Abstract
Authors of Working Papers are normally staff members of the Fund or consultants, although on occasion outside authors may collaborate with a staff member in writing a paper. The views expressed in the Working Papers or their summaries are, however, those of the authors and should not necessarily be interpreted as representing the views of the Fund. Copies of individual Working Papers and information on subscriptions to the annual series of Working Papers may be obtained from IMF Publication Services, International Monetary Fund, 700 19th Street, Washington, D.C. 20431. Telephone: (202) 623-7430 Telefax: (202) 623-7201 This compilation of summaries of Working Papers released during July-December 1994 is being issued as a part of the Working Paper series. It is designed to provide the reader with an overview of the research work performed by the staff during the period.
The relationship between trading volume and stock returns has been the focus of much recent interest. Empirical studies have found relationships between volume and various moments of return, while theoretical studies have sought to explain these findings by modeling traders and the trading environment. This paper introduces a model that links this literature to classical methods for asset pricing.
The model is a simple variant of the standard intertemporal consumption-investment problem under uncertainty in discrete time. When trading assets, the agent pays transaction costs, and these costs depend on the level of market activity or noise trading. This device is consistent with stylized facts about market depth and trading costs. In equilibrium, the marginal cost of transaction--and hence noise trading--is priced risk. Omission of this risk factor could underlie well-known anomalies such as market size and January effects.
The model is estimated with aggregate data on real consumption, real stock market returns, and trading volume for the United States. There is a significant link between trading volume and equilibrium returns, and estimated marginal costs decline with volume, so that changes in volume influence returns more when the average trading volume is lower. Specification tests show that the parameters shift over time, but the parameter for transaction cost is still significant, including when the October 1987 crash is omitted from the sample.
This paper also examines the role of volume in the relationship of risk to return in linear capital asset pricing. Both market and consumption pricing models fit better in high-volume months. This is consistent with the estimates of decreasing marginal costs in the intertemporal model.