Mortgage Delinquency
Who is falling behind on mortgages, how often late borrowers recover, and who is getting new loans. Owners with equity can often sell rather than default, which is one reason mortgage delinquency has stayed low since 2010.
| Measure | Latest | 2019 | vs 2019 |
|---|---|---|---|
| Balance 90+ days late | 0.99% | 0.98% | Up to 10% above |
| Newly 90+ days late, annualized | 1.52% | 1.06% | Over 10% above |
| Newly 30+ days late, annualized | 3.95% | 3.46% | Over 10% above |
| Bank single-family mortgages 30+ days late | 1.86% | 2.52% | At or below |
| Bank single-family mortgages charged off, annualized | 0.00% | -0.01% | No positive base |
Counted: latest more than 10% over a positive 2019 average. Up to 10% above and at or below 2019 are shown, not counted; a series whose 2019 average is zero or negative has no base. Near record: at least 90% of the series high. Counted, never scored. Descriptions, not forecasts.
Latest reading
In Q2 2026, 0.99% of mortgage balances were 90 or more days late in credit files, against a 2019 average of 0.98% and a record of 8.89% in Q1 2010. Over the four quarters to Q2 2026, 1.52% of balances newly became seriously late, against 1.29% a year earlier. At banks, 1.86% of mortgage loans were 30 or more days late in Q2 2026 and 0.00% were charged off at an annual rate. Balances stood at $13.12T, +1.4% on a year and −0.6% on the quarter.
Sources, methodology & freshnessNew York Fed Quarterly Report on Household Debt and Credit (Equifax credit files); Federal Reserve bank delinquency, charge-off, G.19 and SLOOS series via FRED · Quarterly; the Q2 2026 report was published 42 days after the quarter ended. Release dates before the latest report are estimated as quarter end + 40 days; the latest quarter uses the report monthData as of Q2 2026, released August 11, 2026 · Open ↓Close ↑
Late payments in credit files
Balances 90+ days late: 0.99% in Q2 2026, +0.17 pts on a year, the 33rd percentile of 94 readings since 2003. Newly 30+ days late over the year: 3.95% in Q2 2026, +0.27 pts on a year, the 47th percentile of 94 readings since 2003. The flow measures count balances that newly crossed 30 and 90 days late during the year, annualized.
Late loans: cure, roll or foreclose
Of mortgages 30–60 days late, 42.04% returned to current in Q2 2026 and 15.42% rolled on to 90+ days late. New foreclosures: 55.2K consumers in Q2 2026, against a peak of 566.2K in Q2 2009. Each quarter, a small share of current mortgages slips to 30–60 days late, and of those already late, some cure and some roll on to 90+ days. A falling cure rate means fewer late borrowers are catching up; in 2006 and 2007 it fell before foreclosures surged.
| Step, per quarter | Latest | 2019 | Low – high since 2003 |
|---|---|---|---|
| Current to 30–60 days late, per quarter | 0.84% | 0.87% | 0.27% (Q2 2021) – 3.05% (Q4 2008) |
| Current to 90+ days late, per quarter | 0.23% | 0.15% | 0.08% (Q2 2021) – 0.75% (Q2 2009) |
| 30–60 days late back to current | 42.04% | 37.75% | 20.47% (Q3 2009) – 61.05% (Q2 2020) |
| 30–60 days late on to 90+ | 15.42% | 13.65% | 5.51% (Q3 2021) – 44.23% (Q1 2009) |
| Consumers with a new foreclosure | 55.2K | 69.4K | 8.1K (Q2 2021) – 566.2K (Q2 2009) |
Who is falling behind, by age
In Q2 2026, the highest rate was ages 18–29 at 1.89% and the lowest ages 70+ at 0.83%. Each line is the share of that age group’s mortgage balances that newly became 90+ days late over the last four quarters.
On bank books
Bank mortgage loans 30+ days late: 1.86% in Q2 2026, +0.08 pts on a year, the 20th percentile of 142 readings since 1991. Charge-offs: 0.00% in Q2 2026, +0.00 pts on a year, the 22nd percentile of 142 readings since 1991. Banks report single-family mortgages they hold that are 30+ days late, seasonally adjusted, and the share charged off, net of recoveries. Charge-offs near zero mean banks are losing little on the mortgages that do fail.
How much is owed
Balance outstanding: $13.12T in Q2 2026, +1.4% on a year, the 98th percentile of 94 readings since 2003. Open accounts: 86.4M, −0.1% on a year.
Source note. The New York Fed attributes the second quarter’s fall in mortgage balances on credit reports to a temporary gap in reporting while mortgages changed servicer. New York Fed Household Debt and Credit Report, Q2 2026.
New lending and credit scores
New mortgages: $504.6B in Q2 2026, +10.1% on a year. 2.97% went to borrowers scoring below 620; under the previous score model the share peaked at 15.24% in Q1 2007. The median score at origination was 764. New mortgages are counted by the borrower’s credit score at origination. The New York Fed switched from Equifax Riskscore 3.0 to VantageScore 4.0 from the first quarter of 2026, so band volumes and score percentiles before and after that quarter are not comparable; the charts mark the change. Volumes include refinances.
The price and supply of credit
30-year fixed mortgage rate: 7.40% in October 2026, +1.23 pts on a year, the 85th percentile of 406 readings since 1993. The 30-year fixed rate is Freddie Mac’s weekly survey, shown at each month’s last reading.
How the numbers are built
Borrower side. The New York Fed's Quarterly Report on Household Debt and Credit reads a random five percent sample of Equifax credit files, so it covers every lender: banks, credit unions, finance companies and the federal government. Its delinquency shares are of balances, not of borrowers. Each reading covers a quarter and is plotted on the quarter's last day; the New York Fed publishes it weeks later (the Q2 2026 report on August 11, 2026), so the charts show when each period ended, not when it became known.
Lender side. Where the Federal Reserve publishes a bank series for this loan type, it is charted beside the credit files and never combined with them. Banks count loans 30+ days late on their own books, seasonally adjusted.
The answer card counts delinquency measures above their 2019 average and near their record by the rule printed under it, the same rule as the Delinquency Tracker. No forward study ships with this page: the series are quarterly and hold three recessions, too few to test one.
Download the series: mortgages.json.
| Measure | Latest | 2019 | vs 2019 |
|---|---|---|---|
| Balance 90+ days late | 0.99% | 0.98% | Up to 10% above |
| Newly 90+ days late, annualized | 1.52% | 1.06% | Over 10% above |
| Newly 30+ days late, annualized | 3.95% | 3.46% | Over 10% above |
| Bank single-family mortgages 30+ days late | 1.86% | 2.52% | At or below |
| Bank single-family mortgages charged off, annualized | 0.00% | -0.01% | No positive base |
Counted: latest more than 10% over a positive 2019 average. Up to 10% above and at or below 2019 are shown, not counted; a series whose 2019 average is zero or negative has no base. Near record: at least 90% of the series high. Counted, never scored. Descriptions, not forecasts.
How Mortgage Delinquency Works
- 1Start with the answer cardEach row is one measure at its latest quarter, against its 2019 average and its own record, with the printed rule that marks it above 2019 or near its record. The rows are counted, never blended into a score.
- 2Read the borrower sideThe New York Fed reads a five percent sample of Equifax credit files every quarter, so it sees every lender. Its headline is the share of balances 90 or more days late, and its flow measures show how much newly fell behind during the year.
- 3Compare the lender sideWhere the Federal Reserve publishes bank figures, they sit beside the credit-file series rather than inside them: banks count loans 30 or more days late on their own books and lend mostly to prime borrowers.