Reghdfe "invalid format" error with 1500 absorbed variables

[Mounir K]

Hi,
I'm running into a problem with reghdfe, which I have not found documented elsewhere. When I run a regression with about 1500 variables to be absorbed, the regression goes through normally but I receive the error message "invalid format" when it should display the summary of absorbed levels. As a result, I can't store any output in eststo to create output tables.

The problem does not appear with fewer fixed effects. Haven't heard of any max var limit in this program.

If it helps, I'm running Stata MP on a high-end computer.

[Aaditya Dar]

That sounds odd. Could you post a reproducible example (and please consider using -dataex-)?

[Daniel Cooper]

Was this issue ever resolved. I am having a similar issue. I am using a wildcard with user defined dummy variables. I realize that this is slower relative to i.var1##i.var2##i.var3 but there are reasons that I need to do it that are not worth getting into.

This is my estimation statement and the results with the format error are below. I am pretty sure it is tied to the wildcard, because if I do i.var1##i.var2 I do not have the error. Any help is much appreciated.


eststo: reghdfe wagegrowth UR age agesq hs somecoll collplus if wagegrowth <40 & wagegrowth >-40 & mfg ==0 [pw=earnwt4], absorb(FE_1=i.occ20104 FE1= yn*)


------------------------------------------------------------------------------
| Robust
wagegrowth | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
UR | -.1709644 .2245362 -0.76 0.446 -.6110487 .2691199
age | -.2217584 .0302377 -7.33 0.000 -.2810234 -.1624934
agesq | .0020359 .000354 5.75 0.000 .0013421 .0027296
hs | -.2735019 .1389762 -1.97 0.049 -.5458911 -.0011127
somecoll | .030521 .1449549 0.21 0.833 -.2535862 .3146282
collplus | .1690761 .1572636 1.08 0.282 -.1391559 .4773081
------------------------------------------------------------------------------

Absorbed degrees of freedom:
{hline 5292}--------------------------------------------------+
invalid %format
r(120);