부산대학교 도서관

Using tick data of the USD/JPY rate, I propose the method to detect the time of the FX intervention. I use the simple microstructure model and assume that the FX intervention causes regime-switching in the microstructure of the USD/JPY market, changes in adverse selection, and inventory effect. The time of the intervention is estimated endogenously by the Markov-switching model, and the actual starting time is well estimated. I also find that no market orders, except a large U.S. dollar purchase, convey any private information during the period of the intervention.