st: abnormal trading volume when there are a lot of zeros

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

Dear all,

I have an econometric question but I believe it is relevant for people
using Stata to perform event studies.

I have the trading orders of some groups of investors, and I would
like to estimate an abnormal trading volume measure around some
earnings releases. To do so, I need to have a "normal" trading volume
measure for those groups in the pre-announcement period.

The problem is that for some stock/groups I have many days with zero
volume in the pre-announcement period, so I do not know if I must
include those zeros in the computation of the average "normal" trading
volume prior to the event. If I include them, I mechanically have a
smaller measure of normal returns. If I do not include them, then two
stocks with the same average trading volume, but not the same number
of observations in the pre-announcement period, will have the same
normal measure.

Do you have any ideas/references?

Many thanks!
*
*   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/