Using dominance in forming bounds on DEA models: The case of experimental agricultural data

Dominance techniques are used to create optimistic and pessimistic convex free-disposal hull technologies for data sets. These optimistic and pessimistic technologies create bounds within which one can reasonably expect the true technology to lie. The potential differences between these optimistic and pessimistic reference technologies are examined by comparing the reference technologies for data drawn from agricultural field trials. Empirical analysis suggests that there are significant differences between the reference technologies.