I have a TANK model and am trying to replicate a COVID shock (shocking hours and consumption). I can get my head around most of the impulse response functions I get. However, a 300% wage increase on impact seems odd to me. I am assuming the rise in inflation and the interest rate is a consequence of that. A mod file and the respective IRFs are attached. Thank you for your time. IRF2.pdf (5.7 KB)
IRF1.pdf (6.5 KB)
TANK14_revised.mod (7.4 KB)
Have you tried whether the model works well for other types of shocks?
It depends. I have tried numerous specifications/calibrations, and I see one pattern: In particular, for transitory shocks, there often is a strong impact response you would not expect but the other periods seem reasonable. For example, a transitory TFP shock causes GDP (and other procyclical variables) to decrease sharply on impact, but only on impact. From the second period onward, the responses make sense. The impulse responses to permanent shocks are along the lines of what you observe in the literature with similar model specifications.
The try starting from a simpler model where you have more of an idea how things are supposed to look like. What you describe can happen if there is not much endogenous persistence in the model so that most adjustment happens in the first period.