IRFs inconsistent with moments of simulated variables

Dear Dynare team,

I am working on a model based on Gertler et al. (2007) - the open-economy version of Bernanke et al.'s (1999) financial accelerator model. A key difference is that my entrepreneurs are farmers who use fertiliser (instead of capital) with a 100% depreciation rate, and which must be imported from abroad; the production technology is CES. I attach the mod file.

My problem is that there are a number of variables, such as A (TFP) or Y (output) which the IRFs show as always being below their steady-state values, but for which the mean of the simulated variables is above the steady-state values; the MATLAB workspace window also shows maximum values for these variables being significantly above their steady state. Could you please help me understand where this problem arises from?

Many thanks in advance,

S├ębastien
FFFF.mod (15.4 KB)