AboutInsightsToolsSign in
All insightsMortgage finance
Oct 8, 202622 min

When low
mortgage rates
expire.

Who faces an ARM payment shock in 2027?

Most borrowers are sheltered from rate resets. For some, that protection has an expiry date.

Begin the story
An ivory paper house sits on a folded sheet beside a blank date sheet: a conceptual illustration of an expiring mortgage rate.
Illustrative loan / $400,000 / 23 years left
Before / 2.75%$1,957/mo
At 6.5%$2,796/mo
Principal and interest only. Not a reported borrower.

The house is the same. The bill is not.

Nothing has happened to the house.

The kitchen is where it was yesterday. The bedroom still catches the same morning light. No one has bought a larger place, borrowed for an extension or missed a payment.

Then the mortgage changes.

Consider an illustrative loan, not a family we have interviewed: $400,000 still owed, 23 years left to repay it, an interest rate of 2.75%. Its monthly principal-and-interest payment is about $1,957. Recalculate that same balance over that same remaining term at 6.5%, and the payment becomes about $2,796.

An extra $839 every month. Roughly $10,069 over a year if that payment difference persists.

The house has not become more useful. The borrower has not received another dollar. A period of protection has simply ended.

That is the unsettling feature of an adjustable-rate mortgage. For years, it can behave like the fixed-rate loan next door. Both households send the same predictable payment. Both can stop following every announcement from the Federal Reserve. The difference lives in a date on a document.

Eventually, one household keeps its rate. The other discovers what it costs to borrow now.

This is not a prediction that American homeowners as a whole are about to face such an increase. They are not. It is a closer examination of the people for whom the insulation is temporary, and of a question that a national mortgage-rate chart cannot answer: how much of a household's life was built around a payment that would not last?

A low rate can mean two different things

A fixed-rate mortgage promises a rate for the life of that loan. A hybrid adjustable-rate mortgage promises a rate for an opening period, then a method for setting the rate afterward.

That distinction is easy to describe and easy to underestimate.

With a 5/6 ARM, for example, the introductory rate is fixed for five years; the rate can then adjust every six months. The second number is not another five-year grace period. Nor does reaching the end of the fixed period mean the balance becomes due all at once. The debt continues. Its price can change.

The eventual rate is governed by the contract, not by whatever number appears in a headline about new 30-year mortgages. An ARM generally combines a specified market index with a margin, subject to its own limits and other terms. The Consumer Financial Protection Bureau explains that distinction between index and margin.

It is possible to make an entirely deliberate choice here. Someone expecting to sell before the first adjustment might value the initial saving. Someone with abundant cash might accept uncertainty that another household cannot. Whether the initial bargain was sensible depends partly on what the alternatives cost at the time.

But an intention to leave is not a guarantee of leaving. A planned move can be postponed. A job can change. A refinance can turn out to be more expensive than expected.

The important question is therefore not whether the borrower knew the word “adjustable.” It is whether the household has a workable plan for the day the adjective becomes a bill.

Two contracts / a schematic, not a comparison of offers

A fixed rate

Protection for the life of this loan.

Rate stays fixed

A hybrid ARM

Protection for an opening period.

Introductory periodThen: index + margin, within limits
A 5/6 ARM has five introductory years, then adjustments every six months. The balance does not become due simply because the introductory rate ends.

The national picture is smaller than the fear

ICE's October 2026 Mortgage Monitor identifies about 3.1 million active first-lien ARMs, or 5.6% of the first-lien mortgage stock in its estimates. About 1.05 million have already passed their initial fixed period; 2.05 million have not. Its calendar-year projection places 186,000 first adjustments in 2027. These are loan counts, not unique households. ICE, October 2026 report and accompanying release, October 5 release.

Three populations are hiding inside those numbers.

First are the loans whose rates have never moved. Their borrowers may be approaching the first adjustment, or may still have years of protection left.

Second are loans already in the adjustable phase. They can change again. A first-reset calendar leaves them out, even though their next adjustment may arrive sooner than a new entrant's first one.

