Commonality, Information and Return/Return Volatility - Volume Relationship
Syracuse University - Whitman School of Management
National Cheng Kung University
AFA 2004 San Diego Meetings
This paper develops a common-factor model to investigate relationships between security returns/return volatility and trading volume. The model generalizes Tauchen and Pitts' (1983) MDH model by capturing possible interactions among securities. In our model, both price changes and trading volume are governed by three kinds of mutually independent variables: common factor variables, latent information variables and idiosyncratic variables. Despite its similarity to Hasbrouck and Seppi's (2001) model in terms of the form, the model extraordinarily allows us to identify the cause of interactions among securities by decomposing factor loadings into constant and random components. Three key implications are reached from our model. First, common factor structures in returns and trading volume stem from information flows. Second, returns' common factors are not related to trading volume's common factors. This implication directly opposes Hasbrouck and Seppi's (2001) assumption. Finally, cross-firm variations of returns and volume respectively rely on underlying latent information flows. The positive relation between return volatility and volume also results only from underlying latent information flows. Thus, common factor structures in returns and trading volume have no additional explanatory power in cross-firm variations and the positive return volatility-volume relationship. We fit the model for intraday data of Dow Jones 30 stocks using the EM algorithm. The results support specifications of our model. The empirical results demonstrate 3-factor structures in returns and trading volume, respectively. All 30 stocks in our sample are governed by at least one common factor. This fact implies that our model outperforms Tauchen and Pitts' (1983) model because their model is a special case of our model without the presence of common factors. We also show that after controlling the effect of information flows, persistence in return variance disappears.
Number of Pages in PDF File: 36
Date posted: December 10, 2003
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 0.234 seconds