Read an Excerpt
Models of Unemployment, Layoffs, and Wage Dispersion
By Andrew Weiss
PRINCETON UNIVERSITY PRESSCopyright © 1990 Princeton University Press
All rights reserved.
Efficiency Wages Models of Unemployment, Layoffs and Wage Dispersion
Department of Economics, Boston University, MA, USA
1. INTRODUCTION AND OVERVIEW
In discussing models of long-term unemployment we are motivated by the question "If a firm faces an excess supply of labor why doesn't it cut its wages? "This is a somewhat different question than the usual macroeconomic question concerning insufficient aggregate demand. Here we are simply asking, regardless of the level of aggregate demand, why don't markets clear at the micro-economic level? Firms are presumably maximizing their profits. In the conventional micro-economic analysis whatever the level of aggregate demand firms would offer the lowest wage that would satisfy their labor demand. Consequently, as Modigliani pointed out, downward wage rigidity is necessary for an unemployment equilibrium.
Several reasons have been offered for downward wage rigidity. Azariadis , Gordon , and Baily  presented implicit contract models in which they argued that risk aversion by workers would induce firms to offer a fixed wage contract. However, in these models the level of employment is above the level achieved in a Walrasian equilibrium. This is because risk neutral firms would wish to insure workers not just against wage fluctuations, but also against the income fluctuations associated with unemployment. If, as is commonly assumed in the implicit contract literature, a firm cannot commit itself to future wage payments to workers that it no longer employs, then the only way firms can lower the variance of their workers earnings is by employing them in states in which the value of their output is less than their private value of leisure (including the unemployment benefits they receive). On the other hand, if workers could be paid when not employed, the equilibrium contracts would leave workers indifferent between being employed and unemployed. These issues are discussed by Russell Cooper  in his survey of the literature on implicit contracts and will not concern us any further.
The other leading explanations for downward wage rigidity are based on the effect of wage contracts on the quality of labor. In these efficiency wage models, wages affect the quality of labor either through their effect on the distribution of workers hired—the adverse selection effect; or through their effect on the performance of individual workers—the incentive effect.
The basic presumption of these models is that the labor market is substantively different from the model of product markets discussed in the typical elementary economics texts. In those models it is typically assumed that both buyers and sellers act as price takers. Although in most product markets prices affect the distribution of buyers and sellers as well as their behavior, it is usually reasonable to assume that neither buyers nor sellers care about how price changes affect the unobserved characteristics of their trading partners nor how their product will be used. In the typical model of a product market, if the market clears at a given price no buyer would offer to pay a higher price. For example, in the Ricardian model as the price of wheat falls, low productivity land is taken out of production. However, buyers of wheat are unconcerned whether they are buying wheat that was grown on high or low productivity land. This is because the unobservable characteristics of wheat that are correlated with the quality of land on which it was grown are not sufficiently important to negate the conventional Walrasian analysis.
In the labor market, buyers of labor services are faced with considerable uncertainty concerning many aspects of the productivity of the workers they are hiring. If lowering the wage they offer significantly lowers the average ability of the job applicants they face, firms may find that lowering their wage makes them worse off. The adverse selection aspect of labor markets is an essential feature of those markets while it is often of only secondary importance in product markets.
In labor markets (or credit or product markets) in which informational asymmetries are important there may be too little (or no) trade because of the adverse selection effect of wages. A rigorous analysis of markets with important informational asymmetries in which unobserved characteristics are correlated with the reservation price (wage) of sellers shows that the Walrasian market-clearing price may not be consistent with optimizing behavior by agents. Thus defining an equilibrium as a set of prices at which markets clear is inappropriate. It is possible that no firms would choose to charge the market clearing price, and if it were chosen by all firms, one firm could increase its profits by deviating from the Walrasian price.
While adverse selection models of the labor market are motivated by a concern with unobserved differences among workers, these models should not be thought of as only relevant to young workers. The important feature of these models is informational asymmetries. Riordan and Staiger  have presented a model of the labor market in which employers are better informed about the characteristics of their workers than are other firms. In that model adverse selection considerations coupled with sectoral shocks generate unemployment, because firms do not wish to hire the laid off workers in sectors suffering negative demand shocks. Indeed, one could argue that informational asymmetries become more marked as workers gain experience with particular firms, and thus that adverse selection considerations are even more important for older workers than for younger ones.
