I don't see how the article can be interpreted as "not attacking
econometric methods." The article starts off by referring to such methods
as "junk science," follows by arguing that econometricians can obtain any
result they wish by arbitrary manipulations, and finishes with a lament
that academics have no procedure to acknowledge the "failure" of
econometrics. If this doesn't constitue an "attack," it's hard to see
what would. While I certainly agree we should be skeptical of
"junk science," we ought to be similarly skeptical of "junk critiques"
of good science.
Goertzel's piece is of course much more extreme than McCloskey's. There
is some truth to the latter I think -- I have been frustrated after having
papers rejected because referees though my results clashed with their
priors. But one must concede it is rational to have fairly strong priors
in many contexts -- if I estimated, say, that higher education causes
lower incomes, I'd start wondering how I screwed up the statistics rather
than start wondering why education suppresses income. But if it were true
that no one ever believed anything that was in conflict with "blackboard
economics" as McCloskey sometimes puts it, there would be no "empirical
puzzles" in the discipline. Since we can all think of many such puzzles,
our priors are not as strong as McCloskey makes out. Unexpected results
also produce attempts to replicate using alternate data and models,
although notice a hallmark of good research is reporting many different
specifications and attention to robustness.
As for Goertzel's more shrill complaints, well, if he thinks quantitative
work proceeds in the manner he describes, perhaps he should write a paper
or two in that manner and see how they fare at refereed journals.
Cheers,
Chris Auld
Department of Economics
University of Calgary
[EMAIL PROTECTED]
