The Business Lending Boom Went Negative in July. The Mortgage Flatline Never Moved.

by Alina Castellanos · August 16, 2026

Economics Correspondent, US Market Current

There is a clean split running through the Federal Reserve's weekly bank credit data, and it is not the one that gets described most often. It is not a story about credit being loose or tight in aggregate. It is a story about business borrowing and household mortgage borrowing behaving as if they inhabit different economies. But the split has a complication that matters more than the split itself: the business side, which spent two quarters running hot enough to be cited as evidence of an economy that did not need easier policy, stopped in July.

Per the Fed's H.8 release for the week ended Aug 5, published Aug 14, commercial and industrial loans at all commercial banks grew at a 12.2 percent seasonally adjusted annual rate in the first quarter of 2026 and 14.2 percent in the second. Those are large numbers by the standards of the past several years; the same series ran minus 7.5 percent in 2021, minus 0.2 percent in 2023, and 0.9 percent in 2024. Then look at the monthly path inside and after that second quarter, all on the same seasonally adjusted annualized basis: April 15.8 percent, May 10.8 percent, June 4.0 percent, July minus 1.1 percent. The quarterly figures describe a boom. The monthly figures describe a boom that decelerated every single month and then went outright negative.

The household mortgage side never participated in either direction. Residential real estate loans grew 1.8 percent annualized in the first quarter and 0.7 percent in the second, with monthly readings of 0.6 percent in April, 0.3 percent in May, 2.6 percent in June and 0.2 percent in July. That is a flatline with noise on it. The level sits at $2,691.0 billion for the week ended Aug 5. Commercial real estate, by contrast, has been steady but unspectacular, at 3.4 percent in the second quarter and 3.3 percent in July, with a level of $3,124.2 billion.

Reading the household side as uniformly weak would be a mistake, though, and this is where the data resists easy summary. Consumer loans grew 6.0 percent annualized in the second quarter and re-accelerated to 8.5 percent in July after soft May and June readings of 2.7 and 2.6 percent. Credit cards and other revolving plans grew 5.6 percent in the second quarter and 5.0 percent in July, after printing 0.0 percent in May. Households are not absent from the credit data. They are absent from the mortgage market specifically, which is the most rate-sensitive category of household borrowing there is, and present in revolving credit, which is the least discretionary.

Total loans and leases decelerated through the summer as well, from 8.3 percent annualized in the first quarter and 7.9 percent in the second to 3.9 percent in July. Yet overall bank credit grew 5.9 percent in July, faster than in June's 4.7 percent. The gap is explained by securities: the securities portion of bank credit grew 10.6 percent annualized in July, after 3.4 percent in June and minus 1.8 percent in April. Banks did not shrink their balance sheets in July. They bought bonds instead of making loans. Bank credit stood at $19,780.8 billion for the week ended Aug 5, with loans and leases at $13,956.4 billion and C&I at $2,921.6 billion.

The timing of all this is what makes it more than a data curiosity. At the July 28-29 meeting, the FOMC held the target range for the federal funds rate at 3-1/2 to 3-3/4 percent on a nine-to-three vote. The statement records it plainly: "Voting against the monetary policy action were Beth M. Hammack, Neel Kashkari, and Lorie K. Logan, who preferred to raise the target range for the federal funds rate by 1/4 percentage point at this meeting." The statement also described economic activity as expanding at a solid pace despite elevated uncertainty, and noted inflation remaining elevated relative to the Committee's 2 percent goal, in part reflecting supply shocks in certain sectors including energy.

Anyone building a hawkish case in late July from the credit data had two very good quarters of business lending to point at, and monthly data running only through June. What they did not have was the July print. That is not a criticism of the dissenters, who were reasoning from evidence about inflation and activity rather than from the H.8 alone. It is a statement about what the newest month does to one particular strand of that evidence. The strongest credit-side argument that business demand was running hot enough to warrant tightening rests on a series whose most recent observation is negative.

