Efficiency Wages, Interfirm Comparison, and Unemployment.

• Author: Johansen, Kare
• Position: Statistical Data Included

Kare Johansen [*]

Bjarne Strom [+]

The paper presents an efficiency wage model where worker effort depends on own wages relative both to wages of other workers in the firm and to similar workers in other firms. First, we show how the Solow conditions are modified if internal comparison effects are at work. Second, we discuss the effect of internal wage comparison on wage inequality within firms. Third, we study unemployment and relative wage determination within a general equilibrium model, and analyze the effect of technological change and various tax policies on equilibrium unemployment and relative wages. Finally, the short-run effects of aggregate demand shocks are analyzed.

1. Introduction

   A popular argument for the existence of involuntary unemployment is that a worker's effort rises with wages, giving employers incentives to set wages above market clearing levels. It is often assumed that the worker's effort is a function of the wage level, with no reference to the size of the wage in comparison with that received by other workers. Yet, such comparisons are likely to influence an individual in his effort choice. If this is the case, then it seems equally likely that workers compare themselves with others, both internal and external to their place of work. This paper contributes to the efficiency wage literature by providing an understanding of wage setting and unemployment determination within a model where workers' efforts depend on their wage relative to both the wages of other workers at the same firm and to the wages of similar workers at other firms.

   A common explanation of the permanent rise in unemployment rates in European countries since the mid-1970s is that relative wages do not adjust sufficiently in response to relative labor demand shocks (OECD 1994). Strong labor unions fighting for a high degree of wage equality have been blamed for this relative wage rigidity. However, when worker effort depends on relative wages, it might well be in the employers' own interest to pursue an equalizing wage policy. Within our efficiency wage model we demonstrate that under fairly plausible conditions, firms have incentives to reduce within-firm wage inequality when internal wage comparisons are important.

   The present paper is related to the Akerlof and Yellen (1990) fair wage--effort model. Akerlof and Yellen assume that effort reaches a maximum value when the wage paid equals the fair wage, so that further wage rises have no effect on effort. We extend the Akerlof and Yellen (1990) model by letting effort be a continually increasing function of the two relative wage rates. [1] Accordingly, whereas unemployment occurs only for low-skilled workers in the Akerlof and Yellen model, our model generates equilibrium unemployment for both high-skilled and low-skilled workers. According to Nickell and Bell (1996) there has been a substantial increase in skilled as well as unskilled unemployment in Europe since the mid-1970s. At this point our model seems to be more in line with the empirical evidence than the Akerlof and Yellen specification.

   An important question is what factors may cause the equilibrium unemployment to change over time? Technological changes working against unskilled workers combined with insufficient relative wage flexibility have been proposed as an explanation for the permanent rise in European unemployment rates since the mid-1970s (Krugman 1994; OECD 1994). Our model predicts that under fairly plausible conditions, a fall in productivity of low-skilled relative to high-skilled workers increases the aggregate equilibrium unemployment rate. Moreover, we show that economic policy in the form of increased progressivity of the payroll tax system will reduce equilibrium unemployment. Finally, we examine the time path of unemployment after a demand shock. When inflationary expectations are formed at least partially adaptively, actual unemployment and wage inequality will decrease in the short run and then gradually return toward their equilibrium levels in response to a favorable aggregate demand shock.

   The plan of the paper is as follows. Section 2 sets out the basic model. We derive modified Solow conditions and discuss how internal wage comparisons affect relative wages. In section 3 we discuss unemployment and relative wage determination within a general equilibrium model. In section 4 we show how payroll tax policies can be designed to create incentives to wage moderation and thus to reduce equilibrium unemployment in our model. In section 5 we introduce expectation inertia and analyze effects of aggregate demand shocks on unemployment and relative wages in the short and intermediate run. Concluding comments are given in section 6.

2. A Model with Internal and External Comparison

   Consider a firm producing a single output using two types of labor. The production function is given by

   X = F([e.sub.1][N.sub.1], [e.sub.2][N.sub.2]),

   where [N.sub.i] is the number of workers in group i and we assume that the effort of each group of workers, [e.sub.i], enters the production function in a labor-augmenting way. We assume that

   [F.sub.i] [less than] 0, [F.sub.ii] [less than] 0, and [F.sub.11][F.sub.22] - [F.sub.12][F.sub.21] [greater than] 0.

