Re: Interest Rates and inflation

["Alan G. Isaac" <aisaac@xxxxxxxxxxxx>]

On Thu, 21 Dec 2000, John O'Donnell wrote:
> Perhaps you'd like to explain just what use the Phillips curve is?
> As often as it is brought forward as such a major discovery surely
> there must be some use.

Since pedagogy is my profession, I can hardly refuse
such a request.

I'll briefly discuss the contemporary ``price'' version
of the Phillips curve, since that seems to be your
main interest. The key message economists communicate
with this is that we should *NOT* expect a to find a
correlation between inflation and unemployment in the
data, since *any* inflation rate is compatible with
*any* unemployment rate.  This message should be
communicated in any macro principles course.
Related to this, mainstream economists use the
basic reasoning to bolster their case for the
natural rate hypothesis in some form.

Textbook treatments generally focus on expectations
as the reason why we do not expect to find a
correlation between inflation and unemployment in
the data. Changes in inflation ``shift'' the
Phillips curve.  Indeed, if one uses survey expectations
on inflation, one can get pretty good empirical
Phillips curves (for the US).  More common in
PC estimations, however, is to use past inflation
as a proxy for expected inflation.  This is why it
is correct to say that the predicted correlation from
the Phillips curve is between the change in inflation
and the level of unemployment. (Even here the empirical
work needs to be informed by common sense: don't expect
to see this correlation in raw monthly data.)

It is fair to say that no serious empirical researcher believes
that s/he should empirically implement the Phillips curve
by looking for a simple correlation between inflation rates and
unemployment rates, against your apparent belief.
The best representative of the ``policy makers view'' of
the Phillips curve is probably Robert Gordon.
For Gordon, the Phillips curve literature is simply an
attempt to describe price dynamics in the US.
(I think this is a pretty fair characterization.)
The concern is with the **proximate** determinants of
inflation, **not* with the ultimate determinants of
inflation.  Gordon would certainly accept that one
does not see large, sustained increases in prices in
the absence of large, sustained increases in money.
(See below.)

You seem to think economists treat the Phillips curve
as refuting the natural rate hypothesis.
In fact almost the exact opposite is true: most economists
who take the Phillips curve seriously also adhere to some
version of the natural rate theory. (Of course natural
rate theory is not popular in PK circles, and for some
good reasons, but that is a different discussion.)
For example, in justly famous articles, Sargent (1971 JMCB)
showed why standard Phillips curve estimates do not offer
evidence against the natural rate hypothesis
and Lucas (1972) showed why the correlations that led to
the Phillips curve can be far from accidental without
in the least refuting the natural rate hypothesis.

The reference to Lucas points to a serious split in
the Phillips curve literature.  The Lucas story is
that surprise inflation leads to temporary unemployment
declines, so that (surprise) inflation comes first
causally.  It is an equlibrium story.
The traditional story in contrast is a disequilibrium
story: labor market conditions are proximate determinants
of wage movements, which in turn are proximate determinants
of price movements. The traditional story is the one that
motivates the presence of a Phillips curve in many PK
models (and in many macro forecasting models).
And indeed it is a very reasonable story.
In these models too, it is worth noting, the Phillips
curve is a description only of the **proximate**
determinants of inflation.  Generally speaking,
there is no implication of long-run inflation-unemployment
trade-off. Nor is there any implication that we will
see large, sustained changes in prices in the absence
of large, sustained changes in the money supply.

>> Please note the words ``large'' and ``sustained'' in
>> my statement. Then look at the US data (freely available
>> on FRED).

> Grab at straws, don't dare admit you may have simply
> misspoke.

Far from it; I said what I meant.
For cross country evidence see the article by
McCandless, George T., Jr., and Warren E. Weber, 1995
 http://research.mpls.frb.fed.us/research/qr/qr1931.pdf
(I am just pointing to their summary of the stylized facts,
not advocating their causal analysis.)
And the US data, as I noted before, is freely available
at FRED
 http://www.stls.frb.org/fred/index.html
There are many ways one might approach the data, but
to get started take a look at a 5-year moving average
of M2 growth and CPI inflation rates.
You will find your characterization of the 1970s to
be quite misguided.

Alan Isaac