Testes do modelo capm condicional no mercado brasileiro: um estudo dos efeitos momento, tamanho e book-to-market no período de 1995 a 2008.

Abstract

This work tries to verify the ability of the conditional CAPM model to explain anomalous returns of momentum, size and book-to-market using Lewellen and Nagels (2006) proposed methodology in the brazilian stock market. The research sample complains of Bovespas stocks from July of 1995 to June of 2008. It was observed that there is a difference in using value weighted returns and simple average return when one is studying an anomalous return. There was no evidence of a momentum anomaly in the studied sample; the size anomaly only was observable when the portfolios were value weighted and the book-to-market only with simple average portfolios returns. The unconditional CAPM presented problems of heteroskedasticity, serial correlation and normality. The Fama and Frenchs (1993, 1996) tri-factorial model has not beenable to present significant better results than the unconditional CAPM. The conditional CAPM tested from twelve month regression series has not presented significant gain from the unconditional form either. However, it has been noted that betas do vary over time, just not varying enough to aid the conditional model to explain return anomalies.