You should interpret the results like this: the most efficient units operates at an efficiency level which is about 1/1.44=70 of its theorerical maximum. While the numbers are economically surprising (you can't believe that a firm survives working at one seventh of its efficiency level --- but some niche car builders as Ferrari or Lamborghini may not be driven to efficiency by market forces as car builders in more competitive market segments - as Wolkswagen or Ford), the concept is not. Firms usually have some inefficiencies (they are called X-inefficiency). Those who appear to you as inefficiencies are slack resources that firms usually need to face environmental changes/external shocks/unexpected events.
If you really want to rescale, you need to divide each TE by 1.439743, as suggested by Ahmed, not to substract .439743 (remember that TE are exponentiated coefficients)! Then you can interpret the resulting coefficients as units producing their output below the *most* efficient unit.
Nicola
At 02.33 28/06/2007 -0400, "Alexander Kalb" wrote:
>When I estimated my stochastic cost frontier I thought exactly the same.
>But when I predicted the efficiency scores (with the command predict
>efficiency, te), I got estimates lying between 1.439743 (most efficient)
>and 6.81706 (most inefficient). The question now is, how one can
>interpret this results, since the most efficient unit does not have the
>number one (as expected) but the number 1.439743. You simply could
>rescale the efficiency scores by substracting .439743 from all numbers
>to get one unit with an efficiency score with 1. Then a unit with, say
>an efficiency score of 1.20, can be interpreted as producing its output
>with 20% above the efficient level. But I don't know if this is the
>right way.
>
>Alex
*
* 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/
