How Many Factors? Does Adding Momentum and Volatility Improve Performance

Abstract

This paper considers whether adding two established anomalies, momentum and low volatility, will improve our understanding of asset pricing beyond the FF5 model. We do this by considering whether these factors provide economic, as opposed to statistical, significance within the asset pricing model. We measure economic significance in two ways, first, we consider whether the factor coefficient signs and values on the factors are economically meaningful, for example, do the coefficients distinguish between high and low risk portfolios. Second, we consider an out-of-sample trading rule based on expected returns derived from each asset pricing model. Our results suggest that the momentum and volatility factors provide no additional information over the FF5 model. Moreover, it is not clear that the FF5 model itself provides a noticeable improvement over the FF3 model. Of note, the momentum and low volatility factors exhibit limited statistical significance, have similar coefficients across high and low values of different anomalies and big and small firm portfolios. The trading performance of a seven-factor model, while reasonable itself, is worse than both the FF3 and FF5 models. Furthermore, based on the trading results, the FF5 model provides no noticeable contribution over the FF3 model, the latter of which could be regarded as preferred.