I had an exchange with Peter Klein the other day on public funding of science. He had a post up criticizing an APSA email for allegedly ignoring social science findings and methods. It's true they have a weak experimental design (it's a mass email - there's no experimental design!) so I guess I have to agree with him on the methods point. But their plea was very much in line with social science findings. The post raises good points, of course, just not entirely valid criticisms of the APSA. And of course if you don't agree with Klein you haven't read Bastiat (I think this cliche - can I call it a cliche at this point? - is strategic... it would be silly to say "you haven't read Friedman" or "you haven't read Hayek" because most social scientists actually have. By saying "you haven't read Bastiat" you're more likely to get a "who is Bastiat?" response and feel vindicated because he's more obscure. It's kind of silly, though. You can get these points about opportunity costs and unintended consequences from Samuelson for God's sake. But an Austrian saying "you clearly haven't read Samuelson" doesn't have quite the same effect!).
I comment on a specific point he made about some oft-cited research by Austan Goolsbee:
"You have quite high expectations of research design in mass emails!
Part of the issue with Goolsbee is that he’s looking at relatively
short term impacts. Economists studying this labor market since at least
the 50s have noted that one of its most important features is a high
short-run inelasticity of supply. So yes, in the short run you have much
higher wage effects than quantity effects.
This doesn’t hold true in the long run. It’s a good study of
Goolsbee’s, consistent with the rest of the literature, but I think it’s
often poorly (and for that reason over-) cited."
He responds, again with the Samuelson... errr... Bastiat point:
"Right, but I’m talking about the within-scientist effect. Take the
supply of scientists as fixed in the short run. The issue is whether a
grant induces the researcher to work on a project he otherwise wouldn’t
have worked on, or simply increases his compensation for working on the
projects he was already doing or planning to do (salary supplement,
equipment and travel funding that benefits his other activities, etc.).
APSA seems to assume that the former effect dominates the latter (or
that the latter effect is zero). I have plenty of first-hand evidence
that the latter effect is common."
And I clarify:
"But that’s precisely why we use markets for scientific labor that do
respond in the long run. Certainly people will shift what they’re
working on. That’s largely the point of bidding up wages, right!
If there is something they would have worked on in the absence of the
grant that has a lower value than the stuff they would work on on the
grant we don’t want them working on it. We want to bid them away.
The trouble is, we’d like a sure method of valuing science that has
Unfortunately political allocation can’t give us that. Unfortunately, the market can’t either.
So we’ve gotta do our best. But I don’t see why it’s a bad thing to bid people off other projects.
And again – insofar as their is bidding off other work, it’s a short-term
phenomenon. In the long-run, so long as we do our best to approximate
the social value of this research, the bidding away that goes on is
bidding away that we’d like to see happen.
I think the issue ultimately should be attaching the right social
value to publicly funded research, not being concerned about short-run
inelasticities that Goolsbee’s work picks up. If we get the valuation
right none of the bidding away should bother us.
And I think good social science tells us two things: positive
externalities are underfunded by the market and corner solutions are
rarely optimal except in some unusual circumstances. So I think good
social science is on the APSA’s side here. Of course there’s a lot more
legwork to do to determine the right amount of funding."
Using evidence in an "unthinking" way
5 hours ago