Updated August 14, 2026. Quick answer: among US homeowners aged 65 and over, 38.6% still carried debt secured by their home in 2022 — against 20.7% in 1989. Measured against all families 65+, renters included, the 2022 share is 30.2%. Both numbers are in this study, they answer different questions, and they differ by 8.4 points. The sharpest movement is in the oldest band: 75+ homeowners went 9.0% to 34.0%.
What this measures: the share of older families carrying any debt secured by their primary residence — a first mortgage, a second, a home equity loan or a HELOC. The Federal Reserve calls that quantity MRTHEL. The first-lien-only figure is given separately in every table.
As of: the 2022 Survey of Consumer Finances, with all eleven earlier waves back to 1989. The Board’s own survey page, read 14 August 2026, states that “The 2022 Survey of Consumer Finances (SCF) is the most recent survey conducted”. Computed here from the Board’s own public data files. Every share in the all-families column reproduces the Board’s own published Table 13, 24 comparisons out of 24 — see Method.
The findings
Seven columns. On a phone, swipe the table sideways to reach the balance and equity figures.
| Survey year | Homeowners 65+ with home-secured debt | With a first-lien mortgage | All families 65+ with home-secured debt | Homeownership rate, 65+ | Median balance owed, indebted owners (2022$) | Median home equity, owners (2022$) |
|---|---|---|---|---|---|---|
| 1989 | 20.7% | 16.4% | 15.4% | 74.5% | $49,968 | $271,102 |
| 1992 | 18.2% | 16.8% | 14.3% | 78.4% | $74,699 | $281,460 |
| 1995 | 21.8% | 18.4% | 16.6% | 76.5% | $65,178 | $289,931 |
| 1998 | 24.1% | 20.9% | 19.1% | 79.3% | $84,443 | $268,956 |
| 2001 | 26.4% | 22.9% | 20.9% | 79.4% | $121,628 | $280,249 |
| 2004 | 30.4% | 24.4% | 25.3% | 83.3% | $106,163 | $298,737 |
| 2007 | 34.9% | 26.7% | 28.3% | 81.2% | $122,866 | $299,792 |
| 2010 | 39.8% | 34.4% | 32.6% | 81.9% | $116,773 | $234,293 |
| 2013 | 38.5% | 34.3% | 32.0% | 83.2% | $117,944 | $203,161 |
| 2016 | 41.3% | 35.3% | 33.4% | 80.8% | $108,614 | $218,445 |
| 2019 | 41.8% | 34.7% | 33.4% | 80.1% | $115,294 | $228,170 |
| 2022 | 38.6% | 34.5% | 30.2% | 78.2% | $102,240 | $249,520 |
Shares are of the group named in the column heading. Dollar figures are medians, deflated to 2022 dollars with the CPI-U annual average; the median balance is computed among indebted owners only, and the median home equity among all 65+ owners. 1,373 unweighted 65+ families per implicate in 2022.
1. The share roughly doubled, then flattened. 20.7% of homeowners 65+ carried home-secured debt in 1989. That reached 41.8% in 2019 and stood at 38.6% in 2022. The rise is not a recent story. The share was already 34.9% in 2007 — 79% of the entire 1989-to-2022 move was in place before the financial crisis, and the last three waves have moved sideways and slightly down.
2. First-lien mortgages more than doubled. 16.4% of 65+ owners in 1989, 34.5% in 2022. So this is not mainly a home-equity-borrowing story dressed up as a mortgage story — the ordinary mortgage share moved on its own. The HELOC share alone was 5.6% in 2022.
3. The oldest band is where it moved, and the two bands have nearly converged. Among 75+ homeowners the share went 9.0% to 34.0%, a factor of 3.8. The 65–74 band peaked at 50.1% in 2007 and has fallen back to 42.3%. Two bands 19 points apart in 1989 are 8 points apart now. The path between those endpoints is not smooth — the gap is 32.1 points in 2007 and bounces around in between, on subsamples of a few hundred families per band. Read the endpoints as the finding and the middle of the column as noise.
| Survey year | 65–74 homeowners with home-secured debt | 75+ homeowners with home-secured debt | Gap, percentage points |
|---|---|---|---|
| 1989 | 28.1% | 9.0% | 19.0 |
| 1992 | 23.4% | 11.1% | 12.3 |
| 1995 | 31.1% | 9.3% | 21.8 |
| 1998 | 31.9% | 15.0% | 16.9 |
| 2001 | 38.8% | 12.4% | 26.3 |
| 2004 | 39.5% | 22.0% | 17.5 |
| 2007 | 50.1% | 18.1% | 32.1 |
| 2010 | 49.1% | 29.8% | 19.3 |
| 2013 | 49.1% | 24.8% | 24.4 |
| 2016 | 49.2% | 31.9% | 17.3 |
| 2019 | 48.0% | 33.6% | 14.4 |
| 2022 | 42.3% | 34.0% | 8.2 |
4. The balance owed doubled in real terms. Among indebted 65+ owners the median balance was $49,968 in 1989 and $102,240 in 2022, both in 2022 dollars. In the dollars of their own day those figures are $21,672 and $102,240.
The part that cuts against intuition
Median home equity for owners 65+ is lower in 2022 than it was in 1989 — $249,520 against $271,102, in the same 2022 dollars, with a trough of $203,161 in 2013. Home values rose over those thirty-three years. Debt secured against them rose faster, and the homeownership rate itself moved only from 74.5% to 78.2%.
