simulation not working (running only once)

[Aditi Roy]

Hi, Stata people. I am really sorry to post this again. I am assigning random treatment and then want to run the regression 1000 times. However, now my regression is only running for once..

Please, Please help

set more off
set matsize 800


cap program drop model
set seed 6000
set obs1280
program model, rclass

generate random=uniform()
sort year random
generate post=0
replace post=1 in 1 in 1/29 // for year 2010
replace post=1 in 296/337 // for year 2011
replace post=1 in 578/619 // for year 2012


reg wheat post i.country_id i.year i.country#c.line_time_trend [aw=ypop], cluster( country_id )


return scalar tau_coeff = _b[post]
return scalar tau_se = _se[post]
return scalar r2 = e(r2)
exit
end

simulate coeff=r(coeff) se=r(se) r2=r(r2), reps(1000): model


sum

[Joseph Coveney]

You don't show any output, but my guess is that Stata balks the second time it sees this line

generate random=uniform()

And if you fix that, Stata will balk a moment later, when it reaches

generate post=0

Try using temporary variables inside your program so that they aren't left lying around (as already defined variables) afterward.

Also, what is the line

set obs1280

supposed to be doing for what appears to be an already loaded dataset? And what's

exit
end

all about?

You might want to look into your data-management and into the specification of your regression model, too: you're not controlling the sort order of country_id, and you don't include both main effects of the interaction term.

[Aditi Roy]

Thanks for your reply, Joseph.

Yes, I run this on an already loaded dataset. Generate random= uniform () and then I don't control the sort order of country id and main effects of the interaction term and random and years because I want to randomly assign treatment to the different no of observation treated in year n (where the no of the assignment in year n should be same as in the base dataset)

The post is my treatment variable, so I am setting it to zero first and then assigning the treatment. How should I go about it ??