Third are the much larger number of loans outside this ARM population. A rise in market interest rates does not rewrite a fixed-rate first mortgage's contractual rate. Other housing costs can still rise, and a new purchase or refinance is another matter.

Keeping those groups separate does more than improve a chart. It prevents two mistakes in the way we talk about housing.

One mistake turns every interest-rate increase into an imminent payment increase for every homeowner. The other notices that most owners are protected and concludes that the exceptions cannot matter.

A small share of a national balance sheet can contain a large share of a particular household's available cash. The useful unit of analysis changes when we cross the front door.

The year on the mortgage matters

“An ARM is resetting” tells us that a contract has reached a stage. It does not tell us whether the payment will rise, how far it will rise or whether its borrower will struggle.

A loan taken out at a very low introductory rate has a different journey ahead of it from one taken out when rates were already high. Its outstanding balance and remaining term matter too.

The reported 2027 cohorts make that distinction visible:

ICE / first adjustments projected for calendar 2027

The year on the loan
changes the journey ahead.

Loan counts and report-modeled payment changes are different measures. Neither is a foreclosure forecast.

186,000

Loans projected for their first reset in calendar 2027.Not all periodic adjustments. Not unique households.

20207-year fixed period
74,000 loans
+$1,066/mo+36% / P&I
20225-year fixed period
65,000 loans
+$430/mo+22% / P&I
201710-year fixed period
20,000 loans
+$498/mo+25% / P&I
20261-year fixed period
14,000 loans
+$233/mo+7% / P&I
20243-year fixed period
13,000 loans
−$10/mo−1% / P&I
5.6%ARMs in ICE’s active first-lien stock

About 3.1 million ARMs94.4% outside this ARM population

ICE Mortgage Monitor, October 2026, page 16, released October 5. The five cohorts total 186,000 loans. Report-date assumptions, not observed future bills. First resets are not all adjustments; an adjustment need not raise the payment. Counts are loans, not unique households.
Source data CSV
Exact cohort data
ICE’s 2027 first-reset cohorts. Modeled P&I, USD per month.
VintageFixed yearsLoansCurrent rateFirst rateMonthly change
2020774,0002.75%5.79%+$1,066
2022565,0004.5%6.45%+$430
20171020,0003.5%6.02%+$498
2026114,0005.62%6.28%+$233
2024313,0007.5%7.42%−$10

ICE model, not observed future bills. Principal and interest only. The report uses cohort characteristics and report-date index assumptions; individual loans will differ. Mortgage Monitor, October 2026, page 16.

The last row is worth lingering over. An adjustment is not synonymous with an increase. Starting on the other side of the rate cycle can reverse the direction.

The table also resists a second shortcut. The change for the largest low-rate cohort is not the change for everyone. And a modeled cohort figure is not a promise about the notice any one borrower will receive.

What the year captures is exposure to a particular bargain. Borrowers arrived at different moments, accepted different lengths of protection and now carry different amounts of debt. They are meeting today's financing conditions from different starting lines.

That is why a single “average ARM borrower” is such a poor guide to the experience ahead.

Inside the calculation

Return to our illustrative $400,000 loan.

To understand its next payment, we need more than an interest-rate forecast. We need the balance still outstanding, the number of payments left, the index specified in the note, the margin added to it, and the constraints that govern the adjustment. An interest-only feature would add another complication. Our example is a straightforward fully amortizing loan.

Suppose, purely for this calculation, that the relevant index is 4% and the contractual margin is 2.5 percentage points. Together they produce a fully indexed rate of 6.5%, before applying any cap or other contractual restriction. That 4% is an input to the illustration, not a claim about today's index fixing.

If 6.5% applies immediately, the payment becomes $2,796. Yet the interest rate has risen by far more, proportionally, than the payment has. The rate is about 2.36 times its starting level; the payment is about 43% higher.

Why the difference? A principal-and-interest payment does two jobs. It pays interest on the outstanding balance, and it repays that balance over the remaining term. Raising the rate changes the financing cost. It does not multiply the principal still owed by the same proportion.

