This may be a basic question, but what I (still!) find puzzling is that
using the commands:
(a) xi: regress depvar varlist indivdummies timedummies
and
(b) xi: xtreg depvar varlist i.indiv i.time
I get the same fixed effects (which may be somewhat logical) but also
the same estimated coefficients for varlist, which I don't find that
logical... I'd expected to have different effects (at least on
varlist), given that the means are subtracted.
It sounds very fishy to include i.indiv in a fixed effects regression
model (if that is indeed what you are doing: xtreg, fe) as those
effects are already being incorporated in the model (by demeaning). For
the general equivalency of regression with dummies vs the fixed effects
transformation, see any good econometrics text (e.g. Greene).
