CES Technical Notes (Restricted Access)

Constructing a Person-Country Roster Using Tax Return and Social Security Administrative Data
​(W/ J. David Brown)
Abstract (Click here to expand)

This technical note documents the creation of a person-country roster using 2020 address information from the IRS’s 1040 tax returns and the SSA’s Master Beneficiary Records. The roster provides the most comprehensive snapshot available of U.S. persons who are residing abroad or have dual residency in the U.S. and another country. We first describe the inputs and process to assign each administrative record a country of residence. Key demographic and address characteristics of persons with a foreign address are documented. We test the validity of inferring foreign residence from a foreign address by measuring the share of persons with a foreign address are found in the 2020 Census, which we assume to be strong evidence that a person is a U.S. resident on Census Day. The note concludes with a discussion of the current limitations of the person-country roster and recommendations to improve the roster going forward.

Evaluating the Quality of a National Voter Registration File Using Administrative Records
Abstract (Click here to expand)

This technical note presents a fitness-for-use analysis of the 2020 voter registration files provided to the Census Bureau by a commercial data vendor. The analysis covers the PIK and MAFID assignment rate at the national, state, and county levels as well as by demographic characteristics. The quality of the demographic and address data is measured and their impact on the PIK and MAFID assignment rates is examined. The assigned PIKs are used to measure the number of duplicate voter registrations within and across state voter registration files. The number of deceased voters in the voter registration files according to the Numident are measured. The 2020 files are then merged with the 2024 voter registration files using the unique vendor-provided ID to measure the quality of each state’s voter registration files over time. The merged files are also used to measure the quality of the vendor-provided IDs and the PVS process in assigning PIKs. The note concludes by comparing the total number of registrations to the 2020 Current Population Survey official estimates of the number of persons registered to vote by state.

Public Technical Reports