Compare coefficients from reghdfe model on two samples

[Noor Sethi]

Hi all,

I'm trying to compare beta coefficients from the same set of models applied to two different populations (e.g. male and female). My code looks something like:

eststo clear
 
*run regressions on male subpopulation 
reghdfe y x1 if female == 0, absorb(timevar) cluster(personvar)
eststo a0

reghdfe y x1 x2 if female == 0, absorb(timevar) cluster(personvar)
eststo b0

reghdfe y x1 x2 x3 if female == 0, absorb(timevar) cluster(personvar)
eststo c0

*run regressions on female subpopulation 
reghdfe y x1 if female == 1, absorb(timevar) cluster(personvar)
eststo a1

reghdfe y x1 x2 if female == 1, absorb(timevar) cluster(personvar)
eststo b1

reghdfe y x1 x2 x3 if female == 1, absorb(timevar) cluster(personvar)
eststo c1

Then I want to compare a0 to a1, b0 to b1, and c0 to c1, and ideally output these comparisons into a latex table.

I tried using -suest- but it seems incompatible with reghdfe. Any help would be appreciated!

[George Taylor]

I think you want to be using -esttab-.

 eststo clear  

*run regressions on male subpopulation  
reghdfe y x1 if female == 0, absorb(timevar) cluster(personvar)
eststo a0  

reghdfe y x1 x2 if female == 0, absorb(timevar) cluster(personvar)
eststo b0  

reghdfe y x1 x2 x3 if female == 0, absorb(timevar) cluster(personvar)
eststo c0  

*run regressions on female subpopulation  
reghdfe y x1 if female == 1, absorb(timevar) cluster(personvar)
eststo a1  

reghdfe y x1 x2 if female == 1, absorb(timevar) cluster(personvar)
eststo b1

 reghdfe y x1 x2 x3 if female == 1, absorb(timevar) cluster(personvar)
eststo c1  

esttab a0 b0 c0 a1 b1 c1 using "{OUTPUT_PATH}.tex", mgroups("Male" "Female", pattern(1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span erepeat(\cmidrule(lr){@span})) booktabs label

See this thread for titles that span multiple columns https://www.statalist.org/forums/for...umns-in-esttab, as "Male" and "Female" do here.

[FernandoRios]

Here some options
https://friosavila.github.io/chatgpt...023/index.html

[Noor Sethi]

George, I'm not looking to output the regression results, which is what esttab does. I'm looking to formally compare regression results (statistically test whether they're different from one another) from a model applied to two different samples.

Fernando, I think the link got close to an answer to my question. But unfortunately, it provides no examples such as the one I posted above: using reghdfe with the absorb and cluster options on two samples, and comparing the results...

[Andrew Musau]

reghdfe is from https://github.com/sergiocorreia/ivreghdfe (FAQ Advice #12). See https://www.statalist.org/forums/for...erent-outcomes

[FernandoRios]

It is the same idea
i do have only simplest case
but if you have specific suggestions on how to add more info to my post, I’m glad to hear