GK model with uncertainty shock


I am trying to see the effect of uncertainty on capital quality shock in a standard Gertler and Karadi (2011) model. I took the sample code FA.mod and added the uncertainty shock (sigma_ksi in line 157) in FA_uncert.mod and run the simulation following @jpfeifer’s replication on Basu and Bundick (2017).

Maybe it is about theory, but isn’t that uncertainty shock should be contractionary? What I get is increases in output, consumption, and investment. It seems like the asset price goes down, and that decreases net worth of the bank, but I am not sure how to interpret the results. Is this because you want to invest now before you see capital with very bad quality?

Thanks in advance!

FA.mod (7.5 KB)
FA_uncert.mod (10.5 KB)

Maybe there is a Jensen’s Inequality effect due to you using exp(ksi). Have you tried centering ksi at 1?

ksi=(1-rho_ksi)+  rho_ksi*ksi(-1)-sigma_ksi*e_ksi;

Thanks for your reply! I just tried with ksi as suggested in this new file. I guess I get the same results.

By Jensen’s inequality effect, you mean households would do precautionary saving on deposit? It seems like deposit is going down (which makes some sense for an increase in consumption?).

I was thinking I would have decrease in investment and borrowing (or more deposit) with some similarity with Ferandez-Villaverde et al. (2011AER)?

Here is the new file. Would you please take a look? FA_uncert.mod (11.5 KB)

  1. No, I mean that the exp()-formulation results in a lognormal distribution where the mean increases in the variance.
  2. I don’t really have a good intuition for your model. It may be a similar mechanism as in Fernandez-Villaverde et al… Uncertainty can be expansionary depending on where it hits and how strong frictions are. That is the point of Basu/Bundick.

Thank you for your reply and comments! I guess I will have to think about it.