Determining the Number of Factors and Lag Order in Dynamic Factor Models: A Minimum Entropy Approach

Abstract

This article proposes a solution to one of the issues in the rapidly growing literature on dynamic factor models, i.e., how to determine the optimal number of factors. Our formal test, based upon the canonical correlation procedure related to concepts from information theory, produces estimates of the number of factors and the lag order simultaneously. Simulation experiments illustrate the potential of our approach.