[Carlo Lazzaro]

Abebe:
add one predictor at time and see when Stata starst choking.
Moreover, if you wisely use -fvvarlist- notation, the -xi.- prefix is redundant.

[Abebe Desta]

Thanks indeed Carlo! Should i check regressing each explanatory variable separately?

[Carlo Lazzaro]

Abebe:
not quite.
You should start your regression model with one, two, (and so on) predictors, see when Stata starts complains, investigate what's going on and change your model specification accordingly.