I use system-GMM to assess the effects of financial conditions on macroeconomic performances. But I face to some doubts. Some of my regressors are non-stationary. So, how can I deal with?
- Can I just first-differentiate non-stationary regressors in my system-GMM? Do results remain relevant in terms of interpretation?
- Can I just use system-GMM with non-stationary regressors at level? Because it seems System-GMM is consistent with persistent variables?
- Or, is there another way to deal with this issue ?
Best regards!