How I would go about setting up the dummys you describe is:
gen byte f=.
levels firmid, local(lev)
foreach f of local lev {
replace f=firmid==`f'
bys year: egen byte firm_`f'=max(f)
replace firm_`f'=firm_`f'-f
}
drop f
(note: if you want firm_f to be 1 for all records with firmid f, then you don't need the -replace- statement above).
I think this would be quicker than the approach you describe (which could be done with nested -levels- and -foreach- commands as described by Fred Wolfe).
David
-----Original Message-----
From: Robin Luo [mailto:[email protected]]
Sent: 26 February 2005 05:28
To: [email protected]
Subject: Re: st: How to transfer a variable's value into a macro (in a
panel dataset)
Sorry for the confusion. Let me explain the situation in more details:
I have a variable called "firmid" and I generate a set of dummy
variables from it. Now I need to use the value of the varialbe
"firmid" to locate the dummy variables and do some calculations but
the location part confuses me. Basically, my data has the following
structure:
firmid year firm_100 firm_101 firm_102 firm_103 firm_104 ......
100 1980
101 1980
105 1980
121 1980
124 1980
101 1981
103 1981
196 1981
The data has been sorted by year and my goal is to check in each year
which firms co-exist with a certain firm - for those coexisting, their
respective dummy variables will be changed to 1 accordingly. For
example, firm 100, 101, 105, 121, 124 coexist in 1980. Therefore, for
the record where firmid=100 and year=1980, variables firm_101,
firm_105 and likewise will be changed to 1. My plan is that for each
year group check the values of "firmid" and then output these values
to a local macro, which can be used to locate the corresponding firm
dummy variables. But, as stated in my firm mail, I was stuck in how to
do this macro set-up in a groupwise environment.
Many thanks!
Robin Luo
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
