st: xtreg vs feldvreg

[Enrico Cavale <tolacahc@gmail.com>]

Dear Statalist members,

I want to estimate the effect of a treatment T in different months of
the year. For this purpose I created 12 dummy variables corresponding
to the interaction of the treatment T with each month dummy. The model
I want to estimate has individual-by-year  and exact-date fixed
effects.

Given the memory requirements I tried to use the command felsdvreg
with the option cons. In the results one of the regressor is dropped.
Here I report how the command I used looks like:

felsdvreg y (T*January) (T*February) ...  (T*December),
i(individual_by_year) j(exact_date) f(feff) p(pef) m(mover) g(group)
xb(xb) r(res) mnum(mnum) pobs(pobs) cluster(individual) cons


Once, I had the possibility to allocate more memory to my data I tried
to use the standard xtreg command. This time all the regressors were
there. By the way the coefficient  previously dropped was even highly
significant. Here I report how the command I used looks like:

xtreg y (T*January) (T*February) ...  (T*December) _Iexact_date_*, fe
i(individual_by_year) cluster(individual)


Now my question is: what's going on with the felsdvreg command? What's
the difference between feldvreg and xtreg that can explain this
difference in the results?

Thanks,
Enrico
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/