The way of testing is:
Estimation proceeds in three steps:
1. Estimate a model for non-treated potential outcomes using the non-treated (i.e. never-treated or not-yet-treated) observations only. The benchmark model for diff-in-diff designs is
a two-way fixed effect (FE) model: Y_it = a_i + b_t + eps_it, but other FEs, controls, etc., are also allowed.
2. Extrapolate the model from Step 1 to treated observations, imputing non-treated potential outcomes Y_it(0), and obtain an estimate of the treatment effect tau_it = Y_it - Y_it(0)
for each treated observation.
3. Take averages of estimated treatment effects corresponding to the estimand of interest.
1. Estimate a model for non-treated potential outcomes using the non-treated (i.e. never-treated or not-yet-treated) observations only. The benchmark model for diff-in-diff designs is
a two-way fixed effect (FE) model: Y_it = a_i + b_t + eps_it, but other FEs, controls, etc., are also allowed.
2. Extrapolate the model from Step 1 to treated observations, imputing non-treated potential outcomes Y_it(0), and obtain an estimate of the treatment effect tau_it = Y_it - Y_it(0)
for each treated observation.
3. Take averages of estimated treatment effects corresponding to the estimand of interest.
From reading his help file, I saw the way to conduct anticipation effect is
shift(integer): specify to allow for anticipation effects. The command will pretend that treatment happened shift periods earlier for each treated unit.
- Do NOT use this option for pre-trend testing; use it if anticipation effects are expected in your setting. (This option can be used for a placebo test but we recommend a
pretrend test instead; see Section 4.4 of Borusyak et al. 2021.)
- The command's output will be labeled relative to the shifted treated date Ei-shift. For example, with horizons(0/10) shift(3) you will get coefficients _b[tau0]..._b[tau10]
where tauh is the effect h periods after the shifted treatment. That is, tau1 corresponds to the average anticipation effect 2 periods before the actual treatment, while
tau8 to the average effect 5 periods after the actual treatment.
- Do NOT use this option for pre-trend testing; use it if anticipation effects are expected in your setting. (This option can be used for a placebo test but we recommend a
pretrend test instead; see Section 4.4 of Borusyak et al. 2021.)
- The command's output will be labeled relative to the shifted treated date Ei-shift. For example, with horizons(0/10) shift(3) you will get coefficients _b[tau0]..._b[tau10]
where tauh is the effect h periods after the shifted treatment. That is, tau1 corresponds to the average anticipation effect 2 periods before the actual treatment, while
tau8 to the average effect 5 periods after the actual treatment.
Array
Is it tau2 is the average effect of the first and second years before treatment. and tau8 is the average treatment of the year 1 to year 5 after treatment?
And what can we conclude here in wTOT_ASS_TUR? I intend to say there is no anticipation effect two years before the event year but the significant coefficient in the tau0 deter me from concluding that.
Many thanks and warmest regards.
0 Response to How to read the anticipation effect result in difference-in-difference?
Post a Comment