reghdfe: F-test of absorbed heterogeneous slope with two-way clustering

[Biliana Veselinova]

Can someone help me and explain why the F-test is not reported for absorbed variables. I am running a regression with heterogeneous slopes and want to test their joint significance. However, the help says the F test of the absorbed is not reposted when we cluster standard errors. Does anyone have a reference (areg and xtreg have similar behaviour)? How should I proceed if I anyway want to report the F-test? What about postestimation tests with testparm? Many thanks in advance.
Kind regards,Biliana

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a quick answer if you follow the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. Note that reghdfe is user written so it may not fully follow Stata procedures.

As for what happens with xtreg and areg, you could have checked this yourself in two minutes. It took me less than that to check it. xtreg does not give the F test with clustered errors.

When Stata does not report something, there is usually a good reason. I assume that the test on the fixed effects is somehow not as doable with clustered errors as with standard errors. With clustered errors, I suspect you'll have a different error variance associated with each panel and thus each fixed effect. Have you read the papers the documentation cites? They might offer some indication of the reason for no F test.

There are probably ways to approximate such a test. You might use predict to generate the fixed effects and treat them as a separate sample for each panel and test the panels all have the same mean.