Determinants of U.S. Credit Card Delinquency Rates

Abstract

This paper examines what drives U.S. credit-card loan delinquency rates using quarterly time-series data from FRED (2000Q3–2023Q2) and a set of macro/financial variables capturing household behavior, business-cycle conditions, and monetary policy. The analysis estimates linear regression specifications and explicitly models nonlinear and regime effects by including squared unemployment and a COVID interaction term, while using Newey–West robust standard errors to address autocorrelation. Results indicate delinquency rises with higher credit-card plan interest rates, tighter financial conditions, and recession periods, while credit-card spending and the effective federal funds rate are associated with lower delinquency; unemployment exhibits a threshold effect, switching from negative below about 4.19% to positive above it, with the threshold shifting upward to about 7.59% during COVID.

Data

• Dataset details: Quarterly U.S. time-series from FRED, 2000Q3–2023Q2 (N = 92 observations)
• Dependent variable: Credit card delinquency rate (DELRATE, %)
• Independent + control variables (summary): Credit card APR (CCINTREST), revolving balances (CCAMOUNT), unemployment (UNRATE) with a nonlinear term (UNRATE²), financial conditions (Chicago Fed NFCI), personal saving rate (PSAVERATE), real GDP (GDP), effective fed funds rate (FEDFUNDS), and recession/COVID indicators including an unemployment×COVID interaction

Methods

• Methodology: Quarterly time-series OLS regressions to explain delinquency using macro + credit-market drivers
• Model features: Adds nonlinear unemployment (UNRATE²) and a COVID interaction to capture regime shifts in the unemployment–delinquency relationship
• Robustness / inference: Uses Newey–West robust standard errors to address strong autocorrelation typical in macro time series (coefficients unchanged; inference adjusted)

Key Results

• Credit card APR is a major driver: A +1 pp increase in card APR (CCINTREST) is associated with roughly a +0.4 pp increase in delinquency (holding other factors constant)
• Financial conditions matter: A +1 unit increase in NFCI (tighter conditions) is associated with about a +0.565 pp increase in delinquency
• Unemployment is nonlinear (threshold effect): Unemployment becomes a positive delinquency driver only above about 4.19%, and the implied threshold shifts to about 7.59% during COVID

Conclusion

Delinquency is most strongly explained by borrowing costs (APR) and tight financial conditions, while unemployment affects delinquency in a nonlinear, regime-dependent way that changes in the COVID period. The results emphasize why modeling macro credit risk needs both financial-conditions variables and nonlinear labor-market effects, not just a single linear unemployment term.

Key Model Outputs