News shocks and monetary policy: Kobayashi and Nutahara2010

[ludog]

Hi all,

I try to reproduce some results of Kobayashi and Nutahara (2010), "Nominal Rigidities, News-Driven Business Cycles, and Monetary Policy", B.E. Journal of Macroeconomics (note that the published version is different from WP). http://www.degruyter.com/view/j/bejm.20 ... 1.2094.xml

More specifically, I try to reproduce Figure 1 (shock on growth of productivity) and Figure 3 (shock on level productivity) when the news realize (then it should be easy to reproduce graphics when the news does not realize). The only difference is that I did not linearize the model and then there are recurrence and price dispersion for sticky prices. But since dynare linearize with stock_simul, results should be similar.

I found a way to reproduce Figure 3 by using old version of adjustment cost function (sigma_phi=0.2 instead of 1.01)...
For Figure 1, I did not found a way to reproduce it. It is hard to understand wheither the mistake is in my code or in the paper (I tried to change other parameters). If it is in my codeI would guess that it is from the shock on growth of productivity (g_1) since I am able to reprocude results for the other shock.

In case of Figure 1, the dynamics before the news realize are similar but turn to be different when the news realizes.

I did a mod file which allow to choose old parameters (sigma_phi=0.02) or new parameters (sigma_phi=1.01), or choose another adjustment cost function. One can choose also if you want to save graphics (Graphs=1 or 0). The mod file also generate a tex file describing the model (easier to read than model in mod file)

You can find attached the mod file as well as the model in pdf generated by dynare and pdf for graphics:
- Figure3_oldSigmaphi.pdf: level productivity news bshock: Reproducing Figure 3 of the paper by using sigmaphi=0.02 instead of 1.01.
- Figure3_newsigmaphi.pdf: level productivity news bshock: Same model and same shock but with sigma_phi=1.01.
- Figure1_oldsigmaphi.pdf: growth productivity news shock with sigmaphi=0.02: do not reproduce Figure 1.
- Figure1_newsigmaphi.pdf: growth productivity news shock with sigmaphi=1.01.
- Figure1, Figure 3 and Model of Kobayashi and Nutahara 2010 (pages 9, 12 and 22 of the paper).

Can anyone help me?

[ludog]

Since it is not possible to attach more than 5 file by post I attach the file with pages 7, 9, 12 and 22 from the paper to this post.
I also attache here Figure3_Newsigmaphi.pdf

[jpfeifer]

I have only skimmed the paper, but a growth rate shock will permanently shift the steady state. If IRFs are for the log-levels relative to the old steady state, you have to add the trend change back. This might explain the difference. For the growth shock, the IRFs in the paper look permanent when the technology shock actually realizes. Your IRFs, in contrast, are stationary.

You might want to take a look at the AguiarGopinath2007.mod on my homepage on how to generate IRFs after a permanent shock from the detrended model. Moreover, you can find an example on how to generate a pure news shock: https://sites.google.com/site/pfeiferecon/dynare

[ludog]

Dear Johannes,

Thanks a lot.
It explains why it did work for variables with trend before the news arises and then it did not work only for those variables!
I attach to this message the last .mod file for whom may be interested (there is not the false news part but it should be hard to add).

Cheers,
Ludo