Comparing coefficients across subgroups

[Lies Jans]

Hello all,

I am new to Stata and tried to find an answer to my problem but I have not found a solution yet (I am using Stata 15/SE).

I am running four regressions to compare the coefficients on subgroups using two dummy variables (if female=0/1 & edu=0/1)
To test if the coefficients are significantly different from each other between genders ( thus if x1=x1, x2=x2, x3=x3 )

reg job x1 x2 x3 (+ covariates) ,r if female==0 & edu==1
reg job x1 x2 x3 (+ covariates) ,r if female==0 & edu==0

and

reg job x1 x2 x3 (+ covariates) ,r if female==1 & edu==1
reg job x1 x2 x3 (+ covariates) ,r if female==1 & edu==0

I tried using the suest command, since the regression is robust this is not working.
I tried generating new dummies:

gen male_edu = 1 if female==0 & edu==1
gen male_edu_x1 = male_edu*x1

And running new regressions, but I am not sure if this is working the right way.

Are there any solutions to this? Thanks in advance!

[Carlo Lazzaro]

Lies:
the Technical note at page 2664, -sureg- entry, Stata .pdf manual can hopefully shed light on this issue.

[Phil Bromiley]

To add a little to Carlo's suggestion, it is also possible to run models with interactions (see factor variable notation) that allow for different parameters in the different conditions. This may make some assumptions about the error term that differ from separate estimates, but robust standard errors may help there. Run in the same model lets you do conventional tests. Just an alternative to sureg.