Measuring Foreign Residence in U.S. Administrative Records

• In Event: Measurement and Data Collection in Migration Research

Abstract

In an era of declining survey response rates and increasing dependence on administrative records (AR) for population measurement, developing accurate methods to identify the true resident population of the United States has become an urgent priority. AR data offer substantial advantages for improving coverage and reducing measurement error, but they also introduce distinct challenges. One of the most persistent issues is the difficulty of identifying U.S. residents who have emigrated. Because emigration is not consistently captured across federal administrative systems, individuals who leave the country often continue to appear in AR-based frames for years after their departure. This leads to systematic overcounting in population estimates and can distort statistics used for policy evaluation, resource allocation, and demographic research. This study constructs a Foreign Residence Frame designed to identify likely emigrants using foreign address information from IRS Form 1040 filings and the Social Security Administration’s Master Beneficiary Records. For the years 2019–2020, I identify 2.7 million individuals with a documented foreign address, including 1.5 million who simultaneously appear to maintain a U.S. address in other AR sources. This overlap highlights the complexity of inferring residency from AR signals alone. To evaluate the validity of foreign address indicators, I compare these individuals to the 2020 Decennial Census, using enumeration status as a proxy for physical presence in the United States. The analysis shows that having any foreign address in AR data is a strong predictor of residence abroad, though the strength of this relationship varies substantially across regions of foreign residence, age groups, and citizenship status. Finally, I quantify the magnitude and demographic distribution of overcoverage that results from including these likely emigrants in the 2020 Demographic Frame—a composite file drawing on more than twenty federal administrative sources. The findings underscore the need for refined methodologies to better integrate AR data with traditional census operations and offer actionable insights for improving the accuracy of future population estimates.

Author

• Andres Felipe Mira, U.S. Census Bureau
  Presenting Author