Applying GARCH for examining CAPM and APT across time

Soufian, Nasreen (2004) Applying GARCH for examining CAPM and APT across time. UNSPECIFIED. Manchester Metropolitan University. ISSN 1478-8209

Preview

Available under License In Copyright. Download (246kB) | Preview

Abstract

This paper examines the performance of APT and CAPM across time, considering the acquired betas from GARCH-type models. This paper also includes structural breaks for those factors that have undergone significant change in their mean or volatility. Correction for hetero-scedasticity in the estimation of beta in the first-stage of the twostage estimation procedure of CAPM gives an improvement over OLS. Relative to OLS, IGARCH gives smaller estimates of betas, higher risk premia, increased in log-likelihood and higher R2. However, adding the macro-economic factors to the model, the result for market portfolio is different from that of the CAPM. After adding the economics factors coefficient, the power of the market portfolio's beta has either diminished or reduced. Relative to OLS, applying IGARCH improved the performance of APT for the full timeperiod, however, over entire three samples, the correction for hetero-scedasticity by GARCH is not sufficient to lead to satisfactory ressult for APT. This paper similar to Soufian, Joseph and Ritche (2000) shows that different macro-economic factors capture the variation in average returns for different time periods.