The intraday pattern of information asymmetry : evidence from the NYSE
Previous studies (e.g. Benston and Hagerman, 1974, Bagehot, 1971 and Stoll, 1978) suggest that the bid-ask spread consists of three components: asymmetric information cost, inventory holding cost, and order processing cost. Other literature (e.g. Brock and Kleidon, 1992, Hef-lin et al, 2007, and McInish and Van Ness, 2002) reports that the bid-ask spread varies during a trading day following a U-shaped pattern. One explanation for this observation is that it is the result of changes in information asymmetry costs over the trading hours, assuming the other costs are fixed. However, no empirical study directly measures how information asym-metry changes over the trading day. We explore how this measure relates to the spread as well as the quote depth. Our research divides a trading day into 13 half-hour trading intervals and measures in-formation asymmetry during each interval following the model developed by Madhavan and Smidt (1991) and Noronha et al (1996). Their model can directly estimate the level of infor-mation asymmetry in each interval. This enables us to observe the intraday pattern of infor-mation asymmetry directly and compare it to the patterns of the spread and the quote depth. Furthermore, we test the relationship between the spread and the information asymmetry and the relationship between the depth and the information asymmetry in a dynamic context to see how market makers manage information risk over trading hours. We find that the risk of information asymmetry varies significantly during the trading day. There is a large drop over the first interval, and another large drop over the last interval, with smaller fluctuations over the remaining intervals. Moreover, we show that the spread is consistent with an L-shaped pattern as opposed to the U-shaped pattern proposed by previous studies while the depth is increasing throughout the 13 trading intervals. Furthermore, we ob-serve that the variations of the spread and the depth are respectively positively and negatively related to the intraday variations in the degree of information asymmetry across the trading intervals. In particular, a large decline in information asymmetry at the beginning of the day is associated with a large reduction in the spread, whereas a large decline in information asymmetry at the end of the day is associated with a large increase in the quote depth. This emphasises the importance of studying both measures of liquidity simultaneously.
DegreeMaster of Science (M.Sc.)
DepartmentFinance and Management Science
ProgramFinance and Management Science
SupervisorWilson, Craig; Tannous, George
CommitteeMarie, Racine; Sick, Gordon; Dev, Mishra
Copyright DateAugust 2009