Jennifer Ma, TIAA-CREF Institute and
Paula E. Stephan, Andrew Young School of Policy Studies, Georgia State University
April 2004 |
U.S. research universities are increasingly populated by postdoctoral fellows. This trend can be explained by both the increasing number of new Ph.D.s taking a first postdoc position and a lengthening of the duration of the postdoctoral position. In this paper, we first document the growing number of postdoctoral positions and the lengthening duration of these positions at U.S. research universities, drawing data from the biennial Survey of Doctorate Recipients (SDR). We then study factors that affect the duration of time spent in postdoctoral appointments by estimating a hazard model of postdoctoral duration for individuals holding one or more postdoctoral positions.
Our analysis of duration will pay special attention to how the supply of new Ph.D.s and the condition of the academic labor market affects postdoc duration. A goal of the study is to inform the policy debate surrounding the “postdoc issue.” Of particular interest is:
- What is the impact of the supply of new Ph.D.s on the duration of postdoc positions?
- What is the interaction between the condition of the academic labor market and the duration of postdoc positions? In particular, what do the changing hiring patterns in higher education imply for the duration of postdocs?
- How does duration relate to research opportunities outside academe, especially in R&D labs?
Our underlying assumption is that the current market for recent Ph.D.s is dysfunctional and has led to a situation that creates increased tension between universities and young researchers. On one hand, the demand for postdoctorates is fueled primarily by the U.S. federal research budget, a budget largely outside the domain of universities and a budget that has become increasingly large. This growth in federal research budget contributed to increasing numbers of Ph.D. recipients and postdoctoral positions. On the other hand, the university system has few permanent slots for hiring the growing supply of the research-trained doctoral workforce. And, while industrial employment has grown for the doctoral trained in recent years, growth has not been sufficient to provide research positions for all of those with research training, creating what a recent National Research Council report (1998) called a "crisis of expectations."
Data for this study will be drawn from (1) the Survey of Doctorate Recipients (SDR), a biennial survey of individuals who earned their Ph.D. in the United States, and (2) the Survey of Earned Doctorates (SED), a census of all individuals receiving a research doctorate from a U.S. institution, and (3) Integrated Postsecondary Education Data System (IPEDS) Fall Staff Surveys, a biennial survey of postsecondary faculty and staff.
The SDR is a biennial longitudinal survey of doctoral recipients awarded in the U.S. for the period 1973-2001. It is administered by Science Resources Statistics (SRS) of the National Science Foundation (NSF). The sampling frame is the Survey of Earned Doctorates (SED), the census of all Ph.D.s awarded in the United States, which is also administered by SRS and linked to records in the SDR. The longitudinal nature of the survey permits one to study individuals over time, as they move from one position to another. Moreover, the retrospective questions on career history that were included as a special module in the 1995 survey, permit a more intensive study of duration for 1995 respondents. A particularly attractive component of the 1995 special module is the inclusion of questions concerning the type of fringe benefits that were available to the postdoc.
Covariates can also be constructed from the SED to measure labor market conditions by field by year.