« The tortoise and the tortoise | Main | Scientific retractions and competition for jobs »

Perverse incentives in science

Economist Paula Stephans has a stimulating commentary in Nature this week, arguing that "Counterproductive financial incentives divert time and resources from the scientific enterprise."

For example, she says that cash incentives offered in China, South Korea, and Turkey have led to a 46% increase in submissions to Science, but no increase in publications. Officials in offending countries are presumably indifferent to the workload of Science reviewers, so is this really a perverse incentive, from their point of view? It could be, if papers from their country get such a bad reputation that even good papers suffer guilt by association.

She argues that graduate students should be supported mainly by broad training grants rather than research assistantships tied to specific research projects, because the training grants have lower overhead costs (8% versus 50%). This is nonsense. Unless we somehow reduce the actual costs of educating grad students, which would likely reduce the quality of their education, the money has to come from somewhere. Universities subsidize training-grant students by diverting money from 50%-overhead grants to cover the difference between 8% and the actual cost of those students. Eliminate the 50%-overhead grants, and the whole thing falls apart. On the other hand, she may be right that students in training grants get a better education. If so, maybe we should support more training grants, but we'll have to pay for them.

But is the 50% overhead rate fair? It may be too high. I would argue that the socially optimum overhead rate is the rate at which university administrators would accept a grant iff they thought it would increase the university's (or their own) prestige, not just for the money. At 50%, they would probably accept a grant to study almost anything (the healing power of prayer, say), so long as incremental costs of providing facilities for that research were much less than 50% of the grant amount. On the other hand, an administrator at UC Davis told me not to apply for USDA grants, because they only paid 15% overhead. So the fair overhead rate is probably somewhere between 15% and 50%. 8% isn't enough to cover even the marginal costs of research.

Stephans discusses various perverse incentives that make individual labs grow too big -- smaller labs publish more per dollar -- which she attributes to "bonuses" based on external funding. I don't think eliminating such bonuses would much effect. The real problem is that grants tend to go to those who have published the most, not those who have published the most per dollar. So big grants lead to more big grants, a process known as "cumulative advantage." I discuss this problem in more depth in the "Selection Among Ideas" chapter of my forthcoming book, Darwinian Agriculture.

Stephans also discusses perverse incentives that make universities build too many buildings. If they borrow money for new buildings, they can include the interest on the loans in their overhead costs. In addition to this perverse institutional incentive, there may be perverse individual incentives. The administrator who takes credit for the new building puts it on her CV and uses it to get a higher-paid job, leaving the costs of maintaining the building (as state and federal funding declines) to her successors.

The issue of "training more PhDs than there are jobs" is more complex than Stephans implies. What's best for the millions whose taxes support research universities isn't necessarily best for individual students. Click "careers in science" for my past discussions of this topic.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Type the characters you see in the picture above.