The Empirical Foundations of the Arbitrage Pricing Theory I: the Empirical Tests

Abstract

This paper provides a detailed and extensive examination of the validity of the APT based on maximum likelihood factor analysis of large cross-sections of securities. Our empirical implementation of the theory proved in capable of explaining expected returns on portfolios composed of securities with different market capitalizations although it provided an adequate account of the expected returns of portfolios formed on the basis of dividend yield and own variance where risk adjustment with the CAPM employing the usual market proxies failed. In addition, it appears that the zero beta version of the APT is sharply rejected in favor of the riskless rate model and that there is little basis for discriminating among five and ten factor versions of the theory.