Sub-4% US mortgages shed only 0.3 points of share in the second quarter

A one-point decline in a year ^[2] leaves the sub-3% bucket, now 19.2% of the stock ^[1], about 19 years from empty ^[3]. That projection assumes people who sell because of a job in a new city or a divorce keep doing so at the same rate. Wolf Richter, who compiled the FHFA series at Wolf Street, wrote that this kind of selling has come to a near-halt ^[15]. He wrote that the sub-3% decline "has essentially stalled" ^[4].

The FHFA series counts every loan type, 30-year and 15-year fixed as well as adjustable-rate ^[14]. Some ARMs carried rates below 3% and sit in the same bucket, and holders of ultra-low ARMs have seen their rates reset to current levels ^[11]. A reset moves a loan out of the bucket without a house being listed, so sale-driven attrition is, if anything, slower than the one-point figure. ARMs are 4.3% of the stock, unchanged on the quarter, so the effect is small ^[10].

The locked share can shrink three ways, at very different speeds. The first is already running. The 6%-plus bucket has gained about 15 points of share since Q2 2022 ^[5] because most new loans are written there ^[6].

The second is a rate cut, and the order in which borrowers respond matters. The 34.5% of loans at 5% or more ^[4] give up the least by moving, and the 6%-plus group among them would be first to refinance ^[6]. Every loan below 4% would still mean surrendering a cheaper rate at any market rate above 4% ^[1]. A drop in mortgage rates would show up in refinancing well before it shows up in existing-home listings.

The third is attrition picking back up, and the latest move runs against it: regular 15-year mortgages priced into the 6%-plus range in recent weeks, a change that lands in the Q3 data ^[12].

I'd expect the freeze to hold through that release. Richter is blunter. "The housing market, wrecked by ultralow mortgage rates, will stay wrecked for longer," he wrote ^[13]. The counter-case sits in the same table. A third of loans already carry rates close enough to market that selling costs little, and that group grows every quarter new loans are written above 6% ^[4][6]. Sales could recover with the 49% barely moving. The view is wrong if the below-4% share falls well beyond the 0.3 points it lost in Q2 ^[2] while rates hold above 7% ^[5].

Behind the slump in existing-home sales is a decision not to list, or rather the same decision made again and again by the holders of roughly half of all mortgages outstanding ^[1]. Richter wrote that homeowners "don't want to replace that 3% mortgage with an over-7% mortgage" ^[16].