Dear Stata listers,

I would like to estimate a cross-section time-series (in form of pooled cross sections) structural equation model, consisting of four structural equations.

What is quite important for my analysis is that I can estimate each of the four equations with either

- robust standard errors (Huber..) and clusters or
- weights and clusters or
- newey-west standard errors.

Unfortuantely, the comand reg2 does not permit the use of the options 'robust' and / or 'cluster' (If I am wrong, pls correct me !).
With the robust standard erros or weights I intend to correct for heteroscedasticity, and the clusters should allow for correlation within groups, where groups are defined by the observed unit.

Do you know of any alternative comand I could use in my case?

Comments are most welcome !!


Yours sincerely,

Justina





Dipl.Vw. Justina A.V. Fischer,
Research Associate
SIAW - University of St. Gallen
Dufourstr. 48
CH-9000 St. Gallen
Tel.:++41-71-224 2345
Fax: ++41-71-224 2298
www.siaw.unisg.ch