Some tendencies in modern econometrics

Abstract

All analytical subjects, probably all scholarly fields of investigation, for the matter, should be exploring new avenues of research and seeking new findings. In economics, generally, there is an explosion of research activity, associated with new information, new devices, new scholarly journals, and new professional opportunities. Econometrics, which greatly expanded as a sub-dis? cipline of economics in the second half of this century, is presently branching out in many directions, some of which appear to be fruitful and some of which do not. There are now several books and articles on the historical development of econometrics from sparse early roots in the 19th century and much more significant growth activity in the 20th century. It is worth while to reconsider some earlier research, in relation to recent developments. The subject grew quite naturally as formal economics and mathematical statistics expanded, but a breakthrough occurred that came to be known as the 'Haavelmo Revolution'. A manuscript authored by Trygve Haavelmo circulated among academic economists and statisticians in the late 1930s and early 1940s; it was published as a supplement to Econometrica in 19441. An important interpretation of Haavelmo's contribution, viewed largely from the side of mathematical statistics, has been published by Theodore W. Anderson on the occasion of the award of the prize in memory of Alfred Nobel to Haavelmo in 1989.2 In order to assess some recent developments in econometrics, I want to emphasize a few important aspects of Haavelmo's contribution, which indeed has many facets. Essentially, Haavelmo placed econometric methodology in the midst of developments that were taking place, or had already taken place, in statistical inference. Statistics in economics, which was already well established, but under development at the time of Haavelmo's work, was largely concenrned with descriptive statistics. The main problems were those of describing the economy in quantitative terms. Haavelmo's approach showed how to relate mathematical specifications of economic relationships to probability interpretations of statistical estimates of the specified relationships. All econometrics does not deal with the estimation of behavioural relationships or economic decision-making, but a large portion of the discipline is involved in analyzing economic behaviour, and this forms the focus for the present discussion. To show how probability and economic relationships are involved together, let us consider the fundamental, but general, model that formed the basis for Haavelmo's approach.