Manually calculating variance-covariance matrix after regression

[Marta Oliveira]

Dear Stata users,
I would like to calculate the variance-covariance matrix after estimating the following regression:

did_multiplegt y city year treated, robust_dynamic dynamic(3) placebo(3) breps(50) cluster(city)

did_multiplegt is an installed Stata package. Is there any way of estimating manually, given that the authors do not provide it in the saved results?
I also tried the simple OLS and calculate

xtreg y L3treated L2treated L1treated treated F1treated F2treated F3treated i.year, fe
e(V)

but I obtained the error “V not found r(111)”.
I am using Stata14.

[Carlo Lazzaro]

Marta:
yes, there is: https://blog.stata.com/2015/11/17/pr...e-ols-objects/

[FernandoRios]

Hi Marta
While estimating standard errors "manually" is "simple" (follow Carlo Lazzaro link and advice for example), Standard errors for a user written command like did_multiplegt is not.
If you feel you need that information, you could:
a) Hack into the adofile, locate where they use or create the variance covariance matrix, and save it yourself.
b) reimplement the procedure yourself (Recommended to learn what it does, but highly impractical)
c) contact the authors. They may decide to make an update and share it with you and the Stata community. (This i do for csdid for example)
HTH