Financial markets and factor structure

Abstract

Testing, in the financial markets, the hypothesis of a k latent factor structure by the methods of exploratory factor analysis rises some difficulties. In order to get over these problems a new criterion is suggested, in which all a priori known information are used: the sample size and the number of observations. A Monte Carlo simulation shows that the new criterion performs better than the traditional methods. Finally, an application to the Italian stock market indicates the presence of a multifactor structure underlying the stock returns.