   The crucial assumption is that effort of each group of workers is a continually increasing function of wages for this group relative to wages being paid to other workers both within and outside the firm. Survey evidence in favor of interpersonal comparison effects is provided by Blinder and Choi (1990), Agell and Lundborg (1995), Clark and Oswald (1996), Campbell and Kamlani (1997), and Bewley (1998, 1999). [2] The Swedish evidence reported in Agell and Lundborg (1995) indicates that workers pay great attention to the wage distribution both within and across firms. Blue-collar workers are concerned with both the inter- and intrafirm wage structures, whereas white-collar workers put a greater emphasis on the interfirm wage structure. Agell and Lundborg also provide evidence that managers believe that workers will respond to perceived shortcomings in their relative wage by reducing effort. Surveys among U.S. managers suggest that employers mainly care about the wage structure within firms. Campbell and Kamlani (1997) found that notions of fair wages depend on own past wages, firm's profits, and wages of other workers in the same firm. The evidence in Campbell and Kamlani (1997) strongly supports the hypothesis that relative wages affect effort. In particular, changes in internal wage structure would reduce effort for those workers who received relatively less. According to Bewley (1998, p. 478) "The relation of a firm's pay to that at other companies usually has little effect on morale or work effort, because workers know little about pay levels outside their own company." On the other hand "The relation of employees' pay to that of others at the same work site has a major impact on morale and effort, as does the relation of pay to performance."

   We further assume that higher unemployment will increase effort of both groups of workers. Incentive-based efficiency wage models as described for instance in Layard, Nickell, and Jackman (1991) typically assume that individual utility increases with wages and decreases with effort. The expected utility, conditional on being laid off, increases with the alternative wage and decreases with the unemployment rate. Thus, according to incentive models, the effort chosen by the individual will depend on own wages, external wages, and the unemployment rate. Our effort function may be viewed as a generalization of such incentive-based models augmented to take into account considerations of fairness. [3] Econometric evidence supporting the positive effort--unemployment relation can be found in Oster (1980), Green and Weisskopf (1990), Ackum Agell (1994), and in Rebitzer (1987, 1988). Survey evidence is provided in Agell and Lundborg (1995). In the basic version of the model we assume that only aggregate unemployment m atters, whereas in section 3, we also discuss the case where effort depends on group-specific unemployment.

   The assumptions above are formally stated by

   [e.sub.i] = [e.sub.i] ([w.sub.i]/[w.sub.j], [w.sub.i]/[[w.sup.e].sub.i], u) i, j = 1, 2 i [not equal to] j,

   where [w.sub.i] ([w.sub.j]) is the wage rate paid to group i (j), [[w.sup.e].sub.i] is the (expected) external reference wage for group i, u is the aggregate unemployment rate. We assume that

   1. [partial][e.sub.i]/[partial]([w.sub.i]/[w.sub.j]) [greater than] 0,

   2. [partial][e.sub.i]/[partial]([w.sub.i]/[w.sup.e].sub.i]) [greater than] 0,

   3. [partial][e.sub.i]/[partial]u [greater than] 0,

   4. [[partial].sup.2][e.sub.i]/[[partial].sup.2]([w.sub.i]/[w.sub.j]) [less than] 0,

   5. [[partial].sup.2][e.sub.i]/[[partial].sup.2]([w.sub.i]/[[w.sup.e].sub .i]) [less than] 0,

   6. [[partial].sup.2][e.sub.i]/[partial]([w.sub.i]/[w.sub.j])[partial]([w .sub.i]/[[w.sup.e].sub.i]) = 0,

   7. [[partial].sup.2][e.sub.i]/[partial]([w.sub.i]/[w.sub.j])[partial]u [less than] 0,

   8. [[partial].sup.2][e.sub.i]/[partial]([w.sub.i]/[[w.sup.j].sub.i]) [less than] 0.

   Turning to the firm's decision problem, we write down the profit function given by

   Assumption (f) is made for simplicity and states that the marginal effect of internal wage relativities is independent of outside wages, and vice versa. Assumptions (g) and (h) state that high unemployment reduces the marginal gain (from the firm's point of view) of further wage increases. These assumptions are intuitively appealing since assumption (c) means that workers already work very hard if unemployment is high. The remaining assumptions are standard. We also assume that the derivatives of the effort elasticities with respect to [w.sub.i]/[w.sub.j], [w.sub.][[w.sup.e].sub.i], and u are decreasing in the same arguments.

   Our specification of the effort function is a generalization of that used in Akerlof and Yellen (1990) since their effort...