Add the option to calculate confidence intervals and plots using a wild cluster bootstrap, for cases when there is a small number of clusters.
Note that this is already possible using a workaround: Generate the event-time dummies, estimate again and use boottest:
webuse nlswork
xtset idcode year
* Estimate and save eventtime dummies
xtevent ln_wage age c.age#c.age ttl_exp c.ttl_exp#c.ttl_exp tenure , pol(union) w(3) impute(nuchange) savek(ev)
* Estimate again, same results
reghdfe ln_wage age c.age#c.age ttl_exp c.ttl_exp#c.ttl_exp tenure ev_eq_m4-ev_eq_m2 ev_eq_p* i.year, absorb(idcode)
* Wild bootstrap test
boottest ev_eq_p1==0, cluster(idcode)
After we finish #94, we may be able to do the event-time dummie generation without estimating, saving time. Still, we may want to integrate this to be able to use xteventest and xteventplot with the wild-bootstrap CIs.
Add the option to calculate confidence intervals and plots using a wild cluster bootstrap, for cases when there is a small number of clusters.
Note that this is already possible using a workaround: Generate the event-time dummies, estimate again and use
boottest
:After we finish #94, we may be able to do the event-time dummie generation without estimating, saving time. Still, we may want to integrate this to be able to use
xteventest
andxteventplot
with the wild-bootstrap CIs.