st: understanding weights in a -xtreg panel regression

[jean-luc morin-chesnel <jeanluc.morinchesnel@gmail.com>]

Dear all,

I would like to understand what the weights do in a fixed effect
regression using -xtreg in Stata.
So assume I run the following model

Y_it=Alpha + Beta*X_it + Epsilon_i + Epsilon2_it

(where I keep Stata's definition of a fixed effect regression: i.e. a
baseline constant and an individual time-invariant error term)

Consider that Y_it is a return earned on some investment done by
investor i at time t, and that I would like to weight the returns by
the amount engaged on each investment (M_it). That is higher returns
earned when the amount engaged is high should be more "important" than
higher returns earnied when the amount is lower.

Should I type  -xtreg Y X [aw=M], fe ?

Are the weight defined separately for each investor? Which weights
should I use? I am confused by the definition of aweight (or analytic
weights, are weights that are inversely proportional to the variance
ofan observation; that is, the variance of the jth observation is
assumed to be sigma^2/w_j, where w_j are the weights.) Are they
appropriate in that situation?

Many thanks for your help and remarks
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/