Male female disparity in starting pay.

• Author: Becker, Elizabeth

Empirical studies of the earnings of male and female workers persistently reveal a gap that is unexplained by the economic variables in these models.(1) This gap has been given two interpretations in the literature. One holds that the gap reflects non-market forces which distort the equilibrium relation among the wages of competing workers. Discrimination, according to this view, prevents female workers from being offered jobs commensurate with their qualifications and "crowds" them into low productivity occupations.(2) The alternative interpretation holds that some substantial portion of the gap is an artifact, the result of missing variables, biased parameters, or measurement error. These researchers view the labor market as highly competitive, and, were it possible to correct for all these statistical problems, the measured wage gap would largely vanish.(3)

This research belongs to the latter category. We argue that a portion of the measured gap, at least that observed between the earnings of young men and women during their first few years with their employers, reflects sex-related differences in the shares of investment in firm-specific human capital borne by workers of each group. Moreover, these variations in investment levels are freely chosen by women and in no way imply long run deficits in their earning power. On the contrary, the model predicts steeper earnings profiles of women who undertake this firm-specific investment. Thus, women's earnings are predicted eventually to overtake and surpass those of equally qualified men.

This theory was developed in earlier research by Becker and Lindsay [2] based on a model of Hashimoto [11; 12] and extensions by Parsons [21]. We argued that young women, for non-discriminatory reasons, elect to shoulder a larger share of this investment than young men. The explanation for this is developed in two stages. The first derives the terms under which both male and female workers share the investment with their employers. The second argues that female workers will be led by these considerations to bear a larger share.

Workers of both sexes agree to enter into inflexible wage contracts with employers specifying a share of the investment (i.e., a wage deduction from current MVP) and a post-investment wage. The advantage to such contracts lies in the solution to an assymetric information problem. The issue of who bears the cost of firm-specific human capital when the future quasi-rents produced by this investment are uncertain is the source of agency cost. An agreement to share the investment and the quasi-rent works only if the magnitude of the realized quasi-rent may be identified at low cost. Where disagreement about the worker's productivities inside and outside the firm can exist, wasteful bargaining strategies will be adopted. Workers in the post-investment period will have the incentive to claim that opportunities outside the firm are good; hence the offered wage is too small. Employers may claim that productivity inside the firm is low and thus the demanded wage is too high. This gaming is reduced with a contract that specifies ex ante the sharing of the training cost and the wages to be paid in the post-investment period.

However, these wage contracts, designed to reduce inefficiency on one margin, produce inefficiency on another. Fixed wages yield inefficient job separations due to the inability of workers and management to adjust wages to productivity-affecting events. Some workers will be dismissed because their post-investment wages are too high in spite of the fact that their productivity inside the firm exceeds their productivity outside the firm. Others will quit under the same inefficient circumstances because their wages are too low inside the firm. An optimal contract minimizes the expected cost from both types of inefficient separation.

Young women agree to bear a larger share of this firm-specific human capital in order to minimize the costs of these inefficient quits and terminations. The efficient inflexible contract adjusts the share of the investment in firm-specific human capital between workers of either sex and employers so as to (effectively) minimize the combined likelihood of an inefficient separation of either type. We have shown elsewhere [2] that the contract terms that minimize the costs of inefficient separations for women will differ from those of men, involving a larger worker share of the investment in firm-specific human capital. Although this investment must eventually be returned to women workers in the form of higher earnings later in their careers, the practice creates an apparent wage gap for these women in the early years of tenure. This paper seeks to determine how much of the observed wage gap in starting wages can be explained on the basis of such contracting.(4)

Perhaps the most important previous challenge to the "wage gap as discrimination" view lies in the work of Mincer and Polachek [18; 19] and most recently Polachek and Kao [22]. According to this interpretation, experience is measured with error for women. These female workers anticipate only intermittent labor force participation. They expect to leave the labor force to bear and/or raise children; hence they invest less in human capital than workers anticipating full time employment. Male workers with a given number of years' experience will therefore have accumulated more human capital than female workers contemplating an intermittent working career and earn correspondingly higher wages reflecting this real productivity differential.

Evidence in support of this "human capital" explanation of the wage gap is impressive. Various studies have produced estimates of the portion of the gap explained by this approach that range from 70 to 90 percent. However, critics of the human capital explanation point to an uncomfortable empirical finding that raises doubts about this methodology. England [7], for example, has observed that disparity in wages between predominantly male and female occupations is observed very early in the working life, even in the first year when precious little human capital will have been created by workers of either sex.

The appearance of a substantial wage gap in the early years of employment raises serious problems for the human capital theory of the wage gap. If measurement error in human capital acquired on-the-job is responsible for the wage gap, then it seems plausible that the extent of this artificial gap should be related to the amount of human capital acquired by males (and not, by hypothesis, by females). A large gap in the early years of employment can be interpreted as implying that some factor other than systematic errors in variables may be at work in the data. One potential explanatory factor is, of course, discrimination.(5) We propose another: namely, differential investment sharing.

1. The Investment Sharing Model

   The Hashimoto model concerns the sharing of the costs of human capital investments that are specific to employers. Employment contracts must be concerned with three margins. First, such investments create opportunities for hold-ups on the part of workers and employers. With costly information neither will know the real magnitude of the quasi-rents produced by such investments. Resources will thus be dissipated through efforts to gain a larger share of these returns. Workers can be counted upon to exaggerate outside employment opportunities, while firms will dissimulate concerning the real contributions of these workers to the fortunes of their enterprises. One contractual provision which minimizes the costs of such opportunism is an inflexible wage agreement providing both a division of the costs of this investment together with a predetermined future wage, the terms of which can be competitively determined ex ante.

   The second margin concerns inefficiency produced by such inflexible contracts themselves. Workers are optimally separated from their employers when they are worth more in alternative employment, that is, when their marginal value products inside the firm ([MVP.sub.in]) exceed their marginal value products elsewhere ([MVP.sub.out]). However, an inflexible agreement on wage [Mathematical Expression Omitted] can lead to inefficient separations in two cases, depending upon the stipulated wage and the ex post realizations of inside and outside productivity. Separations will be inefficient where [Mathematical Expression Omitted] and also where [Mathematical Expression Omitted]. In the former case workers are dismissed when, in fact, their productivity remains highest inside the firm. In the latter case workers will quit even when they are worth more to their current than their alternative employers.

   Inflexible wages therefore reduce hold-up costs but introduce costs of inefficient separations. It is plausible that on net such inflexible wage agreements may reduce total cost. However, the extent of this reduction depends on adjustment concerning a third margin. The efficiency of such inflexible wage contracts is itself dependent upon how the investment in firm-specific human capital is shared. A larger share for the worker implies a higher future wage. Such an agreement will produce more dismissals including more inefficient ones. On the other hand, a larger share for the firm implies a lower future wage, which can be expected to increase the number of total as well as the number of inefficient quits. The optimal contract is one which equates the costs of these two forms of inefficient separation at the margin.

   Parsons [21] argues that the structure of this contract will depend on the variance in expectations about future productivity inside and outside the firm. More specifically, he predicts that workers with a higher variance in outside expectations will efficiently bear a larger share of the cost of firm-specific human capital investment. Gronau [8] and more recently Lazear and Rosen [15] have...