This is a small point with large consequences. Multiplying today's payment by the ratio of the new rate to the old one produces the wrong answer. So does treating the whole payment as interest.

Time matters as well. A borrower with only a few years left has relatively little time for interest to accumulate, but must retire the remaining principal quickly. A borrower with decades left can spread repayment further, at the cost of paying interest for longer.

The right calculation is tied to the actual remaining loan, not to an imagined new 30-year mortgage.

One invented contract / not a market fixing

The headline rate
is not the whole calculation.

  1. 01Index

    The note’s market reference

    4%
  2. 02Margin

    The contract’s addition

    + 2.5 points
  3. 03Fully indexed rate

    Before contractual limits

    6.5%
  4. 04Payment

    Applied rate, balance, time

    Not the rate alone
The 4% index is illustrative. Caps, floors, rounding rules and adjustment dates intervene before a new payment is calculated. No interest-only feature is modeled here.

A cap can spread the shock across a year

An adjustment cap is often discussed as though it were a shield. Sometimes its more useful description is a schedule.

Take the same illustrative loan and give it a two-percentage-point first-adjustment cap, a one-point cap at subsequent adjustments and a five-point lifetime cap. Let it adjust every six months. Hold its fully indexed rate at 6.5% throughout. This is a controlled comparison of contract mechanics, not a claim that these are the terms of the reported 2020 cohort.

At the first reset, the rate cannot move directly from 2.75% to 6.5%. It reaches 4.75%. The payment becomes about $2,385.

Six months later, the allowed rate reaches 5.75%. After accounting for the principal paid down during those months, the new payment is about $2,612.

At the following adjustment, a year after the first, the rate reaches 6.5%. The borrower now owes less than at the start, and has 22 years left. The payment becomes about $2,786.

Interactive / an isolated cap sequence

The market rate holds steady.
The bill keeps climbing.

Hold the fully indexed rate at 6.5%. Advance the loan by 6 months.

Applicable loan rate (%)0%2%4%6%8%Fully indexed: 6.5%2.75%Before reset4.75%First reset5.75%+6 months6.5%+12 months
Before reset$1,957/mo2.75% applied
First reset$2,385/mo4.75% applied
+6 months$2,612/mo5.75% applied
+12 months$2,786/mo6.5% applied
First reset$2,385/mo

4.75% applicable rate · $400,000 remaining · 276 payments left

Illustrative, fully amortizing loan. The blue steps show the applied rate; the dashed reference is the unchanged index plus margin. Later payments account for principal repaid and the shorter remaining term. No further increase in the reference rate is needed for cap catch-up. Principal and interest only, excluding taxes, insurance, fees, prepayments and interest-only features.
Change the contract assumptions

Symmetric first and later rate caps. Lifetime upper bound is the introductory rate plus the lifetime cap. This study limits the floor to the introductory rate so those bounds do not conflict. Rounding uses the nearest increment, with ties upward; 0 means none. The surrounding worked example uses the default settings; changing this study does not rewrite it. Actual contract priority rules can differ. Check the note.

Return to the result
Export CSV
Balance and payment table
Illustrative schedule, current assumptions. Unrounded values used in calculations.
StageRateBalanceMonths leftMonthly P&I
Before reset2.75%$400,000.00276$1,957.21
First reset4.75%$400,000.00276$2,384.90
+6 months5.75%$395,142.75270$2,611.90
+12 months6.5%$390,779.75264$2,786.02

Oftu illustration. All scheduled payments are made; no fees, escrow changes, refinancing or additional principal payments. The index assumption does not change during the sequence.

No further rate increase was needed for those later steps. The contract was still catching up to a rate it had previously been prevented from reaching.

That changes the question a borrower should ask after the first notice. “Can we manage this payment?” is necessary, but incomplete. “What could the next scheduled adjustment do if the index stays exactly where it is?” can uncover a second problem while there is still time to prepare.

Caps are not interchangeable. The CFPB distinguishes first-adjustment, subsequent-adjustment and lifetime caps. Fannie Mae's July 2020 SOFR product specifications, for example, listed a two-point first cap for its three- and five-year products but a five-point first cap for its seven- and ten-year products. Those are specified Fannie Mae plans, not every ARM in America. Fannie Mae, July 31, 2020.

