Professional Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review
By JANET L. YELLEN*
an identical wage, and if there is full employ- argue that a major part of the productivity
ment, there would be no cost to shirking and slowdown is attributable to loss of employer
it would pay all workers, assumed to get control due to a reduction in the cost of job
pleasure from loafing on the job, to shirk. In loss. The shirking model also offers an inter-
these circumstances, it pays each firm to pretation of hierarchical wage differentials,
raise its wage to eliminate shirking. When all in excess of productivity differences (Calvo
firms do this, average wages rise and employ- and Stanislaw Wellisz, 1979).
ment falls. In equilibrium, all firms pay the All these models suffer from a similar the-
same wage above market clearing, and un- oretical difficulty-that employment con-
employment, which makes job loss costly, tracts more ingenious than the simple wage
serves as a worker-discipline device. Unem- schemes considered, can reduce or eliminate
ployed workers cannot bid for jobs by offer- involuntary unemployment. In the cheat-
ing to work at lower wages. If the firm were threat model, the introduction of employ-
to hire a worker at a lower wage, it would be ment fees allows the market to clear effi-
in the worker's interest to shirk on the job. ciently as long as workers have sufficient
The firm knows this and the worker has no capital to pay them (see Eaton-White and
credible way of promising to work if he is Stoft). Unemployed workers would be will-
hired. ing to pay a fee to gain employment. Fees
The shirking model does not predict, coun- lower labor costs, giving firms an incentive to
terfactually, that the bulk of those unem- hire more workers. If all firms charge fees,
ployed at any time are those who were fired any worker who shirks and is caught knows
for shirking. If the threat associated with that he will have to pay another fee to regain
being fired is effective, little or no shirking employment. This possibility substitutes for
and sacking will actually occur. Instead, the the threat of unemployment in creating work
unemployed are a rotating pool of individu- incentives. Devices which function similarly
als who have quit jobs for personal reasons, are bonds posted by workers when initially
who are new entrants to the labor market, or hired and forfeited if found cheating, and
who have been laid off by firms with declines fines levied on workers caught shirking. The
in demand. Pareto optimality, with costly threat of forfeiting the bond or paying the
monitoring, will entail some unemployment, fine substitutes for the threat of being fired.
since unemployment plays a socially valu- Edward Lazear (1981) has demonstrated the
able role in creating work incentives. But the use of seniority wages to solve the incentive
equilibrium unemployment rate will not be problem. Workers can be paid a wage less
Pareto optimal (see Shapiro-Stiglitz). than their marginal productivity when they
In contrast to the simple efficiency-wage are first hired with a promise that their earn-
model, the shirking model adds new argu- ings will later exceed their marginal produc-
ments to the firm's effort function-the aver- tivity. The upward tilt in the age-earnings
age wage, aggregate unemployment, and the profile provides a penalty for shirking; the
unemployment benefit. The presence of the present value of the wage paid can fall to the
unemployment rate in the effort function market-clearing level, eliminating involun-
yields a mechanism whereby changes in labor tary unemployment.
supply affect equilibrium wages and employ- As a theoretical objection to these schemes,
ment. New workers increase unemployment, employers would be subject to moral hazard
raising the penalty associated with being fired in evaluating workers' effort. Firms would
and inducing higher effort at any given have an obvious incentive to declare workers
wage. Firms accordingly lower wages and shirking and appropriate their bonds, collect
hire more labor as a result. In a provocative fines, or replace them with new fee-paying
recent paper, Thomas Weisskopf, Bowles, workers. In Lazear's model, in which the firm
and David Gordon (1984) have used the pays a wage in excess of marginal product to
presence of the unemployment benefit in the senior workers, there is an incentive for the
effort function to explain the secular decline firm to fire such workers, replacing them
in productivity in the United States; they with young workers, paid less than their pro-
ductivity. The seriousness of this moral ment or training fee scheme could be em-
hazard problem depends on the ability of ployed without the problem of moral hazard.
workers to enforce honesty on the firm's It is no longer in any firm's interest to dis-
part. If effort is observable both by the miss trained workers; explicit contracts could
firm and by the worker, and if it can be veri- probably be written to insure that training is
fied by outside auditors, the firm will be actually provided to fee paying workers. Al-
unable to cheat workers. Even without out- though moral hazard thus appears to be a
side verification, Lazear has shown how the less formidable barrier to achieving neoclas-
firm's concern for its reputation can over- sical outcomes via fees or bonds than in the
come the moral hazard problem. Sudipto shirking model, capital market imperfections
Bhattacharya (1983) has suggested tourna- or institutional or sociological constraints
ment contracts that also overcome the moral may in fact make them impractical.
hazard problem. The firm can commit itself
to a fixed wage plan in which a high wage is C. Adverse Selection
paid to a fraction of workers and a low wage
to the remaining fraction according to an ex Adverse selection yields further reason for
post, possibly random, ranking of their effort a relation between productivity and wages.
levels. By precommitting itself to such a plan Suppose that performance on the job de-
with a fixed wage bill, any moral hazard pends on "ability" and that workers are het-
problem on the firm's part disappears. erogeneous in ability. If ability and workers'
reservation wages are positively correlated,
B. The Labor Turnover Model firms with higher wages will attract more
able job candidates. (See James Malcolmson,
Firms may also offer wages in excess of 1981; Stiglitz, 1976b; Andrew Weiss, 1980.)
