Hi all,
I am doing unbalanced panel analysis where T > N. I have first done static FE/RE panel regressions, applying Driscoll-Kraay se after testing for cross-sectional dependence, multicollinearity, autocorrelation, and heteroscedasticity, and done a Hausman test, as the code shows below.
Then, after this and separate from this, I have done second-generation unit root tests, found a mix of I(0) and I(1), tested for cointegration and detected cointegration, and applied a CCE MG estimator, as the code shows below.
I wanted to ask if I am missing anything from either code for each step, and how to interpret the results of each regression performed i.e. is the static FE/RE results a short-term analysis or a rudimentary analysis? And is the CCE MG results a long-term analysis, or simultaneously a short-term and long-term analysis. Many thanks.
I am doing unbalanced panel analysis where T > N. I have first done static FE/RE panel regressions, applying Driscoll-Kraay se after testing for cross-sectional dependence, multicollinearity, autocorrelation, and heteroscedasticity, and done a Hausman test, as the code shows below.
Code:
xtcd ln_HEpc xtcd ln_GDPpc xtcd ln_Pop65 pwcorr ln_HEpc ln_GDPpc ln_Pop65, sig reg ln_HEpc ln_GDPpc ln_Pop65 estat vif xtreg ln_HEpc ln_GDPpc ln_Pop65, fe xttest3 xtserial ln_HEpc ln_GDPpc ln_Pop65 xtreg ln_HEpc ln_GDPpc ln_Pop65, fe estimates store fe xtreg ln_HEpc ln_GDPpc ln_Pop65, re estimates store re hausman fe re xtscc ln_HEpc ln_GDPpc ln_Pop65, re
Code:
pescadf ln_HEpc, lags(1) trend pescadf ln_GDPpc, lags(1) trend pescadf ln_Pop65, lags(1) trend xtwest ln_HEpc ln_GDPpc, lags(1) leads(1) bootstrap(500) xtcce ln_HEpc ln_GDPpc ln_Pop65
