Hello forum,
I construct a robustness check using the LIML method for models generated through sys-GMM. Upon conducting the LIML analysis, I observed slight variations in the results along with notably low R2 values.
My research is primarily focused on examining the moderating impact of different governance structures on the relationship between economic development and social progress. It is worth noting that there exists potential endogeneity between economic development and social progress, given the reciprocal relationship between development and progress.
I utilized lagged controls as instruments in the LIML and the all the independent variables in the GMM specification, and further incorporated interaction terms multiplied by exogenous variables. However, I am now grappling with several questions:
1. Is it common to obtain divergent results when employing different methodologies?
2. Could heteroscedasticity be influencing the efficiency of LIML, thereby contributing to the observed low R2 values?
3. How reliable is to multiply the lagged controls by the exogenous variables to instrumize the interaction term that contains one endogenous and one exogenous variable.
I construct a robustness check using the LIML method for models generated through sys-GMM. Upon conducting the LIML analysis, I observed slight variations in the results along with notably low R2 values.
My research is primarily focused on examining the moderating impact of different governance structures on the relationship between economic development and social progress. It is worth noting that there exists potential endogeneity between economic development and social progress, given the reciprocal relationship between development and progress.
I utilized lagged controls as instruments in the LIML and the all the independent variables in the GMM specification, and further incorporated interaction terms multiplied by exogenous variables. However, I am now grappling with several questions:
1. Is it common to obtain divergent results when employing different methodologies?
2. Could heteroscedasticity be influencing the efficiency of LIML, thereby contributing to the observed low R2 values?
3. How reliable is to multiply the lagged controls by the exogenous variables to instrumize the interaction term that contains one endogenous and one exogenous variable.
