Measuring Renters in Credit Data: Evidence from Linked Survey and Administrative Data

• In Event: Innovative Rental Market and Neighborhood Data

Abstract

Introduction/Background: With rents rising rapidly, traditional credit bureau data have become increasingly popular for studying renter financial well-being. These datasets combine large samples and more timely information compared to standard household surveys. However, these datasets lack systematic housing tenure information, driving researchers to use proxy measures. A common approach for researchers is to classify individuals without mortgages as renters, which misclassifies nearly 25 percent of U.S. adults who own homes free and clear.  Moreover, compared to the typical renter, homeowners who own free and clear are much better off, leading to mismeasuring and likely understating the impacts of rising rents on renter financial well-being.

Purpose/Research Question: This paper analyzes: (1) How much mismeasurement occurs when using traditional credit data proxies to identify renters? (2) How does this mismeasurement affect analyses of renter financial well-being? (3) What alternative proxies using commonly available information in credit data can better identify renters?

Data: We use anonymously linked data combining the 2024 Survey of Household Economics and Decisionmaking (SHED), a nationally representative survey of over 12,000 U.S. adults, with Experian credit records. The sample includes approximately 6,900 consenting respondents with both self-reported housing tenure and detailed credit bureau information.

Research Design and Methods: We construct a baseline measure of "proxy renters" following existing methodology in the literature: individuals without mortgages in their credit reports over seven years (2016-2023) are considered renters. We compare this group to "survey self-reported renters." We use confusion matrices to analyze accuracy and financial outcome measures to analyze differences in financial well-being. We then test alternative proxies using information commonly available in credit data: age, credit score, and mobility.

Results/Findings: The baseline proxy measure substantially misclassifies renters: only 42 percent of "proxy renters" actually report paying rent. Among the misclassified, 32 percent own homes free and clear, 16 percent report having mortgages, and 11 percent neither own nor rent. This misclassification substantially understates renter financial distress. Among self-reported survey renters, 41 percent have delinquent credit accounts versus 28 percent for proxy renters. Survey renters are also more likely than proxy renters to have high credit card utilization (26 percent versus 17 percent) and have lower self-reported financial well-being. The best alternative proxy, no mortgage and consumer age under 50, improves accuracy to a 56 percent true positive rate while capturing 70 percent of all renters. However, this measure still has a 44 percent false positive rate.

Conclusion/Implications: Researchers using standard proxies for renters in traditional credit data should exercise caution, as these proxies contain critical measurement errors that artificially inflate measures of renter financial well-being. While alternative proxies reduce misclassification, they involve tradeoffs between accuracy and population coverage. Researchers using credit data to study renters must carefully select proxies appropriate to their research questions and explicitly acknowledge measurement limitations to avoid inaccurate conclusions about renter financial well-being.

Author

• Erin Troland, Federal Reserve Board
  Presenting Author