Looping over two variables at once

[salma ktat]

Hi !
I have been using STATA not for so long. And now I'm trying to loop over 2 variables at the same time in order to generate a new variable. My data, as shown in the attached dta file looks like lpermno trt1m (which is the monthly stock returns) year mydatemonthly rf.
The question is that, for every lpermno, I want to generate monthly excess returns=trt1m/100-rf.
I tried doing the following but it's not working:

egen group = group( lpermno mydatemonthly)
su group, meanonly
foreach i of num 1/`r(max)' {
gen ret= trt1m- rf if group == `i'
} //invalid numlist has too many elements

Then, I tried this but in vain:
egen group1 = group(mydatemonthly)
su group1, meanonly
local maxrf = r(max)

egen group2 = group( lpermno)
su group2, meanonly
local maxlpermno = r(max)

foreach i of num 1/`maxlpermno' {
foreach t of num 1/`maxrf' {
gen ret= trt1m/100- rf if group1 == `i' & group2 == `t'
}
}

Any suggestions are welcome !

[Matthew J. Baker]

Salma --

When I try and run your code, I run into one obvious problem: you are trying to generate the variable ret each time through the loop. So, the question is if you intend to generate only one variable, ret, in which case, you should do something like:

egen group = group( lpermno mydatemonthly)
su group, meanonly
gen ret=.
foreach i of num 1/`r(max)' {
replace ret= trt1m- rf if group == `i'
}

Or if you want a separate variable for each, you don't have to pregenerate the variable. You would just do:

egen group = group( lpermno mydatemonthly)
su group, meanonly
foreach i of num 1/`r(max)' {
gen ret`i'= trt1m- rf if group == `i'
}

I don't know if that's the whole problem, but I hope that helps.

Matt Baker

[salma ktat]

Matthew
That works. Thank you very much.

[Sarah Edgington]

As a side note, this is general Stata question, not a Mata question. In the future you'll likely get more readers for your question if you ask questions in the general forum.
(Unless, of course, your question really is about Mata programming)

[Robert Picard]

There's nothing in your inquiry that requires looping or grouping. Stata will perform the same operation on all observations at the same time. All you need is

gen ret = trt1m / 100 - rf

[salma ktat]

Robert !
That's right. I tried it and it just generated the same values, so you're right, it's as simple as that! Thanks!