Reply to B. Mitchell, P. Davidson

[ACSLKS@xxxxxxxxxxxxxxxx id <01HMYIGXU30W8Y70XH@xxxxxxxxxxxxxxxx>; Sun, 12 Feb 1995 11:15:59 -0400 (EDT)]

Here is an excerpt from a well known cointegration text which I think is
relevant to the apparent misconceptions associated with the cointegration
technique:

"The idea that variables hypothesized to be linked by some theoretical economic
relationship should not diverge from each other in the long run is a
fundamental one.  Such variables may drift apart in the short run or because of
seasonal effects, but if they were to diverge without bound, an equilibrium
relationship among such variables could not be said to exist.  The divergence
from a stable equilibrium state must be stochastically bounded and, at some
point, diminishing over time.  'Co-integration' may be viewed as the
statistical expression of the nature of such equilibrium relationships."

Simply put, the high correlations that one observes using "levels" of economic
time series are merely characteristic of the data clustering around an
"attractor," (linear or otherwise), and therefore cointegrated.  As I have
mentioned before, much of the past empirical research that has utilized levels
of variables along with time trends in a regression analysis is
problematic--"spurious" relationships are detected.

I am not sure as to the nature of the resistance to this technique on the part
of P. Davidson.  To dismiss it because of its potential of misuse is clearly
thowing the baby out with the bathwater.  Is the reference to an "equilibrium"
relationship offensive to some?--or should I say a neo-classical equilibrium
relationship? (e.g., permanent income hypothesis of consumption, the Fischer
effect, purchasing power parity, etc.)  While I have problems with some of
these hypotheses, the cointegration technique should not be exclusively linked
to the orthodoxy.  Moreover, results from error correction models have gone
against the prevailing deterrence view in the neo-classical crime literature.

If anything, cointegration has forced us to face the reality of integrated
processes in economics and has resulted in a significant re-evaluation of
econometric modelling.

Lonnie K. Stevans