تاثیر اطلاعیه شورای بانک مرکزی اروپا در بازار ارز خارجی : بررسی میکرو ساختاری

We examine the evidence regarding systematic patterns in the euro-dollar foreign exchange market on days when the Governing Council (GC) of the European Central Bank announces its interest rate decisions versus other days. We examine 5-minute data in a non-linear framework allowing for switching between a high-volatility, informed-trading state and a low-volatility, liquidity-trading state. We find strong evidence that the GC policy announcements contain significant news content. Although there is some evidence of positioning in the hour prior to the announcement, this probably reflects dealers minimizing their exposure rather than evidence of information leakage.

The European Central Bank (ECB) was officially created on 1 June 1998 in the run-up to European Monetary Union (EMU) which took effect on January 1 1999. The primary objective of the ECB is to promote price stability throughout the Euro Area and to design and implement monetary policy that is consistent with this objective.1 The ECB operates with the assistance of the national central banks in each of the Euro Area countries. While the implementation of monetary policy by the ECB is under the control of the Executive Board, the actual design of monetary policy is the responsibility of the Governing Council (GC), which comprises the whole of the Executive Board plus the governors of the national central banks in the Eurosystem—at present a total of eighteen members. The GC, which meets every two weeks, is in this sense the most important decision-making body of the ECB and decides, in particular, on the level of the ECB’s key interest rate, the refinancing rate.2 Since the end of 2001, the GC has adopted the policy of announcing interest rate decisions only on the first of its two monthly meetings. Since the market knows that the interest rate decision is given in a press announcement from the President of the ECB at 13:45 Central European Time (CET) on the day of the first monthly meeting, there may be positioning prior to the announcement and news effects after the announcement that result in systematic patterns in exchange rate behavior on GC meeting days that differ from other days. We examine the evidence in the foreign exchange market to analyze the pattern of exchange rate changes and volatility surrounding the announcement. We use 5-minute data and apply a Markov-switching model where the data generating process of exchange rate returns switches between a high-volatility, informed-trading state, and a low-volatility, uninformed or liquidity trading state, with the probability of switching between regimes depending endogenously on time-of-day indicator variables. The remainder of the paper is set out as follows. In Section 2 we provide a discussion of our econometric techniques, while in Section 3 we describe the data. In Section 4 report our empirical findings and in a final section we provide a summary and conclusion.

The European Central Bank was created in 1998 in order to foster monetary policy consistent with stable inflation and economic growth in the Euro Area, which became operational on January 1 1999. The interest rate decisions of the ECB are taken by its Governing Council (GC) at regularly scheduled, pre-announced dates and the policy decision is always announced at a pre-set time. We exploited this institutional framework in order to examine the effect of monetary policy announcements on the foreign exchange market, using an endogenous-probability Markov-switching analysis and a high-frequency sample of 5-minute observations on euro-dollar exchange rate returns over the trading days when the GC met during 2002 and 2003, as well as on a set of control days during the same period. We found that the probability of switching into a high-volatility, informed trading state rose significantly on GC meeting days when interest rates decisions were announced (i.e. on the first of each of the two monthly GC meetings) and that the probability of remaining in the low volatility state was found to fall significantly at the same time. Our evidence therefore indicates that there is a statistically and economically significant news effect related to the noon announcement. The effect of an announcement of an interest rate change took about fifteen minutes for its full impact to be felt, however, and the announcement effect overall, regardless of whether or not a change was announced, took about an hour to dissipate. This is consistent with a microstructural analysis of information acquisition and dissemination in the foreign exchange market, whereby agents learn about the significance of new information by trading with one another, rather than jumping immediately to a new equilibrium as a more traditional rational expectations analysis would suggest. In addition, however, we also found significant evidence of an increase in the probability of being in the informed state beginning an hour before the interest rate announcement is made. But the fact that this occurs whether or not an interest rate change is announced strongly suggests that the effect is due to dealers closing out positions in order to minimize their risk exposure immediately before the announcement, rather than due to information leakage concerning the policy decision.4