I have a positive correlation in my original data between inflation and output. I as well have a posiitve correlation in the Bayesian Impulse Response Functions between Inflation and Output except for a productivity shock (which is one out of the seven shocks I have). However, in the matrix of correlations output and inflation appear negatively correlated. What can be a posible explanation of this?
Are you considering the same transformation for the empirical and model variables? E.g. the same filters?
Yes, I have transformed the variables in the same way.
When you say
what is the variance contribution of that shock? Maybe it explains most of the data.