Market microstructure and transaction costs

In this dissertation, I show that the estimators of the adverse selection cost component of the bid-ask spread developed in the extant literature are downward biased due to intertemporal variation in bid-ask spreads as well as serially correlated expected returns. I provide a modified estimator that relieves these biases. My empirical results are consistent with this prediction. Using the NYSE's Trade and Quote (TAQ) data for NYSE and AMEX stocks for 1993, my method yields estimates of the adverse selection cost component that, on average, account for over 50-60% of the total bid-ask spread, compared with 7-20% from the extant estimator.;My time-series analysis indicates that volatility is positively related to the bid-ask spread for all size-sorted quintiles, while frequency of trades and order size tend to be correlated with the bid-ask spread only in the case of smaller quintiles. Granger causality tests indicate that volatility is the driving force among these market microstructure variables. In the cross-section, frequency of trades and order size carry information that is not contained in volatility of firm size. I argue that information carried solely by frequency of trades and order size may be regarded as the intensity of transactions. My evidence supports both the order processing and adverse selection cost hypotheses.;Finally, the bid-ask spread has the U-shaped pattern across trading hours of the day, but shows a minimal and insignificant variation across weekdays. Furthermore, the U-shaped intraday pattern in the bid-ask spread is solely due to the variation in adverse selection costs. No evidence is found for U-shaped interday pattern for the bid-ask spread components. Therefore, this evidence does not support strategic behavior models that emphasize discretionary liquidity traders.