Asset pricing models have been a constant theme in finance research. Since the Capital Asset Pricing Model (CAPM) proposed by Sharpe (1964), such models linearly relate the expected rate of returns of an asset or portfolio of assets with systemic risk factors. This research presents a test of a pricing model with risk factors based on statistical comoments and using a Brazilian dataset. The proposed model is an extension of the original CAPM with the addition of coskewness and cokurtosis between stock rates of return and the market portfolio rates of return. The effects of other variables such as market to book value, financial leverage and a negotiability index served as control variables. The sample consisted of 179 non-financial Brazilian companies traded on BM&FBovespa with data available from 2003 to 2007. Annual systemic moments were calculated from weekly rates of return. They were then tested on a pricing model in order to check for the existence of a risk premium associated with each of these risk measures. We employed a Generalized Method of Moments (GMM) panel data analysis. The use of GMM aims to address potential problems of endogeneity and simultaneous determination of the data, avoiding the occurrence of bias in the estimates. The estimation results show that the relationship between rates of return and covariance and cokurtosis are statistically significant. The results were robust to alternative model specifications. This research contributes to the literature by presenting empirical evidence that there is a Brazilian risk premium associated with systemic moments.
Coskewness; Cokurtosis; Rates of return; Panel data; GMM
