[Carlo Lazzaro]

Katherine:y
can't you simply interact -treat- with -peak- in an unique regression?

[Katherine Adams]

Carlo,

Unfortunately, no, I can't. I can't change the regression specification.

[Eric de Souza]

You should run one estimation command instead of two (one for peak =1, and one for peak =0) . You can adapt the following example for reghdfe
(the regression with the option -coefl- is just to give you the names with which Stata stores the values):

. webuse auto
(1978 Automobile Data)

. reg mpg c.weight#i.foreign, robust

Linear regression                               Number of obs     =         74
                                                F(2, 71)          =      76.10
                                                Prob > F          =     0.0000
                                                R-squared         =     0.6714
                                                Root MSE          =     3.3626

----------------------------------------------------------------------------------
                 |               Robust
             mpg |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
foreign#c.weight |
       Domestic  |  -.0066409   .0005421   -12.25   0.000    -.0077218     -.00556
        Foreign  |  -.0075145   .0006839   -10.99   0.000    -.0088782   -.0061509
                 |
           _cons |   41.95073   1.794367    23.38   0.000     38.37286    45.52859
----------------------------------------------------------------------------------

. reg mpg c.weight#i.foreign, robust coefl

Linear regression                               Number of obs     =         74
                                                F(2, 71)          =      76.10
                                                Prob > F          =     0.0000
                                                R-squared         =     0.6714
                                                Root MSE          =     3.3626

----------------------------------------------------------------------------------
             mpg |      Coef.  Legend
-----------------+----------------------------------------------------------------
foreign#c.weight |
       Domestic  |  -.0066409  _b[0b.foreign#c.weight]
        Foreign  |  -.0075145  _b[1.foreign#c.weight]
                 |
           _cons |   41.95073  _b[_cons]
----------------------------------------------------------------------------------

. test _b[0b.foreign#c.weight] = _b[1.foreign#c.weight]

 ( 1)  0b.foreign#c.weight - 1.foreign#c.weight = 0

       F(  1,    71) =    5.24
            Prob > F =    0.0251

[Eric de Souza]

My post #18 crossed Carlo's. This is what he meant, I think. It doesn't change your model

gen treat1 = treat*peak
gen treat0 = treat*(1-peak)
reghdfe cons treat0 treat1 , absorb (i.location#i.hour i.hour#i.calday) vce(cluster i.location#i.hour)
test treat0 = treat1

On edit: And is more intuitive than my first post
On further edit: I am assuming that treat only takes on two values: zero and one

[Carlo Lazzaro]

Eric is correct.
Actually, I replied from my mobile and I could not add an example.

[Katherine Adams]

Eric and Carlo,

Thank you for your help! I have run the code successfully.

I apologize for my late reply.

[Raja Hasan]

I have similar problems.
The model looks like the following:

leghdfe inv gender salary rent sch opl if inc==1, absorb (country fyear) vce(cluster firm fyear)
estimate store inc1

leghdfe inv gender salary rent sch opl if inc==0, absorb (country fyear) vce(cluster firm fyear)
estimate store inc0

I want to test whether there is any significant difference between the coefficient of gender. Here, gender is a dummy variable and inc is also dummy variable but sometimes I group inc into three. if inc is 3 group then I want to test the difference between the coefficient of gender between high and low inc group.
.
By looking at the solution from Eric (#19) , I have come up with the following solution for my models.

gen gender1 = gender*inc
gen gender0 = gender*(1-inc)
reghdfe inc gender0 gender1 salary rent sch opl, absorb (country fyear) vce(cluster firm fyear)
test gender0 = gender1

Could anyone please tell me whether I am doing things correctly or I am making mistakes? Also, how will I test the equality if inc is grouped into three (I want to test the equality between high and low groups for the coefficient of gender, which is a dummy variable)