From: John Conover <john@email.johncon.com>
Subject: Re: Friedman VS Keyneslan
Date: 23 May 1999 09:07:27 -0000
JOHN J. WEATHERBY writes:
> The hard way and the way to really prove it is through statistical test.
> There are a series of unit root test with structural breaks included. In
> these test a set of dummy variables are added to find the break date. The
> break year is set endogenously and the test are run. These are a pain to run
> but worth the time. I spent a good part of last semester running them for a
> class on time series analysis. I found as the other researchers who have
> done this have found, that almost all countries tested first seemed to a
> random walk under traditional ADF test. Yet when structural breaks are
> included the results are reversed. In all countries tested, a sample of
> about forty(developed and developing included) countries showed all had
> structural breaks where the process was stationary on each side of the
> break. For instance the big break in Britain was WWI. FOr the US around the
> great depression. When dummy variables are added you can see that the
> procress was trend stationary and not a random walk on each side of the
> break. The problem with normal unit root test and your method is that you
> can't distinguish between a random walk and stationarity if there is one
> time endogenous shock to the system.
Hi John. The graphs can detect between a random walk and
stationarity. That's why I chose to do the distribution of run lengths
of expansion and contractions of GDPs, (and showed them to be the same
at different time scales.) If the US GDP was characterized by a "one
time endogenous shock to the system", then it would be a Levy flight
distribution, over time, and the graph would be a horizontal line.
The empirical data does not show that to be the case.
What the empirical data does show is a fractal dimension that lies
between a random walk, and a Levy flight, about half way.
But, since either random walk or Levy flight is constructed via a
random mechanism, it still does not explain how:
John Conover wrote:
>>
>> How can the paradigms of monetarism and/or Keynesianism be reconciled
>> with something that appears to be produced by a random mechanism?
>>
John
--
John Conover, john@email.johncon.com, http://www.johncon.com/