is commonly understood. Indeed, the world is ruled by little else. Practical men, who believe themselves to be quite exempt from any intellectual influences, are usually slaves of some defunct economist.
Because this is obvious, it is surprising that so little is done to explore the predictive power of a theory before using it as a basis for policy formulation. This survey is a tentative first step toward evaluating unemployment policies in this light.
It goes without saying that such an evaluation alone is not sufficient for the design of unemployment policies, but, as has been demonstrated, it can provide a variety of useful insights about where promising policy approaches are to be found. For example, this paper has examined how differences in labor turnover costs across sectors (such as services versus manufacturing) and regions (such as Europe versus the United States) may help account for differences in the level, variability, duration, persistence, and distribution of unemployment. This analysis suggests that policies to reduce the harmful effects of these labor turnover costs—such as reductions in statutory severance pay, training and recruitment subsidies, benefit transfers, and policies to lower the barriers to the entry of new firms—may have a significant role to play in combating unemployment. These and the variety of other insights adduced above show why it is important to evaluate unemployment policies through the predictions of the underlying theories.
Abraham, Katharine G., and Lawrence F.Katz, “Cyclical Unemployment: Sectoral Shifts or Aggregate Disturbances?”Journal of Political Economy, Vol. 94 (June1986), pp. 507–22.
Akerlof, George A., and Janet L.Yellen, “A Near-Rational Model of the Business Cycle, with Wage and Price Inertia,”Quarterly Journal of Economics, Vol. 100 (Supplement, 1985), pp. 823–38.10.1093/qje/100.Supplement.823)| false
Blanchard, Olivier Jean, “Price Asynchronizalion and Price Level Inertia,” in Inflation, Debt, and Indexation, ed. by RudigerDombuseh and Mario H.Simonsen (Cambridge, Massachusetts: MIT Press, 1983).
Blanchard, Olivier Jean, “Price Asynchronizalion and Price Level Inertia,” in Inflation, Debt, and Indexation, ed. by RudigerDombuseh and Mario H.Simonsen (Cambridge, Massachusetts: MIT Press, 1983).)| false
Bover, Olympia, JohnMucllbaucr, and AnthonyMurphy, “Housing, Wages, and U.K. Labour Markets,”Oxford Bulletin of Economics and Statistics, Vol. 51 (March1989), pp. 97–136.10.1111/j.1468-0084.1989.mp51002001.x)| false
Chatterjee, Satyajit, and Russell W.Cooper, “Muliiplicity of Equilibria and Fluctuations in Dynamic Imperfectly Competitive Economies,”American Economic Review, Vol, 79 (Papers and Proceedings, May1989), pp. 353–57.
Chatterjee, Satyajit, and Russell W.Cooper, “Muliiplicity of Equilibria and Fluctuations in Dynamic Imperfectly Competitive Economies,”American Economic Review, Vol, 79 (Papers and Proceedings, May1989), pp. 353–57.)| false
Dreze, Jacques, EdmondMalinvaud, and others, “Growth and Employment: The Scope of an European Initiative,”European Economic Review, Vol, 38 (April1994), pp. 489–504.10.1016/0014-2921(94)90086-8)| false
King, Robert G., Charles I.Plosser, and SergioRebelo (1988a), “Production, Growth and Business Cycles I: The Basic Neoclassical Model,”Journal of Monetary Economics, Vol. 21 (March/May1988), pp. 195–232.
King, Robert G., Charles I.Plosser, and SergioRebelo (1988a), “Production, Growth and Business Cycles I: The Basic Neoclassical Model,”Journal of Monetary Economics, Vol. 21 (March/May1988), pp. 195–232.10.1016/0304-3932(88)90030-X)| false
King, Robert G., Charles I.Plosser, and SergioRebelo (1988b), “Production, Growth and Business Cycles II: New Directions,”Journal of Monetary Economics, Vol.21 (March/May1988), pp. 309–41.10.1016/0304-3932(88)90034-7)| false
LindbeckAssar, and Dennis J.Snower, (1987a), “Union Activity, Unemployment Persistence, and Wage-Employment Ratchets,”European Economic Review, Vol. 31 (February/March1987), pp. 157–67.10.1016/0014-2921(87)90027-4)| false
LindbeckAssar, and Dennis J.Snower, (1987c), “Transmission Mechanisms from the Product to the Labor Market,”Institute for International Economic Studies Seminar Paper No. 403 (Stockholm, Sweden: University of Stockholm, December 1987).
LindbeckAssar, and Dennis J.Snower, (1987c), “Transmission Mechanisms from the Product to the Labor Market,” Institute for International Economic Studies Seminar Paper No. 403 (Stockholm, Sweden: University of Stockholm, December 1987).)| false
Mortensen, Dale T., “Job Search and Labor Market Analysis,” in Handbook of Labor Economics, Vol. 2, ed. by OrleyAshenfelterandP.RichardG.Layard (Amsterdam and New York: North-Holland, 1986).10.1016/S1573-4463(86)02005-9)| false
PhelpsEdmund S.“Wage Subsidy Programs: Alternative Designs.” in Unemployment Policy, ed. by Guillcrmo de laDehesa and Dennis J.Snower (Cambridge: Cambridge University Press, 1995, forthcoming).
