Undercounting Medicaid Enrollment in Maryland: Testing the Accuracy of the Current Population Survey

The U.S. Census Bureau’s Current Population Survey is the most commonly cited source for estimates of the insurance status of Americans, but there are concerns that the data may undercount participation in public programs such as Medicaid. Such a Medicaid undercount may result from survey respondents not acknowledging Medicaid coverage because they are unaware that they are enrolled in Medicaid, because they have not recently received health services, due to the stigma associated with receiving public assistance programs, or due to simple recall bias. This paper estimates the extent to which the Current Population Survey undercounts Medicaid participation in a single mid-Atlantic state, Maryland. We administered the Current Population Survey questionnaire to a random selection of known Medicaid participants. We find evidence that the Current Population Survey significantly undercounts Medicaid participation in Maryland and that much of the undercount could be corrected if the survey better identified the Maryland Medicaid program. We also find that recall bias may contribute to the undercount as well. There was no indication that stigma contributed to any undercount. Though not an aim of this study, we find that the Medicaid undercount may contribute to an over count of the number of uninsured.