ȁȂȅ Partʺ: Economics“Why?” the student asked. “I don’t know,” was the reported reply, “you’ve
just got to have a model.” Ļe other advisor reportedly went on to say that
if the student expected to get his dissertation past certain members of the
department, he would have to do work of the kind they expected.
Such sermonizing seldom appears in print and fully articulated, which
is why it can be so insidious. Although influential, it escapes critical exam-
ination. I wish economists would drag it into the open by recounting their
experiences with it. (Mayer makes a good beginning in hisȀȈȈȂ, chap.Ȉ,
entitled “Model or Die.”)
One little episode involved me directly. During the discussion period
at a conference, I remarked that a particular monetary reform would elim-
inate the contagion of bank runs, and I briefly explained why. During the
further discussion, and at greater length during the coffee break, another
conferee objected that if I and my coauthor expected anyone to understand
what we were saying, we would have to argue in the context of a model
of bank runs, complete with specification ofȀǿǿpersons,ȃȆcommodities
(or whatever the numbers might be), and so forth. If I had thought fast on
my feet, I would have pressed the question “Why?”. I would have asked
my interlocutor to make his methodological sermon explicit and support
it with reasons. Unfortunately, the conversation wandered off.
Months later, in conversation with Donald McCloskey, I wondered
about the claims of some economists not to understand arguments pre-
sented outside of formal models. McCloskey conjectured that they mean
what they say: some of them are so wrapped up in their own models
and favorite symbols that they actually cannot understand arguments pre-
sented in an unexpected language, English.
Peter N. Ireland (ȀȈȈȃ) provides another example of what bothers me.
After making sensible remarks about relations between money and eco-
nomic growth, he goes on to give his argument supposed rigor with mathe-
matics and numerical simulations. He elaborates a model of a large number
of identical, infinitely lived households possessing perfect foresight. Each
consists of a worker and a shopper. Production functions have specific spe-
cial properties. Perfect competition prevails. A cash-in-advance constraint
applies to purchases made without the assistance of a financial intermedi-
ary. Yet if these and other special assumptions spelled out in great detail
(about transactions costs and so forth) are not necessary for the conclusions
reached, what is the point of making them? And if theyarenecessary, is it
not a great lapse from the trumpeted rigor to convey the impression that
the conclusions reached apply to the messy real world anyway?