superefficiency

Superefficiency is a term used in several fields to describe performance that exceeds a given standard or benchmark of efficiency. In general, it denotes a situation where a process, estimator, or system delivers better performance than the conventional efficiency measure under a specified model or dataset. The meaning and implications vary by field.

In statistics, superefficiency refers to an estimator whose asymptotic risk is smaller than the Cramér-Rao lower bound at some parameter values. The classic example is Hodges' estimator for the mean of a normal distribution with known variance, which is superefficient at zero but has higher risk elsewhere. Superefficiency is rare in practice and often involves pointwise improvement at the cost of worse performance in other regions; it is inherently an asymptotic and non-robust phenomenon, and cannot be achieved uniformly across all parameter values.

In operations research and data envelopment analysis (DEA), the term super-efficiency is used to describe a class of efficiency scores where a decision-making unit (DMU) can score above 1 on an efficiency frontier, typically by removing the unit under evaluation from the reference set. This facilitates ranking among efficient units but can be highly sensitive to data and may require cross-validation or secondary measures to ensure stability.

Other uses are more informal, including references to systems or processes that perform better than a conventional efficiency benchmark, such as energy or computation where the measured efficiency exceeds 100 percent under certain definitions.

In summary, superefficiency is a contextual term signaling atypical, nonuniform advantages rather than a universal improvement.