Spurious moments when using pruning

Stochastic simulations
1

Hi everyone,
Maybe this question has already been asked, but I can’t find the answer.
I have a DSGE model which when running first order approximation returns me correct theoretical moments (I mean at least the mean of my variables is equal to their steady state values). But when I run second order approximation using pruning the moments for my variables that I get make no sense. What could be the problem?

The results that I get using pruning:

THEORETICAL MOMENTS BASED ON PRUNED STATE SPACE
VARIABLE MEAN STD. DEV. VARIANCE
pie -1.9643 0.0979 0.0096
R -1.9842 0.1004 0.0101
GDP 64.1307 1.6299 2.6565
y_ex -7.3825 0.3194 0.1020

2

At second order, the quadratic terms will move the mean away from the steady state.

3

Thanks professor, so if I get the mean of nominal interest rate is -1.9842, does it mean that I have a mistake in my model?
And if it is not a mistake, is there a way to fix it anyway?, because I need to use the simulated points from this model in another model as an initial values.

4

The particular numbers are strange. They suggest a scaling issue in your shock standard deviations.