Information Flow Dependence in Return and Trading Volume Across Different Stocks

We develop a multivariate return and trading volume model, where each stock’s system is driven by latent information arrivals in continuous time. The arrivals contain idiosyncratic and cross-relevant information, which provides both return and trading volume dependence. Conditional on the accumulated information, returns are jointly normal and correlated, which implies a second layer of dependence in the return dimension. Using a sample of nine common stocks, we show that trading volume significantly adds to the operationalization of the latent information flow process driving the contemporaneous return distribution. The dependence parameter estimates provide significant and interpretable degrees of information flow dependence across all results. Portfolio risk measurement applications are extended by conditioning on the level of trading volume, e.g. reflecting stress, leading to an accurate risk quantification.