Education policy and energy R&D
Posted by Evan Herrnstadt on June 30, 2008
In the run up to America’s first comprehensive piece of climate litigation, the idea of government R&D subsidies has been bandied around quite a bit. As images of the Manhattan Project and the Space Race have been invoked, much of the debate has centered around how much money the feds should be throwing at energy R&D. In this rush to find a magic number, a major question has been overlooked. That being, does increasing R&D funding actually increase the amount of R&D performed?
Salaries for R&D labor comprise over half of total R&D spending, making it by far the dominant input in the process. Although better equipment can augment a laboratory, it really comes down to the quantity and quality of brainpower in the room. Consider the labor market for scientists and engineers, wage versus total employment.
Now suppose R&D expenditures are subsidized or given special tax treatment, resulting in an outward shift in S&E labor demand. If supply of labor is perfectly elastic (top), we would see more R&D activity accompanied by constant S&E wages. This outcome would be considered a rousing success, as all the new R&D money went toward more R&D. However, if the labor supply is perfectly inelastic (bottom), all the additional funds will go to increased wages. Austan Goolsbee ( AER, 1998 ) attempts to explore this question econometrically and finds that, in the short run, the world is somewhere in between, but inelastic enough that we should take pause. His results suggest that the 11% increase in federal R&D spending from 1980-1984 increased the salaries of S&E by 3.3% on aggregate — in terms of energy-relevant fields, the numbers are: physicists 6.2%, mechanical engineers over 4%, electrical engineers 2%.
The upside of these increased wages is that perhaps they will affect long-run employment decisions. If an entering undergraduate student is given information that a particular field (e.g. high-energy physics) is paying well these days, he/she might be more likely to enter that discipline. However, although there is concern about potentially binding level of S&E, we still hear about the bruisingly competitive buyers market for new PhD’s. More and more, a PhD must spend time in a postdoctoral position before entering academia as a professor. However, this simply indicates that PhD programs may be training and/or selecting for students who myopically pursue an academic career.
Paul Romer ( NBER WP #7723, 2000 ) discusses some potential policies in the context of increasing S&E training to bolster the greater macroeconomy.
- Provide training grants to undergraduate insitutions designed to increase the proportion of students receiving S&E degrees.
- Create a system of objective tests to measure mastery of such areas.
- Fund a new class of fellowships to fund graduate-level study.
The first suggestion seems straightforward enough: provide financial incentives to students for choosing the desired fields. However, Romer points out that this might simply lead to reclassification of majors (e.g. environmental studies + one ecology course = environmental science). Also, there are additionality problems, i.e. universities might simply reduce the number of institutional scholarships they provide.
The second policy is designed to alleviate signaling problems by providing graduate institutions and firms with more information. It’s hard to tell how a 4.0 from the University of Florida stacks up against a 3.6 from University of Chicago. Maybe favorably, maybe not. Luckily, we’re talking about some of the most easily-objectively tested subjects in academics. Literary ability is tough to quantify. Ability to solve a physics problem is less so.
Finally, the third suggestion is my favorite, especially from an economist/policy analyst standpoint. Romer proposes taking a random sample from a subset of promising entering undergrads, such as those passing AP Calc, and offering them a portable fellowship if they enter graduate school in S&E. This policy will give students a chance to take appropriate prerequisites during undergrad and might lead some students considering S&E to go ahead and try it out. The random nature of the allocation could be exploited to see if the policy makes a systematic difference in decisionmaking among rising students.
These policies are designed to affect the overall proportion of S&E in the workforce and level of R&D output for the economy. For the narrower energy case, they may need to be tweaked. The reason is that we are trying to get more energy R&D without depriving the rest of economy of innovation. If we start massively subsidizing energy R&D, we may see some crowding-out effect in which S&E choose to work on energy R&D instead of other R&D they would have otherwise performed.
We can begin thinking about this in terms of pure R&D fund crowdout. David Popp recently presented ongoing work with Richard Newell to investigate potential crowding out by environmentally-friendly R&D at the EPA. Their preliminary results fail to suggest that R&D in energy sectors crowds out R&D in other sectors. They also find that there is some evidence that energy R&D crowds out other R&D within energy sectors as well as amongst alternative energy firms. If this is an issue, it increases the opportunity cost of energy R&D. However, it’s difficult to say what level of crowding out is acceptable. As the cost of not doing energy R&D increases, crowdout becomes less important, relatively speaking. Perhaps our world is moving in a direction in which we need to shift some R&D resources to energy, even if it means sacrificing some other type of innovation.