Wages and the intensity of labor effort: efficiency wages versus compensating payments.

• Author: Fairris, David

1. Introduction

   The distinction between efficiency wages and compensating payments is, at best, rather vague in both the theoretical and empirical literatures positing a relationship between wages and the intensity of labor effort.(1) While efficiency wages and compensating payments are both related to work intensity, the mechanisms that generate these relationships are in each case quite distinct. Compensating payments exist in competitive labor markets to equalize overall compensation across identical workers with dissimilar working conditions. Workers who expend more labor effort must be paid more or they will seek employment elsewhere. Efficiency wages are employment rents paid by firms in order to ensure adequate labor effort. These rents vary with the costs firms face in monitoring labor effort. They cannot be competed away because without employment rents, profit-maximizing levels of work intensity would not be forthcoming.

   In this paper we distinguish between the efficiency wage and compensating payments components in wage differences across workers and firms. In the first section of the paper, we use a theoretical model of optimal labor effort choice by workers and firms to make these two components of wage differences distinct. The theoretical analysis points to the need for estimating simultaneously a set of structural equations in order to empirically test the efficiency wage hypothesis. In the second section of the paper, we employ this approach to test for the effect of employment rents on labor intensity. The results offer tentative support for the efficiency wage hypothesis, but virtually none for the existence of positive compensating payments for labor effort.

2. Theoretical Analysis

   The earliest theoretical work on efficiency wages by Shapiro and Stiglitz [23] and Bowles [3] contends that competitive firms may rationally pay wages greater than workers' opportunity costs if labor intensity is a positive function of wages. Relative to a nonexistent counterfactual in which firms do not pay efficiency wages, labor intensity, wages, and profits are greater in the world of efficiency wage payments, but so too is unemployment.(2) Note, however, that because labor effort is a working condition which leads to worker disutility, any increase in labor intensity will call forth a compensating payment for the increased disutility of work. Thus, the full magnitude of the wage difference between the (counterfactual) market-clearing and (actual) nonmarket-clearing equilibria incorporates both efficiency wage and compensating payments components.(3) This point has not been fully appreciated in the literature.

   Even when efficiency wage theorists discuss the issue of wage differences across firms or industries they often make the mistake of ignoring the theory of equalizing differences.(4) For example, in a discussion of efficiency wages, Stiglitz begins by cautioning the reader that "we are comparing the wages paid by firms for workers with a given set of observable qualifications."(5) He then goes on to argue that if firms face different internal costs of generating labor intensity, then they will be found to pay different wages for workers of identical characteristics.

   However, at this level of analysis such wage differences are indistinguishable from compensating payments, because compensating payments are also premised on the differing characteristics of work places, worker qualifications constant. And, indeed, an example Stiglitz appeals to immediately following the above comments is that larger firms may pay higher wages because of their higher costs of monitoring. But this empirical observation has also been explained with reference to compensating payments. For example, Hamermesh and Rees write that "these wage differences may reflect compensating differentials for the disamenity of working in the more rigid environment that often characterizes larger firms and plants" [10, 287].

   Sorting out the confusion surrounding the relationship between wages and the intensity of labor effort requires an analysis of efficiency wages that explicitly allows for the existence of compensating payments. Efficiency wages must be seen as payments over and above any compensation which merely offsets the disutility associated with increased labor intensity. Put more succinctly, and in the context of our model below, efficiency wages are the differences in wage payments of equally qualified workers, intensity held constant.

   Model

   We begin our model by noting that both the efficiency wage and compensating payments literatures are concerned with the qualitative characteristics of the work place [24; 25]. The authors in both literatures posit worker utility functions that depend on wages and work place characteristics, and firm production functions that depend on these same characteristics as well as the more familiar quantities of capital and labor inputs.

   We shall assume that worker utility functions are of the form:

   U = u(W,i), (1)

   where W is labor earnings and i is individual labor intensity per hour of labor. We assume that firm production functions are of the form:

   Q = q(K,L [center dot] i), (2)

   where L [center dot] i is the total work done by labor during the purchased labor hours, L.(6) We also assume that workers are homogeneous with respect to human capital endowments, although not necessarily with respect to preferences over intensity, and that they face similar working conditions other than intensity. Finally, we assume that the value firms place on labor intensity and the costs they face in monitoring labor effort vary along with the technology and organization of production.

   The efficiency wage literature of interest to us is concerned with the problem of eliciting labor effort from purchased labor inputs. In that literature, workers are assumed to give forth more labor effort the greater the cost of job loss and the larger the threat of dismissal. A worker's utility-maximizing choice of intensity, given the cost of job loss and the probability of being dismissed for insufficient work effort (i.e., shirking), can be expressed as follows [3, 23]:

   i = i(W - W*,p), (3)

   where W is the wage, W* is the opportunity wage in the general labor market, and p is the probability of being caught shirking (and therefore dismissed), which depends on the extent of supervision and plant size, among other things.

   The theory of compensating payments is concerned with how wages differ across working conditions. Competitive labor markets produce outcomes in which the compensation packages of identical workers are equal. However, because not all nonmonetary preferences of workers are identical, equilibrium compensating payments will not equalize compensation across all workers, but will instead reflect a worker-employer matching process in which the magnitude of compensating wage differentials depends on the distribution of tastes and production technologies, inter alia, among the respective agents. The equilibrium relationship between wages and intensity can be expressed as follows:

   W* = w(i).(7) (4)

   The intensity of labor effort is, of course, only one of many elements of the work environment that concerns workers; others include safety, noise levels, creative work, and perhaps firm size.(8)

   We combine these two theories by positing the existence of a labor market setting in which some finns can monitor workers costlessly while others face positive costs of monitoring. The only binding constraint on the first group of firms is equation (4) above, which represents the locus of wage/intensity combinations existing in this segment of the labor market, and from which one can derive the equilibrium compensating payments for labor intensity. Firms in the second labor market segment find it optimal to pay efficiency wages, and thus have equations (3) and (4) as binding constraints.(9)

   A typical firm's profit-maximizing decision can be characterized as follows:

   MaxII = VA - W [center dot] L (5)

   subject to i = i(W - W*,p) and W* = w(i), where VA is value added in production. For ease of exposition, let us hold constant input levels, input and output prices, and p (the probability of shirking detection) in order to concentrate on the firm's choice of intensity of labor...