However, analysis of promotions across time periods indicates considerable stability. For this analysis, the percentage of employees who were promoted in each 3-4-year interval was considered, again limiting analysis to jobs with five or more employees. The correlations indicate moderately strong stability for the two lower levels. The correlations clearly indicate that individuals tend to get promoted from the same set of jobs over these very different periods. Economic recessions, changing organization priorities, and affirmative action led to many kinds of changes in this organization over this 13-year period; however, to a great extent, the highest promotion rates are consistently associated with the same set of jobs.
What Determines the Promotion Chances Offered by Jobs?
Having found that the promotion chances associated with jobs are stable properties of jobs, one may wonder what it is about jobs that leads some jobs to offer greater promotion chances than others? By repeating the analyses previously carried out for job salaries on promotion chances, demographic composition is often found to explain promotions as well or better than it explains job salaries, judged in terms of total explained variance. The percentage of college graduates in a job has the largest influence, while tenure ha$ a negligible influence throughout. Minority and female percentages are never significant in nonmanagement level. Female percentage in foreman level jobs shifts from an almost significant negative impact to a significant positive impact over these periods. Apparently, the affirmative action program, which did not markedly increase the compensation of foreman level female jobs, did increase the promotion opportunities out of these jobs.
Although the findings of these analyses indicate that jobs are created and terminated more often than is commonly assumed, the jobs which do persist over time tend to have stable attributes-stable job salaries, stable demographic composition, and stable promotion chances-even across periods of changing growth and during which affirmative action programs were instituted. Since many individuals have either changed jobs or left the firm over the period from 1962 to 1965, it may be concluded that this stability is not merely due to the same individuals remaining in jobs but rather that jobs per se have stable attributes.
The notion that jobs have attributes contradicts an implicit assumption of individualistic research. When research on individuals' attainments shows an association between individuals' attributes (e.g., education) and their earnings, the relationship is thought to indicate a direct causal link. However, if jobs define the individuals' attributes required for entry and if they define the rewards they confer, then jobs and the job structure may be important determinants of the patterns of compensation and career attainment. While this interpretation does not reduce the value of analyzing individuals' attainments, it does suggest that such analyses must consider to what extent structural factors are contributing to individuals' attainments.
On the central theoretical issue of this article, whether human capital alone is sufficient to explain job salaries, the results are necessarily ambiguous. The concept of human capital includes a diversity of individual attributes, many of which are difficult, if not impossible, to measure. This makes the theory virtually impossible to refute. However, the limited explanatory power of the two human capital variables considered here is important, for these are the two most emphasized variables in the human capital theory and the variables which have the strongest explanatory power in empirical analyses of earnings. Consequently, although the findings here about human capital effects on job value cannot be definitive, they strongly suggest that human capital composition has a minor impact in determining job value.
The structural contention about the job status effect on job salaries is unambiguously answered by these analyses. The impact of job status is complete and leaves very little room for other factors to have separate influences on job salaries. The compensation for jobs is nearly totally determined by the status to which the jobs are allocated, and other factors have minimal additional impact. The analyses suggest that the weak effect of human capital and the strong effects of sex and race composition are largely mediated by the job status system, and this is true when job status is determined by traditional status conception and after it was presumably rationalized by the job evaluation system.
Job status explains nearly all the variance in job salaries; it mediates nearly all of the influence of demographic variables and has a large unique influence besides. This strong mediating influence casts considerable doubt on the human capital model; for it suggests that the human capital composition of jobs is not compensated directly, but only insofar as it is reflected in the status of jobs. The salary for a job occupied predominantly by individuals with valuable human capital is determined only by the status value of the job, not by its occupants.
Policy Implications: The Reexamination of Job Status
These empirical findings clearly indicate the need for more careful analysis of the job status system. Efforts directed at revising job status systems are aimed at the right target: the system which is central to the determination of salaries, which perpetuates the negative value of women's jobs, and which has been largely immune from examination. It is a system created by particularistic political influences and based on traditional values which may not reflect current equity norms and may not even be responsive to current economic conditions. Job evaluation, being an effort to reexamine traditional job status and create a more rational system, would seem to be a hopeful reform for creating equal pay for "equivalent" work.
Job evaluation's most impressive accomplishment is the creation of a process of determining job value which bridges the gaps among com-peting advocates and which removes the process from particularistic political considerations. It does this by getting competing interest groups to express their views in terms of numbers, by considering abstract factors instead of concrete jobs, and by using "experts" to reconcile conflicts (Sargent 1972, p. 2-34). However, these procedures also make the factor weights less susceptible to criticism and make their biases difficult to discover.
This risk is not just hypothetical; it may be built into the job evaluation procedure. As Treiman (1979, p. 7) notes, "some job evaluation analysts argue that the principal criterion for the validity of a job evaluation plan should be how closely the job worth hierarchy produced by the plan matches the existing wage hierarchy (Fitzpatrick 1949; Fox1962)." Perceptions of equity, or standards of work value, are so variable and changing among people, among professions, and among various organizations that a single organization must develop its own set of value standards and sustain those standards over a period of time if they are to have meaning and are to influence perceptions of equity. Thus, the organization should use caution in attacking or changing a long-established set of job relationships, regardless of the source of the equity perceptions which originally generated the relationships. (Wing 1972, p. 2-21)
Experimental studies find that, while compensation specialists are not influenced by the percentage of females holding a job, they are substantially influenced by the current salary reported (Schwab 1983). Consequently, the present findings of remarkable stability of job status and job salaries over the 13 years spanning traditional and job evaluation systems indicate a serious defect in job evaluation as a way of remedying past discrimination. Rather than being a means for reforming traditional practices; job evaluation may be preserving and legitimating these traditions and making them less subject to criticism.
The present findings have several implications for assessing the potential value of job evaluation for advancing affirmative action goals. First, they indicate the crucial importance of formulating a coherent and fair job evaluation system if reform efforts are to endure. If the present findings about the enormous numbers of new jobs which appear over time are generalizable, then one-shot efforts at reevaluating jobs will have only short-term benefits as old jobs disappear and others are created. An enduring system for fairly evaluating jobs is necessary if reforms are to have persisting effects.
Second, reform efforts to implement job evaluation, even in the context of a strong affirmative action program, may not overcome the salary disadvantage of female jobs. The conclusion to the interim report of the Committee on Occupational Classification warned that job evaluation procedures use several procedures which raise questions about whether they can be effective in fairly assessing sex-segregated jobs (Treiman 1979, p. 48). In particular, their reliance on market wage rates and it should be noted that these findings indicate the emptying and filling of particular job titles as defined by the corporation. Organizational policy in this corporation states that when a job acquires or loses one or more significant tasks, a new job title is assigned. How small a change can constitute a "significant task"? Official policy is not explicit. One of the managers in charge of making such decisions stressed that his office looks at requests for job redefinitions critically in order to guard against arbitrary changes being made solely to increase salaries. However, it is just this kind of subjective judgmental processes which have contributed to past discrimination.
