The behavior of U.S. credit card borrowers varies considerably, with the riskiest cardholders about 20 times more likely to become seriously delinquent on payments than the least risky cardholders. Lenders can predict cardholder behavior using credit scores and income, which lenders can use to set credit terms (that is, the credit limit and interest rate) on newly originated credit card contracts.1 Because lenders can anticipate differences in delinquency risk, large differences exist in credit terms between the least and most risky cardholders.

In their paper, “Explaining Contract Heterogeneity in the Credit Card Market,” Satyajit Chatterjee and Burcu Eyigungor of the Philadelphia Fed investigate differences in borrower behavior to explain the wide variations in credit terms, usage, and default risk. They also examine the implications of proposed U.S. legislation to cap credit card interest rates aimed to help borrowers lower their cost of funds.2

Using anonymized U.S. bank supervisory data, they identified patterns in newly originated credit card accounts.3 To explain these patterns, they extended a standard (heterogeneous-agent) macro model to include the credit card market and then compared it with a traditional model. They also measured individual cardholders’ marginal propensity to consume (MPC) to better understand cardholders’ spending habits.4 Finally, they analyzed the puzzlingly large spread between the credit card interest rates lenders charge (net of their cost of funds) and the observed default rates.

They found that cardholders’ credit terms, usage, and default rates “vary systematically” with income and credit score. Specifically, cardholders with both higher scores and higher incomes received lower interest rates,5 and those with higher scores received higher limits relative to their incomes and those with higher incomes received lower limits relative to their incomes.6 Default rates increased sharply as credit scores and income fell,7 and borrowers with credit scores in the lowest quintile had an average credit utilization rate of about 55 percent compared with less than 4 percent for those in the top quintile.8

Next, the authors concluded that the traditional model, which attributes differences in credit card contracts solely to income, can’t explain the patterns in the data. They instead allowed individuals to differ in discount factors to capture how they view consuming today versus saving for the future; this reflects their degree of “patience.” And they allowed people to differ in default costs to capture individuals’ aversion to the consequences of defaulting.9 They show that more-patient individuals have higher default costs (which means these two factors are correlated) and that individuals are, generally speaking, “quite impatient.”

These insights have important real-world implications. Less-patient borrowers are riskier to lenders, which incentivizes lenders to offer them lower credit limits and higher interest rates. Similarly, borrowers who are less averse to defaulting receive less-favorable credit terms.

Their model also explains why credit card interest rates may appear high relative to observed default rates. First, many borrowers receive promotional rates at credit origination, typically for the first year, which lowers the effective interest rate. Second, although many consumers carry revolving balances, defaults typically occur only when balances approach credit limits. If the average utilization rate is low across all borrowers, lenders require higher interest rates to recover their costs from occasional defaults. For example, if the average utilization rate is 25 percent (which is a typical figure), “the break-even spread” on the card is four times the default frequency. Simply put, this spikes interest rates.

Reflecting on the proposed 10 percent interest rate cap, the authors write that “the motivation for such caps seems to derive from the perception that credit card interest rates are too far above the cost of funds to be competitive,” but “spreads exceeding default probabilities by a large margin can easily arise in a competitive credit card market.” They show that, if faced with a 10 percent interest rate cap, lenders would significantly reduce credit limits for high-risk borrowers, making the policy “welfare-reducing” for this group.

Finally, the authors found that many individuals have a high MPC: When income rises, even temporarily, many consumers spend a large fraction of that income rather than saving it. High MPCs, they show, contribute to both high default rates among high-risk borrowers and high credit card utilization rates, even in the face of the high interest rates.

Chatterjee and Eyigungor’s paper shows us that credit card contract heterogeneity reflects more than income alone. Differences in patience and default aversion play important roles in determining credit terms, usage, and default risk, providing valuable insights for policymakers. The authors also provide a timely policy contribution by showing that interest rate caps benefit some consumers but restrict credit access for others.

  1. The views expressed here are solely those of the author and do not necessarily reflect the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System.
  2. Specifically, the interest rate is on revolving debt, which is the portion of outstanding debt that is not paid in full by the statement due date.
  3. See Bills H.R.1944 and S.381, which were introduced in the 119th Congress (2025–2026) and titled “10 Percent Credit Card Interest Rate Cap Act.”
  4. The data are from the Federal Reserve System and cover about 80 percent of all U.S. credit card accounts between mid-2014 and mid-2015.
  5. Marginal propensity to consume (MPC) is defined as the fraction of additional income that an individual spends on consumption rather than saving it.
  6. Higher credit scores were more important than income in explaining interest rates.
  7. However, credit limits increased less than proportionally as incomes rose.
  8. Credit score was found to be more important than income in predicting defaults.
  9. A lower income was also associated with higher credit utilization, but income was a less important factor than credit scores.
  10. Credit card holders in default typically experience a damaged credit record, reduced access to future borrowing, and potential collection activity.