101. Randomly Specified Macroeconomic Models: Some Implications for Model Selection.
- Author
-
Peach, James T. and Webb, James L.
- Subjects
MACROECONOMICS ,ECONOMIC models ,PROBABILITY theory ,TIME series analysis ,ECONOMETRICS ,ECONOMIC statistics - Abstract
The main thrust of this article is simply to present the results of an experiment that demonstrate the practical inconclusiveness of econometric testing of macroeconomic models. The brief discussion of the philosophical issues anticipates the reaction of some econometricians and economic practitioners who would dismiss the results of the experiment on the grounds that more sophisticated econometric techniques will ultimately eliminate the inconclusiveness of econometric tests of competing hypotheses. This article focuses on the use of econometric testing as a means of choosing among competing economic theories, and does not examine the use of econometrics as a tool for economic engineering. This article is meant to contribute to making economics a more genuinely progressive empirical science. This article differs from related efforts in that the authors are not explicitly considering the nature of probability distributions of economic time-series, nor are they addressing the effects of mathematical transformations of model specification on statistical fit to a given set of observations, nor do they make exhortations for personal integrity and institutional changes.
- Published
- 1983
- Full Text
- View/download PDF