Dynare forums

Hi All,

I solved my model with Dynare and with Klein P.(i think 2000 JEDC) i got different results. Not so much qualitatitatively but for sure quantitatively.
Which method for LRE models Dynare use? My results should be the same? (In terms of eigenvalues determinacy conditions etc?)

Thanks.

Petros

Hi,
could you be more precise about the differences you mention? Where do they occur, e.g. the eigenvalues, the policy functions etc.? If you mean in the stoch_simul command, which order of approximation do you use? The Klein code usually uses a first order approximation while Dynare uses second order by default.
Also note that different solution techniques are known to give different but largely similar results. The difference should not matter quantitatively (see http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.126.2298&rep=rep1&type=pdf).

Hi,

Thanks a lot for your answer.I will check your reference.Dynare uses Schmitt-Grohe and Uribe method 2004 JEDC.Their code up to a first-order approximation as they mention is very close to Klein. However, when i use Dynare first order i come up with different eigenvalues and policy functions.The most crucial is that i come up also with different critical values for some policy coefficients (i.e. fiscal rules) but may i had made a mistake.I will check it again,and report.Thanks again for your time and the reference.

Petros.

Another issue that might give rise to differences it the way Dynare substitutes lags/leads larger than 1 with auxiliary variables. This may make it hard to compare policy functions as some states may appear in a different for, i.e. already plugged in.