This paper investigates the Information content of daily trading volume with respect to the long-run or high persistent and the short-run or transitory components of the volatility of daily stock market returns using bivariate mixture models. For this purpose, the Standard bivariate mixture model of Tauchen and Pitts (1983) in which volatility and volume are directed by one latent process of Information arrivals is generalized to the extent that two types of information processes each endowed with their own dynamic behavior are allowed to direct volatility and volume. Since the latent information processes are assumed to be autocorrelated which makes standard estimation methods infeasible, a simulated maximum Iikelihood approach is applied to estimate the mixture models. The results based on German stock market data reveal that volume mainly provides information about the transitory com-ponent of volatility, and contains only little information about the high persistent volatility component.
Volatility persistence Bivariate mixture model Long memory Latent dynamic variables Simulated maximum Iikelihood