Skewed shocks workaround

Hi all,

I have read the previous posts on the forum, but still have a question about the feasibility of skewed shocks in dynare.

Suppose I have an exogenous variable, X, that is mean zero with variance sigma. I appreciate that the shocks, call them epsilon, in dynare have to be Gaussian, but could I express the variable as:

X_t = f(epsilon_t)

Where the f(.) function introduces skewness into X. I understand that I would have to go to a higher order perturbation when solving the model for this to have any impact. If this is possible, does anyone know of a functional form for f(.) which can transform a normally distributed variable to a variable distributed with non-zero skewness? I found this post online which looked promising, but the inverse hyperbolic sine function is not compatible with dynare it seems.

Many thanks for any help in advance,


@wmutschl What do you think?