Incorrect results with `absorb`? about ivreghdfe HOT 6 CLOSED

Yup, looking good now!

PS: xi was there bc ivreg didn't support factor variables and my mind really disliked the code alignment otherwise. I mean, the third regression didn't even use factor variables and I still put a xi in front.

from ivreghdfe.

I think I know the problem. Yesterday I changed the reghdfe code to keep by default the variables standardized after reghdfe (it improve numerical accuracy).

The fix should be relatively simple (changing an argument in the call to partial_out()) but we are swamped with recruiting this week so I'm only going to be able to update it tomorrow evening,. at the earliest.

(on the meantime, You can use reghdfe with the old option to run the IV though)

from ivreghdfe.

Actually, a better soln osnto change the reghdfe code so it does not returned standardized values by default. There might be other programs out there using the internals of reghdfe that might get affected otherwise

from ivreghdfe.

i just tested the release from the 27th and it works fine, but the release from the 28th introduces the problem (reghdfe).

What is the old option to run the IV? No idea this was a thing.

from ivreghdfe.

from ivreghdfe.

Just pushed a commit to reghdfe that should fix this issue. Let me know if it works on your side.

(BTW, you don't need to use the xi: prefix anymore, as both ivreg2 and reghdfe support factor variables! see help xi for more details)

from ivreghdfe.