st: cumulative moving sum when there are repeated date values

["P. Rao Sahib" <p.rao.sahib@rug.nl>]

Dear Statalist users,
I have a panel dataset of acquisition attempts made by firms in which I 
would like to calculate the cumulative number of acquisitions a firm 
completes in the past 5 years prior to
the current acquisition.
Firms enter the dataset if they make an acquisition attempt but as some 
acquisitions are later abandoned, I have an acquisition completion dummy 
“acdum” that is 1 if the
acquisition is completed, 0 otherwise” .

My data looks like this:
Acdum is the Acquisition completion dummy.



Firmid Date Acdum Cumulative

Shell 24 Jan 1993 1 0

Shell 1 Dec 1995 0 1

Shell 3 Aug 1996 1 1

Shell 3 Aug 1996 1 2

Mittal Steel 3 Jan 1993 1 0

Mittal Steel 4 March 2000 1 0

Mittal Steel 5 April 2001 1 1


I would like to count the cumulative number of acquisitions made by the 
firm in the past 5 years prior to the current acquisition (even if they 
occured in the same year ).
By searching Statalist, I found a solution using “ tssmooth, egen and a 
filter. However, these solutions involving tsseting the data and as it 
can be seen from the example above,
I have firms that sometimes announce two acquisitions as being completed 
on the same day (and I don’t have any finer time details like hours 
etc.) to tsset the data properly.
As a result I get an error message saying that I have repeated time 
values within my panel. The actual ordering of acquisitions that occur 
on the same day does not matter,
I simply need to sum up the number of acquisitions prior to the current 
one as long as they occurred in a 5 year interval, and it is OK if the 
ones on the same day are
off by one acquisition.

I tried something like this:
bysort firmid (date):gen prevacq = sum(acdum)
by firmid: gen prevacq5 = prevacq - prevacq[_n-4)

(also borrowed and adapted from a solution to a related question on 
statalist) but this
does not quite work and seems to cap the number of previous acquisitions 
at 4.

Out of desperation, I thought of manually adding a day to the date 
announced as it is the year that is important
but there are too many observations for this to be feasible…I think I 
need a loop of some sort but my STATA loop knowledge is somewhat deficient..

If anyone has any suggestions on how I can solve this problem, I would 
be very grateful!


Padma Rao Sahib
Department of International Economics and Business
Faculty of Economics and Business
University of Groningen
The Netherlands

*
*   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/