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

a somewhat broader range for the numbers. The Grilli-Roubini specification
delivers very different results, depending on the country pair chosen: less
than 5 percent (at the median) is explained in the US-German case, nearly
30 percent is explained at a horizon one to three years out in the US-UK
case and there is a sharp peak one year out in the US-Japan case.

These rather different behaviours for the Grilli-Roubinin specifications
as well as the sharp peaks in the Eichenbaum-Evans specifications may be
viewed as a further odity, when applying these conventional identification
strategies, which are avoided with the sign restriction approach.

For the sign restriction approach, the 86% quantile rarely moves beyond
30% for any country or specification. When considering the joint contribution
of both monetary policy shocks the median estimate ranges between 10 and
20 percent, and the 86% quantile reaches up to 45 percent of the exchange
rate movements.

All in all, our results are in contrast to Eichenbaum and Evans (1995)
who estimated a percentage of 42, 26 and 23 for Germany, UK and Japan
at lags 31 - 36. Other examples for studies which find that monetary shocks
have a substantial contribution in explaining real exchange rate fluctuations
are Clarida and Gali (1994) and Rogers (1999). Our result that monetary
policy shocks do not seem to be important for exchange rate fluctuations
is compatible with the weaker result of Faust and Rogers (2003) who state
that the percentage might be anything between 8 and 56. However, since we
are imposing more identifying assumptions, we find a narrower range than
they do. Our results are also compatible with Kim and Roubini (2000) who
estimate a percentage of 5, 16 and 17 at long horizons.

22

<<   22   23   24   25   26   27   28   >>

More intriguing information