Anomalia de ações de baixo risco no mercado brasileiro

Abstract

This dissertation aims to study the presence in the Brazilian stock market of the low-risk stocks anomaly, originally reported by Black, Jensen, and Scholes (1972), in which stocks with low beta coefficients present higher results than those of high-beta stocks, contrary to the CAPM predictions, under which the return is a direct and linear function of the investment’s systematic risk. For this purpose, this study is based on the daily prices (adjusted for dividends) of stocks listed on BM&FBOVESPA for a period of 18 years (1995-2012), and consisted of several portfolio studies in which the portfolios were assembled by grouping stocks through the order of their betas (estimated based on 750 days prior to portfolio selection). In order to allow the analysis of the anomaly on different liquidity constraints, five alternative criteria for the stocks minimum trading presence on the 750 days prior to the selection of assets were considered, totaling 86,350 portfolios analyses. Evidence shows that low-beta stocks’ portfolios have higher returns than high-beta stocks’ portfolios, in about eighty percent of the cases, contrary to predictions of the CAPM. There was also a reversal of the anomalous behavior of the portfolios during the economic crises started in 1999, 2003 and 2007. Through t-tests of differences in mean returns, it was found that the differences between the low-beta and the high beta portfolios are not statistically significant in most of studies performed. Through a nonparametric statistical test (Kolmogorov-Smirnov test), the hypothesis that low-beta and high-beta stocks portfolios returns were extracted from different statistical distributions was not rejected, raising questions and suggesting further studies as to the most appropriate approach to test mean differences in samples as large as the one adopted in the present study. Regardless of the issue on the results’ statistical significance, the practical significance due to the higher returns of low-beta portfolios when compared to high-beta portfolios, in about 80% of the asset selection moments, points to the need of further studies on the reasons that could explain the anomaly, considering alternative approaches on how to measure it.