Estimating and Testing Two Consumption-Based Asset Pricing Models for Brazil: An Information-Theoretic Approach

Since Brazilian data sets for consumption and asset returns are short and the standard GMM-based overidentifying restrictions test has low power in small samples, a GMM approach imposes difficulties to the evaluation of asset pricing kernels better suited to describe asset pricing phenomena in Brazil. This paper addresses the question of estimating and testing two asset pricing models, using an information-theoretic method of moments estimator, which minimizes the Kullback-Leibler Information Criterion (KLIC). The goal is to compare the traditional GMM method with the alternative information-theoretic approach, which has promising finite sample properties, focusing on over identifying restrictions test and parameter estimate.