The RATS Software Forum

I am trying to replicate Hilde C. Bjørnland's Monetary policy and exchange rate overshooting: Dornbusch was right after all, Journal of International Economics 79 (2009) 64–77. Attached are the codes for Sweden along with the data. I am able to replicate everything using the program but not quite getting a significant impulse responses of real exchange rate (RER) to a shock in monetary policy as in the paper. This is also true for the other set of countries used in the analysis with this program. I am not sure why this is happening? Is the program alright?

Secondly why I am able to generate good impulse responses with MCProcessIRF but not with MCGraphIRF in this case?

If the published empirical work used the same basic RATS code as Bjornland and Leitemo, then

@ShortAndLong(lr=lr,sr=sr,masum=inv(%modellagsums(model))) sigmad factor

was done using the lag sums from the OLS estimates rather than the recomputed lag sums for the drawn coefficients (where the latter, reflected in what you did, is the correct procedure). That almost certainly makes the correct error bands wider than they would be when done incorrectly.

Dear Tom, Thanks very much for the reply! The paper was all about the response of real exchange rate to a monetary policy shock which with the corrected codes are becoming insignificant for all the countries. Is there a way to get a significant impulse responses to real exchange rate with the same kind of short run and long run restrictions?

If the empirical work is done correctly, apparently not. You would have to explore different choices for measurements or embed in a larger model, or ...