I am little bit confused about the ranking of the application of diff and FE/RE estimations. Shall I start with FE/RE first and then switch to "diff-in-diff analysis" as the interaction phase of the FE estimation? Or shall I apply two approaches separately?

My Model : Y= B0+B1MILEX_GS+B2GDP_GS +.time*treat+time+treat+ (other covariates)+e

total observations (balanced) 1411
54 countries
4 regimes
5 income_class groups

time variances: post-coldwar, postGFC(Global Financial Crisis)
treatment : 1 for Emerging Powers, 0 for Great Powers

puzzle: the impact of GFC on Military Expenditures by country/powergroups/income-class groups/ time
DV: Military Expenditures.

Thxs for your time & help