Is Foreign Exchange Intervention Effective? Some Micro-Analytical Evidence from Central Europe

Abstract

We test the effectiveness of the interventions performed by the Czech National Bank in the EUR/CZK within the framework of the Evans-Lyons (JME, 2002) microstructure model of the forex market. Employing time-stamped quotes and transactions on the Reuters Spot Matching market, we estimate a two-equation system on the rate change and order flow at hourly intervals. We find a significant impact of order flow on the exchange rate, equal on average to 7.6 basis points per 10 million Euro, of which 80 percent persists through the day. The existence of distinct subperiods enables us to estimate the potential effects of intervention in three different states of the market. First, the central bank might perform secret interventions in orderly market conditions, when dealers perceive a low likelihood of their occurrence. Second, secret interventions might take place in a state characterized by a high likelihood of their occurrence. The third state is characterized by the market's knowledge that intervention is indeed taking place. Our results lend support to the theoretical model and provide useful insight on the conditions for effective central bank intervention in the future.