Third, we redo the estimations using only events that do not coincide with intervention operations
conducted (unilaterally as well as coordinated) by the Fed, the Bundesbank and the Bank of Japan. 14
Again, the results are unchanged.
Fourth, we expand equation (1) to incorporate various dummies in order to test for asymmetric
exchange rate responses to the unexpected component of a monetary policy change. In particular, we
separate monetary tightening from monetary loosening. We find, however, no evidence of asymmetries.
Fifth, in order to allow for the possibility of a slow absorption process in regards to other macro
news announcements the lead-models are also estimated including all leads of the news control variables
ranging from 1 to 60 (as opposed to the baseline approach where each lead model includes only the
associated lead of the news control variables). This alteration has no impact on the delayed effect results.
4. Conclusion
We investigate whether exchange rates respond to only the surprise component of actual monetary policy
changes and we assess the importance of isolating the surprise component from the actual news
announcement. Furthermore, we investigate whether the exchange rate adjustment associated with
monetary policy surprises is instantaneous or delayed. We focus our investigation on the 42 US monetary
policy changes that occurred during the 1989 to 2000 time-period and we follow Kuttner (2001) in using
Fed funds futures data in order to isolate the surprise component of each of these actual policy changes. In
addition, we incorporate several control variables that capture the surprise element of US macroeconomic
news and policy developments.
Our main findings are the following: First, we show that the expected component of a monetary
policy change has no impact on the exchange rate while the unexpected component of a tightening
(loosening) of US monetary policy is associated with a same-day appreciation (depreciation) of the USD
and, importantly, that failure to disentangle the surprise component from the actual monetary policy
14 This leads to exclusion of one event from the DEM/USD sample and one event from the JPY/USD sample. Due to
unavailability of Bank of England intervention data, we are not able to control for intervention in the GBP/USD
exchange rate.
16