Hi,

I am running a triple difference estimator to look at the effect of the EPA's pollution non-attainment status for counties on wage in 1990. My triple difference estimator is Non-attainment Status X Post Treatment Dummy X Sector. This gives me the effect on wage in a non-attainment county after 1990 for the sector. My controls are the first-order interaction terms Non-attainment X Post Treatment Dummy, Post Treatment Dummy X Sector and Non-attainment X Sector.

My question is - do we include the individual dummies Non-attainment and Sector in the regression? I have checked at least four different papers looking at the effect of non-attainment status - they do not do this. These papers were published at the AER and QJE.

Can someone please shed light on what is the normal practice here? Thank you!