3.6 Additional Robustness Checks
In order to test the robustness of our results, we also carry out the analysis using different exchange rate
quotes time-stamps, redo the estimations on various sub-samples, and address issues pertaining to the
possibility of asymmetric effects and slow absorption processes of the announcement surprises contained
in the news control variables.
First, as noted earlier, the announcement routine of the Fed has changed during the sample period
and, therefore, no exchange rate quotes time-stamp is ideal for the entire period under study. We test the
robustness of our baseline results (employing only noon Eastern exchange rate quotes) by redoing all the
estimations using instead exchange rates recorded at 3 pm Eastern time (ideal for the target rate changes
that occurred during the last part of the sample). In addition, we use a mix of exchange rate quotes
(exchange rates recorded at noon Eastern time for the 24 target rate changes that occurred prior to 1994
and exchange rates recorded at 3 pm Eastern time for the target rate changes that occurred after 1994).
Our baseline results are not affected by using these alternative exchange rate quotes.12
Second, we redo the estimations using only US target rate change events that do not coincide with
interest rate changes conducted by the relevant foreign central bank. This is done to ensure that the
estimated exchange rate responses to US monetary policy surprises are not partly driven by simultaneous
monetary policy changes carried out by foreign central banks. The results are identical to those based on
the full sample. 13
12 Another way of addressing the timing issue would be to split up the sample into two sub-samples, conduct the
event study analysis on these sub-samples separately, and use exchange rates with a noon Eastern time-stamp for the
1989-1994 sample and exchange rates with a 3 pm Eastern time-stamp for the second part of the sample. However,
this is unappealing due to the relatively few events in the full sample (42 for the JPY/USD and GBP/USD exchange
rates and 38 for the DEM/USD exchange rate).
13 Fed funds target rate change events coinciding with changes in the German Discount and Lombard rates, the
Japanese Discount Rate of Commercial Bills and the British Minimum Lending Rate and Repo Rate are excluded.
This leads to exclusion of two events from the DEM/USD sample and one event from the GBP/USD sample.
15