skewed distribution for shocks

[wjgatt]

Hi all,

I'm new to dsge modelling and my knowledge on the subject is very limited. I'm building a model and I want to incorporate shocks whose distribution are skewed, i.e negative shocks should occur with a higher probability than positive shocks, yet the expected value of the shocks should be 0. Can this be done in dynare, if so which distribution would be suitable (beta?) ?

Thanks in advance
William

[SébastienVillemot]

Hi,

Dynare is not able to simulate such a model. Dynare assumes a zero skewness for all shocks.

Note that for 1st and 2nd order simulations, it makes no difference whether the distribution is skewed or not, because only the first and second moments of the shocks distribution enter the approximation formula.

The difference only appears when you are doing a 3rd order simulation (which Dynare can do). But again the skewness is not modifiable from the Dynare MOD file.

Best,

[wjgatt]

Thank you very much for your reply.

[MichaelCreel]

What would happen if one used a normal shock, say e distributed N(0,1) and then created a variable that depends on the shock? For example, the variable e2 below is distributed Chi-square(1) (with mean 1 and variance 2). z is a technological shock that depends on the chi-square variable, centered and normalized. How would Dynare deal with this setup? I can confirm that dynare will apparently solve the model using order=3, and will give stochastic simulations. Is this output meaningful and reliable?.

Thanks,
Michael

...
varexo e;
var e2, z;
...

e2 = e*e;
z = rho*z(-1) + sigma*(e2-1)/sqrt(2);

[GilBurt]

Thanks for this post. I have no knowledge about model.

[ambropo]

Hi Micheal,

Thanks for your post. Can tou please post your mod file, I would like to invstigate this issue further.

Thanks!