Quarterly US macro-financial data from 1995Q1–2023Q1 are drawn from the Federal Reserve, Bureau of Labor Statistics, and Bureau of Economic Analysis. Key variables include credit-card charge-off rates, delinquency rates, interest rates on credit-card plans, the unemployment rate, and a principal-component measure of real personal income growth. A multi-variable Structural Vector Autoregression (SVAR) model with exact identification and multiple quarterly lags is employed to examine the transmission channels linking macroeconomic shocks to consumer-credit performance. Estimated over 121 quarterly observations, the model satisfies rigorous contemporaneous restrictions and demonstrates strong statistical robustness. Results indicate asymmetric and nonlinear effects. Positive income shocks paradoxically raise charge-off and delinquency rates immediately, suggesting heightened credit use, greater borrower risk-taking, or delayed recognition of repayment distress. Conversely, unemployment shocks contemporaneously reduce charge-offs, consistent with lender forbearance or deferred loss recognition. Lending-rate shocks display no contemporaneous effect, revealing a lag in monetary- policy transmission. These structural dynamics challenge the efficacy of static policy tools, such as a uniform 10% Annual Percentage Rate (APR) cap. While such caps can curb pricing, they risk constraining credit access for high-risk borrowers, inducing adverse selection, and diminishing value-added services for lower-risk segments. A dynamic macroprudential framework is therefore warranted, incorporating risk-tiered interest corridors, countercyclical buffers for unsecured lending, and granular supervisory monitoring aligned with labor-market and income-support programs. By integrating SVAR-based structural clarity with scenario-driven stress testing, this methodology provides central banks a rigorous foundation for adaptive, risk-sensitive regulation that safeguards both credit access and financial stability.
Masaaki Yoshimori (Tue,) studied this question.