Wages (and prices) also affect behavior. These effects are both direct and indirect. The most direct effect of wages on behavior is on nutrition. In less developed countries the wages of workers directly affect their health, mental alertness, and physical well being. These effects are not limited to wages that are below the level needed to satisfy the worker's caloric requirements for the job. Workers, even at very low wages, will use some of their income to support family members and for consumption of goods and services other than food, shelter and medical care. Consequently there may be a significant range of wages at which increasing the wage increases the labor endowment of the worker through improved nutrition and health care. Indeed, the prevalence of subsidized meals, housing and medical care in the industrial sector of less developed countries suggests the importance of these effects. These direct effects of wages on worker productivity have been modeled by Leibenstein , Bliss and Stern [1978a, b] and Dasgupta and Ray [1986, 1987],
The first models of the indirect effect of wages on worker productivity investigated the effect of wages on quit propensity (see Hammermesh and Goldfarb , Salop [1973a, b], and Stiglitz [1974b]). Firms for which quits are costly are reluctant to cut wages when faced with an excess supply of workers. Lowering their wages would increase their turnover costs. Of course, the firm could lower the wage it pays newly hired workers while maintaining high wages for previously hired workers. However, multi-tiered wage structures of that form often generate serious morale problems. To the extent that newly hired and experienced workers need to co-operate, the frictions caused by a multitiered wage structure could impose serious costs on the firm. Note that we are not arguing against firms reducing quits by paying wages that rise steeply with tenure. Rather we are arguing against models in which firms respond to job queues by lowering the expected life-time wage of newly hired workers.
Analyses of the effect of wages on quits differ markedly from analyses of the direct effect of wages on nutrition. Quits are affected not only by the wage contract the worker is receiving, but also by the wages and hiring rates of the other firms in the economy as well as the unemployment rate. The unemployment rate affects the responsiveness of quits to changes in a firm's wage offer even if quits do not involve intervening spells of unemployment. This is because the unemployment rate affects the ratio of applicants to vacancies at different wages, and hence the probability of a worker receiving a wage offer above his current wage.
In addition to their effect on quits and nutrition, wages also affect effort. If workers were to receive wages that make them indifferent between staying with the firm and becoming unemployed, workers would always choose the levels of effort they find most enjoyable or least onerous. Neither the threat of being fired nor the threat of a wage cut would induce a higher level of effort: Since the worker would be indifferent between employment and unemployment, being fired is not a punishment. Similarly, since the worker would respond to a wage cut by quitting the firm or shirking (quitting on the job), wage cuts would have no incentive effects. The lower the wage, the more resources the firm would have to expend on supervision to maintain a given level of effort from its workers. In the limit, if the wage made workers indifferent between working and unemployment, the firm would have to constantly monitor each worker.
The effect of wages on effort depends on the expected utility from being unemployed and the degree of monitoring chosen by firms. The expected utility from unemployment depends, in turn, on the joint distribution of wage offers, the arrival times of those wage offers for an unemployed worker, and the expected duration of employment at each new job, as well as the worker's income while unemployed and his value of leisure. In particular, if no firms were hiring workers who had previously been fired for low effort then obviously the utility of unemployment would be independent of the wages offered to new entrants and the unemployment rate.
1.1. Criticisms of efficiency wage models
Readers who have no previous acquaintance with efficiency wage models may wish to skip this section and return to it after familiarizing themselves with some of the models presented in Sections 2–8.
If equilibrium is characterized by job queues, one question that naturally arises is why firms don't "sell" jobs either by requiring lump sum payments from newly hired workers, or requiring those workers to work at low wages during some initial "apprenticeship" period. Carmichael  emphasizes that, although capital market imperfections prevent workers from being able to pay a bond that equals the monetary value of the surplus they will gain from the job, if workers are sufficiently risk averse, they can be charged fees that are large enough to make them indifferent between employment and unemployment. Alternatively workers can be initially assigned to low wage jobs at which, even if effort were not monitored, the workers would be contributing some profit to the firm. These low wage job assignments would be long enough to eliminate the surplus accruing to the worker from his eventual employment at the high wage job. By having these payments made by newly hired workers, the firm is still able to have low quit rates and high levels of effort by paying high wages to experienced workers. Thus Carmichael argues that the need to pay high wages to experienced workers does not prevent firms from capturing the associated surplus, thereby eliminated job queues.
