ON LABOR AGGREGATION

抄録

The aggregation of labor has been considered with emphasis on the order of aggregation in the framework of, what I call, a "multi-level CES function." To investigate the order of aggregation among three factors (namely sex, occupation and education), a rigorous statistical test has been attempted. A test procedure has been performed in the framework of linear constraints on parameters of different equations. At the same time, the estimation procedure which was applied is capable of overcoming the problem of serial correlation and contemporaneous correlation in the disturbance terms, and the result seems to be reasonably successful.
The empirical result seems to indicate that the aggregation of labor proceeds by occupation→sex→education in the Japanese manufacturing industries. One economic implication of this result is that enterpreneurs allocate labor input by education at the first-stage. In other words, they decide how many educated or less educated workers should be hired at the first-stage in order to fulfill the total requirements of labor. Next stage allocates by sex, and the final stage allocates by occupation. This allocation procedure ensures the multi-stage maximization procedure which is one of the conditions for aggregation (or grouping). This is equivalent to say that employer ranks three factors by the order of education, sex and occupation in the hiring process. Also, it was found that the most substitutable factor among them is occupation. Next is education. Substitution possibility between male and female is relatively lower.
There are two remaining problems; one is that there may be an overlap between skill (occupation) and education (possibly sex also) in the actual hiring process. If all the classified labors are aggregated only once, this problem does not occur. So long as the multi-stage maximization is a condition for aggregation, this overlap problem may not be easily handled at least in empirical stages. Second is that we ignored age (or experience) of workers as a factor. Age or experience may be influencial in the hiring process. Technically speaking, this problem does not impose any difficulty. These two are the future subjects to be investigated.