Forum: help

RE: Translog Cost Function [ Reply ] By: Arne Henningsen on 2012-05-24 18:17	[forum:5720]
using argument "farrell": Assume, you have named the object returned by sfa() or frontier() "model", then you can get the efficiency estimates by R> efficiencies( model, farrell = FALSE ) or R> efficiencies( model, farrell = TRUE ) You can access the documentation by the command R> ?efficiencies.frontier

RE: Translog Cost Function [ Reply ] By: Muhammad Ali Asghar on 2012-05-24 15:48	[forum:5719]
Sorry for asking you again! Can you please tell me what is farrell argument?

RE: Translog Cost Function [ Reply ]
By: Muhammad Ali Asghar on 2012-05-24 15:27

[forum:5718]

You mean to say i have to switch to cobb-douglas form;
Farrell (1957) suggested the use of either: (i) a nonparametric piecewise-linear convex isoquant constructed such that no observed point should lie to the left or below it (known as the mathematical programming approach to the construction of frontiers); or (ii) a parametric function, such as the Cobb-Douglas form, fitted to the data, again such that no observed point should lie to the left or below it (known as the conometric approach). These approaches use different techniques to envelop the observed data, and therefore make different accommodations for random noise and for flexibility in the structure of the production technology.

RE: Translog Cost Function [ Reply ] By: Arne Henningsen on 2012-05-24 15:17	[forum:5717]
No, taking the inverse is not correct as E[1/eff] is not equal to 1/E[eff]. The cost efficiency can be either C_observed / C_minimum or C_minimum / C_observed. If you want to switch between these two concepts, you have to use the "farrell" argument (as I wrote before).

RE: Translog Cost Function [ Reply ] By: Muhammad Ali Asghar on 2012-05-24 15:04	[forum:5716]
if i do following process; is it right? 1/efficiency score e.g. 1/1.5 = 0.667 and then 1 - 0.667 = 0.333 efficiency by following this The cost inefficiency score can be estimated as the ratio of observed cost Cit to frontier or minimum cost Cf

RE: Translog Cost Function [ Reply ] By: Muhammad Ali Asghar on 2012-05-24 14:05	[forum:5714]
Thank you

RE: Translog Cost Function [ Reply ] By: Arne Henningsen on 2012-05-24 14:02	[forum:5713]
Yes, you can specify whether you want to get Farrell-type or Shepard-type efficiency measures using the "farrell" argument.

RE: Translog Cost Function [ Reply ] By: Muhammad Ali Asghar on 2012-05-24 13:54	[forum:5712]
Thank you very much for your quick response. Can i transform these scores into 0-1 . how? plz

RE: Translog Cost Function [ Reply ]
By: Arne Henningsen on 2012-05-24 13:49

[forum:5711]

These cost efficiencies indicate the ratio between observed costs and minimum costs and hence, must always be larger than or equal to one.

Which version of the "frontier" package do you use? Please make sure that you are using version 0.997-8.

You can use argument "farrell" of the efficiencies() method for objects of class "frontier" to decide whether you want Farrel-type or Shepard-type elasticities (see documentation of efficiencies.frontier).

Translog Cost Function [ Reply ]
By: Muhammad Ali Asghar on 2012-05-24 13:19

[forum:5710]

I have computed my cost efficiency scores but surprisingly they are more than 1 instead of 0-1. how that is possible (2outputs and 3 inputs in my case).
Please help me.. how i can interpret it

cost efficiency estimates :

firm eff.-est.

1 0.76157915E+01
2 0.27770330E+01
3 0.23623843E+02
4 0.26691737E+01
5 0.32127416E+02
6 0.35028812E+04
7 0.20355003E+06
8 0.10268270E+01
9 0.45905461E+01
10 0.43948418E+01
11 0.20324918E+04
12 0.14903700E+01
13 0.52577625E+01
14 0.37785936E+01
15 0.21809874E+01
16 0.12150592E+01
17 0.34416517E+01
18 0.15638163E+01
19 0.10073941E+01
20 0.13482742E+01
21 0.55436186E+01
22 0.56933269E+01
23 0.76157915E+01
24 0.27770330E+01
25 0.23623843E+02
26 0.26691737E+01
27 0.32127416E+02
28 0.35028812E+04
29 0.20355003E+06
30 0.10268270E+01
31 0.45905461E+01
32 0.43948418E+01
33 0.20324918E+04
34 0.14903700E+01
35 0.52577625E+01
36 0.37785936E+01
37 0.21809874E+01
38 0.12150592E+01
39 0.34416517E+01
40 0.15638163E+01
41 0.10073941E+01
42 0.13482742E+01
43 0.55436186E+01
44 0.56933269E+01

mean efficiency = 0.95089430E+04