Doctoral production, supply and demand: A dynamic simulation of the academic labor market for humanities Ph.D.'s.

Abstract

Doctoral production in the humanities increased steadily throughout the 1980's and 1990's, even as the academic employment prospects for Ph.D.'s deteriorated. This created the appearance of universities, responsible for both the supply and the bulk of demand for Ph.D.'s, manipulating the system to maintain a surplus of academic labor. Yet, researchers have tended to employ occupational choice, wage and human capital theories to explain the supply of educated workers, ignoring institutional behavior in the universities that produce Ph.D.'s. An exception is Goldman & Massy's (2001) model of the academic labor market in science and engineering, in which the supply of Ph.D.'s was shown to be driven not by the market for graduates, but by the market for graduate students: the need for TAs and RAs. This study seeks to adapt the Goldman & Massy model to the humanities, viewing Ph.D. labor markets as systems where institutional behavior plays as important a role as individual choice. The economic behavior of English, history and foreign language departments is characterized as the maximization of faculty utility and prestige through cross-subsidization of research and graduate education with profits from undergraduate education. The allocation of instructional demand among full time, part time and TA labor is simulated as a process of utility maximization subject to budget and production constraints using nonlinear optimization techniques. Cross-sectional data are taken from the 1999 program staffing surveys of the Coalition on the Academic Workforce, leading to simulation of Ph.D. production and employment using a dynamic market outcomes model. The financial regressions produced predictable results for the effects of enrollments at the BA, MA and doctoral levels on departmental revenues. The regression results for the structural equations, however, indicated that key parameters were either not significant or of the wrong sign, contra-indicating the theoretical framework. The conclusion reached is that the available data do not support the model constructed. Analysis of the shortcomings of the data and the model leads to recommendations for improved data collection and modeling methods, as well as insights into how the academic production system in the humanities differs from that in science and engineering disciplines.