Hi all,

I am conducting an analysis with T=32 (quarters) and N=52 using difference GMM. I have my results, but have now realised I didn't consider if my variables were stationary.

Does the stationarity of my variables matter for diff-GMM? I have tried running the fisher unitroot test and it only rejects the null of unit roots in all panels for one of my variables. Should I be concerned with this?

Most other papers I have seen doing the same analysis as me do not conduct unit root tests, but I am not sure why. Ideally, I suspect I should do one, but I'm not sure of the significance.

Many thanks in advance,
Jaspal