Specification Analysis in Dynamic Models.

The theoretical and empirical implications of omitted variables, particularly dynamic adjustment effects, are studied. In particular, the attempt to model for such omission by including possibly irrelevant variables is investigated. This extends the existing knowledge of misspecification analysis in several directions. Ordinary least squares is the estimation technique under study, as has been the case in several recent and related studies. In our empirical example, the question of seasonal variation in interest rates is addressed. We deal with the related issue of deterministic versus stochastic detrending and demonstrate that it can be usefully cast in the context of "misspecification analysis" in dynamic models developed in this article.