There are several problems with this critique of efficiency wage models. First it does not apply to the adverse selection or nutrition models. For those models, requiring workers to buy their jobs would negatively affect the expected productivity of the firm's work force in much the same way as would offering low lifetime wages. Thus the Carmichael critique is not valid if firms are offering high wages as a means of improving the quality of their applicant pool or as a means of directly improving the fitness of their workers. Similarly the Carmichael critique is irrelevant if high wages are paid for their morale enhancing effects or to generate good will among the firm's workers.
Second, the elimination of unemployment by having workers "suffer" (either by accepting low initial wealth or low initial wages) does not eliminate and may aggravate the social problems normally associated with unemployment. Although the level of suffering required to get a job makes the marginal worker indifferent between working and not working, there are many workers that are made strictly better off by being able to buy a job, rather than facing a lottery. These will, in general, be workers with large wealth holdings or rich parents for whom the cost of buying a job has a small effect on their consumption during that initial period. Thus the allocation of jobs and of lifetime income may be less equitable than in the pure efficiency wage model with random job allocations.
Third, because capital markets are imperfect, forcing workers to buy jobs by working at low wages when they are young imposes large social costs. These costs are due to the same informational asymmetries that cause unemployment in the efficiency wage—effort models when workers cannot buy jobs. Indeed the notion that involuntary unemployment is eliminated by having workers buy their jobs by working at low wages for long periods does not seem to be a very useful way to frame the issues. An individual that is being paid close to zero, is contributing almost no effort and is indifferent between working and staying home (as in the Akerlof and Katz model) may be labeled as employed, but that labeling will tend to cloud our understanding of the effects of information asymmetries on labor markets. Whether the incentive effects of wages result in unemployment or in long periods of low wage employment during which effort levels are low, the associated social and economic problems are likely to be similar.
Finally, although contracts that require workers to buy their jobs by working at very low wages during some initial (possibly lengthy) period of employment eliminate so-called involuntary unemployment, the level of employment is below that which would maximize output or social welfare.
On the other hand, in the effort inducing version of the efficiency wage model, requiring workers to buy their jobs, could result in an efficient level of employment being achieved. This would be the case if workers are risk neutral and had sufficient wealth to pay firms the expected value of the surplus they will receive from future employment. Of course, workers are not risk neutral and do not have unlimited wealth, but the question might still be asked as to whether outside financing of performance bonds would generate the same result. The answer is no! The same informational asymmetries that give rise to efficiency wages being offered, combined with the possibility of personal bankruptcy, will deter bank financing of performance bonds.
To see this, suppose a bank were to finance a performance bond. Then in states in which the borrower is employed, the value of the job is decreased by the interest the borrower must pay the bank as a financing charge for the bond. In states in which the worker is unemployed, the worker declares bankruptcy and defaults on the bond. Consequently, if bankruptcy is permitted the penalty for being fired might be smaller with bank financed performance bonds than without them: the fired worker loses his wage, but also no longer has to pay the finance charges on the performance bond.
If workers default when fired, not only do bank financed performance bonds accentuate the incentive and adverse selection problems described above, but banks are unlikely to finance performance bonds for workers. (Usually discussions of performance bonds implicitly assume that these bonds are self financed, or that personal bankruptcy is not permitted.)
In contrast to the effort models, in the typical quit models, where workers are risk neutral and all quits are associated with job changes, bank financed performance bonds do not affect behavior, even if bankruptcy is not permitted. The performance bond serves as a lump sum cost. Workers still seek to maximize their life-time income.
On the other hand, if some quits are quits into unemployment, a bank financed performance bond would increase the probability of a worker quitting. To see this consider a long spell of unemployment. The borrower will not pay the interest on the loan. Thus in some unemployment states the loss of income from being unemployed is partially offset from the savings gained from no longer paying interest on the bank debt. Hence quits into unemployment are made more attractive.
Excerpted from Efficiency Wages by Andrew Weiss. Copyright © 1990 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.