Getu Hailu, Scott Jeffrey, Jim Unterschultz
In this study, two non‐homothetic translog stochastic meta‐frontier cost functions—with and without local concavity imposed—are estimated using a nonlinear maximum likelihood estimation procedure to compare the cost efficiency of Alberta and Ontario dairy farms for the period 1984–96. The resulting cost efficiency estimates are not very sensitive to whether or not curvature is imposed. In contrast, the properties of the cost and input demand functions (e.g., elasticities) are sensitive to imposition of local concavity during estimation. The implication is that if an inappropriate model that does not satisfy the properties required by the economic theory is used, the estimated input demand functions may not be reliable. Average cost efficiency for the pooled sample, with local concavity imposed, is approximately 89%. This suggests some potential for improved performance in the sector. The results also suggest that Ontario dairy farms may be more cost efficient than Alberta dairy farms, but the statistical evidence is inconclusive.
Publication: Canadian Journal of Agricultural Economics
Date: January, 2005