Counter argument from the pool of ignorance where I live: I'm afraid it's going to take me longer to learn how to implement this rather than just running reghdfe ...
David Roodman has a new paper documenting how to use Julia as a backend in Stata. Main advantage at the moment is gaining access to `reghdfejl`, which is up to 10x (!) faster than `reghdfe` #EconSkyarxiv.org/pdf/2404.093...
Reg with i.x has always shown it - the old version of reghdfe, when absorbing, did not. Not the issue here likely, but it hid all issues with two-way-FEs back then, was easy to just “oh ok, that’s it then, job done” without thinking of double normalizations, weights, all we have learned since
TFW you've been working on a project for so long that the reghdfe and ppmlhdfe versions no longer play nicely with each other.
David Roodman has written a new Stata package that replicates `reghdfe` using Julia. The new implementation requires installing Julia but is about 10x faster. Details available here: www.statalist.org/forums/forum...#EconSky
I’ve been (somewhat) trying to figure this out for a while now: can one of the built-in Stata commands do it, or are folks calculating that by hand? I was under the impression the reghdfe option for sum statistics gives you the mean for everyone, not just the control group
A new update for my @Stata command #xtdcce2#reghdfejanditzen.github.io/xtdcce2/janditzen.github.io/xtdcce2/
A new update for my @Stata command #nwxtregress#reghdfejanditzen.github.io/nwxtregress/janditzen.github.io/nwxtregress/
Niche post-thing. As much as I enjoy `reghdfe` in Stata, I've become a complete `fixest` in R convert. It's not even close.