Event Study Interpretation

[Kate Ritchie]

I am using an panel event study method for my dissertation and wish to confirm the interpretation. My panel spans from 2000-2023 and the treatment year is 2014, with 2013 being dropped as the base reference year. Does each coefficient represent the estimated treatment effect relative to 2013 or to 2014 (i.e., is it relative to the dropped, final pre-treatment year or the treatment year)? I have searched online but found conflicting answers, any help would be greatly appreciated.

[Maxence Morlet]

Relative to 2013, i.e. the omitted year, so -1 if 0 is treatment year. Generally this is convention: in event studies, ssc install eventdd will by default omit the period before the start of treatment.