A longer opening period is more time before the first adjustment. It need not be a gentler first adjustment when that day arrives.

Falling rates might not restore the old payment

There is a hopeful sentence often attached to an expiring low rate: perhaps rates will come down first.

They might. But “down” needs a starting point.

In our illustration, lowering the fully indexed rate from 6.5% to 5.5%, with the original $400,000 balance and 23-year remaining term held unchanged, reduces the recalculated payment to about $2,557. That is welcome relief compared with $2,796. It is still about $600 above the payment at 2.75%.

A falling index and a rising household payment can therefore coexist. The market can improve relative to last month while remaining expensive relative to the rate promised years ago. Whether an actual payment moves, and when, also depends on the note's adjustment date and restrictions.

The Federal Reserve raised its target range by a quarter of a percentage point on September 16, 2026, to 3.75% to 4%. That is a documented decision, not a commitment about the rest of the year. Nor is the federal funds target the rate on every ARM. Federal Reserve, September 16, 2026.

A borrower needs the correct index and the correct observation date. A policy announcement is useful context, but it cannot substitute for either.

Loans that have already reset face a different question: the effect of another index movement, rather than the loss of an entire introductory-rate period. ICE models a median monthly increase of about $14 among affected post-reset borrowers if a quarter-point change passes fully through their indexes; the figure is about $53 for more recent loans. Neither is an immediate, guaranteed Fed-to-bill effect. ICE, October 5, 2026.

That is a different experiment from moving our example all the way from 2.75% to 6.5%. Combining the two under one average would conceal the reason some payments barely move while others move sharply.

This is one reason the story should not be told as a bet on the next Fed meeting. Some of the distance between an old introductory rate and a future bill already exists. The calendar, rather than a fresh policy surprise, can bring it home.

Same $400,000 / same 23 remaining years / P&I only

The market can come down.
The old bill need not come back.

Old introductory rate$1,957/mo

2.75% applicable rate

Illustrative reset$2,796/mo

6.5% applicable rate

A lower reset rate$2,557/mo

5.5% applicable rate

Payment bars share a $0 to $3,000 scale. At 5.5%, the bill is about $600 above its 2.75% starting point. A controlled comparison, not a rate forecast.

The missing number is the money left over

An extra $839 sounds like a complete description of a shock. It is only the beginning.

Imagine two households with our illustrative loan. Neither is a reported borrower, and neither profile represents a national average. Before the reset, one has $300 left in its monthly budget after all outgoings, including the old mortgage payment. The other has $1,500.

Apply the immediate increase from $1,957 to $2,796, keeping everything else unchanged.

The first household now runs a monthly deficit of about $539. The second still has about $661 left. The same debt, the same rate movement, the same payment increase: one has lost its surplus and must find cash; the other has lost room but remains in surplus.

This is not a model of how people will respond. Spending might be reduced. Earnings might increase or fall. Someone might have substantial savings, or almost none. The comparison isolates a point that a payment percentage cannot convey.

The consequential denominator is often the money that was left over.

A household can accommodate a substantial rise by saving less. That is still a change in its economic life, but it is different from falling behind. Another household can exhaust its whole surplus under a much smaller increase.

Savings add time, not immunity. Put $10,000 of liquid reserves beside the first illustrative household's $539 monthly deficit. In a world with no other changes, those savings cover about 18.6 months. A car repair, medical expense or income interruption would shorten that runway. An effective budget adjustment would lengthen it.

The number is not a countdown to default. People do not wait motionless for a savings account to reach zero. It shows why the date of the reset is only one date worth watching. The other is the point at which a household's coping plan has to change.

Illustrative households / same mortgage, different room

The missing number is
the money left over.

An extra $839 a month means something different in two budgets.

