Hi all, I am quite confused with the panel econometric methods concerned, and would appreciate your help.
My dataset is an unbalanced panel where T > N.
I have done a cross-sectional dependence test (detected cross-sectional dependence in all my variables), a second-generation unit-root test (my dependent variable is nonstationary, and I have a mix of I(0) and I(1) explanatory variables), and I have done a cointegration test (detected cointegration). From here, I am not sure what to do.
I have lightly read into CCE and DCCE using the commands: xtmg; xtmg ..., cce; xtpmg; xtcce; xtdcce2.
I am confused on which command is best to use, so given the findings I have outlined above of my data, which is recommended? I understand for xtdcce2 that it is dynamic rather than static but it also provides a result when a lagged dependent variable is not included as an explanatory variable, so I am further confused by what the results actually mean for that.
I initially was wanting to just do static fe/re estimations, applying Driscoll-Kraay se to mitigate cross-sectional dependence, but have seemed to find papers going way out of my depth econometrically, and I would quite like to try find more robust results rather than settling for results that are not accounting for unit-roots and nonstationary and the consequent issues.
Many thanks.
My dataset is an unbalanced panel where T > N.
I have done a cross-sectional dependence test (detected cross-sectional dependence in all my variables), a second-generation unit-root test (my dependent variable is nonstationary, and I have a mix of I(0) and I(1) explanatory variables), and I have done a cointegration test (detected cointegration). From here, I am not sure what to do.
I have lightly read into CCE and DCCE using the commands: xtmg; xtmg ..., cce; xtpmg; xtcce; xtdcce2.
I am confused on which command is best to use, so given the findings I have outlined above of my data, which is recommended? I understand for xtdcce2 that it is dynamic rather than static but it also provides a result when a lagged dependent variable is not included as an explanatory variable, so I am further confused by what the results actually mean for that.
I initially was wanting to just do static fe/re estimations, applying Driscoll-Kraay se to mitigate cross-sectional dependence, but have seemed to find papers going way out of my depth econometrically, and I would quite like to try find more robust results rather than settling for results that are not accounting for unit-roots and nonstationary and the consequent issues.
Many thanks.
