Hello, I have a panel dataset with N>T where single variable prior-estimation are cross-sectional dependent when running xtcd. They do not have a unit root and are not cointegrated (in panel).
At the same time, the residual of a basic regression including of all variables as well as a time dummy does not show cross-sectional dependence when running xtcd.
The residual of the same regression without time dummies shows cross sectional dependence.
In this context, can i run a system GMM with xtabond2 including time dummies and be relatively "safe" on the results interpretation and significance ?
Thank you,
Luca
At the same time, the residual of a basic regression including of all variables as well as a time dummy does not show cross-sectional dependence when running xtcd.
The residual of the same regression without time dummies shows cross sectional dependence.
In this context, can i run a system GMM with xtabond2 including time dummies and be relatively "safe" on the results interpretation and significance ?
Thank you,
Luca
