Source: Dissertation Abstracts International, Volume: 65-03, Section: A, page: 1053.

Director: William N. Goetzmann.

This thesis investigates causes and effects of systematic liquidity variation for the U.S. stock market. Using a vector autoregression approach, Chapter 1 examines the dynamic relation between market liquidity and macroeconomic factors over the past four decades. The chapter finds that the macroeconomic influence on liquidity is stronger before the mid 1980's when business cycle dynamics are more volatile. During the pre-1983 period, market liquidity improves significantly in response to a positive shock in nonborrowed reserves and to negative shocks in supply-side inflation and in the federal funds rate. The macroeconomic shocks also affect market-level liquidity drivers, such as market return, volatility, and share turnover. After 1983, liquidity is less responsive and more resilient to market-level and economy-wide shocks, leaving volatility the only significant predictor of liquidity.

Chapter 2 studies the effect of market liquidity on excess market returns using four different liquidity proxies. In contrast to past empirical findings, there is only weak evidence of stock return predictability. However, the negative contemporaneous effect of illiquidity shocks on excess market returns is robust, and market return responds significantly to various illiquidity shocks simultaneously. The negative illiquidity-return relation is strong when the economy is in recessions and when the short-term interest rate is high. This suggests that investors depress prices more in response to illiquidity shocks when they anticipate higher market illiquidity to be accompanied by a sluggish economy.

Chapter 3 finds a significant positive relation between illiquidity and conditional variance of stock returns both at the individual and aggregate levels. For each of the largest two hundred stocks on the NYSE and NASDAQ, GARCH model in which share turnover and proportional spread enter the conditional variance equation is estimated. The results show that for 75% of the stocks examined, proportional spread is a significant and positive determinant of conditional heteroscedasticity. Illiquidity has an even stronger positive effect on the variability of aggregate market return. A simple microstructural model to explain these findings is provided.