Why ARMs Disappeared and When They Still Make Sense
Lender pricing puts the fixed-to-ARM discount near 37 basis points, roughly one third of its level when Freddie Mac last surveyed ARMs in November 2022. With a 7/6 ARM priced 3 basis points from its fully indexed rate, the product now works only for borrowers with a defined exit.
Executive Summary
Adjustable-rate mortgages did not fall out of favor because borrowers stopped understanding them. They fell out of favor because the pricing advantage that justified their reset risk collapsed.
- Fixed-to-ARM discount: roughly 37 basis points in June 9, 2026 lender pricing, versus 102 basis points in the final Freddie Mac ARM survey reading on November 10, 2022.
- ARM share of mortgage applications: 8.5% for the week ending May 29, 2026, and in single digits across every recent MBA reading.
- A typical 7/6 SOFR ARM's initial rate (6.31%) sits 3 basis points below its fully indexed rate (6.34%). There is effectively no teaser discount in current ARM pricing.
- ARM savings on a $320,000 loan: $78 per month, $6,539 over the 84-month fixed period.
Why ARMs Disappeared
The fixed-rate backdrop is the starting point. The 30-year fixed averaged 6.48% in the Freddie Mac survey week ending June 4, 2026, down from 6.53% the prior week. Over the trailing 12 months the survey has moved between a 6.72% high in July 2025 and a 5.98% low in the week of February 26, 2026, then climbed back through the spring. Fixed rates have been rangebound; what changed for ARMs is not the level of fixed rates but the spread beneath them.
The clearest institutional signal came earlier: Freddie Mac discontinued its 5/1 ARM survey series in November 2022. The final observation, on November 10, 2022, recorded the 5/1 ARM at 6.06% against a 30-year fixed at 7.08%, a 102 basis point discount. When the benchmark national rate survey stops tracking a product, the product has stopped mattering at survey scale. For the structural differences between the two products, see fixed-rate versus adjustable-rate mortgages.
Current lender pricing on June 9, 2026 shows a 30-year fixed at 6.68% and a 7/6 SOFR ARM at 6.31%, a discount of roughly 37 basis points. That continues the narrowing flagged in our June 4 rate update, which tracked the 7/6 ARM discount compressing from roughly 50 basis points to roughly 40 week over week.
Application data confirms it. In Mortgage Bankers Association weekly survey readings, ARMs accounted for 8.1% of applications in the week ending December 19, 2025, 8.2% on February 20, 2026, 8.9% on March 6, 9.4% on May 22, and 8.5% on May 29, a week in which the ARM applications index fell 12%. These are point-in-time press-release readings rather than a continuous series, but the picture is consistent: ARM share has run in single digits across every recent observation.
The mechanism is the yield curve. ARM initial pricing keys off short-term funding costs through the SOFR index; fixed pricing keys off long-term yields. With the 10-year Treasury at 4.56% on June 8, 2026, and the ARM index math (a 3.59% 30-day average SOFR plus a standard 2.75% margin) landing near 6.3%, lenders have no room to manufacture a deep initial discount. A flat curve produces a thin spread, and a thin spread removes the only structural reason to accept reset risk.
The ARM Math Today
The standard conforming product illustrates the math. Under the Fannie Mae Selling Guide, a 7/6 SOFR ARM is indexed to the 30-day Average SOFR published by the Federal Reserve Bank of New York, which stood at 3.59% on June 9, 2026. The standard margin is 2.75% (Freddie Mac permits 100 to 300 basis points). Caps follow a 5/1/5 structure: a 5% limit at the first adjustment, 1% per subsequent six-month adjustment, and 5% over the life of the loan. The first reset arrives at month 85, after 84 fixed payments.
On a $320,000 loan, the spread is worth $78 per month: $1,983 against $2,061 for the 30-year fixed. That compounds to $934 per year and $6,539 across the 84-month fixed period, during which the ARM borrower also pays $8,325 less interest and retires more principal. Borrowers can test other loan amounts with the monthly payment calculator; the proportions scale with loan size.
| Measure | 30-Year Fixed (6.68%) | 7/6 SOFR ARM (6.31%) |
|---|---|---|
| Monthly principal and interest | $2,061 | $1,983 |
| Interest paid, first 84 months | $143,285 | $134,960 |
| Principal paid, first 84 months | $29,809 | $31,595 |
The reset scenarios frame the risk. After 84 payments at 6.31%, the ARM balance stands at $288,405. If the loan reset at today's fully indexed rate of 6.34%, the payment would move from $1,983 to $1,988, an essentially unchanged obligation. If short-term rates instead rose enough to trigger the full 5% initial cap, the rate would reach 11.31% and the payment $2,939, against the fixed borrower's unchanged $2,061. Translated to the household budget: at the capped rate the ARM payment runs $878 per month above the fixed payment, enough to erase the full $6,539 of seven-year savings in under eight months. The cap structure defines the envelope; the question is whether $78 per month compensates for carrying it.
