Adjustable-rate mortgages have a reputation problem. The 2008 financial crisis was driven in part by ARMs that reset into payment shock — and the cultural memory has stuck. The reality in 2026 is more nuanced: ARMs can be the right choice for the right borrower in the right rate environment, and the wrong choice for everyone else.
How ARMs actually work
An ARM has two phases. The fixed period — typically 5, 7, or 10 years — has a locked rate, usually below the comparable 30-year fixed rate. After the fixed period ends, the rate adjusts annually based on a market index (often SOFR or the constant-maturity Treasury) plus a margin set by the lender.
A 5/1 ARM means: 5 years fixed, then adjusts every 1 year. A 7/6 ARM means: 7 years fixed, then adjusts every 6 months. A 10/1 ARM: 10 years fixed, then adjusts annually. The first number is the comfortable part; the second is the variable.
Caps
Modern ARMs have rate caps in three places:
- Initial cap. Maximum rate change at the first adjustment. Typically 2% on a 5/1, 5% on a 7/1 or 10/1.
- Periodic cap. Maximum change at each subsequent adjustment. Typically 2%.
- Lifetime cap. Maximum total increase over the life of the loan. Typically 5% above the start rate.
So a 5/1 ARM at 5.5% with 2/2/5 caps could go to 7.5% at year 6, to 9.5% at year 7, and never exceed 10.5% over the life of the loan. The caps are real — they're written into federal regulation.
The case for ARMs
Lower starting rate
In a normal yield curve, ARM rates run 50–125 basis points below 30-year fixed. On a $340,000 loan, that's $115–$285/month in savings during the fixed period. Over a 5-year hold that's $7,000–$17,000 — real money that compounds if invested.
Time-horizon match
Most U.S. homeowners don't keep their original mortgage for 30 years. The median tenure for a homeowner is roughly 8 years (longer in expensive metros, shorter in younger demographics). For a borrower whose realistic horizon is 5–7 years before they move, refinance, or sell, paying for 30-year rate certainty is paying for insurance against an event that won't happen.
Rate-environment optionality
If rates fall over the fixed period, an ARM borrower can refinance into a new fixed loan at the lower rate, paying only the closing costs. If rates rise, the ARM borrower at least had the lower starting rate during the fixed period, which front-loaded the equity build.
The case for fixed
Predictability
The 30-year fixed exists because Americans value knowing what their housing payment will be in 2046. This is a real form of value, even if it doesn't show up in a payment-only comparison. A household that needs predictability for budgeting and stress-management reasons should pay for it.
Worst-case modeling
The ARM math depends on rate environment after the fixed period — and that's unknowable. A borrower who can comfortably handle the worst-case post-reset payment can take the ARM bet rationally. A borrower who can't is gambling, even if the most likely outcome is fine.
Long-hold cost
For borrowers planning to hold 15+ years without refinancing, the fixed rate compounds favorably. The ARM's lower starting rate matters less when it's followed by 20+ years of unknown adjustments.
The decision framework
Choose an ARM if
- Your realistic horizon in the home is ≤ the fixed period (e.g., 5/1 ARM if you'll be there ≤ 5 years)
- Your budget can absorb the lifetime-cap payment without distress (worst-case test)
- The rate spread vs. fixed is meaningful (50+ basis points)
- You have stable, growing income that can absorb a higher payment if rates rise
Choose fixed if
- You plan to hold 10+ years without refinancing
- Your income is stable but not growing, or is at risk of decline
- Predictability is a high-value good for your household
- The ARM-fixed spread is narrow (under 50 basis points) — the upside isn't worth the variability
The worst-case test
The single most important test before taking an ARM: can your household afford the lifetime-cap payment? If your starting rate is 5.5% and your lifetime cap is +5%, model the payment at 10.5%. If your budget can absorb that — even barely — the ARM is rational. If your budget can't absorb it, you're betting against the cap, and the cap exists because it gets hit.
The mistake most ARM borrowers make
The mistake is not "taking an ARM." The mistake is taking an ARM without modeling the reset. The 2007 cohort of ARM borrowers were largely qualified on the teaser rate, with no analysis of what would happen at year 3 or year 5. When the reset came and home prices had stopped rising, the borrowers couldn't refinance and couldn't afford the new payment. The structure didn't fail; the qualification process did.
A borrower who takes a 5/1 ARM after running the math at lifetime cap, with a documented refinance or sale plan inside the fixed period, and stable income to absorb a worst-case reset, is making a rational financial decision. That borrower wasn't the problem in 2008 and isn't a problem now.
The rate environment matters
ARMs are most attractive when:
- The yield curve is steeply normal (short rates well below long rates) — large ARM-fixed spread
- Rates are at recent highs and likely to fall in the medium term — refinance optionality is real
- The borrower's horizon is short relative to the fixed period
ARMs are least attractive when:
- The yield curve is flat or inverted — ARM rates are not meaningfully lower than fixed
- Rates are at recent lows and could only rise — locking the low fixed makes more sense
- The borrower's horizon is long
The 2026 environment has elements of both — rates are off their 2023–2024 highs but not at 2020–2021 lows. The ARM-fixed spread varies meaningfully by lender and product. Run both quotes; don't assume the conventional wisdom is right for your specific deal.