The Normal distributions for the two error components of a Stochastic Frontier approach can be seen as a result of computational convenience. Several authors suggest using more flexible distributions than the Normal distribution in order to capture situations of symmetry and asymmetry or kurtosis with one or more parameters. In any case, such generalizations do not exclude the possibility that such distributions may be Gaussian.
What do you think about this statement?
The normal is within a more generalized pearson density. Attempts to do wald testing on a more general density failed theoretically in the 1980s. See work by Chris CornwallCornwall
Thank you for the answer, but I am not sure about the paper you mentioned. Is it Cornwall, Schmidt and Sickles (1990) on Journal of Econometrics or Cornwell and Rupert (1988) on Journal of Applied Econometrics?
https://www.researchgate.net/publication/279053691_The_Wrong_Skewness_Problem_A_Re-Specification_Of_Stochastic_Frontiers
Conference Paper The “Wrong Skewness” Problem: A Re-Specification Of Stochast...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size