Stigum, Bernt P.: Rationality in Econometrics
World Conference Econometric Society, 2000, Seattle

Bernt P. Stigum, University of Oslo
Rationality in Econometrics
Session: C-8-17  Monday 14 August 2000  by Stigum, Bernt P.
The idea of rationality enters an econometrician's work in many ways; e.g., in his presuppositions about sample populations, in his model selections and data analyses, and in his choice of projects. I shall consider some of these ways and their ramifications for the econometrician's own life and for the development of econometrics. I begin with a discussion of rationality that I have found in the writings of Aristotle and other leading philosophers. My aim here is to establish the characteristics that we in good faith can expect rational members of a sample population to possess. The characteristics with which I end up have no definite meaning. Instead they are like undefined terms in mathematics that an econometrician can interpret in ways that suit the purposes of his research and seem appropriate for the population he is studying. When interpreted, the pertinent characteristics of the rational members of a given population become hypotheses whose empirical relevance must be tested. In rationally designed econometric studies the interpretation of 'rationality' that seems appropriate for a given study is usually an interpretation that the pertinent econometrician extracts from various economic theories. I look at some of these interpretations and discuss their empirical relevance. The interpretations of particular interest concern consumer choice under certainty, choice under risky and uncertain conditions, and choice in game-theoretic situations. These interpretations appear in various representations in the ways econometricians model rationality. I single out for discussion microeconometric models of consumer choice and macroeconometric rational expectations models. In the last section of the paper I consider two lacunas in Kuhn's and Lakatos' theories, and see how econometricians go about solving puzzles and extending positive heuristics. I begin by discussing the considerations that guide an econometrician in his choice of research projects. Then, I argue about the determinants of rational choice in model selection. Finally, I consider the politics of writing research reports. The contents of these sections concern aspects of an econometrician's rational choice that are relevant for the orderly development of econometrics.
Submitted paper full-text in .pdf


File created by Jurgen Doornik with eswc2000.ox on 2-01-2001