Robust stochastic frontier analysis applied to the Brazilian electricity distribution benchmarking method

Campos, Magno Silvério; Costa, Marcelo Azevedo; Gontijo, Tiago Silveira; Lopes-Ahn, Ana Lúcia

doi:10.1016/j.dajour.2022.100051

Cited by 5 publications

(1 citation statement)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The standard SPF model assumes the noise component follows a Normal distribution and the inefficiency component follows a half-Normal distribution. However, the standard SPF model has limitations in estimating technical efficiency when outliers are present in data [5]. The presence of outliers in the data will affect the inaccuracy in estimating the parameters, leading to an exaggerated spread of efficiency predictions [6].…”

Section: Introductionmentioning

confidence: 99%

Robust Stochastic Production Frontier to Estimate Technical Efficiency of Rice Farming in Sulawesi Selatan

Pranata

Djuraidah

Aidi

2023

BAREKENG: J. Math. & App.

View full text Add to dashboard Cite

The stochastic production frontier (SPF) is the stochastic frontier analysis (SFA) method used to estimate the production frontier by accounting for the existence of inefficiency. The standard SPF assumes that the noise component follows a Normal distribution and the inefficiency component follows a half-Normal distribution. The presence of outliers in the data will affect the inaccuracy in estimating the parameters and leads to an exaggerated spread of efficiency predictions. This study uses two alternative models, the first with SPF Normal-Gamma and the second with SPF Student's t-half Normal, then the results are compared with standard SPF. This study uses data from statistics Indonesia on the cost structure of paddy cultivation household survey in 2014. This study aims to examine the effect of changes in distribution assumptions on the standard SPF model in estimating parameter value and the technical efficiency score in the presence of outliers. The parameter coefficient estimates similar results that apply to three SPF models. Only the standard error value in the alternative SPF model tends to be smaller than the standard SPF model. The Normal-Gamma model performs better in assessing residual with smaller root mean square error (RMSE) than the others, but the results of the estimated technical efficiency still contain outliers. The Student's t-half Normal model estimates technical efficiency no longer contains outliers, the range is shorter than the other models, and the results of estimating technical efficiency are not monotonous in the distribution of residual tails. The SPF Student's t-half Normal model is more robust in presence outliers than SPF Normal-half Normal and SPF Normal-Gamma.

show abstract

Section: Introductionmentioning

confidence: 99%