Banks Told the Fed Their Business Loan Standards Now Sit Easier Than Their Post-2005 Midpoint. Subprime Card Standards Went the Other Way.

by Alina Castellanos · August 17, 2026

Economics Correspondent, US Market Current

Four times a year, the Federal Reserve's Senior Loan Officer Opinion Survey asks banks a question about motion: did you tighten, ease, or leave alone your lending standards over the past three months. Once a year, in the July round, it adds a different question about position. Banks are asked, in the release's words, to "consider the range over which their lending standards have varied since 2005 and to report where the current level of standards is relative to the midpoint of that range." The answers in the July 2026 survey, published by the Board of Governors on Aug. 3, describe two credit markets that are not in the same place.

The mechanics first, because they bound everything that follows. The July 2026 survey went out to banks on June 17 and responses were due by July 2, so it reflects conditions at the turn of the third quarter and asks about changes over the second. The release states that responses were received from 56 domestic banks and 18 U.S. branches and agencies of foreign banks. The level-of-standards questions are not part of the main questionnaire; the release groups them under a heading, "Special Questions on Current Level of Banks' Lending Standards," and cites them as table 1, questions 27-28 and table 2, questions 9-10. That block carries two sets of questions this year: the long-running one measured against each bank's range since 2005, and a second one on loans to non-depository financial institutions measured against the range since 2011.

On commercial and industrial lending, the special questions produced a result that is easy to skim past. The release reports that the current levels of standards were "easier than the midpoints of their historical ranges" for syndicated or club loans to investment-grade firms and for non-syndicated loans to large and middle-market firms and to small firms. For syndicated or club loans to below-investment-grade firms, and for loans to very small firms, the release places standards near their midpoints. The C&I sentence carries no exception clause, and nowhere in that block does the survey report a C&I category sitting on the tight side.

That is not the whole of the business-lending picture, and the same section says so. In the separate paragraph on non-depository financial institutions, banks reported that the current levels of standards for every type of NDFI loan the survey breaks out - lending to mortgage credit intermediaries, business credit intermediaries, private equity funds, consumer credit intermediaries and other NDFIs - sit at the tighter ends of the ranges that have prevailed since 2011. Banks describe their direct corporate book as loose against its own history and their lending to the financial intermediaries beside them as tight against its own.

The C&I level reading is not an artifact of one quarter's drift. The regular change questions in the same release report that banks left C&I standards basically unchanged for firms of all sizes during the second quarter, so the level did not get there by a recent burst of easing. Terms did move: the release reports that significant and moderate net shares of banks charged narrower loan rate spreads to large and middle-market firms and to small firms, respectively. On the demand side, a moderate net share reported stronger demand for C&I loans from large and middle-market firms, while small-firm demand was basically unchanged.

The household side of the same document points the other way. Over the second quarter, banks left standards basically unchanged for most residential mortgage categories, with a modest net share reporting easier standards on jumbo mortgages and HELOC standards basically unchanged. Demand for residential mortgages weakened on balance, with moderate net shares reporting weaker demand for GSE-eligible, government, non-qualified-mortgage jumbo and non-QM non-jumbo loans, and modest net shares reporting weaker demand for QM jumbo and QM non-jumbo non-GSE-eligible loans. In consumer credit, a modest net share of banks reported tighter standards on credit card loans over the quarter, while auto and other consumer standards were basically unchanged and a moderate net share reported weaker auto loan demand.

Then come the special questions on consumer loans, and the sentence is unqualified. The release states: "Regarding consumer loans, standards were at the tighter ends of their historical ranges for all categories." It continues: "A major net share of banks reported standards at the tighter end of their range for subprime credit card loans, and significant net shares reported so for subprime auto loans and other consumer loans." Moderate and modest net shares placed prime credit card and prime auto standards at the tighter ends of their ranges.

Those adjectives are not decoration, and this is where the release rewards a literal reading. The survey carries its own glossary. A "modest" net share means a net percentage above 5 and no more than 10; "moderate" means above 10 and no more than 20; "significant" means above 20 and below 50; and "major" means 50 percent or greater. The release also defines the arithmetic: a net fraction is the share of banks reporting tightening minus the share reporting easing. So "major" is a threshold, not an intensifier. It says that on subprime credit cards, banks reporting tight-end standards outnumbered those reporting the opposite by at least half the respondents to that question.

The comparison that gives the 2026 answers their shape is the July 2025 survey, released Aug. 4, 2025, which ran the same level question against the same post-2005 range and published the same four-band glossary, including the same "major" threshold of 50 percent or more. A year ago, banks put standards for syndicated or club loans to investment-grade firms and for non-syndicated loans to large and middle-market firms near the midpoints of their historical ranges, and a significant net share reported that standards for syndicated or club loans to below-investment-grade firms were tighter, on net, than the midpoints. Standards for non-syndicated loans to small firms and to very small firms also read tighter than midpoint in 2025. In 2026 the investment-grade, large-and-middle-market and small-firm categories read easier than midpoint, and the below-investment-grade and very-small-firm categories read near it. Every C&I category the question covers moved at least one notch toward looser.

Subprime credit cards moved the opposite way, and across a band boundary. The July 2025 release put significant net shares of banks at the tighter ends of their ranges for subprime credit card loans, subprime auto loans and other consumer loans. In July 2026, subprime auto and other consumer stayed at "significant" while subprime credit card was described with "major." Because both releases define the bands identically, that is the net share crossing from somewhere below 50 percent to 50 percent or above. It is the only category anywhere in the 2026 level questions to carry the top descriptor.

Two cautions belong on this reading. The first is that the level questions and the change questions measure different things, and a bank can answer them in ways that look contradictory without being so. A modest net share eased jumbo mortgage standards during the second quarter while significant net shares report that the level of jumbo standards is still at the tighter end of its range; both can be true, because one is a rate of change and the other is a position. The same pattern holds on commercial real estate, where a significant net share places construction and land development standards at the tighter end and moderate net shares do so for nonfarm nonresidential and multifamily loans.

The second caution is about the ruler. Each bank is asked about its own range since 2005 and its own midpoint. There is no common unit, no published spread or credit score attached to the answer, and no way to compare one respondent's midpoint with another's. The survey supports the statement that banks describe business standards as sitting on the easy side of their own history and consumer standards on the tight side of theirs. It does not support a claim about how loose business credit is in absolute terms, or that standards are at any kind of extreme, and the release makes no such claim.

Panel composition is worth noting for the same reason. Net percentages are shares of the banks that answered a given question, and the panel is not fixed: the July 2026 release reports responses received from 56 domestic banks and 18 U.S. branches and agencies of foreign banks, against 65 domestic banks and 19 foreign branches and agencies a year earlier. The foreign branches and agencies answer the business-lending questions in table 2; the household questions appear only in table 1, so the consumer results rest on the domestic respondents alone.

For anyone holding U.S. listed securities, the survey is the closest thing to a direct read on the supply side of bank credit, and it is describing a divergence rather than a cycle turn. Corporate borrowers, particularly investment-grade and larger middle-market ones, are being met by banks that say their standards sit easier than their own two-decade midpoint and that cut spreads last quarter while demand strengthened. Card borrowers, and subprime card borrowers above all, are meeting banks that say the opposite and that tightened card standards over the quarter. For listed lenders, the same two sentences describe where the underwriting risk is being added and where it is being held back. The next survey will repeat the change questions. The level questions are a July fixture - the release notes that every July survey since 2011 has carried a set of special questions - and will not come back until next summer.