Abstract
This paper examines and compares the technical efficiency measures of Ontario and New York dairy producers for the period 1992 to 1998. A nonparametric stochastic frontier model is introduced to estimate technical efficiency. The backfitting algorithm of Breiman and Friedman is used to estimate the frontier. Empirical results indicate that during the period of study, New York dairy farmers produced milk more efficiently than Ontario dairy producers, but the magnitude of the difference was small. The estimated mean technical efficiency for the former group is 0.602 as compared to 0.532 for the latter. The results also indicated that over time, dairy farms in both regions improved their level of technical efficiency. Furthermore, no correlation was found between farm size and estimated technical efficiency.
Acknowledgements
The authors would like to thank Robert Romain to many helpful comments and suggestions that led to substantial improvement of the paper. All the remaining errors are their own.
Notes
DEA (Charnes et al., Citation1978) is a linear programming methodology used to construct a piece-wise convex surface (or frontier) which ‘ envelops ’ the data. FDH (Deprins et al., Citation1984) is similar to DEA, but the convexity assumption is relaxed. With both of these deterministic frontier methods, the distance from each observation to the computed frontier is the measure of inefficiency.
It is intended to show the non-robustness of efficiency score estimates to the choice of various distribution assumptions of error terms in the authors ’ next research.
The structure of the assumed production function can be easily tested. See Section VI for more details.
The new independent variable can be introduced because with the type of methodology proposed, the problem of the curse of dimensionality is not applicable to the model.