Weird responses to monetary policy shocks

[cesk]

Dear all,

I am working on a very standard search model which is similar to that published by Faia (2008) in the JEDC except for the fact that I have endogenous participation, informal self-employment (i.e., the shadow economy) and money through a CIA constraint on the purchases of the informal consumption good. Basically, the idea is that the representative household has to decide how many members to employ as informal self-employed, taking into account that doing so the latter do not participate to the matching process in the formal labor market and are not entitled to receive the unemployment benefit.
Now, when I run stochastic simulations with respect to a monetary policy shock, I obtain weird results which I am not able to explain.

1) If I look at the variance decomposition, a shock to the nominal interest rate does explain only the variance of inflation, wages and marginal costs. The contribution of a monetary policy shock to the variance of R is zero, despite plots of IRFs suggest this is not the case. Also, and probably even more strangely, for the same monetary policy shock some variables (inflation, wages, marginal costs and lump-sum taxes) are plotted twice with completely different responses.

2) Altough public spending in this model is completely exogenous (G=GBAR*exp(Z_G)), monetary policy shocks do affect G.

I am not new to DYNARE, but I never experienced such strange results. I suspect there is something wrong in the model, but I cannot figure out what.
Any comment or suggestion would be very appreciated. Attached there is the mod file.

Best,
Claudio

[jpfeifer]

Something with your IRF-plots is off, but I can't tell you what. When I run your files, everything looks as expected. Monetary policy shocks do not move R and G so that the results from the variance decomposition are consistent. Which Dynare version are you using?

[cesk]

Dear Johannes,

thank you for replying. I was using dynare 4.5, then today I tried to set again the path and this weird response to a monetary policy shock disappeared.
I still have that a monetary policy shock does not affect R, but I think it is due to the fact that given that new matches become productive only in t+1, then employment and output are predetermined. I will change the model to make job matches productive already in t.

Best,
Claudio