David B wrote:

>> > this seems to be rather strong statement. You can treat
>> > regressors as non-stochastic if you have control over it. So, it seems to
>> > me that the only case when you can treat regressors as fixed is when your
>> > data is coming from some designed experiment.
>>
>> That is precisely what I wanted opinions about. It seems to me it is a
>> philosophy of probability problem (to be pompous), which is overlooked in
>> basic econometrics/statistics textbooks (or even more advanced one, I would
>> say).
>> Why would one be obliged to carefully test systematically for unit roots,
>> since integrated process do not "really" exist ?

Vadim and Oxana Marmer:

>if you regress log of agregate consumption on log of GDP, woul you like to
>treat log GDP as fixed regressor? I guess not. Fixed regressor implies a
>lot of strong properties which is not reasonable to assume in this case.

If the only "problem" is that the regressors are stochastically
generated from a non-stationary process and/or that the residuals are
autocorrelated, there is no reason not to use standard regression
methods (properly taking account of autocorrelation of the residuals
if necessary).  This is completely conventional practice.  There is no
philosphical problem to deal with.

Now, when regressing log consumption on log GDP, you may have another
problem, which is that consumption may be influenced not just by
current GDP but also by changes to GDP in the recent past.  If these
changes are not included as predictors in the regression, the effect
is that the residuals are not independent of the predictors.
Alternatively, if you choose to regard the predictors as fixed, the
mean of the residuals is not always zero, but varies from case to case.

Perhaps this whole discussion arises from the mistaken assumption that
methods which are invalid in this situaton (where the residuals are
not independent of the predictors) must also be invalid in other
situations where the predictors are stochastic, even when they are
independent of the residuals.

   Radford Neal

----------------------------------------------------------------------------
Radford M. Neal                                       [EMAIL PROTECTED]
Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED]
University of Toronto                     http://www.cs.utoronto.ca/~radford
----------------------------------------------------------------------------


=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
                  http://jse.stat.ncsu.edu/
=================================================================

Reply via email to