Stochastic Dominance, Estimation and Inference for Censored Distributions with Nuisance Parameter

This note investigates the behavior of stochastic dominance tests of censored distributions which are dependent on nuisance parameters. In particular, we consider finite mixture distributions that are subject to exogenous censoring. To deal with this potential problem, critical values of the proposed tests statistics are calculated using a parametric bootstrap. The tests are then applied to compare differences between distributions of incomplete employment spells with different levels of censoring obtained from Canadian General Social Survey data. The size of the proposed test statistics is computed using fitted GSS data.