Cross-market interactions, price discovery dynamics, and market quality measurement

In three chapters, my dissertation studies the cross-market interactions in foreign exchange markets at the microstructure level. In the first chapter, we analyze the economic and econometric determinants of two widely used measures of price discovery leadership between multiple markets that trade closely-related securities. Using a structural cointegration model, we show that both the information share (IS) and component share (CS) account for the relative avoidance of noise trading and liquidity shocks, but only the IS can measure the relative informativeness of individual markets. Because the IS only captures the contemporaneous responses of market prices to the news innovation, it is static. The IS estimates based on higher sampling frequency will forgo more pricing discovery dynamics of interest. We illustrate these results with analytical and simulation examples.; In the second chapter, we investigate the information vehicle role of the dollar in the price discovery process of the euro and yen. We propose a new approach to measure the dollar's information contribution using the relative speeds of price convergence toward the new equilibrium by the direct yen/euro rate and the rate implied by the dollar/euro and yen/dollar. Using tradable quotes from spot exchange rates, we show that substantial price discovery of yen/euro occurs through the dollar. The dollar's price leadership positively depends on the liquidity of the dollar markets relative to the cross-rate market, and reaches a minimum during the European business hours when the competition from the euro is most intense.; In the third chapter, we examine the role of cross-market prices and order flows in market quality measurement. We revise the econometric analysis in Hasbrouck (1993) and show that cross-market information variables are relevant to the inference of pricing error variance as a market quality measure. Using a sample of five spot exchange rates, we find that the cross-market variables account for substantial increases of the pricing error variance even after controlling for own-market order flow. The results imply that when the cross-market information is ignored, the information quality of markets, particularly those minor ones with smaller size, is overstated.