Monthly cash / USDHousehold AHousehold B
Left before reset$300$1,500
Additional mortgage payment−$839−$839
Left after reset−$539+$661

Household A

Before$300

Extra payment−$839

After−$539

Household B

Before$1,500

Extra payment−$839

After+$661

Same invented $400,000 loan. Immediate change from 2.75% to 6.5%, with 23 years left. All other cash flows unchanged. This is separate from the cap-sequence study above.
18.6 months

Of this isolated deficit covered by the reserve. Not a default or depletion prediction.

Export CSV

Equity cannot make the payment by itself

A homeowner can be wealthy on paper and short of money on Friday.

That is not a contradiction. Equity is the difference between an asset's value and the debt secured on it. A mortgage payment requires cash. Turning one into the other usually involves another transaction: borrowing, refinancing or selling.

Each route asks its own questions. What terms are available? Does the borrower qualify today? How much would closing cost? Is there enough equity after those costs? If selling, how long might it take, and where would the household live next?

Suppose the house in our example has appreciated substantially. That could create choices that an underwater owner lacks. It does not mean the higher monthly payment can be ignored. The owner still has to access the value, and an exit that works on a balance sheet may be awkward on a calendar.

Conversely, a borrower can have little equity and continue paying reliably if the monthly obligation remains manageable. A weak asset position and a weak cash position are different vulnerabilities. When they occur together, the routes out can narrow.

Research from the Philadelphia Fed found that negative equity and measures of illiquidity both helped explain default in its historical sample, with an interaction between them. It did not establish that a payment increase alone determines a household's fate. Elul and coauthors, April 2010.

For an ARM borrower, that distinction moves the conversation away from one dramatic rate and toward a more revealing set of circumstances: the coming payment, reliable income, accessible reserves and realistic alternatives.

Value and cash / schematic paths, not guaranteed options

A house can hold wealth.
A payment needs money.

Equity

Market value less debt

Borrow Terms + eligibility

Refinance Costs + qualification

Sell Time + net proceeds

Cash

Available by the due date

Every bridge is conditional. More equity can create choices; it does not put money in the payment account by itself.

Refinancing can change the deadline, not just the rate

The neatest proposed solution is usually two words: refinance it.

As an instruction, it omits most of the work. There must be a loan worth taking, a borrower eligible to take it and enough benefit to justify the transaction. Today's credit conditions and the household's current circumstances matter, not just its situation at the original closing.

There is also a subtler possibility. A new mortgage can lower the payment without lowering the rate.

Keep our illustrative $400,000 balance and 6.5% rate. Over 23 remaining years, principal and interest are about $2,796 a month. Spread the same debt over a new 30-year term, and the monthly figure is about $2,528.

The apparent relief is about $268 a month. No interest-rate saving has occurred. The borrower has purchased more time to repay the principal.

If both hypothetical loans were held to their scheduled maturity at that rate, with no fees or extra repayments, the 23-year version would incur about $371,767 in interest. The 30-year version would incur about $510,178, roughly $138,411 more. Real borrowers can sell, refinance again or repay early, so these are contractual-path comparisons, not predictions of their realized costs.

For someone trying to keep a home through a difficult period, lower monthly outgoings can be genuinely valuable. The point is not to dismiss that relief. It is to identify what paid for it.

A better rate, a longer term and a lower balance are different mechanisms. They should not all be filed under “a cheaper mortgage.”

Even a sensible bridge is easier to judge when both shores are visible.

Same principal / same rate / a different maturity

A smaller bill
can buy more time.

Payment relief can come from lengthening the term, even with no interest-rate saving.

Remaining loan23 years · $2,796/mo
New term30 years · $2,528/mo

Interest over 23 scheduled years$371,767

Interest over 30 scheduled years$510,178

Additional scheduled interest$138,411

Duration strips share a 0 to 30-year scale. Fully amortizing $400,000 at a constant 6.5%, held to scheduled maturity. Fees, escrow, early repayments and future refinancing excluded. These are contractual paths, not predictions of realized costs.

$268/mo less, with 7 more years of scheduled repayment.

Export CSV
Historical evidence / not a 2027 default forecast

The bill matters.
History is not a conversion table.

