出版社：Associação Nacional de Pós-Graduação e Pesquisa em Administração
摘要：This article tests and compares three alternative models for predicting expected returns on the Brazilian stock market: 1) the Sharpe-Litner-Mossin CAPM version; 2) the Fama and French 3-factors model; 3) and the Reward Beta Model, presented by Bornholt (2007). It was used as a two-step empirical procedure methodology for models of general equilibrium: the first step to base regard the parameters of the models from time series regressions, and in the second step the regard parameters are used as explanatory variables in cross section regressions. The tests were preceded on portfolios, in agreement with Fama and French (1993) and Bornholt (2007) methodologies, and applied in two sub-samples of stocks the data of which were available from Bovespa: the within-sample covers the period 07/1995-06/2001 and the out-of-sample 07/2001-06/2006. As well as other evidence found in Brazilian market, the results to support the Fama and French 3-factors model to explain future returns, albeit without the factor that captures the book-to-market effect being significant. In this way, a twofactor model is recommended for the prediction of expected returns on the Brazilian stock market: 1) one that captures the market excess return; and 2) another that captures the size effect.
关键词：CAPM ; APT ; Fama and French 3-factors model ; Reward Beta Approach.
