In a simple diff-in-diff model, such as:

$$Y_{it}=\beta_0+\beta_1S_{it}+\beta T_{it}+\beta_3S_{it}T_{it}+\epsilon_{it}$$

In which $S_{it}$ represents a dummy for the units that received the treatment and $T_{it}$ is a dummy for the time in which ocurred. Consider the situation in which $S_{it}$ is time-invariant. If one was to run a fixed-effects version of this model on a common software package such as Stata, in which one would add a fixed effect term in the equation above, $S_{it}$ would be dropped out due to perfect multicollinearity, I believe. If that were to happen, does the model still 'works'?

The model makes sense to me, but I wonder if there is something that I should be worrying about.

  • $\begingroup$ The $S_{it}$ is still in the model. It's just wiped out by the demeaning transformation of xtreg, fe. You don't need to worry. $\endgroup$
    – dimitriy
    Commented Oct 23, 2015 at 2:09
  • $\begingroup$ It wipes out every variable that is not time-variant, right? By 'demeaning' you mean exactly what? $\endgroup$
    – John Doe
    Commented Oct 23, 2015 at 15:37
  • $\begingroup$ Yes. Demeaning is subtracting the mean of y for that person from each observation. $\endgroup$
    – dimitriy
    Commented Oct 23, 2015 at 15:41
  • $\begingroup$ Are there multiple periods across which the effect was measured? Why not establish a simpler treatment effect by, for the moment, collapsing time into pre and post? A pre-post ratio test would be a robust stake in the ground that the more granular (and sparser) temporal relationships might lose. $\endgroup$
    – user78229
    Commented Oct 24, 2015 at 12:57

1 Answer 1


In your difference in differences model some of the subscripts seem to be a little confused. For instance, group membership is something fixed over time, hence it should be $S_i$. Likewise the treatment (supposedly) starts at some time for everyone and therefore you can drop the $i$ subscript.

Whether you then use OLS to estimate $$Y_{it} = \beta_0 + \beta_1 S_i + \beta_2 T_t + \beta_3 (S_i\cdot T_t) + \epsilon_{it}$$ or if you estimate $$Y_{it} = \beta_2 T_t + \beta_3 (S_i\cdot T_t) + c_i + \epsilon_{it}$$ via fixed effects will give you the same result for the difference in differences coefficient. In the second model your $S_i$ is going to be absorbed by the individual fixed effects $c_i$ which would be differenced out anyway in the difference in differences setting.

Estimating DiD via a fixed effects model comes in handy when you have a non-binary treatment and/or the treatment comes into effect for some people at difference points in time.

  • $\begingroup$ why it is become handy to estimate DID with a fixed effect specification when individuals have different treatment time? And what's the difference it makes compare with the first equation you gave.I'm happen to have a problem like this.(different treatment time ) I do not have enough reputations to leave an comment, sorry for the mess,but I really need to know this? $\endgroup$
    – Jason Goal
    Commented Mar 14, 2017 at 13:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.