Local influence analysis of stochastic frontier estimation: A case-weights perturbation approach

The local influence analysis of the stochastic frontier model estimation is explored under the case-weights perturbation scheme, which is based on the theory of differential geometry. These local influence analysis procedures are illustrated using the half-normal stochastic frontier model. The likelihood displacement is derived, which is the crucial influence diagnostics. This kind of local influence analysis has many practical implications on the stochastic frontier model, such as data sensitivity analysis. A simulation study is provided which will be useful for practitioners.