Estimation et spécification Estimation and specification

Abstract

In his presidential address to the Société Canadienne de Science Economique, at the 1978 meeting, the author had chosen to talk about the organizational problems of the Société because of the pressing nature of the situation. Should the author have decided to discuss issues related to the science of economics, as is usuall done in such occasions, he would have taken this opportunity to point out that one of the major problems of econometric research is that of specifying correctly the structural models utilized. Econometric textbooks discuss thoroughly the methods of estimations under the assumption that the structure of the econometric model is given. However, it is well known that in practice, trial and error procedures are extensively used to find "acceptable" functional forms for the equations of the models. Efforts have been made to develop systematic techniques of choice between functional forms, but the results available until now are very limited in scope. Much greater research efforts should be devoted to this fundamental topic.