DETERMINANTS OF DAILY FLUCTUATIONS IN LIQUIDITY AND TRADING ACTIVITY

Abstract

Transaction expenses can accumulate to a relatively large decrement in total return when portfolios are turned over frequently. Since money managers often trade several securities simultaneously, it is important to know whether trading costs are correlated across securities. Yet research on trading costs has focused almost exclusively on individual securities. Typically, we do not think of illiquidity in a market-wide context, and the classic models of market microstructure involve a dealer in a single stock who provides immediacy at a cost that arises due to inventory holding risk (Stoll 1978) or because of the specter of trading with an investor with superior information (Glosten and Milgrom 1985). Empirical work also deals solely with the trading patterns of individual assets, most often equities sampled at high frequencies (see, for example, Wood, Mclnish, and Ord 1985). There are a variety of reasons why illiquidity-induced trading costs should be correlated across securities. For example, if trading volume exhibits correlated changes in response to broad market movements, this should induce a correlation in liquidity costs. Similarly, the cost of holding inventory could be correlated across securities because it depends, in part, on market interest rates. Within the asymmetric information view of the spread, there might be types of information pertinent for most firms in an industry sector whose imminent revelation could influence the liquidity of several securities simultaneously. Sudden changes in system-wide liquidity appear to have been important in some well-known financial episodes. The international stock market crash of October, 1987 was associated with no identifiable major news event (see Roll 1988), but was characterized by a temporary reduction in liquidity. During the