PhelpsEdmund S. “Wage Subsidy Programs: Alternative Designs.” in Unemployment Policy, ed. by Guillcrmo de laDehesa and Dennis J.Snower (Cambridge: Cambridge University Press, 1995, forthcoming).)| false
Snower, Dennis J. (1983a), “Search, Flexible Wages and Involuntary Unemployment,” Birkbeck College Discussion Paper in Economics No. 132 (London: Birkbeck College, University of London, January1983).
Snower, Dennis J. (1983a), “Search, Flexible Wages and Involuntary Unemployment,” Birkbeck College Discussion Paper in Economics No. 132 (London: Birkbeck College, University of London, January1983).)| false
The author wishes to thank David T. Coe and Bert Hickman for their perceptive comments and is particularly indebted to Robert P. Ford for his insightful suggestions. The paper was written while the author was a visiting scholar at the Research Department of the International Monetary Fund, and the research is part of the Labour Market Imperfections Program, funded by the U.K. Employment Department and organized by the Centre for Economic Policy Research.
However, these failures need not call for government intervention at all, as it may not be feasible to correct some market failures through government unemployment policy. Even when it is possible to do so, the gains from correcting the market failures may fall short of the losses from the “government failures,” namely, policy-induced inefficiencies.
See, for example, Lucas (1972) and (1975). Some economists use the term “natural rate of unemployment” more broadly, letting it stand for any short-term equilibrium unemployment rate, regardless of whether the labor market clears (for example, Phelps (1970) and (1994)) and regardless of the underlying institutional structure (for example, Friedman (1968)). In that view, the natural rate clearly rests on much more than tastes, technologies, and endowments: it could also depend on the existence of credit constraints, the degree of competition in labor and product markets, the nature of wage-bargaining institutions, the level of labor turnover costs, and the size of the incumbent workforces, just to give a few examples. Then, however, the natural rate theory becomes so all-inclusive that it can no longer be distinguished from labor union, insider-outsider, efficiency wage, and other theories.
Taking the wider view of the natural rate theory, as summarized in the previous footnote, it is worth observing that the degree of competition and the economic institutions governing behavior in the labor, product, credit, and international markets are generally not subject to cyclical fluctuations either. Thus, cyclical fluctuations in unemployment remain to be explained by fluctuations in expectational errors.
However, measures of the elasticity or labor supply can be raised substantially if one assumes that the choice between work and inactivity is usually a discrete one. Then, to account for observed cyclical swings, the theory requires that decisions about whether to participate in the labor force be very sensitive to variations in real wages and real interest rates.
The technological shocks in the real business cycle models are measured by Solow residuals, for example, the differences between the growth rate of output and a weighted average of the growth rates of factor inputs. But technological regress is not the only conceivable explanation of the negative Solow residuals; labor and capital hoarding is another.
See, for example, Carlton (1986), who finds significant price rigidities in manufacturing. Gordon (1990, p. 1150) has argued that, in the context of a complex input-output system, complete indexation may be difficult, owing to “the informational problem of trying to anticipate the effect of a currently perceived nominal demand change on the weighted average costs”; but it is hard to see why some (albeit imperfect) indexing should not be belter than none.
In practice, some rules for setting wages and prices appear to involve both time and state dependence, such as the provision in wage contracts to renegotiate at specified intervals but only under specified conditions—for example, if the inflation rate exceeds a certain magnitude. It has been suggested that, if the major cost is that of learning, a time-dependent rule is desirable; whereas, if the major cost is a menu cost, a state-dependent rule will be chosen. However, menu and learning costs are notoriously difficult to measure.
The European Community Commission laid stress on these measures in combating European unemployment. For example, the Council Resolution of May 29, 1990 recommended that counseling interviews be made available to all long-term unemployed. There is also wide recognition that these measures have a chance of being particularly effective only if they are combined with other active labor market policies, such as training programs.
This is, of course, not the only conceivable explanation of unemployment persistence. Other, comparably important, causes are employment adjustment costs, wage-price staggering effects, insider membership effects, and adjustment costs of labor force participation.
In Weiss (1980), a higher wage offer encourages workers of high skill, who were previously self-employed, to join the firm. In Shapiro and Stiglitz (1984), the firm randomly samples workers’ effort and fires those who shirk; thus a higher wage offer raises effort by raising the expected penalty for shirking. In Snower (1983a), a higher wage offer discourages workers from searching on the job and thereby promotes productivity. In Akerlof (1982), workers agree to work more than is specified in their contract, and firms, in return, pay more than the minimum that would be necessary to attract them.
