Three Essays in Finance

Author:

Glidewell, Edward

Graduate Program:

Business Administration

Degree:

Doctor of Philosophy

Document Type:

Dissertation

Date of Defense:

April 06, 2004

Committee Members:

Herman J Bierens, Committee Chair/Co-Chair
Quanwei Cao, Committee Chair/Co-Chair
Oliver Hansch, Committee Member
Timothy T Simin, Committee Member

Keywords:

S&P
spread decomposition
trading costs

Abstract:

I discuss three topics in this thesis. The first two concern topics centering on addition to the S&P 500 Index. In chapters two and three, I examine the price and liquidity effects (respectively) of additions to the S&P500 Index from January 1993 to December 2000. The results from chapter two indicate that previous inferences about long run buy and hold returns are fallaciously obtained from a test statistic with low power. Consequently, while the short run demand curve for newly added S&P stocks appears to slope down, the data suggest that no S&P effect exists in the form of a permanently downward sloping demand curve. In chapter three, I present the lasting liquidity improvements that arose from index addition. Permanent reductions in several static and dynamic measures of liquidity are statistically significant but are confined to NYSE firms. The spread decomposition and static trade analyses agree with previous research and imply a perennial increase in liquidity through a 15% reduction in the quoted and effective spreads. The dynamic liquidity measure indicates a permanent increase in market resilience. Post addition, smaller quote revisions obtain. Cumulative mid-quote revisions to buy side shocks fell by 25% and revisions to sell side shocks fell by 27%, implying that private information in the order flow fell after addition. The final chapter is a spread decomposition model that allows for nonlinearities in the data. Most spread decomposition models utilize volume and/or trade signs in a linear specification to explain observed price changes. This paper illustrates the importance of another variable, unexpected trade intensity, in explaining and forecasting effective spreads. I use Engle's (1996) WACD(p,q) model and other proxies for unexpected trade intensity in a generalized version of Glosten and Harris' (1988) spread decomposition model. I adapt Hansen's (1999) technique for testing for the presence of nonlinearities in the relationship between price changes and trade volume and between price changes and unexpected trade intensity. The data suggest that volume and intensity are priced independently and on average place a 22% premium on large volume and a 8% premium on immediacy conditional on a given volume. Selecting a suitable proxy for trade intensity and incorporating the threshold values into the model not only allows for enhanced estimation of the effective spread, but also for more precise forecasts.