Hi, is there any ways to check whether estimated parameters (in my case, Bayesian DSGE estimation) are statistically significant?

According to dynare output, we can see t-statistics. I am not sure whether I just check them or not.

Also, I am confused of how they are computed given that we assume that each parameter follows distribution that is different from each other.

Significance testing in the context of Bayesian estimation is not well defined. The highest posterior density interval will deliver you a “Bayesian confidence” interval in which the parameter lies.

Two additional remarks:

- The t-statistics are only there after mode-finding and assume asymptotic normality.
- If your prior assigns zero mass to a parameter being 0, it will by construction be “significantly different from 0”