[Limdep Nlogit List] efficiency scores above 1
dgiantsios at uom.gr
dgiantsios at uom.gr
Mon Jul 18 22:51:30 AEST 2016
Hi All,
I estimated cost efficiency for Battesse-Coelli (1995) model with panel
data. I added the comand Costeff=....; with the List; and i had the
following results:
Data listing for stochastic frontier model
Observ Pd. Data row Observed Y Fitted Y Y - Xb E[u|e]
1 1 1 12.6523 8.0012 4.6511 4.6912
1 2 2 12.2788 7.6433 4.6355 4.6507
1 3 3 12.6505 7.7804 4.8700 4.6105
1 4 4 12.5984 7.8458 4.7526 4.5707
1 5 5 13.4217 8.5396 4.8821 4.5312
1 6 6 13.7691 9.1428 4.6263 4.4920
1 7 7 13.0413 8.9292 4.1121 4.4532
1 8 8 12.8631 8.6091 4.2540 4.4147
2 1 10 12.0794 7.2354 4.8441 4.6294
2 2 11 12.5092 7.8969 4.6123 4.5894
2 3 12 12.6131 8.2284 4.3847 4.5497
2 4 13 13.7244 9.4242 4.3002 4.5104
2 5 14 13.3188 9.0529 4.2659 4.4714
2 6 15 13.2917 8.6882 4.6035 4.4328
2 7 16 13.2386 8.7104 4.5282 4.3945
2 8 17 13.2547 8.6791 4.5757 4.3565
3 1 19 10.4277 5.5600 4.8677 4.5799
3 2 20 10.9968 6.3027 4.6941 4.5403
3 3 21 11.1416 6.8385 4.3031 4.5011
3 4 22 12.2998 8.1706 4.1291 4.4622
3 5 23 12.2120 7.9780 4.2341 4.4236
3 6 24 12.1462 7.7386 4.4077 4.3854
3 7 25 12.2588 7.8629 4.3959 4.3475
3 8 26 12.4791 7.8318 4.6472 4.3099
4 1 28 11.0053 6.3753 4.6300 4.5413
4 2 29 11.9644 7.4316 4.5327 4.5021
4 3 30 12.3282 7.9215 4.4067 4.4632
4 4 31 13.6965 9.5069 4.1896 4.4246
4 5 32 13.4434 9.1795 4.2639 4.3864
4 6 33 13.3726 8.9541 4.4185 4.3485
4 7 34 13.4541 9.0695 4.3846 4.3109
4 8 35 13.5175 8.9951 4.5224 4.2736
5 1 37 10.4945 4.2519 6.2426 4.6274
5 2 38 10.0848 4.9867 5.0981 4.5874
5 3 39 13.0467 8.7164 4.3303 4.5477
5 4 40 11.7206 7.5400 4.1806 4.5084
5 5 41 12.4345 8.2190 4.2155 4.4695
5 6 42 12.7461 8.8359 3.9101 4.4308
5 7 43 12.9505 8.8448 4.1057 4.3925
5 8 44 12.8575 8.8661 3.9914 4.3546
Does somebody know why the column E[u/e] give efficiency estimates largely
above 1??? The theory states that efficiency scores must take values
between 0 and 1.
Best Regards
Dimitrios Giantsios
Phd Candidate
More information about the Limdep
mailing list