Alas, just when I thought I might actually get some work done, I
have been summoned to make an appearance on PEN-L. I have no idea
whether I delivered a knockout blow to the AD curve (or AS--read
on), but, instead of referring people to the archive, maybe it
will be a public service to repeat my posting of a year or so ago.
BTW, at the moment I am not teaching macro (or any conventional
economics, for that matter) so my blood pressure has dropped a bit
on this topic. What follows is my higher-pressure state of mind
circa 1992-93. [Original post follows.]
Just for openers, here are a few of the problems I have with AS-
AD:
1. To begin with, the price axis is mislabled: it should measure
the *rate of change* in prices, not the price level itself. True,
the logic (such as it is) of AS-AD should still hold if all the
variables are given their dots, but the mislabling is symptomatic
of the general sloppiness surrounding this model.
2. AD: Yes, if the nominal money supply is held constant AND if
the price level rises, then, ceteris paribus, real activity must
fall. But the assumptions packed into this claim render the curve
irrelevant if not impossible:
A1: Nominal MS is exogenous. No it isn't. I am willing to accept
the idea that there are constraints on the endogenous creation of
monetary assets, but admitting any endogeneity throws the AD
argument into disarray.
A2: Nominal MS can be changed while all else is held constant.
NO! First, this is wrong as a historical claim, since in every
period nominal MS is not constant, yet the AD curve is applied to
explanations of historical change. To put it differently, EVERY
time nominal MS changes a new AD curve should be drawn, but who
ever does this, and why would we want an apparatus that forces us
to do this? Second, from a logical standpoint the very change in
nominal MS changes other variables that in turn impinge on both P
and Y. For instance, nominal interest rates and long-term con
tracts are sensitive to perceptions of central bank activity. So,
as they say, the ceteris are not paribus.
Instead of focusing on money assets only, of course, one can drag
in the whole Pigou effect. But the real stock of nonmonetary
assets, if their prices reflect capitalized flows, should remain
unaffected. (Or, more precisely, the effect is not systematic,
since the flows to asset owners depend on the level of activity,
and once again the relationship between P and Y is underdeter-
mined.)
3. But what really galls me is not AD but AS--why should firms
want to supply more when prices rise? Above all, this should
depend on the time structure of contracts: are their contracts
with suppliers more or less flexible in nominal terms than their
contracts with customers? This is an empirical question, but it
is typical of the AS-AD "literature" that no one has bothered to
investigate it. One would expect that the supply behavior of
firms would differ from one industry to the next...
4. The whole AS-AD approach undermines the circular flow model
which we struggle to convey to students. It is a real insight to
see that, in the aggregate, incomes and expenditures are two sides
of the same coin, and that, by itself, a change in the price level
holds no implication for real activity. What AS-AD does is to
make a large number of questionable, or at best highly contingent,
assumptions and then create the impression that prices and activ
ity are logically related, ever and always. Bad thinking driving
out good.
So the question arises, in light of these arguments, which are
accessible to any inquisitive undergraduate, why has the profes
sion jumped on the AS-AD bandwagon? My guess is this: if you
construct a diagram with real output on one axis and prices on the
other (or, better, rates of change of these variables), and if you
introduce one upward-sloping curve and one downward-sloping curve,
you can "explain" any possible event or sequence of events in the
economy. In just a half hour or so and with virtually no intel
lectual effort--indeed, as I have argued, with the need to *avoid*
intellectual effort--you can create a model to occupy your stu
dents and pretend that the economic gyrations of the past two
decades remain within the purview of your expertise. This is
extremely convenient, especially when you consider the alterna
tive: trying to explain why some inflations seem to be associated
with excess demand, others with stagnation, and why we have
difficulty predicting which will be which. So, in the final
analysis, I think most teachers in intro courses are simply lazy--
the primordial sin, according to Kafka. If this is true, the
willingness of economists to cut corners on the truth in order to
make their jobs a little easier constitutes an intellectual
scandal.
Personally, I don't teach this stuff at all. I do circular
flows and Keynesian crosses, a little standard algebra, some
financial market analysis, and offer different scenarios for
changes in the price level. I don't assign the chapters in the
text that use AS-AD, and so a criterion for me is whether a text
integrates AS-AD or just sticks it into a chapter or two. For now
I use Dornbusch, Fischer, & Smalensee because they quarantine AS-
AD. By the way, here is a little anecdote, which may be of
interest. When D,F,&S first came out the publisher put on a
little show at the meetings. There was Stanley Fischer, auto
graphing copies of his book. I came up to him and said I had
anticipated his intro text since I had used the intermediate book
(Dornbusch & Fischer) and thought it OK. (That may be stretching
it, but that's what I said.) But, I added, why does the intro
book have a chapter on AS-AD? Why not, asked Fischer? Well, I
went on, the intermediate book has a long section that demolishes
the basis for the AD curve. (I cited chapter & verse.) Fischer
looked both ways rather furtively, then turned back to me and said
in a low voice, "The publisher made us do it."
Yuck.
Peter Dorman
