Hi,
I have panel data model for 31 years period . In the model almost all the control variables have unit root process.
I'm suggested to estimate the model in log differences (growth rates) as it removes the unit root process.
I have used fixed effects model than I'm moving to GMM due to endogeneity in the model.
When I'm using DifGMM or GMM Sys on log differenced model it distorts the results however, the results of GMM estimator on log model are consistent with fixed effect model results and theory as well.
As far as i understand the (default) differencing in GMM removes the unit root.
Do I still need to run the GMM on the log difference model or I can use GMM on log model ?
I will highly appreciate your guidance.
I have panel data model for 31 years period . In the model almost all the control variables have unit root process.
I'm suggested to estimate the model in log differences (growth rates) as it removes the unit root process.
I have used fixed effects model than I'm moving to GMM due to endogeneity in the model.
When I'm using DifGMM or GMM Sys on log differenced model it distorts the results however, the results of GMM estimator on log model are consistent with fixed effect model results and theory as well.
As far as i understand the (default) differencing in GMM removes the unit root.
Do I still need to run the GMM on the log difference model or I can use GMM on log model ?
I will highly appreciate your guidance.
