Help: Replication of Jaimovich and Rebelo

[bigbigben]

Dear All:

Attached is my code trying to replicating the paper "Can News Drive Up Business Cycles" by Jaimovich and Rebelo. I log-linearze the system by hand. Since they don't specify the functional form of investment and utilization adjustment cost, I use simple quadratic setting. The parameters are named following traditions. I checked the steady states and the co-efficients in the linearized system, and they seem to be right, except some rounding errors (at the magnitude of 1e-15).

However, the generated IRFs are quite different from what JR get in the paper. I looked into details. The most problematic part of my code is the NEGATIVE eigenvlue in the system, which looks weird. The transition matrix shows huge response to technology shocks.

If you can find any bugs in my code, I really appreciate you help..

[bigbigben]

Here is another version of the code, but it doesn't work either. The funny part is that there are only 15 endogneous variables, but dynare thinks there are 17. I don't know how to fix it.

[bcarton]

An error on line 96-99. Put :

shocks;
var e_ng;stderr 0.01;
var e_nz;stderr 0.01;
end;

[J.Höffler]

If you are interested in replication you may also want to take a look at this:
viewtopic.php?f=2&t=5926