Because the discount drives everything, its sensitivity is worth tabulating. Holding the fixed leg at 6.68%, each step in the spread changes what the ARM is worth over its fixed period.
| Discount | ARM rate | ARM payment | Saves per month | Saves over 84 months |
|---|---|---|---|---|
| 0.25% | 6.43% | $2,008 | $53 | $4,430 |
| 0.37% (current) | 6.31% | $1,983 | $78 | $6,539 |
| 0.50% | 6.18% | $1,956 | $105 | $8,811 |
| 0.75% | 5.93% | $1,904 | $156 | $13,143 |
| 1.00% | 5.68% | $1,853 | $207 | $17,423 |
The Discount Is the Decision Variable
The historical comparison makes the collapse concrete. On November 10, 2022, on a PMMS survey basis for both legs, the 30-year fixed stood at 7.08% ($2,146 per month on $320,000) and the 5/1 ARM at 6.06% ($1,931), a 102 basis point discount worth $215 per month. In June 2026, on a lender-pricing basis for both legs, the discount is 37 basis points worth $78. The discount is roughly one third of its level when Freddie Mac last surveyed the product.
This reframes the borrower question. The decision input is not the level of rates, and not the ARM rate in isolation; it is the spread between the two products at the moment of the lock. The sensitivity table above is, in effect, the whole decision: at 25 basis points the ARM saves $53 per month, at 100 basis points it saves $207. The product did not change between those rows. Only the compensation for reset risk changed.
Scenario framing applies in both directions. If short-term rates fall relative to long-term yields, ARM initial pricing could reopen a meaningful discount without any move in fixed rates. If the curve stays flat, ARM pricing stays pinned near its fully indexed level and the discount stays thin. Neither path is knowable from current data; the spread itself is the signal to watch.
What This Means for Borrowers
At a 37 basis point discount, the ARM is not broadly compelling. It remains rational for specific profiles, all of which share one property: the reset either never arrives or was already part of the plan.
| Borrower profile | ARM case at 37 bps | Why |
|---|---|---|
| Defined exit within the fixed period (military PCS, fixed-term relocation, planned sale) | Holds | Captures $78 per month and $6,539 over 84 months; never faces the reset |
| Jumbo borrowers | Sometimes holds | Jumbo fixed pricing (6.84%) runs above conforming, and bank portfolio ARMs sometimes price more aggressively in that segment |
| Would refinance on any meaningful rate drop anyway | Conditionally holds | The reset risk overlaps an existing refinance intention rather than adding a new one |
| Long hold, no defined exit | Does not hold | Accepts 5/1/5 reset exposure in exchange for $78 per month |
Two refinements sharpen the framework. First, borrowers in the third profile should understand conversion features before relying on a future refinance; convertible ARMs formalize that exit at a known cost. Second, the threshold matters more than the profile: the rule is visible in the sensitivity table, where compensation rises from $53 per month at a 25 basis point discount to $207 at 100 basis points.
Outlook
For ARMs to make sense broadly again, the discount would need to widen past roughly 75 to 100 basis points, the range where the sensitivity table shows $156 to $207 per month on a $320,000 loan. That requires the yield curve to steepen: short-term rates falling relative to long-term yields. The federal funds target range has sat at 3.50% to 3.75% since the December 11, 2025 cut, while the 10-year Treasury stood at 4.56% on June 8, 2026. If the Fed resumes cutting and long yields hold, SOFR-linked ARM pricing could reopen a discount. If long yields fall alongside short rates, fixed pricing improves in parallel and the spread stays compressed.
If the flat curve persists, expect ARM share to remain in single digits and the product to remain a planning tool for defined-horizon borrowers rather than a rate play. That is not a market failure; it is pricing doing its job. A thin discount is the market's statement that reset risk is not currently worth paying borrowers to take.
Looking Ahead
Three catalysts will test this analysis. The FOMC meets June 16-17, 2026, the first read on the short-rate path that drives the SOFR index. Weekly Freddie Mac PMMS releases track the fixed leg of the spread. And MBA weekly survey readings show whether ARM share holds its single-digit range or breaks out of it. Borrowers mid-process should treat the spread as a moving input; rate locks fix the chosen product's rate but do not freeze the fixed-to-ARM comparison while shopping.
Whether a 37 basis point discount justifies a 7/6 ARM depends on your loan amount and holding period; run both sides of the comparison before deciding.
Compare ARM vs Fixed