The Fed held for a fifth consecutive meeting and the long end sold off anyway. The vote was 9-3, with Hammack, Kashkari and Logan dissenting in favor of a quarter-point hike against pricing that had roughly one-in-three odds on it. Bonds initially rallied on the no-hike, then reversed through the press conference: Kevin Warsh said the bond market was already "doing the Fed's job" and that he was looking past PCE to a broader set of inflation data, and traders sold the long end on it. The 10-year closed at 4.693%, up 8.6 basis points, and UMBS 5.5 gave back 9 ticks. Selling stopped when the press conference did. Bankrate's national 30-year prints 6.76% this morning, a basis point above yesterday's 6.75% and 3 above last week.
Core PCE lands tomorrow, and Warsh just told the market he is weighing more than that series — which cuts both ways, because a soft print now buys less relief than it would have a month ago. Jobless claims follow on August 6 at a level that has been creeping (197,000 on the July 25 week against 188,000 prior), then the jobs report on August 7. There is no Fed event until September 15-16, and that meeting does carry a Summary of Economic Projections — the first dot plot since June, and the first one that has to reconcile a three-member hawkish bloc. Between now and then the data is the only thing moving pricing.
Today's 6.76% sits 33 basis points above the 30-day low of 6.43% and 6 below the 30-day high of 6.82% set on Tuesday. The 90-day band runs 6.36% to 6.82% with a 6.56% average, so we are at the expensive end of the quarter, not the cheap end — up 3 basis points on the week and 9 on the month. There is no downtrend here to point a borrower at, and the MBA's application data shows borrowers already know it: applications off 6.4% on the week, refinances off 10%.
The segment worth a fresh look today is the one at the conforming ceiling. Jumbo is printing 6.90% against 6.76% conventional — a 14 basis point spread, unusually thin — which quietly changes the arithmetic on the split-loan reflex most files default to. A borrower you were about to structure as a conforming first plus a second may price better as a single jumbo once you account for the second's rate and the added closing costs, and that comparison takes five minutes to run. Do this today: on your next file above the conforming limit, price the single jumbo and the split structure side by side before you recommend either.