Error message: -suest- with -xtfmb-

[Jae Li]

Dear Statalists,

I want to use -suest- to estimate if the FM models' coefficients are significantly different. But using -suest- gives the error that the model "was estimated with a nonstandard vce (Fama-MacBeth)".

Do you possibly know how to correct this error or any other ways to estimate the coefficient difference? Many thanks for your help in advance!

Here is the trial code:

xtfmb y x control if w <= med
est sto small
xtfmb y x control if w > med
est sto large
. suest small large
small was estimated with a nonstandard vce (Fama-MacBeth)
r(322);

[Phil Bromiley]

One of the downsides of using a user written procedure is that it often doesn't play nicely with procedures that handle almost all Stata estimators. Fama McBeth is estimating variances in a very different way than most estimators.

What you can do at least in most estimators is simply create a dummy variable for small versus large and interact it with the right-hand side variables.

[Phil Bromiley]

By the way, it is really bad form to post the same question three times in a row. I strongly suggest you get it right and post it once. We are helping people we don't know with no real reward so it behooves you in asking for our help to not waste our time.

[Jae Li]

@Phil Bromiley Hi Phil, thank you for your response and comments! There was a web crash when I was trying to post on Statalist, and somehow it posted more than once at one time. However, I don't know how to delete those two posts on Statalist, do you know how to delete a post here? Anyway, thank you for your time!

[Hadeer Mounir]

Jae Li Hello, I am facing the same problem. Have you figured out a solution to this? Thank you

[Yz Li]

Jae Li Hi, I am also facing the same. How did you solve it? Thanks!