A job evaluation compensation system clearly differs from this model; job evaluation does not reward individuals per se, but only pays them according to the jobs they occupy. Job evaluation policy stipulates that jobs are evaluated based on social judgments about what the job demands from its occupants (rather than what the jobholders have to offer), and the pay levels of jobs are set in a way that will reinforce the authority and status differences in the organization hierarchy. Little mention is made of market rates; internal equity comparisons are the dominant considerations in setting wage levels.
Job statuses are based on three main factors: requirements, working conditions, and responsibility, none of which is adequately described by human capital theory. The first, job requirements, is likely to be related to individuals' human capital, but individuals' capabilities may sometimes fall short of job requirements (e.g., during tight labor markets) and sometimes exceed job requirements (e.g., during loose labor markets), and, in both circumstances, job evaluation dictates that individuals be paid according to job demands, not their individual human capital: Engineers forced to do janitorial work are paid as janitors. The second, harsh working conditions, is likely to be negatively related to human capital. The third, responsibility, which is primarily indicated by a job's position in the authority hierarchy, may be positively related to human capital, but the incremental human capital needed for a promotion is likely to be much smaller than the pay gains conferred by the promotion. Indeed, managers report that promotion selections are generally difficult decisions made about high-ly similar competitors, but the promotion decision leads to these marginally superior individuals receiving much greater earnings on the day that they move into their new jobs. The human capital they bring with them is clearly not what determines their increased earnings. Moreover, Thurow (1975) contends that human capital theory reverses the causal direction: Jobs determine the value of individuals' human cap-ital, rather than human capital determining jobs and salaries. Clearly, earnings differences across statuses are difficult to attribute solely to human capital.
Of course, individuals are assigned to job statuses in part based on their personal resources (human capital), but, given the difficulties of assessing human capital per se, even these assignments are likely to be based on social signals of ability and features of their job histories, such as original and subsequent job statuses, and the recency of job moves. Although one may interpret earnings conferred by job statuses as vaguely indicative of individual human capital, this relationship is at best a very indirect one, mediated by diverse social processes.
Of course, after individuals have occupied a higher-level job for some time; they will have acquired the skills necessary to perform its functions, and they will have the human capital commensurate with the value attributed to the job. But that acquired human capital is less a property of the individual than it is the property of the job which conferred the skills. As Thurow (1975) states in his job competition theory, fairly equivalent individuals compete for jobs, and the winners win both the better jobs and the on-the-job training required to permit them to do the job.
But there is another aspect of compensation which does resemble the economic theories in being individually based. According to company policy, part of compensation is based on merit considerations; individuals are paid within jobs according to their personal productivity as judged by their supervisors. Since job statuses often contain many different jobs, and since jobs often contain many individuals; it is quite possible that considerable earnings variance could occur within structurally equivalent jobs, and this unstructured earnings variance could resemble the pure, individually based process described by human capital theory. To what extent is the variation in individuals' earnings mediated by structural positions and to what extent does it occur within structurally equivalent positions?
This question can be tested here for random samples of the corporation's employees over several periods, and the job status scale provides a better estimate of the full extent of structural effects. More important, by comparing analyses over different periods, the extent to which social and economic forces affect earnings through the status structure and the extent to which their influence is independent of the status structure can be discerned. Structural theory generally posits that the component of earnings mediated by status is less responsive to external forces than the component of earnings which is independent of status. The dynamic structural theory proposed here, while contending that structurally mediated earnings may be somewhat responsive to external forces, still posits that their responsiveness is subject to constraints, such as those described by the selectivity and developmental hypotheses and by structural discrimination. In contrast, earnings differences within statuses, being determined at the discretion of supervisors, would be unfettered by these structural constraints and may be fully responsive to market influences and individual contributions.
Elaborating the regression model discussed previously permits an investigation of this issue. The empirical analysis is modified by making earnings (in logs) the dependent variable, and by introducing job status as an additional independent variable.
As has already been seen, changes in job status attainments support the predictions of the dynamic structural model; however, since economic theory is meant to explain earnings, the earnings changes pro-vide a better test. Indeed, within job statuses, the results provide some support for economic theory. Within statuses, the earnings advantages of M.B.A.'s and high-status colleges mostly change in the directions predicted by economic theory. However, while the earnings advantages of M.B.A.'s and high-status colleges within statuses are small and insignificant, their earnings advantages across statuses are much larger; and these advantages change-not according to the predictions of economic theory, but according to the predictions of the selectivity and developmental hypotheses. The advantages of the elite college-educated groups (M.B.A.'s and high-status college alumni) decline with organization growth and increase with declining growth; and college-educated men's gains in 1965-1969 are largest for those with low seniority, while their declines in the final period are also greatest for those with low seniority. The greatest part of earnings variance for college-educated men is across statuses, and the changes in the effects of these factors are subject to the constraints described by the dynamic structural model.
Economic theory and the dynamic structural model make the same predictions about the changing effects of college degrees for men's earnings: Both predict increased influence with growth and declining influence with declining growth, contraction, and affirmative action.
