Abstract: This article studies tail behavior for the error components in the stochastic frontier model, where one component has bounded support on one side, and the other has un-bounded support on both sides. Under weak assumptions on the error components, we derive nonparametric tests for thin-tailed distributional assumptions imposed on these two components. The tests are useful diagnostic tools for stochastic frontier analysis and kernel deconvolution density estimation. A simulation study and applications to four previously studied datasets are provided. In two of these applications, the new tests reject the thin-tailed distributional assumptions such as normal or Laplace, which are commonly imposed in the existing literature.
Link to work