New Evidence on the Puzzles. Results from Agnostic Identification on Monetary Policy and Exchange Rates.

between the result of the EE and BIG specification, showing that our results
are fairly robust and the identification method indeed easily generalizable,
as claimed above. Also, the evidence looks rather similar across the three
country pairs. The difference to the results from the Eichenbaum-Evans
identification is considerable, though.

The reason for this difference can be understood from figure 8 (two pages),
where we have compared all the impulse response functions for the variables
of the EE specification in the US-UK case. Note in particular the huge price
puzzle emerging for the Eichenbaum-Evans identification. This price puzzle
is considerably more pronounced than in the original source, because the re-
cursive identification procedures have been applied here to our updated data
set, using data from 1975 to 2002. We believe that this strong and long posi-
tive reaction of the price level casts considerable doubt on this identification
strategy and therefore on the results for the exchange rate response (and we
doubt that Eichenbaum and Evans would have stuck to this strategy in light
of the new data and the new results). Obviously, there are additional impulse
response functions for the BIG specification and similar comparisons could
be made for the US-Germany and the US-Japan case. We show the impulse
responses for the BIG specification in figure 13 (two pages). These do not
change the key insights, however.

The third and forth line of figure 4 shows the impulse response for the
forward discount premium and the resulting Sharpe ratio for a Bayesian in-
vestor. A comparison for our three identification procedures can be found
in figures 9 and 10. The impulse response for the forward premium should
be compared to the theory figure 2. The results can be described as follows.
There is a forward discount premium, but there is also considerable uncer-
tainty regarding its size or whether it is even positive. When taking this
uncertainty into account and calculating the Sharpe ratio, one finds values
between 1 and all the way up to 2.5 for the EE specification and investment

18

<<   18   19   20   21   22   23   24   >>

More intriguing information