Making sense of July’s faster consumer-credit growth

US consumer credit—the amount remaining owed—rose by approximately $18.1 billion in July 2026, versus $14.6 billion in June. The Federal Reserve’s September 8 release expresses July’s pace as 4.2% annualized growth, up from June’s 3.4%. July is preliminary and subject to updating; June is revised.13

The release also lists July’s dollar flow at an annual rate: $216.7 billion. Dividing by 12 converts that to approximately $18.1 billion for the month. June’s outstanding balance was $5,168.1 billion; July’s was $5,186.2 billion. Divide the monthly addition by June’s balance to express growth relative to what was already owed. Multiplying that monthly growth rate by 12 puts it on an annual scale without compounding—without assuming growth builds on earlier growth. This is how the dollar addition translates into the published annual rate. It describes one month’s pace, rather than growth over the past year or a forecast. A faster month alone does not establish a lasting acceleration.13

These figures are seasonally adjusted, meaning they account for recurring seasonal patterns to make months more comparable. Published rates use unrounded data and remove the effects of technical changes in reporting. Subtracting displayed balances therefore need not reproduce an official net change exactly.213

The report excludes loans secured by real estate. Its revolving category allows repeated borrowing within a limit and consists mostly of credit cards. Nonrevolving credit follows a prearranged repayment schedule; vehicle, education and personal loans are familiar examples.2

Larger nonrevolving additions outweighed the slowdown in revolving additions, supplying approximately 84.5% of July’s net increase. That is a share of the increase, not of all outstanding debt. Revolving balances still rose. The data do not identify autos versus education as July’s driver.1

The category dollar changes contribute to the same total, although independent rounding can keep the displayed figures from adding exactly. Their percentage growth rates cannot simply be added: each uses a different starting balance. A category’s dollar contribution depends on both its size and its growth rate.312

To interpret those additions, consider what changes the amount owed. New borrowing adds to it; paying down debt reduces it. More borrowing alongside more repayment can leave the same net increase as less borrowing alongside less repayment. The net figure cannot separately reveal how much consumers newly borrowed, what they purchased or how quickly they repaid debt.23

Reporting coverage also matters. The Fed describes a lender selling loans to an institution outside the report’s scope: measured balances can fall even though households still owe the debt. Such technical breaks are removed from flows so that a change in who reports the loans does not masquerade as changed household obligations.2

Earlier bank responses help test whether faster nonrevolving growth signals easier access or stronger auto demand. The July Senior Loan Officer Opinion Survey generally covered the second quarter, despite its title. On balance, banks reported basically unchanged auto and other consumer lending standards, weaker auto demand and tighter credit-card standards. Standards are policies for approving applications: they describe access to credit, before any resulting lending is counted.61

These results balance banks reporting opposite changes. They count banks, rather than loan dollars, and an unchanged net result can conceal differences among lenders. This earlier evidence provides no basis for assuming broad easing or stronger auto demand behind July’s increase. Its period and bank-only coverage also prevent it from resolving July’s actual borrowing and repayment flows.61

Borrowing cost answers another question. The annual credit-card rate for accounts charged interest was 22.15% in the second quarter, versus 21.52% in the first. At reporting banks, it relates finance charges to average daily balances charged interest, excluding accounts without those charges. It measures cost, not whether payments arrive on time, and cannot establish an individual household’s burden.1

For repayment evidence within nonrevolving debt, the New York Fed’s auto measure tracks balances newly becoming at least 90 days late. The comparison base is balances that were current or less than 90 days late in the preceding quarter. Its annualized transition rate was 3.00% in the second quarter of 2026, versus 2.93% a year earlier. That measures balances crossing a lateness threshold, not the percentage of people struggling or all debt already overdue. Drawn from credit-report data, it adds earlier payment context but cannot establish July borrowers’ condition.42

Accumulated overdue balances can move differently from new delinquencies. In their August 11 analysis, New York Fed researchers attribute rising credit-card delinquency stocks—the share of reported balances already overdue—to charged-off debts being reported for longer. A charge-off removes debt from lender books but generally leaves the borrower’s obligation intact. Continued bureau reporting lets unresolved balances accumulate despite stable new delinquency. Why lenders report these debts longer remains unestablished.5

Their lender comparison measures balances on lenders’ books at least 30 days late; the bureau stock uses 90 days. For new transitions, they sum four successive quarterly rates. This captures a year’s entries into delinquency, whereas the consumer-credit release scales one month’s growth to an annual pace. The annual labels therefore describe different time windows.53

The analysis shares Consumer Credit Panel evidence with the New York Fed’s Quarterly Report on Household Debt and Credit; it is not independent corroboration. The researchers find new credit-card delinquency elevated but largely stable since 2024. Stable entry can coexist with hardship from unresolved debts. These reporting differences explain divergent repayment indicators without establishing July’s growth causes.541

Report an error · Corrections policy