Empirical Analysis Of The VAR Model On Exchange Rate Microstructure

Before the 1970s, the dominant approach to exchange rate determination was the goods market approach. In the 1970s, the asset market approach emerged. Disconcertingly, empirical work dose support neither the goods market approach nor the asset market approach. In the late 1970s, the microstructure approach was developed. What distinguishes the microstructure approach with the traditional macro approach is that it relaxes some the macro approach's assumptions. The microstructure models recognize that the market structure, the market participants and the trading mechanism do affect prices. Order flow conveys information that is not publicly available. The empirical work shows that the microstructure models are significantly better than the macro models in predicting the short and medium term exchange rate movement.In this article, we firstly present and compare four types of the microstructure models, that is, the G-S rational expectation model, the Kyle batch-trade model, the G-M sequential-trade model and the simultaneous model. Then we develop a Structural Vector Auto Regression (SVAR) model to examine the microstructure characteristics of the direct inter-dealer market of DEM/USD and YEN/USD using the data set provide by Professor Evans (1997).Firstly, we present and compare four distinct microstructure models. We thoroughly analyze the insights, the model, the equilibrium solve and the conclusion of each model.Secondly, we summarize the microstructure characteristics of the direct inter-dealer market of DEM/USD and YEN/USD and empirically examine how the market intensity affects the exchange rate.Last but the most importantly, we develop a Structural Vector Auto Regression (SVAR) model to examine how the order flow contributes to the volatility of DEM/USD and YEN/USD.