Modelo de precificação de ativos com erros t de Student

Abstract

The Capital Asset Pricing Model is widely used in financial market applications assisting in decision-making under risk conditions as it links the basic components of an asset evaluation: risk and shares returns. The assumption of normality was always very attractive to evaluate the share returns in Capital Asset Pricing Models. On the other hand, besides the estimates obtained for the normal models being sensitive to huge values, there are empirical evidences that share returns have distributions with heavier tails then the normal ones. In order to develop flexible methodologies in relation to atypical observations, we will study the CAPM considering that the errors follow Student-t distribution. In this paper as an application, the estimation of systematic risk is developed (an important measure of risk for both financial analysts and for portfolio administrators), considering the CAPM model of the Ogx shares with Student-t errors and also normal errors only for comparison purposes. The maximum likelihood estimation of the CAPM model parameters with Student-t error was obtained through the algorithm EM. A residuals analysis was performed to verify the suitability of the models under standard errors and Student-t.