st: Constructing Portfolios for Time-Series Regression

["Tim Streibel" <Tim.Streibel@gmx.de>]

Hello all,

I am facing the following problem:
I want to conduct an event study. I have monthly return data for several companies. Now I want to form deciles regarding several characteristics of theses companies like for example marketcapitalization. The returns on the difference of these deciles should be the independent variables of my time series regressions. So far I only want to use a subsample of alle the firms to form the mentioned deciles (only non-event-firms). The dependent variable of my regression should be the average monthly returns of the other subsample (only event-firms).

I can distinguish between event and non-event firms by a dummy variable. I will certainly be able to form the deciles and compute the returns of the resulting portfolios. However, I don't know how to "store" the computed returns in a way that I could easily use them for conducting a time-series regression...I mean I doesn't seem sensible to generate a variable since this would only result in a mass of duplicate values, which I could drop again, but there should be an easier way to reach my goal. But if I stored them in local variables I cannot use them for my regression?!

I would really appreciate every hint you could give me.
*
*   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/