Issue running pooled regression

[kcolombe]

Hello,

I am trying to run a pooled regression to see if there is a significant change in the coefficient of my independent variable of interest (VOI) pre- to post-treatment. Just to get an idea of what I might expect to see I first them the pre-treatment separate from the post-treatment. The former's coefficient for the VOI was positive and not significant, the latter was negative and highly significant.

Now for the part, I do not understand...running them in the same regression yields a negative insignificant coefficient pre and a negative insignificant coefficient post. Have I done something wrong?

`* Regression for the pre-treatment period (2004-2006)
reghdfe depvar VOI $pg_controls $tg_controls [aw=min] if period=1, absorb(pl gi) cluster(pl gi)

• Regression for the post-treatment period (2007-2010)
  reghdfe depvar VOI $pg_controls $tg_controls [aw=min] if period=2, absorb(pl gi) cluster(pl gi)

• Regression with reghdfe and its interaction with the post-treatment period

• Create a post-treatment period dummy (period 2 is post-treatment: 2007-2010)
  gen period2 = 0
  replace period2 = 1 if year >= 2007

• Interaction term for bl_fw and post-treatment period (period2)
  gen VOI_post = VOI * period2

reghdfe depvar VOI VOI_post $pg_controls $tg_controls [aw=min], absorb(pl gi) cluster(pl gi)`