market clearing to reduce costly labor turn- In such a model, each firm pays an efficiency
over. (See Steven Salop, 1979; Ekkehart wage and optimally turns away applicants
Schlicht, 1978; and Stiglitz, 1974.) The for- offering to work for less than that wage. The
mal structure of the labor turnover model is willingness of an individual to work for less
identical to that of the shirking model. than the going wage places an upper bound
Workers will be more reluctant to quit the on his ability, raising the firm's estimate that
higher the relative wage paid by the current he is a lemon. The model provides an ex-
firm, and the higher the aggregate unemploy- planation of wage differentials and different
ment rate. If all firms are identical, one layoff probabilities for observationally dis-
possible equilibrium has all firms paying a tinct groups due to statistical discrimination
common wage above market clearing with if it is known that different groups have even
involuntary unemployment serving to di- slight differences in the joint distributions of
minish turnover. ability and acceptance wages. However, for
The theoretical objection to the prediction the adverse-selection model to provide a con-
of involuntary unemployment in this model vincing account of involuntary unemploy-
again concerns the potential for more sophis- ment, firms must be unable to measure effort
ticated employment contracts to provide and pay piece rates after workers are hired,
Pareto-superior solutions. As Salop explains, or to fire workers whose output is too low.
the market for new hires fails to clear be- Clever firms may also be able to mitigate
cause an identical wage is paid to both trained adverse selection in hiring by designing self-
and untrained workers. Instead, new workers selection or screening devices which induce
could be paid a wage equal to the difference workers to reveal their true characteristics.
between their marginal product and their
training cost. A seniority wage scheme might D. Sociological Models
accomplish this, although, if training costs
are large and occur quickly it might prove The theories reviewed above are neoclassi-
necessary to charge a fee to new workers. In cal in their assumption of individualistic
contrast to the shirking model, an employ- maximization by all agents. Solow (1980) has
argued, however, that wage rigidity may more cause significant business cycle fluctuations,
plausibly be due to social conventions and is consistent with near rationality in an econ-
principles of appropriate behavior that are omy with efficiency wage setting. Any firm
not entirely individualistic in origin. George that normally chooses its wage as part of an
Akerlof (1982) has provided the first ex- optimizing decision will incur losses that are
plicitly sociological model leading to the only second-order if it follows a rule of thumb
efficiency-wage hypothesis. He uses a variety in adjusting nominal wages which leads to a
of interesting evidence from sociological real wage error. At the point of maximum
studies to argue that each worker's effort profits, the profit function relating wages to
depends on the work norms of his group. In profits is flat. Thus, in the neighborhood of
Akerlof's partial gift exchange model, the the optimum wage, the loss from wage errors
firm can succeed in raising group work norms is second-order small. This implies that firms
and average effort by paying workers a gift with sticky wages have profits that are insig-
of wages in excess of the minimum required, nificantly different from firms with maximiz-
in return for their gift of effort above the ing behavior. Furthermore, if firms have
minimum required. The sociological model price-setting power because of downward-
can explain phenomena which seem inexpli- sloping demand curves, for similar reasons,
cable in neoclassical terms-why firms don't price-setting errors also lead to insignificant
fire workers who turn out to be less produc- losses.
tive, why piece rates are avoided even when In the Akerlof-Yellen model, firms are
feasible, and why firms set work standards efficiency-wage setters and monopolistic
exceeded by most workers. Akerlof s paper competitors. In the long run, wages and prices
in this issue explores alternative sociological are set by all firms in an optimal way. In the
foundations for the efficiency wage hypothe- short run, in response to aggregate demand
sis. Sociological considerations governing the shocks, some firms keep nominal wages and
effort decisions of workers are also em- prices constant, while other firms choose
phasized in Marxian discussions of the ex- these variables optimally. In this model, a
traction of labor from labor power (see, for cut in the money supply causes a first-order
example, Bowles, 1983). change in employment, output, and profits.
But the behavior of nonmaximizers is near
III. Explaining the Business Cycle rational in the sense that the potential gain
any individual firm could experience by
Any model of the business cycle must abandoning rule of thumb behavior is sec-
explain why changes in aggregate demand ond-order small. And thus the efficiency-wage
cause changes in aggregate employment and hypothesis can be extended into a full-fledged
output. A potential problem of the efficiency- Keynesian model of the business cycle gener-
wage hypothesis in this regard is the absence ated by sticky prices and wages.
of a link between aggregate demand and
economic activity. In an economy with IV. Concluding Remarks
efficiency-wage setting, there is a positive
natural rate of unemployment and real wage It has been widely observed that the ex-
rigidity. But the economy's aggregate output istence of excess labor supply does not lead
is independent of price at this natural rate. to aggressive wage cutting by workers and
These models have no wage or price sticki- firms. Firms appear content to pay workers
ness to cause real consequences from aggre- more than the wages required by their poten-
gate demand shocks. However, for a natural tial replacements. The models surveyed here
but subtle reason, the efficiency-wage model offer several different and plausible explana-
is consistent with nominal wage rigidity and tions of this seemingly paradoxical fact. In
cyclical unemployment. This reason (sug- addition to accounting for the persistence of
gested by Stoft), is explored in depth by involuntary unemployment in competitive
Akerlof and myself (1983), where we argue markets, these efficiency wage models can
that sticky wage and price behavior, that will explain why unemployment varies in re-