Stochastic Volatility, Trading Volume, and the Daily Flow of Information

Abstract

We use state-space methods to investigate the relation between volume, volatility, and ARCH effects within a Mixture-of-Distributions Hypothesis (MDH) framework. In most recent studies of the MDH, the information flow or its logarithm is modeled as an AR(1) process. We argue that this is too restrictive because it requires the information flow to be highly persistent. Using a more general process, we find evidence of a large nonpersistent component of return volatility that is closely related to the contemporaneous nonpersistent component of trading volume. However, unlike previous studies that fit volume-augmented GARCH models, we find no evidence that trading volume subsumes ARCH effects. Because volume-augmented GARCH models are subject to simultaneity bias, our findings should be more robust than these prior results.

Keywords: GARCH, mixture-of-distributions hypothesis, bivariate mixture models, Kalman filter

JEL Classification: G12, C22