Sufficiently steep intertemporal wage scales or sufficiently large “exit fees” for employees may help firms stimulate productivity and discourage quitting, thereby giving firms less of an incentive to raise wages above market-clearing levels. It can be shown that, if these devices are practicable, they can eliminate some—but not all—sources of unemployment arising from the imperfect information analyzed by the efficiency wage theory.
In general, training programs, whether in the public or private sector, may be divided into two broad categories: vocational training and “employability” training. In the latter category, a limited number of basic skills are developed that enable people to adjust to a worker environment and adapt to the requirements of semi-skilled jobs. In some countries, Germany in particular, vocational training is integrated within a formal system of basic education.
There are a variety of market failures in training provision that apply to all classes of workers. See, for example, Becker, Murphy, and Tamura (1990) and Booth and Snower (1995, forthcoming). Some of these market failures fall with particular severity on the unemployed and thus make the case for using training subsidies as an instrument to combat unemployment.
Formally, the labor demand curve is given by F · (1 − m) · hn = w, where the left-hand term is the real marginal revenue product of labor and w is the real wage. Specifically, F is the number of firms, hn = hn (n, k) is the marginal product of labor (where n and k are each firm’s use of labor and capital, respectively), and m is Lerner’s index of monopoly power, defined as c/·ηF (where c is the conjectural variation coefficient and η is the price elasticity of product demand). Thus, channels (i) and (ii) work, through the degree of monopoly power, channels (iii) and (iv) work through the effect of the capital stock on the marginal product of labor, channel (v) deals with shifts of the labor demand curve due to changes in the number of firms (which also affects the degree of monopoly power), and channel (vi) is concerned with the direct effect of product demand on the marginal product of labor. Lindbeck and Snower (1994) provide a formal analysis of all these channels of transmission.
These policies involve measures to dismantle government regulations restricting the creation of new firms; reform the system of profit, income, capital gains, and wealth taxes to put new firms at less of a disadvantage in comparison with established firms; increase competition among financial institutions so as to reduce credit constraints on new firms; and reduce the coverage of collective bargaining wage agreements so as to permit new firms to hire new recruits on competitive terms.
The effectiveness of these policies clearly depends on the elasticity of labor demand. The greater the elasticity, the more the unemployment rates of the low-wage workers will fall, and the less their lake-home pay will rise.
This question is answered by the insider-outsider theory, discussed below. But if the answer of the insider-outsider theory is accepted, namely, that it is labor turnover costs that prevent firms from replacing union members by nonmembers, the traditional union theories must undergo substantial revision; see, for example, Lindbeck and Snower (1987a) and (1987b).
This issue can be addressed through labor-market-bargaining theory, which deals with the question of how employers and employees split the economic rent from employment activity. There are two broad approaches. In one approach, employers and employees bargain over wages, and, once the wages have been set, the employers make the employment decisions unilaterally. In the other approach, employers and employees bargain over wages and employment simultaneously. The former are called “right-to-manage” models (because the firms make the employment decisions by themselves), and the latter are usually known as “efficient bargaining” models. There are also models that straddle these two extremes (see, for example, Manning (1987)). It can be shown that the bargaining outcome from the right-to-man age models is inefficient, in the sense that it is possible to find wage-employment combinations that make one party to the negotiations better off without making the other party worse off. This is a common feature of institutional setups in which the price and quantity decisions are made by different agents. This inefficiency, of course, does not arise in the latter models, which are therefore called “efficient bargaining” models.
However, in a more recent article, Calmfors (1993) distances himself somewhat from this simple policy conclusion. He acknowledges that centralization is a multifaceted feature of bargaining systems and that labor market performance is likely to respond quite differently to changes in the degree of centralization across occupations, sectors, unions, employers’ confederations, and geographic regions. He also notes that the degree of centralization is likely to be particularly significant for labor market performance—only in the nontradable sectors, where foreign competition is weak.
Under free market conditions, the private sector generally has deficient incentives to provide unemployment insurance, owing to moral hazard and adverse selection problems: providing unemployment insurance increases the chances of being unemployed, and those with a greater chance of being unemployed will tend to purchase more insurance. Also, under free market conditions, credit constraints prevent workers from purchasing the optimal amount of insurance.-
Handicapped people and those who are likely to be more productive in the household sector than in the labor market (such as single mothers with several infants) would be exempted from this condition; see Snower (1994c).
Of course, economies may generate something like a “lump of labor” over the very short run, that is, over a time span short enough to preclude readjustments in the size of firms’ workforces. But this time span is of little interest for the design of unemployment policy.
Of course, a reduction in labor turnover costs also has a direct effect on employment. This effect could be either positive (when a reduction in hiring costs stimulates hiring) or negative (when a reduction in firing costs leads to more firing). See, for example, Bentolila and Bertola (1990).