ȁȃȃ Partʺ: Economicsthe how the phenomena of recession and recovery may be causally related
to contemporaneous and earlier events.Ȇ
If one would rather not find or see something—like monetary distur-
bances and their consequences—ways are available. Ļe story comes to
mind of Admiral Nelson putting his telescope to his blind eye at the bat-
tle of Copenhagen.
More fundamentally, what reason is there to suppose that a definite
“structure” of the economy, describable by definite functions and definite
coefficients, exists in the first place as an object amenable to econometric
investigation? Furthermore, even if a mathematically formulated system
were deterministic, with known and fixed parameters, the simplest kinds
of nonlinearity could render even its qualitative behavior after several or
many “rounds” extremely sensitive to parameter sizes and initial condi-
tions. Still further, the real system being modeled, instead of being iso-
lated, is exposed to innumerable perturbations (including “noneconomic”
ones) that in principle require being taken into account. In such a system,
numerical prediction is impossible (except, perhaps, for short-run extrap-
olation); the best that can be done is qualitative prediction, or recognition
of patterns. Similar remarks apply to attempts to characterize the “pro-
cesses” supposedly at work in real economies—whether or not they have
unit roots, and so forth.
Ļese are among the lessons for economics of the recently popular
mathematical theory of “chaos” or “catastrophe.” (See EkelandȀȈȇȇ, who
brings E.N. Lorenz’s “butterfly effect” and Henri Poincaré’s reservations
about quantitative modeling into the story.) I do not want to be misunder-
stood, however, as issuing methodological taboos of my own. Economet-
ric research into recent or earlier economic history can indeed be informa-
tive, and its techniques are worth cultivating for applications outside as
well as within economics. I merely want to question insistence on them
as both obligatory and decisive across practically the whole broad range
of economics. Especially where human action is the subject matter, much
can be said for observations described and reasoning conducted largely in
ȆWilliam Poole (ȀȈȈȃ, pp.ȅǿ–ȅȁ) makes a related point: an optimal policy should
abolish any observable relation between money growth andGDPgrowth. He offers an
analogy about trying to determine the relation between a car’s speed and its gasoline con-
sumption by muddling together observations made at moments when the car was going
uphill, going downhill, and proceeding on level ground, even though the driver was trying
to hold the car’s speed steady throughout.