Fuster + Willen / 2012, revised 2015

50%Required payment reduction
in the studied hybrid-ARM sample

55%Approximately lower
delinquency hazard in that sample

We cannot reverse this historical result to calculate defaults from a future payment increase.

What history can tell us, and what it cannot

Adjustable mortgages carry the memory of the housing crash. It is tempting to invoke that memory whenever rates reset. It is equally tempting to wave it away by declaring that lending has changed.

Neither response tells us enough about the loans in front of us.

One useful piece of history runs in the opposite direction from today's concern. Andreas Fuster and Paul Willen studied hybrid ARMs whose payments fell substantially. Their research found that halving the required payment reduced the delinquency hazard by about 55% in the sample, including a relationship among deeply underwater borrowers. New York Fed staff report, November 2012, revised January 2015.

That is evidence that the monthly bill can matter independently of the home's equity position. It is not a license to reverse the finding and invent a default forecast for payment increases in 2027. A treatment effect in a particular historical population is not a universal conversion table.

Underwriting deserves the same care. It would be comforting to say that every modern ARM borrower was tested at the highest payment they could ever face. Current Fannie Mae rules do not support that blanket claim.

Its qualifying rules differ with the initial fixed period. Five-year ARMs generally require the higher of the fully indexed rate or the note rate plus the first-adjustment cap. For terms longer than five years, the general minimum is the note rate, with exceptions; its listed generic seven- and ten-year plans use the note rate. Fannie Mae qualifying-payment requirements, accessed October 8, 2026.

These are current rules for a particular lending channel. They do not describe every jumbo loan or prove how every older mortgage was assessed. They do show why “modern underwriting” is too broad an answer to a specific payment question.

A useful assessment would examine the actual contracts and the borrowers' circumstances, then watch payment performance as adjustments arrive. This article does not have that linked, loan-level evidence. It cannot honestly turn a reset schedule into a foreclosure schedule.

Can we count the next twelve months?

The natural question is also the hardest one: how many mortgages will adjust in the next twelve months?

We found a useful calendar-year estimate of first adjustments. We did not find a public, nationally complete count of loans scheduled for any adjustment between October 8, 2026 and October 8, 2027.

Those are different requests. An exact rolling window needs the next adjustment date on each active loan. Including later adjustments requires loans already beyond their fixed period. Estimating payment effects then needs each loan's index, margin, limits, current rate, balance, remaining term and relevant payment features. Refinancing and payoff can remove loans before their scheduled change.

A calculation that divides an annual cohort evenly across twelve months can produce an approximation. It cannot reveal the actual monthly distribution. Adding an already-adjusting population can create a broader screening pool, but without the next dates and survival assumptions it still cannot produce a verified scheduled count.

The sources also disagree about the size of the ARM stock. FHFA's National Mortgage Database reports about 51.4 million outstanding first liens in the second quarter of 2026, with published ARM shares totaling 4.3%. Applying those rounded shares suggests about 2.2 million loans, below ICE's 3.1 million estimate. The datasets differ in timing and coverage; we have not reconciled the gap. We do not average them or present one as confirmation of the other. FHFA National Mortgage Database, technical notes.

Precision has to be earned. Writing “about” before an unsupported count does not earn it.

For now, the defensible article is about a documented first-reset cohort, transparent payment mechanisms and the information still needed to understand the full population. That is less tidy than a single alarming number. It is considerably more useful.

ICE / calendar-year first-reset projections

A useful calendar.
Not the next twelve months.

148k2026
186k2027
155k2028
252k2029
293k2030
281k2031
425k2032
ICE, October 2026. Counts in thousands of loans. These calendar-year first resets do not count every subsequent adjustment or reveal the exact October 8, 2026 to October 8, 2027 schedule.
What an exact rolling count would need
  • Each active loan’s next adjustment date, not just its first one.
  • Index, margin, caps, floor, rounding, balance and remaining term.
  • Payoffs and refinancing before the scheduled date.
  • Linked household cash flow to study payment stress, not just resets.

The notice is a window, not just a warning

For a household approaching its first reset, the most important document may be neither a news story nor a chart. It may be the notice from its servicer.

