Background and previous literature

66 T.C. Buchmueller et al. Economics of Education Review 14 1999 65–77 itions and, conditional on job placement, with a higher rate of publication.

2. Background and previous literature

2.1. Educational production functions The concept of an educational production function provides a useful framework for studying the effective- ness of graduate education. Educational production func- tions relate educational outputs to the inputs in the edu- cational process. Inputs include the raw materials incoming graduate students, labor faculty members and support staff and capital classroom, computing, and library resources. The goal of this analysis is to learn how the productivity of Ph.D. economists varies with attributes of the graduate training process. To succeed, we must be able to measure not only the inputs to the production process, but also the outputs. Although some educators deny the legitimacy of any quantitative measure of educational outputs, 1 the unde- niable emphasis of economics graduate programs on training students to publish articles eases the task some- what. Hogan 1981, in perhaps the only prior attempt to estimate the relationship between economic education inputs and outputs, justifies his publications-based output measure by the use of “publishing performance” in pro- motion, tenure, and salary decisions in the academic sec- tor. We also employ publications of graduates as a meas- ure of program effectiveness. 2 Suppose we accept published research as the appropri- ate yardstick for assessing graduate training effective- ness. What are the limits of such an analysis? The pri- mary problem we see is the strong relationship between publications and the job one holds. An economist’s pub- lication record will be greatly affected by the resources and incentives for publishing provided by the job. For example, an economist employed at a leading research university will be offered greater supporting resources e.g. research assistance, computing services, and a lighter teaching load and will obtain greater rewards e.g. salary and tenure for publishing in economic jour- nals than will an economist of equal productive capacity who happens to be employed at a Wall Street investment firm or a small liberal arts college. Thus, a structural 1 See Hanushek 1986 for a discussion of this issue. 2 Obviously, increasing the publishing capacity of students is not the only desirable trait of graduate training. Only the harsh- est critics of graduate education would deny that programs attempt to impart teaching skills. Unfortunately, we cannot measure this output with the available data. Nor can the earn- ings of graduates be utilized as a measure of program effective- ness for the simple reason that earnings data for the COGEE survey sample are not available. analysis of the effect of graduate training on individual productivity requires information on both job character- istics and the selection process into these jobs. 2.2. Previous empirical findings While many social scientists have studied the determi- nants of scientific productivity in a variety of disciplines, relatively little research has been done to link graduate program attributes to subsequent productivity. We dis- cuss several aspects of this literature, concentrating on research that estimates the direct and indirect effects of graduate training on subsequent research productivity. What we label the direct effect of graduate training on publishing proficiency is the accumulation of human capital associated with the increased capacity of the stud- ent to publish. The indirect effect is the increased utiliz- ation of this human capital via placement in a job that supports and rewards publishing. Hogan 1981 examines the relationships between inputs in graduate education — the quality and number of faculty members and students — and the output- graduates’ average rate of publication in four top eco- nomics journals. He finds a positive and statistically sig- nificant effect of faculty publishing performance on stu- dents’ subsequent publication rates and argues that this result provides “indirect evidence of jointness in the pro- duction of graduate training and research output.” That is, by sharing the research experience with students, fac- ulty members produce published articles and simul- taneously impart research skills to students. Hansen et al. 1978 estimate the relationship between the graduate department’s “quality” and the publishing proficiency of its graduates. For a sample of economists employed at Ph.D.-granting institutions in 1966, they estimate a system of three simultaneous equations that determine job placement, publishing proficiency, and earnings. As expected, graduate department quality as measured by the Cartter, 1966, index has a positive effect on placement quality. Controlling for job place- ment, however, Hansen et al. find a negative relationship between research output and graduate department qual- ity. They argue that this counter-intuitive finding may result from the omission of any measure of the quality of publications, such as journal rank or number of citations. Long 1978, who studies research productivity in the discipline of biochemistry, tries to separate the direct effect of training from the indirect effect working through job placement. Because of superior data, Long is better able to isolate the two processes. Controlling for prior publications and job quality, he estimates the effect of various training-related variables on research output. Whereas Hansen et al. were able to use only the departmental ranking as an educational input, Long also includes the research output of the dissertation super- visor, a dichotomous variable indicating whether or not 67 T.C. Buchmueller et al. Economics of Education Review 14 1999 65–77 the student collaborated with this individual, and a meas- ure of the “selectivity” of the school from which the baccalaureate was obtained. Each graduate education variable has the expected relationship with the quality of the first post-Ph.D. job as well as with subsequent research productivity. Although prior-to-Ph.D. research output shows a strong, positive effect on subsequent out- put, it is only weakly related to job placement. Long’s results are essentially corroborated by Long and McGinnis 1981.

3. Empirical specification