EP simulation: GDP only fluctuates in the first bit of the sample


#1

Dear community,

I would like to simulate an asymmetric search-and-matching model. While I don’t encounter any problems with stoch_simul, Dynare seems to be unable to solve the perfect foresight model when I try to run a deterministic simulation via simul(). Since simul() does not work I tried the EP method and had more success. However, when I run a simulation using extended_path(), the generated path of GDP only fluctuates during the first 10% of the simulation period, and is implausibly flat thereafter (I attached a screenshot of the simulated GDP series, please find it below). This is a bit suspicious because my exogenous shocks fluctuate during the entire simulation period. Does anyone by chance understand why this would come about? Might it be a big, or is there something that I miss? I get the same result when I simulate over a longer/short time horizon, and use a different calibration.

Kind regards and many thanks!
Paul

gdp_fail


#2

An extended path simulation is based on a sequence of perfect foresight models, so your result for EP is not surprising here. You need to understand why simul fails with your model. It is hard to tell without having a look at your model.

Best,
Stéphane.


#3

Many thanks for your help! I am using the search and matching model of Abbritti and Fehr (JME 2013), which features downward wage rigidity and is written in levels. The original code is available on the JME website, but I attached a mod file that tries to run the modul with simul(). I hope there is no stupid mistake in there :slight_smile: asym_SAM.mod (8.0 KB)


#4

This is hard to tell. Even very small shocks lead to not finding a solution. Have you tried whether simple transitions from outside of the steady state work?