Nonparametric Bounds on the Income Distribution in the Presence of Item Nonresponse

Item nonresponse in micro surveys can lead to biased estimates of the parameters of interest if such nonresponse is nonrandom. Selection models can be used to correct for this, but parametric and semiparametric selection models require additional assumptions. Manski has recently developed a new approach, showing that, without additional assumptions, the parameters of interest are identified up to some bounding interval. In this paper, we apply Manski’s approach to estimate the distribution function and quantiles of personal income, conditional on given covariates, taking account of item nonresponse on income. Nonparametric techniques are used to estimate the bounding intervals. We consider worst case bounds, as well as bounds which are valid under nonparametric assumptions on monotonicity or under exclusion restrictions.