The fact that you can specify a panel variable as well as a time variable is also quite explicit in the help for -tsset-.
Nick
[email protected]
Martin Weiss
h xtset
Kaspar Dardas
I have a data set with 4 variables (date ret market_ret id) which
looks as follows:
date ret market_ret id
01/02/2009 1% 2% 1
02/02/2009 1.20% 2% 1
03/02/2009 -1% -4% 1
04/02/2009 -5% - 3% 1
01/02/2009 8% 2% 2
02/02/2009 9.00% 2% 2
03/02/2009 -16% -4% 2
04/02/2009 0.3% -3% 2
Please note that date repeats/overlaps - (not for all ids but for some).
I try to use a GARCH process to estimate abnormal returns for each id
with the following loop:
forvalues i=1(1) 1137 {
l id gvkey if id==`i' & dif==0
arch ret market_ret, arch(1) garch(1) if id==`i' & estimation_window==1
predict p if id==`i'
replace predicted_return = p if id==`i'
drop p
}
My loop breaks because there is no time variable set. Whenever, I try
to set "date" as time variable with the command tsset I get the
following error message:
. tsset date
repeated time values in sample
Since this is an event study I will have repeated/overlapping dates in
my data set which is unavoidable. However there is NO repetition of
the same dates within any single id set. Thus, is there any way to set
a time variable for subsamples within a large data set such as
"by id: tsset date"?
The above obviously does not work but I hope that it somehow describes
what I need. Any help would be appreciated.
*
* 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/
