Hello everyone!

I construct a DSGE model which have two product sectors and incorporate financial accelerator, my research is focusing on the endogenous variables’ response to the government consumption shock e_ig and government investment shock e_cg, and I have run the dynare code, but the impulse response graphs are strange, a lot of variables are not converging to 0, especially for the gross inflation(pi), inflation in sector1 (pis) and inflation in sector2(pin), and also the relative price st(p1/p2) is fluctuating.

I have upload the mod file, can someone help me to check what’s wrong with the DSGE model, is the parameter value not appropriate or are there other problems?

Thank you very much!!Mymodel.mod (12.8 KB)

Your model simply has a lot of endogenous persistence. If you increase to `irf=100`

, you will see the mean reversion.

Dear professor Jpfeifer

Thank you for your reply. According to your suggestion, I increase the irf periods to 100, and most variable have converge to 0, but there are some variables such as st(relative price) still not converging to 0. So i think maybe there are still some problems in my model which I can not find out.

The reason I set the irf periods=30 are: (1)In general, we observe the fiscal policy effect in short terms, if the irf periods=100, that means 25 years, and it is not realistic; (2) In most papers, I see the author often set the irf periods less than 40.

Besides, I have a question about the parameter in financial accelerator, in my model, I calibrate the value of elasticity of the external finance premium(psi_s, psi_n), but someone tell me this value can not be calibrated, it must derived from the equilibrium steady state value of the model, is that true?

Looking forward for your reply, thank you very much!

Best wishes!

As I said, there is large persistence in your model. If that bothers you, you need to find out why that is the case. Regarding the elasticity, it typically depends on the functional form assumed in the nonlinear version of the model. But in a linearized model, it is usually a free parameter you can choose. What you need to pin down is the steady state level of the finance premium, but not its elasticity.

Thank you for your reply!

The large persistence you mentioned is the government investment spending shock persistence(rho_ig) and government consumption spending shock persistence(rho_cg)?

Best wishes!

Not necessarily. It can either come from the persistence of the exogenous processes or the features of the model that give rise to endogenous persistence (e.g. very strong habits in consumption or investment adjustment costs)

Thank you for your reply!

I will adjust the relevant parameters as well as the model, then try to run the mod file again, I hope it can get a better result~

Thank you for your help, best wishes!