On the robustness of goodness-of-fit criteria for factor identification: simulation and some Korean evidence

Abstract

A recent issue in empirical tests of the Arbitrage Pricing Theory developed by Ross (1978) has been factor identification. Our paper focuses on factor identification by the LR statistic in maximum likelihood(ML) estimation. Several financial researchers have cautioned against using the formal LR statistic because violations of the assumptions in financial applications typically occur with actual return data. As an alternative to the strict test about the number of factors, several goodness of fit criteria can be used. Among these criteria, this paper deals with Cross-Validation(CV), Akaike Information Criterion(AIC), and Schwartz's Baysean Criterion(SBC).