Partially Identifying the Prevalence of Health InsuranceGiven Contaminated Sampling Response Error

This paper derives simple closed-form identi…cation regions for the U.S. nonelderlypopulation's prevalence of health insurance coverage in the presence of household reporting errors.The methods extend Horowitz and Manski's (1995) nonparametric analysis of contaminated samplesfor the case that the outcome is binary. In this case, draws from the alternative distribution (i.e., notthe distribution of interest) might naturally be de…ned as response errors. The derived identi…cationregions can dramatically reduce the degree of uncertainty about the outcome distribution comparedwith the contaminated sampling bounds. These regions are estimated using data from the MedicalExpenditure Panel Survey (MEPS) combined with health insurance validation data available for anonrandom portion of the sample.[...]