Entropy Criteria in the Econometric Model Identification

Abstract

A minmax problem statement of a problem of the parameter identification of econometric models is considered, based on the entropy (in the Shannon sense) of the identification error. An interconnection of the minmax criterion proposed and a criterion of the maximum of the mutual information (in the sense of the Kullback-Leibler divergence) of the output data generating process and econometric model output process is shown. Simulation results confirming the validity of the entropy-based approach proposed are presented.