Am analyzing FDI effect on Women empowerment dimensions by taking 105 developing countries data from 1990 to 2017 (2940 total observations). Would you please suggest best statistic technique from ARDL, FE or SUR.
It depends on your data. If your data are stationary at I(0) and I(1) you may use an ARDL model.
You may also check the compatibility with the FEM using redundant fixed effect test, where you can also use the option Period SUR or Cross-section SUR in order to remove ex-ante the heteroskedasticity. Usually, Period SUR / Cross-Section SUR improves the results of the model.