Estimates of Labor Aggregation Functions

Abstract

This paper first describes the estimation of detailed aggregation functions for labor, using both occupational and educational classifications, following the lines pioneered by Bowles but using more powerful statistical techniques and cross-section data of higher quality (for states within the United States, as opposed to countries) and greater quantity. It then shows how the effect of technical progress on the demand for different kinds of labor may be modeled by means of shift parameters within the aggregation function, and provides estimates for them and the other parameters of the aggregation function using time series data for the United States as a whole. The results assembled in the paper indicate that, for the purpose of estimating total growth in output due to labor, a time-dependent CES function with a fine classification of labor should yield the most accurate predictions, provided that the shift parameters can be estimated. Otherwise, there is little to be gained by going beyond wage-weighted linear aggregation. For educational planners, the most obvious implication of the results is that the manpower requirements approach should be abandoned. They suggest that rate-ofreturn analysis with constant relative wage rates should be a less misleading tool for planning, at least in the short run. However, since such analysis is sensitive to changes in wage differentials, and since these in turn are sensitive to the specification of the aggregation structure, it would be much better still to use the CES function approach to obtain a feedback relationship between the growth of the different levels of the educational system and the rates of return to them, via the growth of the labor force and the structure of wages.