The Robustness of the Volatility Factor: Linear versus Nonlinear Factor Model

Abstract

This article investigates the trade-off between an extension of the standard three-factor model including a new volatility factor compared to a parsimonious Markov switching model in the context of performance and risk analysis for a set of popular alternative beta strategies. The authors use Bayesian techniques to estimate a two-state (bull and bear) regime-switching model. Over the period of 1969–2014, they show that the inclusion of a time-varying feature in the standard model is as good as the extension of the volatility factor, at least in explaining the alphas for some alternative beta strategies.

TOPICS: Factor-based models, statistical methods, volatility measures