Auto loan default sensitivity to unemployment declines sharply
A Federal Reserve Bank of Cleveland working paper published in July 2026 reveals persistent instability in the sensitivity of auto loan defaults to unemployment. Authors Nicholas T. Fritsch and Edward S. Prescott show that pre-pandemic models overstated Covid-era defaults by up to 100 basis points.
The shrinking impact of unemployment
Researchers Nicholas T. Fritsch and Edward S. Prescott estimate a discrete-time Markov transition model of auto loan performance using semi-annual data from 2000 to 2025.
Applying multinomial logistic regressions and rolling pseudo out-of-sample forecasts, they document a persistent decline in the sensitivity of default probabilities to unemployment shocks.
The estimated effect of a 1 percentage point increase in unemployment on default risk fell from 16 percent in 2006 to just 3 percent post-2020.
Two-year cumulative default forecasts over 2020-2021 using pre-pandemic parameters overstated actual defaults by 100 basis points, representing a 25 percent overstatement relative to actual outcomes.
Lender segmentation and subprime risk
The study highlights significant segmentation across lending channels, examining loans originated by banks, credit unions, and specialized auto finance companies.
While banks' market share declined and credit unions expanded since 2000, auto finance companies concentrated heavily in subprime lending.
Even after accounting for observable characteristics, loan performance differs markedly by lender type, with auto finance subprime loans carrying substantially higher risk than those originated by depository institutions.