Apophenia? Data Under-Mining the Volatility Leverage-Effect.

Abstract

The inverse relation between stock returns and their volatility, known as volatility leverage-effect (VLE), is documented as a strikingly robust empirical regularity. This paper argues that existing explanations of the phenomenon either suffer from logical inconsistencies or have secondary implications that contradict empirical evidence. Robustness of the empirical findings is re-examined by conducting a thorough investigation of VLE in S&P500 data. Combining misspecification analysis with a novel approach to outlier detection reveals that the VLE relation is indeed very fragile. Implications range from the empirical validity of VLE itself to its use as a moment condition for structural models.

Keywords: Volatility, Leverage Effect, Outliers, Omitted Variable Bias, Realized Measures.

JEL Classification: C01, C14, G00