Comparing Survey and Administrative Measures of Self-employment Income: New Implications from the Health and Retirement Study

An ample literature has found that self-employment income is under-identified in both administrative earnings records and major household surveys, potentially limiting accurate inference for tax policy. This paper uses a different survey, the 2004-2016 Health and Retirement Study, linked to administrative earnings records and identifies substantially more self-employment income in the survey at both the extensive and intensive margins, in line with the Internal Revenue Service's estimated compliance rate. These findings highlight (1) the capacity for survey data to identify self-employment income otherwise missed in analyses using only administrative records; and (2) heterogeneity in reporting self-employment income across surveys.