From: John Conover <john@email.johncon.com>
Subject: Re: Why are economists reviled?
Date: 19 Aug 1999 05:20:17 -0000
John J. Weatherby writes:
>
> In short it like this. The old forecasting are gone. They don't predict.
> Forecasting models try to predict almost everything for the next year or so.
> Policy implications can be somewhat predicted. Partial solutions are
> available. For instance the new growth theory has a lot to say about what
> kind of returns certain policy can have. Unlike the days of Solow when his
> model said X K in year t means Y K in year t+1, therefore z% amount of
> growth. The Keynesian models gave no policy implications to growth. The new
> growth theory does. For instance taking product creation or R&D an accurate
> estimation for the return of X government dollars in basic research can
> made. This doesn't mean the New Growth models predict the growth rate for
> the next period, well not all of them. The models predict what effects
> increased education, research, worker training programs, etc. may have. A
> very much better result than the old forecasting models that said we can
> predict X% growth but don't ask me why.
>
Oh, perturbation theory.
The good news is-if the system is sufficiently complex-that it will
produce desired results somewhat over 50% of the time. The bad news
is that it won't somewhat less than 50% of the time (ie., Pareto-Levy
distribution.). Which, would be would be decided by lottery.
The question is, is an economy a sufficiently complex system?
Anyone measured it?
John
BTW, a good way to verify it would be to apply some means to achieve a
desired economic end, but do so at different time scales, measuring
the successes and failures, and assembling them into a frequency
distribution. If the distribution doesn't change WRT time scales, it
is sufficiently complex. I'll bet it converges into a Pareto-Levy
distribution, in only a very few iterations, (like two, to assert its
bell shaped characteristic.)
--
John Conover, john@email.johncon.com, http://www.johncon.com/