The CFPB says that, generally, an estimate arrives seven to eight months before the first payment at the initially adjusted rate. For loans already adjusting, notice generally arrives two to four months before a changed payment is due. The first estimate may not be the final amount. Specific requirements and exceptions depend on the loan. CFPB, reviewed February 2, 2024.

That time is useful precisely because the available choices need time too. Budget changes, loan shopping, counseling and a potential sale do not happen at the speed of changing a rate input on a screen.

Start with a small set of concrete questions:

  1. What is the date of the first payment at the new rate, and how soon can the rate change again?
  2. Which index, margin and observation date set the rate? What rounding rule applies?
  3. What are the first, later and lifetime limits? Are there floors or payment features that change the calculation?
  4. Is the notice an estimate or a finalized change? How much of the total bill is principal and interest, rather than taxes, insurance or other amounts?
  5. If the index stays unchanged, what could the following adjustment do?
  6. What happens to the household's actual cash surplus under those payments, and which alternatives are genuinely available?

If the coming payment is unaffordable, the CFPB points borrowers toward their servicer and HUD-approved housing counseling. That is a better place to resolve the terms of an individual loan than a generalized article. This feature is educational, not personalized financial or legal advice.

Read the notice / check the individual contract

The date gives you
time to ask.

7 to 8 monthsGenerally, an estimate before the first payment at an initially adjusted rate.

2 to 4 monthsGenerally, notice before a later changed payment is due.

CFPB adjustment-notice guidance. Exceptions apply. The first estimate may not be the final payment.
Back to the house

The date
was always there.

Return to the beginning

The date was always there

Return to the house at the beginning.

It is still the same house. Its owner has not necessarily made a mistake, and a larger bill does not decide what happens next. There may be ample savings, better earnings or a refinance that makes sense. There may be a plan that needs revising before its assumptions expire.

For years, a low rate can recede into the background. It becomes part of what a household regards as normal: the amount sent each month, the space left for saving, the cost of remaining in one place.

A reset brings the contract back into view. What felt permanent was, for this loan, a period of time.

The national question is how many such periods are ending. The household question is what was built inside them.

Answering the second well requires more than counting the first.

Methods and reading notes

Reported figures versus our calculations

ICE's population estimates and modeled vintage results are attributed to its October 2026 Mortgage Monitor. A first adjustment is not every adjustment, not necessarily a payment increase and not a delinquency event. Its 2027 estimate is a calendar-year projection, not an exact count for the twelve months following publication. We have not multiplied a modeled median payment change by the cohort count to claim a measured national cash burden.

The $400,000 loan, indexed-rate assumptions, cap sequence, household budgets and term-extension comparison are our own illustrations. They are not actual customers, observed mortgage notices, lender offers or estimates of the typical household. No scenario is a validated Oftu product forecast.

Payment calculation

For a fully amortizing loan, monthly principal and interest are calculated from outstanding principal, a nominal annual rate divided by twelve and the remaining number of monthly payments. At each subsequent adjustment in the cap sequence, the model deducts principal repaid during the prior six months and reduces the remaining term. We do not keep resetting the loan to a new 30-year term.

The model excludes escrow, insurance, taxes, fees, interest-only periods, negative amortization, principal prepayments, refinancing and delinquency. Values are calculated without intermediate rounding and shown rounded to the nearest dollar. The annualized opening difference uses unrounded monthly values. The term-extension comparison assumes the same fixed rate throughout each contractual path and no early payoff.

Consequential payment outputs were cross-checked using an independent month-by-month recurrence and a bisection solver. Zero-rate and maturity-balance checks also passed.

What the illustrations do not establish

The cap staircase is an intentionally isolated example, not a representation of the average 2020 seven-year ARM. Its terms should not be substituted for any borrower's note. The reserve illustration is a cash-flow identity with other circumstances held unchanged, not a prediction of default or behavior. Historical research informs the questions we ask; it does not supply a 2027 default rate.

Source trail

Read as Markdown Explore refinancing
More insights