There is a competing explanation for the surge that has been on the record since it was still underway, and it cuts against reading the boom as evidence of underlying demand at all. In an American Banker piece published Apr 30, Samuel Tombs, chief U.S. economist at Pantheon Macroeconomics, argued that stress in private credit was pushing borrowers toward banks. "That's really what's driving bank lending higher," he said. "There are no otherwise obvious explanations why lending would be picking up, now, given the macroeconomic backdrop." He also said: "We hear all the time about problems in the private credit market. So any small restriction there forces companies to source funding from banks."

If that reading is right, the surge was substitution rather than demand: the same borrowing, rerouted through a different channel, showing up as bank loan growth because of where it was booked rather than because there was more of it. Substitution of that kind is a one-time level shift, and a level shift produces exactly the pattern the H.8 now shows, a burst that fades as the migration completes. That is consistent with the July figure. It is not proof of it, and one month of weekly-sourced data is a thin foundation for any conclusion.

The same article carried the counter-argument. Maureen Levelis, an analyst at Morningstar DBRS, allowed that private credit stress "could maybe be a small portion driving this loan demand, but I think it's just one of many factors," and noted that bank executives on earnings calls had cited higher credit utilization rates and the fruits of banks' increased investment in C&I hiring. Brian Foran, an analyst at Truist Securities, likewise called it one item on the list, noting that "there's relatively little visibility into private credit trends." Both readings can be partly true at once, and neither can be settled from H.8 line items, which record what was lent and by whom but not why.

One line does speak to the bank-nonbank plumbing directly. "All other loans and leases" stood at $3,304.4 billion for the week ended Aug 5 and grew 17.8 percent annualized in the first quarter and 14.0 percent in the second, moderating to 9.6 percent in July. This is the category that carries bank lending to nondepository financial institutions, which the H.8 breaks out at $2,016.1 billion for the same week. Banks lending to the entities that compete with them for corporate credit is a structural feature of this cycle, and it means the boundary between bank and nonbank credit is considerably blurrier than the C&I line alone suggests.

The July Senior Loan Officer Opinion Survey, released Aug 3 and covering 56 domestic banks and 18 U.S. branches and agencies of foreign banks, is worth reading with its timing in mind. On the business side, it found that "banks reported having left standards basically unchanged, on net, for C&I loans to firms of all sizes," with a moderate net share reporting stronger demand from large and middle-market firms and demand from small firms basically unchanged. On the household side, it found that "Moderate net shares of banks reported weaker demand for government-sponsored enterprise (GSE)-eligible, government, non-qualified mortgage (QM) jumbo, and non-QM non-jumbo residential mortgages." A modest net share reported tighter credit card standards. The survey asks about the quarter just ended, which means its stronger-C&I-demand finding describes the period in which the monthly H.8 data were already rolling over.

Two arithmetic cautions belong with all of this. First, the American Banker piece cited C&I growth of 12.7 percent quarter over quarter for the first quarter of 2026; the H.8's own Table 1 prints 12.2 percent for that quarter at a seasonally adjusted annual rate. The figures are close but differently constructed, and they should not be treated as the same number or quoted interchangeably. Second, Table 4 of the H.8, covering domestically chartered commercial banks, shows C&I loans of $2,303.0 billion against $2,921.6 billion for all commercial banks in Table 2. The roughly $619 billion difference sits at foreign-related institutions, but that residual is subtraction performed on two printed levels, not a figure the Fed publishes. Residential real estate is the mirror image: $2,690.0 billion at domestically chartered banks against $2,691.0 billion for all commercial banks, meaning essentially the entire book sits domestically.

Deposits are the last piece, and they moved the other way from the popular assumption. They grew 9.0 percent annualized in the second quarter but slowed to 5.3 percent in June and 3.2 percent in July, ending at $19,496.3 billion. A funding base decelerating while securities holdings accelerate and loan growth fades is a coherent picture of banks in a wait-and-see posture, not a system straining against a lending boom. The July FOMC minutes arrive Wednesday, Aug 19 at 2:00 p.m. ET and will show how the Committee characterized credit conditions three weeks before this release landed. What the release itself establishes is narrower and sturdier: the business-versus-household split in bank credit is real, the mortgage side of it has been flat all year, and the business side that made the split dramatic stopped growing in the most recent month on record. None of that determines what the Fed does next, and it should not be read as though it does.