Dynare forums

Dear all,

I am trying to replicate in a simple way the model of S. Brissimis and I. Skotida (2008, JIMF): http://www.sciencedirect.com/science/article/pii/S0261560607001362
It consists in a simple 3-equation model: Phillips curve(5), AD curve and implicit instrument rule ((5),(6),(10)). For the multicountry version: the model consists in 5 equations: for each of the 2 countries('e','d') a Phillips and AD curve and a common instrument rule ((14)-(17),(23)).

While the version of the monetary union with aggregate data works fine, the multicountry version does not: It only incorporates 3 eigenvalues larger than 1 for 4 forward looking variables [x_e(+1), ppi_e(+1), x_d(+1), ppi_d(+1)].

Does anyone have an idea what could be the origin of the problem?

Many thanks in advance!

Replicating a paper is hard. The only way I see is rechecking all equations in your paper and their paper as well as the parameterization. You could use the TeX-capabilities of the Dynare unstable version to make your life easier. See e.g. https://github.com/DynareTeam/dynare/blob/master/tests/TeX/fs2000_corr_ME.mod

Thank you for the tip and the fast reply, Johannes! The Tex option basically enables me to see the outputs and e.g. parameter tables in latex form, rigth? Does it also return me something if the model is not working yet?
But you think, the problem can only be due to adopting their equations or paramters incorrectly, so that the necessary condition for the eigenvalues to be bigger than 1 is not fullfilled? Or could it also be due to the fact that they assume for example the natural rate of interest as constant, while I have seen it usually as a time dependent variable in similar models replicated here? Many Thanks again!

You can generate TeX output for the model equations and the parameter tables independently of the Blanchard-Kahn conditions.

The natural rate of interest is typically an exogenous process. It does not matter whether it is constant or time-varying. So this should not be an issue. The problem is most likely due to i) you incorrectly implementing their model or ii) their presentation of the model in the paper being wrong.

Ok! Thank you for your help!