I can't understand your question. Are you complaining about the absence of TE equal to one?
This is ok. Said brutally, as far as I remember (sorry, no references for what I am saying), it is DEA which takes some observations as (most) efficients (i.e. TE=1) while the remainings are not, because it uses linear programming. On the contrary, SFE is more a statistical method, and it tries to "imagine" the frontier from your data, but does not require that some of your observations lie on the frontier.
As a proof, download the FRONTIER 4.1 package (still somewhere around http://www.une.edu.au/econometrics/cepa.htm, I hope), which uses some different notations and methods, and run the examples: no observations have TE=1.
Nicola
At 02.33 21/06/2007 -0400, "Alexander Kalb" wrote:
>Dear Statalist-members,
>
>I estimated a time-varying panel data cost frontier model with the
>command "xtfrontier depvar [indepvars] [if] [in] [weight] , tvd". After
>the estimation I wanted to get the efficiency scores with the command
>"predict efficiency, te". Now I wonder why this estimates are not
>bounded above 1, because they range from 1.439743 to 6.81706. I thought
>the most efficient unit should get a 1 and all other (more inefficient)
>units should get a number higher than 1. Can anybody explain this to
>me?
*
* 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/
