Dear community,
I have estimated a gravity model with ppmlhdfe and want to conduct an event study now. Unfortunately, it seems that the command eventdd (Clarke, D. & Tapia Schythe, K. (2020): Implementing the Panel Event Study, IZA, Institute of Labor Economics) cannot be applied with a ppml(hdfe) regression but only with "normal" Fixed Effects regressions or OLS.
So, I was wondering if there is any other command to estimate such an Event Study/Diff-in-Diff showing the time development of the treatment effect of my event. I also considered the jwdid command (Nagengast, A., Rios-Avila, F. & Yotov, Y. (2024): The European Single Market and Intra-EU Trade: An Assessment with Heterogeneity-Robust Difference-in-Differences Methods, Drexel Economics, Working Paper series.). However, I'm not sure if this is suitable as my treatment is not staggered but there is only one event occuring at one point in time.
When trying to implement it with jwdid, nevertheless, I struggle with the following:
This command works with the following code:
1. Do I understand the variable "g" (to be included in gvar) correctly, that it should be = 0 for all observations which are never treated, and for the observations that are treated, it should be equal to the year of treatment, no matter in which year the observation takes place? So if the event is in 2015, for a treated unit, g should be = 2015 already in year = 2010 as treatment onset for this observation is in 2015?
2. Where would I include the traditional gravity variables (GDP, distance)? Do I have to include them in exovar, xtvar or xgvar?
I would appreciate any help.
Best
Noemi
I have estimated a gravity model with ppmlhdfe and want to conduct an event study now. Unfortunately, it seems that the command eventdd (Clarke, D. & Tapia Schythe, K. (2020): Implementing the Panel Event Study, IZA, Institute of Labor Economics) cannot be applied with a ppml(hdfe) regression but only with "normal" Fixed Effects regressions or OLS.
So, I was wondering if there is any other command to estimate such an Event Study/Diff-in-Diff showing the time development of the treatment effect of my event. I also considered the jwdid command (Nagengast, A., Rios-Avila, F. & Yotov, Y. (2024): The European Single Market and Intra-EU Trade: An Assessment with Heterogeneity-Robust Difference-in-Differences Methods, Drexel Economics, Working Paper series.). However, I'm not sure if this is suitable as my treatment is not staggered but there is only one event occuring at one point in time.
When trying to implement it with jwdid, nevertheless, I struggle with the following:
This command works with the following code:
Code:
jwdid y, ivar(i) tvar(t) gvar(g) method() never exovar(x_ex) xtvar(x_t) xgvar(x_g) fevar() hettype(option) cluster(cvar)
2. Where would I include the traditional gravity variables (GDP, distance)? Do I have to include them in exovar, xtvar or xgvar?
I would appreciate any help.
Best
Noemi
