Jumpy IRFs in SOE Model

Hi all, I am developing an SOE model that incorporates de banking sector of Gerali et al (2010). I also included the possibility of banks to borrow from abroad. The model works but the IRFs look funny, as in some variables jump a lot at the beginning. Is this a problem? I believe that this could be related to the collateral constraint of the entrepreneurs, that are the ones that borrow from the banks. My concern is that it might be encountering an occasionally binding constraint and that would be the reason for the IRFs to jump. Best
Mariosolvess8b.m (590 Bytes) twocountrymodel008bghh.mod (8.3 KB)

Hi magonfru1,

Do you have smoothing in your interest rate rule?

Those IRFs indeed look a bit strange at the beginning. In contrast to what @cmarch suggests, it is very unlikely to derive from the policy rule. Something else must be going on, because there is a spike at time 2 that most probably should not be there. The most likely candidate is a timing error somewhere, for example for bank capital. Usually the best way is to work with a simpler model to see where the problem comes from.

Yes it does, and is not related to that. Thanks!

Thanks for your answer. I have actually developed the model starting from a very simple version. I include now a version in which the entrepreneurs do not borrow to finance its capital purchases (version 5d), and another version in which the entrepreneur borrows subject to a collateral constraint (version 5b). As you can see, without borrowing the irfs are not jumpy, but by including borrowing the iris turn jumpy.

I have not been able to develop a version in which the entrepreneur borrows without collateral constraint as it leads to entrepreneur consumption equal to 0.

solvess5b.m (558 Bytes) twocountrymodel005bghh.mod (6.7 KB) twocountrymodel005dghh.mod (6.3 KB)

I seemed to have solved the problem, I have forgotten to multiply the foreign interest rate by 1+ the expected depreciation of the exchange rate.

I have one question still. What would happen in Dynare when I have an occasional binding constraint. Would dynare be able to solve it? I know that Dynare is not designed to solve that. I am concerned because I don’t know how to recognize when that happens.

There are toolkits like Occbin to handle this.