Conditional Volatility, Volume Shocks, and GARCH Effects

Abstract

Using daily closing level of S&P500 index and daily trading volume from 1/2/1990 to 12/31/2010, this study examines the relationship between conditional volatility and trading volume at the market level. It is shown that linking volatility to total volume does not extract all information. In fact it is found that at the market level total volume is not a good proxy for the information flow, the latent mixing variable in the Mixture of Distribution Hypothesis, and is unable to explain GARCH effects and has no impact on conditional volatility. However after partitioning volume into expected and unexpected components, private information proxied by unexpected volume shocks, although unable to reduce volatility persistence due to its random nature, is positively associated with conditional volatility. This relation is shown to be asymmetric in the sense that only positive unexpected volume shocks are the contributing factor and negative shocks have no impact. Moreover, evidence suggests that this often reported positive relationship between volume and volatility in equity market is mainly driven by two factors: Costly short selling constraint, and Maturity of the US equity market. Thus the positive relation between volatility and volume in a mature equity market where the number of traders can reasonably be assumed to be constant, is a manifestation of differing cost of taking long and short position. Results indicate that in futures market where the cost of taking long and short positions is symmetric, such a relationship cease to exist. Furthermore, it is documented that in an expanding market characterized by growing number of traders, consistent with the theory, this relationship is still significant but in reversed direction.