Contrast in results

The results of POLICY AND TRANSITION FUNCTIONS and CORRELATION OF SIMULATED VARIABLES are not match, for example based on the first one m and sp have positive and strong correlation and based on the second one they have negative correlation.
Thanks for help.
aliye.mod (3.0 KB)

That is not surprising. The coefficients in the policy functions are the ceteris paribus effect, i.e. holding all other variables constant. The unconditional correlations reflect the movement of all variables to shocks happening all the time.

so which one should I rely on for making a conclusion?

It depends on the question you are trying to answer. It’s like the difference between conditional and unconditional moments.

Thanks a lot for your time.