Econometrics is neither the source of
scientific nirvana nor the ultimate in deluded
alienation. It has its usefulness if kept in
perspective. By itself it neither proves nor
disproves anything. However, just as repeated
findings pointing in a certain direction by
experimental economists, such as that of the
asymmetry between willingness to pay for
improvements and willingness to accept
compensation for degradations, eventually
come to influence the views of the profession and
even of policymakers, likewise repeated findings
by econometric studies without contradicting
findings have similar influence.
Now, I find it a bit amusing that Doug Henwood
has climbed the barricades against econometrics.
It is one thing to criticize all empirical approaches to
economics, and there are serious philosophical
grounds for doing so. But Doug is our ultimate
datameister/wonk. He presumably takes empirical
observation seriously. However, he wishes to restrict
the analysis of such data to eyeballing tables or maybe
at the most scatter diagrams. As I have already noted,
such approaches can be very misleading and are essentially
equivalent to applying very rudimentary econometric tests,
such as bivariate OLS regressions.
Now, there is a horrible fascination in the profession
with ever more "sophisticated" econometric methods. Much
of this reflects the game of publishing to avoid perishing.
Thus, every time a new technique pops up, such as cointegration
(to Di Nardo, I agree with Peter Dorman that checking that one
out more thoroughly would be useful), a bunch of fresh-out-of-
grad-school-and-seeking-tenure economists begin applying
it to all kinds of already-tortured-past-death (but updated
slightly) data sets to study all kinds of also already beaten
to death old questions. Ah ha! Publications! Tenure!
Actually I think that Doug is at least partly right. Whenever
there is a real shift in policy or broad viewpoint it is generally
because something has happened in the data that is so
screamingly obvious that it does not need any fancy econometric
tests to establish it. If it can't be seen by the eyeball, the Board
of Governors will not believe it.
Indeed, the issue of monetarism is a good example. There
was a long and gory econometric debate over "Keynesian fiscal
policy" versus "Friedmanite monetarism" back in the 1960s,
using the then currently fashionable sets of supposedly advanced
econometric methods (sorry to whomever asked, I don't have the
exact cites), mostly in leading mainstream journals. Donald
Hester was a leader on the "Keynesian" side while Friedman
with his former grad student, David Meiselman, were leaders
on the monetarist side. The issue was: which is more stable,
the marginal propensity to consume or the velocity of money
(M1)? The debate was long and essentially inconclusive.
Then in 1968 LBJ attempted to fight inflationary pressures
by instituting a one-year income tax surcharge. It failed and this
was used by observers to declare a victory for Friedman and a
defeat for Keynes. In the same year, Friedman gave his AEA
address on the "natural rate of unemployment" which has since
uttlerly infiltrated all the macro textbooks, and which fed easily
into the new classical/rational expectations movement that took
over much of mainstream macroeconomics in the 1970s, further
stimulated by the oil price shock inflation.
Of course, ironically, Lucas actually explained why the LBJ tax
increase did not stifle inflation. Policymakers relying on a
structural econometric model for policy forecasting cannot do
so if their actions change people's expectations and thus
(in effect) change the values of the coefficients in the models.
The announcement that the tax increase was only temporary
had such an expectational effect, and one useful outcome of
the whole ratex exercise was indeed to make economists think
more seriously about how expectations operate and influence
things, even if the ratex folks themselves were hopelessly off.
As for the death of monetarism, it is true that it was abandoned
as a policy in the US in August 1982 when the Mexicans threatened
to default and the fear of the New York money center bankers
(transmitted through New York Fed President Corrigan to Volcker)
overcame their love of its (successful) anti-inflationary, worker
depressing aspect. Face it, Volcker's monetarism was very
successful at what it was intended to achieve. But the pain got
too great and scary, and so it was abandoned. Reagan agreed
to dump his last round of tax cuts and Volcker agreed to cut
interest rates (and dump monetarism as an active policy). That
was the beginning of the long boom of the US stock market, the
DJIA being in the 700s in August, 1982.
But economists did not more seriously abandon monetarism
until later in the 1980s. It was not in the 1970s that the velocity
of money destabilized. The reasonable stability of the velocity
of M1 was a major support for monetarism all through the 1950s,
1960s, and the 1970s. It was after Volcker had already changed
policy, and major institutional changes began to occur in US
banking, that in the mid-1980s the velocity of M1 went wacko and
began bouncing all over kingdom come. This did not require
any sophisticated econometrics at all to observe. One could
see it with good old Doug Henwood-type eyeballing. And the
Fed FOMC did exactly that and stopped paying much attention
to M1 and still does not (although M2 is another thing altogether).
Barkley Rosser
