The latest Critical Review is a full-issue symposium on The Myth of the Rational Voter. Stephen Bennett and CR editor Jeff Friedman lead off with a novella-length frontal assault on the book. Then there's seven more critiques from David Colander, Elster & Landmore, Gerald Gaus, Kiewiet & Mattozzi, Arjo Klamer, Paul Quirk, and Wittman. I then respond to one and all. My favorite parts:
From my reply to Bennett and Friedman
We don't need to know everything about something to estimate
the value of learning more about it. In many cases, we barely need to know
anything at all. For example, almost everyone deletes spam email unread. If we
accept BF's argument, this practice is a deep philosophical error. After all,
how do we know that a given piece of spam isn't worth our time unless we
actually read it? Indeed, unless we send him the money, how do we know that a
random Nigerian's offer to split an inheritance isn't the path to untold riches?
N.B. Jeff Friedman's not too happy with my interpretation on his position. If he likes, I'd be happy to post his response to my response on EconLog.
From my reply to Gaus
Gaus has high standards for useful advice. First, useful advice requires detailed empirical study: "Unless policy
makers have a good way to predict the magnitude of the costs and benefits, they
cannot have a sound basis for advocating or opposing the policy" (Gaus 2008,
15). Second, useful advice requires detailed empirical study of many things: Advisors must "factor in
all the relevant, real-world variables" (ibid., 10) as well as "the complex
interaction of different variables" (ibid., 12)
By Gaus's standards, economists (and experts generally) do not measure up. Even on relatively simple questions, "the
experts' poor absolute levels of prediction . . . is really striking" (Gaus
2008, 17). If you take the problems of omitted variables and interaction
effects seriously, good advice is effectively impossible: "We seldom if ever
know all the relevant variables" (ibid., 10), and even if we did know all the
relevant variables, "it is usually impossible to predict how they will interact"
Gaus's arguments prove too much. They don't just rule out useful policy advice; they rule out useful advice of any kind. For
example, suppose a student asks me if he should go to grad school. By Gaus's
standards, to "have a sound basis" for my advice, I would need fairly precise
estimates of the magnitude of the costs and benefits of all the relevant, real-world variables and their interactions. I
don't have that. No one does.
I take this to be a reductio ad absurdum: If your standards imply that useful advice is impossible, something is wrong
with your standards.
From my reply to Wittman
Wittman could grant all these points, but
respond, "I'm still not convinced that the disagreement between economists and
the public is big." To get some perspective, I calculated belief gaps between
laymen of the far left (very liberal Democrats) and laymen of the far right
(very conservative Republicans). For the SAEE's 37 questions, the average
absolute value of this belief gap is .30 on a 0-2 scale. I also calculated belief gaps between Ph.D. economists and the general public.
All else equal, the average absolute value of this belief gap is .52. In other
words, the belief gap between economists and the public is more than 70 percent larger than the belief gap
between America's far left and far right. If that isn't big, what is?
If a steady diet of snack-sized blog debates have left you hungry for heartier fare, the latest issue of CR is for you. But you'll probably have to actually buy the journal to partake - it's not online.