With stochastic frontier analysis you have heteroskedasticity in either or
error components. Heteroskedasticity is more serious in than in OLS
potentially causing biased parameter estimates and biased estimates of
technical efficiency. If both v and u are heteroskedastic, the biases are
in opposite directions so overall effect may be small. The -vhet()- and
-uhet()- options are the way to control for heteroskedasticity.
Kumbhaker and Lovell ("Stochastic Frontier Analysis", 2000, Cambridge
University Press New York, NY, USA) suggest starting with a model that
incorporates heteroskedasticity in both components and test down to a
simpler model.
Out of curiosity, why are you constraining the intercept to be zero?
Scott
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of s.bhaumik@queens-
> belfast.ac.uk
> Sent: Sunday, January 16, 2005 10:51 AM
> To: [email protected]
> Subject: st: frontier command in Stata
>
> Hi,
>
> I am using the "froniter" command in Stata to estimate log linear
> production functions. I have suppressed the constant term using the
> "noconstant" option. I would really appreciate it if someone can tell me
> whether I can correct for heteroskedasticity? I tried using the "uhet" and
> "vhet" options, but to begin with I do not know whether they correct for
> heteroskedasticity in the same way that "robust" takes care of
> heteroskedasticity in the "regress" command, and these options seem to be
> inconsistent with the "noconstant" option.
>
> Regards,
>
> Sumon
>
> __________________________________
> Dr. Sumon Kumar Bhaumik
> School of Management and Economics
> Queen's University Belfast
> 25 University Square
> Belfast BT7 1NN
> UK
> Phone: +44 28 9097-3273
> http://www.qub-efrg.com/staff/sbhaumik/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
