Random effects and two-way clustered standard errors in the same model?

[Anna Lindgren]

Hi all,

I would greatly appreciate some help on how to run a regression including both random effects and two-way clustered standard errors. So far, I have not found a stata command that allows me to do this (I’m using stata 14).

Some context:

I am estimating the effects of x1 and x2 on y. The data include three different “id variables” that characterize each observation: ID_1, ID_2, and ID_3 (ID_1 is nested within ID_2). I would like to include random effects at the level of ID_1 and fixed effects at the levels of ID_2 and ID_3. I would also like to cluster the standard errors on both ID_2 and ID_3.

If I only cluster on one level (ID_3), I can estimate this model in the following way:

xtset ID_1
 
xtreg y x1 x2 i.ID_2 i.ID_3, re vce(cluster ID_3) nonest

Is there any way of estimating the exact same model but clustering on both ID_2 and ID_3?

I very much appreciate any help I can get!

Thanks in advance,

Anna