Endogeneity issues
Posted: Fri Aug 12, 2022 2:47 pm
Dear all,
I have a longitudinal dataset of firms and apply a 3-level logistic regression with time as level 1, firms as level 2 and regions as level 3 using Bayesian methods with the runmlwin command in Stata. I now wonder how I can account for possible endogeneity issues in my model that might arise from correlations between level 2 firm characteristics and omitted regional variables at level 3. For example, I have a variable on the R&D expenditures of firms. However, the decision of firms on how much money they spend on R&D is likely to be influenced by regional factors that I cannot observe. Is there any established procedure how I can deal with these kind of endogeneity issues? And if so, how can I implement this with the runmlwin command?
I am happy about any help!
Best,
Thore
I have a longitudinal dataset of firms and apply a 3-level logistic regression with time as level 1, firms as level 2 and regions as level 3 using Bayesian methods with the runmlwin command in Stata. I now wonder how I can account for possible endogeneity issues in my model that might arise from correlations between level 2 firm characteristics and omitted regional variables at level 3. For example, I have a variable on the R&D expenditures of firms. However, the decision of firms on how much money they spend on R&D is likely to be influenced by regional factors that I cannot observe. Is there any established procedure how I can deal with these kind of endogeneity issues? And if so, how can I implement this with the runmlwin command?
I am happy about any help!
Best,
Thore