Slow generating interaction FE

[Chengmou Lei]

Hi all,

I'm trying to add commuting zone-occupation FE to my regression. What I tried to do is:
1) generate the interaction term:
[/CODE]
gen cz_occ_fe=cz*occ
[/CODE]
2) put it in reghdfe
[/CODE]
reghdfe y x, absorb(cz_occ_fe)
[/CODE]
But since I have 740+ commuting zone, and 968 occupation, it took me forever to generate the interaction term (I'm already using a powerful server).
Is there anyway to make things faster?

Thank you

[Andreas Backhaus]

Yes, there is: Factor-variable notation. You can include the interactions with this notation directly into the absorb() option of reghdfe.
https://www.stata.com/links/stata-ba...able-notation/

[Chengmou Lei]

Yes, there is: Factor-variable notation. You can include the interactions with this notation directly into the absorb() option of reghdfe.
https://www.stata.com/links/stata-ba...able-notation/

Hi Andreas,

Thanks, it works perfectly.