Re: Son of unit roots -Reply -Reply

[BILL MITCHELL <ECWFM@xxxxxxxxxxxxxxxxxxx>]

Paul at his ebullient best coaxes Laura on the latest CON in statistical
modelling of time series:

>: Your diatribe may have been misplaced but the use of
>differencing is not! Cointegration is the latest con in eCONometrics.  If
>some staunch defenders would read some mathematical Tretises on Stochastic
>Processes by Diggle, Yaglom, Billinsley and others they might understandthat th
>e tool they are using can't bang square pegs into the round holes they are
>trying to fill.

Which others? There are many treatises on stochastic processes which say
otherwise. Also you do not differentiate b/tw theory and praxis. None of us who
use CI analysis or any econometric analysis would suggest the praxis is 100 per
cent. Any researcher worth their salt qualifies the results, avoids talking
about discovery or truth, emphasises adequacy and tentatitiveness.

At a theoretical level, CI analysis is well grounded in classical statistical
theory. The axioms are reasonable. The insights ingenious. In its application
to economic time series, CI has a plausibility that other approaches do not
have, and does provide researchers with a meaning to a steady-state that was
absent in the first attempts at dealing with stochastic non-stationarity. It
makes intuitive sense that two trending processes (wages and prices, for
example), trend in such a way that their combination is stable (the real wage).
when have we observed the real wage in any major nation explode indefinately?
never is the answer. the applications along this line of reasoning are many.

CI does not assert anything more than that. It also formalises ECMs which help
us understand dynamic adjustments. If there is a notion of a steady state
transitory, unique, multi, or otherwise, determined by class struggle, or
technological things and endowments, then we have to have a concept of
adjustment to advance any further than comparative static analysis. ECM helps
here by linking the steady state to the dynamic process. What determines the
steady state is untestable - it is the Tower of Babel stuff and has to rely on
other authorities for its longevity - economic theory is not testable, and
econometrics only tests econometric models. How we map the latter into the
former is another matter.

but to say that CI is a con, and econometricians are con artists by referring
to 3 works out of a 100 or more is ultimately a very arrogant and misinformed
position.

as always Paul, i disagree with you on basic substance.

kind regards
bill
*******************************************************************************
 William F. Mitchell            Telephone: +61-49-215027      .-_|\
 Department of Economics                   +61-49-705133     /     \    about
 The University of Newcastle    Fax:       +61-49-216919     \.--._/*<-- here
 Callaghan   NSW  2308                                            v
 Australia                      Email : ecwfm@xxxxxxxxxxxxxxxxxxx

 World Wide Web Home Page: http://econ-www.newcastle.edu.au/~bill/billyhp.html
*******************************************************************************



Have a good day! Paul
email address: PB108928@xxxxxxxxxxxxxx
fax# (615)974-1686
phone# (615)974-4221