Using 2015 National Beneficiary Survey (NBS) data matched to Social Security Administration (SSA) administrative records for 4,062 respondents (weighted: 12.9M beneficiaries), Walker and Roessel document the prevalence and characteristics of Disability Insurance (DI) and Supplemental Security Income (SSI) beneficiaries with multiple disabling impairments. They show that survey data reveal substantially higher prevalence of multimorbidity than SSA administrative records — which are capped at two impairments — and that the two data sources exhibit systematically different patterns by impairment type due to coding conventions and self-report biases.
"Although two-thirds of beneficiaries reported multiple impairments, 8 percent reported no limiting impairments." "The administrative data, although complete for that purpose, thus understate the presence of individual impairments as well as the total number and burden of beneficiaries' disabling impairments."
A thorough descriptive paper that makes a point more subtle than it first appears: admin data aren't just incomplete — they're systematically biased in opposite directions for physical vs. mental impairments. Physical conditions are underrepresented in admin (only the primary documenting impairment gets recorded); mental conditions are overrepresented in admin relative to self-report (stigma suppresses survey self-disclosure). This creates a complex picture where "the" truth lies somewhere between the two sources. The paper's main policy message — use survey or electronic Claims Analysis Tool (eCAT) data for targeting employment supports — is well-supported by the evidence. The "no-impairments" group (8%) is an interesting discovery: often mentally ill young SSI recipients who don't self-report their diagnoses, yet the admin data records them clearly.