The purpose of this dissertation is to study an important component of transaction costs, bid-ask spreads. It consists of three essays.; In the first essay, a microstructure model of a foreign currency dealer's spread decision is developed. The dealer maximizes expected utility of wealth in a two-period model and, at the end of the trading period, is assumed to exit the market without any liquidation costs. It is shown that the optimal spread is a function of the dealer's risk aversion parameters, price and volume volatility, the current value of the currency traded, domestic and foreign interest rates, and the opportunity costs of committing funds to currency trading.; The comparative statics of the spread estimator are analyzed. In equilibrium, the percentage spread varies positively with price risk, the foreign interest rate, and the amount and rate of the line of credit, but varies negatively with the domestic interest rate and the expected currency price level. Moreover, the model allows a richer set of interaction among other variables. For instance, it allows for spreads to increase or decrease as trading volume and volume volatility increase. Nevertheless, as uncertainty is resolved, via the shrinking of the trading period, it is shown that increases in trading volume are associated with lower spreads.; The second essay empirically examines the hypotheses derived in the first essay using the exchange rate of U.S. Dollar for six foreign currencies in both the spot and forward markets. The methodology we use to proxy price risk also allows us to investigate the currency return and volatility patterns. Although there is a day-of-the-week return pattern in the currency markets, we find that this pattern is sensitive to the currency, sample period, and model chosen. Also, currency volatility is more predictable than currency return.; The analysis shows that transaction costs in foreign exchange markets vary over time and currencies and that dealers usually charge higher spreads on forward contracts than on spot contracts. The empirical results provide satisfactory support for our model. The evidence indicates that increases in exchange rate volatility cause the spread to widen. Most of the data series have the expected sign between the spread and the domestic and foreign interest rates. Our results also document that on Friday the spread is significantly higher than the average spread on other weekdays. However, only the British Pound has the expected negative relationship between the percentage spreads and the expected price level.; In the third essay we improve Roll's (1984) and Stoll's (1989) serial covariance spread estimators by taking into account not only the magnitude of the price reversal but also a two-period serial correlation of transaction types. The existing spread models are unified under our new model and the three cost components of spread--order processing, adverse information, and inventory holding costs--are considered. We also propose a methodology to calculate the input parameters.
展开▼