Some final observations from Paula Stephan’s provocative book, How Economics Shapes Science (Harvard University Press, 2012):
1) The current incentive structure is creating an inefficient system. The job market for biomedical PhDs has been generally poor for some time now, and it has been getting worse. From the perspective of Deans and established investigators, the system is working beautifully because established scientists are highly productive. But from an economic perspective (and from the perspective of newly trained PhDs), this is a highly inefficient system that relies on cheap, temporary, highly skilled workers with future job prospects that are unlikely to repay the opportunity costs of PhD and postdoc training.
The university research system has a tendency to produce more scientists end engineers than can possibly find jobs as independent researchers. In most fields, the the percentage of recently trained PhDs holding faculty positions is half or less than what it was thirty-three years ago; the percentage holding postdoc positions and non-tenure-track positions (including staff scientists) has more than doubled. In the biological sciences it has more than tripled. Industry has been slow to absorb the excess. A growing percentage of new PhDs find themselves unemployed, out of the labor force, or working part time.
Inefficiency arises from the fact that substantial resources have been invested in training these scientists and engineers. The trained have foregone other careers – and the salary that they would have earned – along the way. The public has invested resources in tuition and stipends. If these ‘investments’ are then forced to enter careers that require less training, resources have not been efficiently deployed. Surely there are less expensive ways to train high school science teachers than to turn PhDs who cannot find a research position into teachers. Yet this is exactly what a recent report suggested. Many of these PhDs may not even have characteristics that make them good teachers. Surely there are better ways to create venture capitalists with a knowledge of science than for PhDs to become venture capitalists – or better ways to create journalists who write about science than for PhDs to become journalists. Yet such careers are often put forward as appropriate alternatives for new PhDs. There is also the question of incidence, the term used by economists to refer to who bears the cost. The current system may be “incredibly successful” from the perspective of faculty, as a recent report described it, but at whose cost? It is the PhD students and postdocs who are bearing the cost of the system – and the U.S. taxpayer – not the principal investigators. (p. 230-231)
2) Science graduate programs don’t respond to poor job outcomes. Universities are still able to attract excess students into the system, in part because graduate programs can recruit substantial numbers of international students, but also because graduate programs have no incentive to respond to poor job placement outcomes. Graduate students lack information about those outcomes. Compared with law schools and business schools, which recruit students in large part on the basis of job placement reputation, science graduate programs do a terrible job tracking the career outcomes of their PhD graduates. Most graduates go on to postdocs, and postdoc outcomes are notoriously poorly tracked. But it doesn’t matter for graduate programs, because they recruit graduate students based on the research reputation of the faculty, not on career outcomes. Stipends and paid tuitions are an additional recruitment tool – unlike law or MBA students, science grad students get paid a stipend and don’t have to take out huge loans to pay tuition, which makes them less likely to consider the economics consequences of getting a science PhD.
3) Our current system has evolved to become highly risk-averse. A risk-averse research system is economically inefficient. It also undermines one of the primary justifications for public funding of scientific research: that good basic research is too risky and has a time horizon too long to be produced in sufficient quantities by industry. Unfortunately, funding agencies have managed to set up a system with incentives to perform non-risky, incremental research with short time horizons.
Stephan suggests several reforms, among them:
A) Require universities to provide job placement data as part of grant applications. I’m not sure how this would work for R01s, but it may be appropriate for institutional grant proposals and grants for training.
B) Limit the percentage of faculty salary that can be charged to grants (but change this very, very slowly to avoid catastrophic disruption). Universities need to move away from the ‘shopping mall’ model, in which the University is like a mall owner that rents attractive space to researchers who eventually have to cover most of the costs. In other words, get rid of those economic incentives that lead universities to overproduce lab space and PhDs.
C) Decouple training from research by encouraging more research institutes…
There is more – go read the book.