Essays on the application of the market microstructure approach in exchange rate economics

Essay one. Exchange Rate Dynamics and Order Flow: An Alternative Model. Realizing and overcoming the disadvantages of the well-known model by Evans and Lyons (2002), this essay presents an alternative model to connect the exchange rate dynamics and foreign exchange market order flow from the perspective of the dealer's quoting behavior in a more realistic environment. Agreeing with previous works, the essay shows theoretically and empirically that order flow has significant positive effects on the short-run exchange rate change.; Essay two. Market Structure and Dealer's Quoting Behavior in the Foreign Exchange Market. Realizing that the structure of inter-dealer and customer foreign exchange markets is different in terms of market transparency and information asymmetry, this essay examines how the differences affect the same dealer's quotes in the two markets. It is shown that customer spreads are generally wider than inter-dealer ones, while their differential tends to fall with the rise in order size. Meanwhile, the same dealer's mid-quotes are found to be equal to each other in the two markets. In contrast to the preceding models claiming that the order size has a positive effect on the spread, the impact is found to be ambiguous in our study. Supportive empirical evidence can be provided by the new data set applied in the essay.; Essay three. The Term Structure of the Forward Premium Puzzle. In contrast to many previous studies that examined the efficiency of the forward market in one particular horizon, this essay tests the unbiased hypothesis by using forward rates in the same period but with various maturities. The hypothesis continues to be rejected as before in the horizon of 1-week and 1-month, but is found to be accepted in the 1-day horizon. It is also found that the magnitude of the slope coefficient increases as the horizon rises. A model in the spirit of the term structure model of interest rates is proposed to reconcile the inconsistency, and the results of empirical testing support the model.