0
$\begingroup$

I am trying to control for pre-treatment trends in my DiD style regression. My baseline regression is as follows (using Stata): reg y c.treat*post i.year treat othercontrols Note that my treatment variable is continuous and I have other specifications where I interact it with years rather than a post indicator. Anyways, this is my idea to incorporate pre-treatment trends. keep only pre-treatment data

1) run the same regression but add a c.year*c.treat variable: reg y c.treat*post i.year treat othercontrols c.year*c.treat variable

2) save the coefficient on the c.year*c.treat as "coef"

3) Using the full sample, run reg y c.treat*post i.year treat othercontrols c.year*c.treat variable*coef

Can anyone tell me if this approach seems right/makes any sense? If not, other ideas?

$\endgroup$
1
  • $\begingroup$ What do you exactly mean by controlling for pre treatment trend? You mean testing the common trend assumptions? In DiD the pre treatment trends have to be the same otherwise your model will not be identified properly, if they are same for both treatment and control there is not really much reason to “control for them”. $\endgroup$ – 1muflon1 Mar 6 '20 at 7:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.