Testing equality of Poisson regression coefficients in two groups using ppmlhdfe

[Jacob Helium]

Dear all,

How do you test Poisson coefficients in two groups? Because I have multi-way fixed effects and want to do two-way clustering, I have to use ppmlhdfe. For example, in the following codes:

ppmlhdfe y x1 x2 if group==1, absorb(year month) vce(cluster industry state) 

ppmlhdfe y x1 x2 if group==0, absorb(year month) vce(cluster industry state)

I regress y on x1 and x2 in two groups using Poisson model. My goal is to test whether coefficient for x1 in Group 1 is different from that in Group 2. However, suest does not work with ppmlhdfe. Therefore, I use the following:

ppmlhdfe y i.group#(c.x1 c.x2), absorb(year month) vce(cluster industry state) 

test 0.group#c.x1 =1.group#c.x1

Can you guys help me determine whether this is correct? I think this approach works with reghdfe. I'm not sure about Poisson because it is not a linear model.

Thanks,
Jacob

[Clara Berestycki]

Can someone answer this question? That would be very helpful, many thanks.

[Joao Santos Silva]

You also need to allow the fixed effects to differ across the two groups.

Best wishes,

Joao

[Jacob Helium]

You also need to allow the fixed effects to differ across the two groups.

Best wishes,

Joao

Thank you so much Joao. I think you are right.

[Jacob Helium]

Can someone answer this question? That would be very helpful, many thanks.

I figured it out. You would need to do:

ppmlhdfe y i.group#(c.x1 c.x2), absorb(year#group month#group) vce(cluster industry#group state#group)

Basically, interacting fixed effects with group indicator. Then, you can

test 0.group#c.x1 =1.group#c.x1