Frequentist and Subjectivist Perspectives on the Problems of Model Building in Economics

Abstract

I plan to discuss, in as simple and nontechnical a fashion as possible, the subjectivist-Bayesian attitude toward model building in econometrics and to contrast it with the standard frequentist attitude. To convey what I believe is the principle distinguishing attitude between Bayesians and non-Bayesians, I refer to their respective positions as “subjectivist” and “frequentist.” The basic differences between these positions arise from different interpretations of “probability.” Frequentists interpret probability as a property of the external world, i.e., the limiting relative frequency of the occurrence of an event as the number of suitably defined trials goes to infinity. For a subjectivist, probability is interpreted as a degree of belief fundamentally internal to the individual as opposed to some characteristic of the external world. Subjective probability measures a relationship between the observer and events (not necessarily “repetitive”) of the outside world, expressing the observer's personal uncertainty about those events. The subjectivist paradigm is designed to produce “coherent” revisions in beliefs about future observables in light of observed data. Most of the issues I raise are familiar to statisticians but not to economists. Rather than give the suspicious reader a menu of Bayesian techniques, I hope to create an interest in acquiring a taste for the Bayesian cuisine by recommending five pragmatic principles.