The Board defines the equity figure the obvious way, in its own published macro: “Home equity equals home value less all home secured debt”. So the equity column and the debt columns above are two views of one balance sheet, not two independent series.
What “has a mortgage” actually means here
It means three different things, and the difference is large enough that quoting the wrong one is a real error. All three are defined in the Board’s bulletin.macro.txt, which is published alongside the data:
MRTHEL— all debt secured by the primary residence, which the macro labels “housing debt (mortgage, home equity loans and HELOCs — mopup LOCs divided between HE and other)”. This is the headline: 38.6% of 65+ owners in 2022.HPRIM_MORT— a first-lien mortgage, defined asHPRIM_MORT=(X805>0). 34.5% of 65+ owners in 2022.- The HELOC portion alone. 5.6% of 65+ owners in 2022.
The 4.1-point gap between the first two is real people with debt secured by their home and no first mortgage. They are not counted by a “retirees with a mortgage” figure and they do owe money against the house. We publish the wider measure as the headline and the narrower one beside it, in every row of every table, so either can be quoted correctly.
And the denominator is the other half of the sentence. 38.6% is of 65+ homeowners. 30.2% is of all families 65+, which includes renters, who by construction have no mortgage. Same survey, same year, 8.4 points apart. Neither is wrong; quoting one as the other is.
Method
Every figure is computed from the Federal Reserve Board’s Summary Extract Public Data files for the Survey of Consumer Finances, one file per wave, 1989 through 2022. These are the Board’s own files, downloaded directly from federalreserve.gov, not a research file or a summary.
- Implicates. The SCF ships five multiply-imputed records per family. Every statistic here is computed within each implicate and the five results averaged. The file is never pooled into one dataset of five times as many families.
- The weight is not the weight. The extract’s
WGTis the survey weight divided by five. The Board says so itself, in the same macro, which assignsWGT=X42001/5;— its comment lines above that assignment explain the division as making totals estimated on the five implicates jointly come out correct (our wording, from the macro’s comment). Shares, medians and percentiles are unaffected by that scaling. Any population count is not, and every count in this study is multiplied back by five. - 1989 names its identifiers differently. That wave uses
X1andXX1where every later wave usesY1andYY1. A wave loop that assumes the later names silently drops 1989. - Real dollars. Nominal figures are deflated with the CPI-U annual average for all items, US city average, not seasonally adjusted, published by the Bureau of Labor Statistics, on a 2022 base.
The external check, and it is the reason this study exists in this form. The Board publishes its own historical table set, whose Table 13 reports the percentage of families holding debt secured by their primary residence, by age of the reference person, for every wave. That is a direct check on the all-families column above. Rebuilding it here reproduces the Board’s published figure 24 times out of 24 — both published age bands, all twelve waves — with a worst difference of 0.0049 percentage points. Matching to four decimal places across twelve waves confirms the weighting, the implicate handling and the MRTHEL definition simultaneously.
And an independent recomputation. A second code path, written separately, re-derives thirty-two figures from the raw files by a different route. All thirty-two agree.
What this study does not establish
No standard errors, so no figure here carries a margin of error. Correct SCF standard errors need the Board’s replicate-weight files combined with imputation variance, and that is not computed here. Do not attach a plus-or-minus to any number on this page, and treat small wave-to-wave moves as noise rather than trend.
No state or metro cut, ever, from this data. The public Survey of Consumer Finances carries no geography at all. Any “retiree mortgage rate in your state” figure is not coming from this survey.
It does not say why. Buying later, moving later, refinancing, borrowing against the house, or simply carrying a longer term into retirement would all produce this pattern, and this study separates none of them. It is a prevalence measure and nothing more.
It is a survey, not a register. Balances and home values are what families reported to interviewers. The Board imputes missing answers, which is what the five implicates are for, and the imputation is why the underlying estimates move slightly between code paths.
Age is the age of the reference person of the SCF family unit, which is not identical to a Census household head. A couple appears once.
Nothing here is advice about your own mortgage. That a share went up says nothing about whether any particular household should carry a loan into retirement. That question is a different one, and it turns on your numbers, not on a national share.
The data
Everything above is built from files anyone can download without an account: the Board’s Survey of Consumer Finances Summary Extract Public Data for each wave, its published bulletin macro for the variable definitions quoted here, its historical table set for the Table 13 reconciliation, and the Bureau of Labor Statistics CPI-U series for the deflator. The URL, byte count and SHA-256 of every source file used are recorded so the figures stay reproducible, and the Board republishes these files from time to time. The complete result set is the two tables above.
Cite this study
“Among US homeowners aged 65 and over, 38.6% carried debt secured by their home in 2022, against 20.7% in 1989; among homeowners 75 and over the share went from 9.0% to 34.0% (Clear Money Guide, 2026, from Federal Reserve Survey of Consumer Finances public data, 1989–2022).”
Suggested citation: “How Many Homeowners 65+ Still Owe on Their Home,” Clear Money Guide, 2026, clearmoneyguide.com/retirement-debt-statistics/. Free to reuse with attribution under CC BY 4.0. No advertising appears on this page and we earn nothing from it.
Related research
Whether to pay off the mortgage before retiring is the decision this prevalence sets up, and the calculator runs it on your own numbers. What housing costs in retirement and what retirees actually spend come from a different federal survey. Retirement savings by age is our other study from this same survey.