Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • CSDID with pre-treatment covariates

    Hello everyone

    Hope you all are doing well. I would very much appreciate your assistance in clarifying the following:
    I work panel data consisting of more than thousand firms yearly accounting data across 6 year period, where firms are treated with heterogeneity in treatment timing (staggered treatment). Some firms are also never treated. I would very much like to use the csdid-command in stata with covariates such as assets and revenue.
    I have noted that one must use pre-treatment covariates for time-varying covariates, however it is unclear to me if the csdid automatically accounts for this, when the covariates are added to the command, or if I should remove the after-treatment values for the covariates before running command?

    Thank you very much. Forgive me if this is already available.

    Have a nice weekend :-)

    Best,
    Rasmus

  • #2
    according to the help file, csdid uses the earliest covariate values for propensity score and outcome regressions.

    Comment


    • #3
      If you want to include time-varying covariates and have them appear in the outcome equation at time t, you can use the jwdid command. As George said, CS use the covariates dated in the pre-intervention period only. Of course, if you do include the contemporaneously you want to make sure they aren't "bad controls" -- influenced themselves by the intervention.

      Comment

      Working...
      X