I am using panel data with 7 countries and banks related data over the period of 2009-2021, however some counties have some banks established later making it unbalanced panel. I have multilayer panel i.e. I am analyzing banks specific factors on performance of banks for all selected country with each country having number of different banks. I am using Stata software. I have 11 regressor. As I am exploring impact of factors so none of my variables are main variable. Please guide about the following issues:
1. what are the compulsory steps for panel data analysis for unbalanced data?
2. How many variables should turn out to be significant out of 11 is necessary in panel regression with fixed and random effect model ( xtreg with fe and re)? Initially I applied pooled regression, but my all variables appear insignificant except two variables, then I applied multicollinearity test “VIF” (after settling it by throwing few regressors, it appears to have value less than 10) but my variables are still insignificant (6 out of 11 are significant or even lesser).
3. While settling or removing multicollinearity, although my no. of significant variables increases, I start facing issue of heteroskedasticity and autocorrelation. How to settle these two issues?
4. Generally, after applying pooled regression I have more significant variable then in panel regression with fixed and random effect model. how do I know when I have to shift from pool to fix or random effect modelling?
Please tell me how I go through all these issues. And tell me what the required tests and their command on are to be used in stata? I am new to it.
1. what are the compulsory steps for panel data analysis for unbalanced data?
2. How many variables should turn out to be significant out of 11 is necessary in panel regression with fixed and random effect model ( xtreg with fe and re)? Initially I applied pooled regression, but my all variables appear insignificant except two variables, then I applied multicollinearity test “VIF” (after settling it by throwing few regressors, it appears to have value less than 10) but my variables are still insignificant (6 out of 11 are significant or even lesser).
3. While settling or removing multicollinearity, although my no. of significant variables increases, I start facing issue of heteroskedasticity and autocorrelation. How to settle these two issues?
4. Generally, after applying pooled regression I have more significant variable then in panel regression with fixed and random effect model. how do I know when I have to shift from pool to fix or random effect modelling?
Please tell me how I go through all these issues. And tell me what the required tests and their command on are to be used in stata? I am new to it.
Comment