Scenario B: Credit Repricing, Subchapter V Growth and What Refinancing Failure Looks Like

1 The scenario and the population it concerns

This is the second of three follow-ups to our analysis of what would actually qualify as a black swan. It takes the credit scenario named in that article and works through it in detail. It is a labelled scenario. It is not a forecast, it is not a prediction, and no probability is assigned to it anywhere in what follows. The parent article framed this material as Scenario B - credit, and the framing rule carries over unchanged: the content below describes a mechanism that could operate and describes what it would look like if it did.

The population is specific and it matters. The operators in question are the same thin-margin freight, farm and construction businesses that Scenario A concerns. They buy diesel at retail and absorb or pass through cost per mile and cost per acre. They also carry equipment debt - trucks, trailers, tractors, combines, excavators, machine tools - and a substantial share of that debt is floating-rate. They borrow from banks, from captive and independent equipment finance companies, and from private lenders. They do not issue bonds. That last sentence is the hinge on which this entire article turns, and we will come back to it in section 5.

The parent article recorded a set of counts and a set of prices, and its own conclusion about what those counts mean. That conclusion is worth restating plainly here, because it governs everything else. Rising bankruptcy filings and rising Subchapter V elections are observed and reported facts. Refinancing stress among small floating-rate borrowers is a catalogued risk, meaning it is written down in public documents and discussed in public data. Something that is catalogued and observable fails a surprise condition by construction. So this is not a black swan, and this article does not argue that it is one. It is a documented vulnerability whose open question is timing and which borrowers, not whether the mechanism exists.

One further distinction before we turn to the data. In this scenario, the harm does not arrive as a single failed institution or a single headline. It arrives as a sequence of elections on court dockets, each one dated, each one identifiable. That is the opposite of a diffuse squeeze. A credit event that shows up as filings can be counted, dated and attributed to particular borrowers in particular counties. That is what makes it analytically tractable, and it is also what makes it, by the parent article's own test, unsurprising.

2 Filing data with dates and composition

Start with the counts, because they are the most concrete thing in this scenario. Epiq AACER and the American Bankruptcy Institute counted 302 Subchapter V elections in August 2026. The same compilation put the August 2025 figure at 185. That is an increase of 63 percent year over year. The month-over-month move is also large: July 2026 recorded 236 elections, so the August reading was up 28 percent from the month immediately before. Both figures are reported claims in the precise sense that they are compilations of court records assembled by a private data provider and a professional association. The underlying dockets are public. The tally is theirs.

The broader filing picture comes from the Administrative Office of the U.S. Courts, which is the official reporter for federal court caseload statistics. For the twelve months ended June 30, 2026, total filings were 608,511, an increase of 12.2 percent. Business filings within that total were 26,941, up 16.9 percent. Those two numbers should be read together, and read honestly. Business filings are roughly 4.4 percent of all filings. The headline total is therefore dominated by consumer cases, and a reader who sees only the 12.2 percent figure is seeing mostly households, not mostly firms. The reason to report the total at all is that it establishes the direction and the scale of the movement. The reason to separate the business line is that it is the one this scenario is about, and it is growing faster than the total - 16.9 percent against 12.2 percent.

The series behind those figures makes the direction unambiguous over a five-year window. Total filings for the year ending June 30 were 380,634 in 2022, 418,724 in 2023, 486,613 in 2024, 542,529 in 2025 and 608,511 in 2026. Business filings over the same five years were 12,748, 15,724, 22,060, 23,043 and 26,941. Every year in the window is higher than the one before it on both lines. Subchapter V, at 302 elections in a single month, remains a small sliver of the business line, but it is the fastest-rising sliver within it.

What this data does not tell us is how much of the increase is genuine distress and how much is ordinary churn - the normal rate at which small businesses fail, close or reorganise for reasons that have nothing to do with credit conditions. That is an unknown, and it belongs in the list of unknowns. A rising count is a fact. A rising count caused by credit repricing is a causal inference, and the inference is stronger when the population, the instrument and the timing all line up. We turn next to the instrument.

3 Floating-rate equipment debt and the repricing clock

A floating-rate loan is priced at a reference rate plus a spread. For small-business equipment debt the reference is typically a short-term benchmark, and the spread is set by the lender based on the borrower. Both halves can move, but the reference half moves with policy and market rates. As of September 23, 2026, the two-year Treasury yield was 4.85 percent, the ten-year was 5.11 percent and the thirty-year was 5.40 percent. A little over two years earlier, on September 16, 2024, the same three yields were 3.56, 3.63 and 3.94 percent. Those are observed facts with dates. The relevant one for this scenario is the front of the curve, because it is short rates that feed the reference leg of an equipment loan.

The important point about the repricing clock is that it does not require rates to rise further. A floating-rate borrower who took on debt when short rates were materially lower is already carrying a higher debt-service cost than the one the business was underwritten against. What has not yet happened, in many cases, is the reset or the maturity. A loan with a fixed reference period, or a term loan with a maturity two or three years out, can sit quietly at the original cost even while the market rate beneath it has moved. The borrower notices at reset. The lender re-underwrites at renewal. The mechanism is therefore calendar-driven rather than price-driven, which is precisely what makes the scenario's outcome discrete and dated rather than gradual.

This is also why the scenario is a bridge from Scenario A rather than a duplicate of it. The fuel shock compresses operating cash flow. The credit structure determines when that compression meets a fixed obligation. A trucking operator whose cost per mile has risen and whose rate per mile has not yet repriced is, in accounting terms, generating less cash against an unchanged debt-service requirement. If the debt-service requirement is itself about to step up because a reset date is approaching, the two effects compound. In this scenario the compounding is the transmission channel. It is a causal inference, not a measured outcome, and we label it as such.

4 Repricing versus declining: two different harms

When a lender faces a borrower whose coverage has tightened, it has a menu. It can reprice - hold the relationship and charge more. It can tighten terms within the existing facility, adding covenants or shortening required reporting. It can decline renewal outright. These are not variations on one theme. They are different states of the world for the borrower, and only one of them produces the outcome this scenario is about.

Repricing keeps the borrower alive at a higher cost. That is painful, and it consumes cash that would otherwise fund maintenance, hiring or fuel inventory, but it is survivable in a large number of cases. Tightening covenants within an existing facility is a middle case: it preserves the credit but transfers optionality from the borrower to the lender, and it makes a subsequent breach more likely if operating performance slips. Declining renewal is the terminal case. A borrower whose facility is not renewed does not merely pay more; it must find a replacement lender, refinance on whatever terms the market will offer, sell assets, or restructure. If none of those paths closes in time, it files.

The collateral mechanics push lenders toward the third option more often than intuition suggests. Equipment collateral depreciates. A truck, a trailer, an excavator or a machine tool has a market value that declines with use and with age, and that market value is itself cyclical. In a downturn, when many operators are selling or surrendering the same kinds of assets, remarketing values fall at the same moment lenders would need to realise them. A lender looking at a borrower whose coverage has tightened and whose collateral is depreciating has an incentive to act at renewal rather than at default, because acting early protects recovery. Declining to renew is the cleanest way to act early.

This is how a credit scenario becomes a refinancing failure. Nothing has gone wrong at the firm in the ordinary sense. It has not missed a payment. It has arrived at a date, and the lender has decided not to extend. That decision is invisible in index spreads, unattributable to any single market event, and completely visible in a court filing weeks or months later. It is, in the vocabulary of the parent article, discrete and dated.

5 Why record-tight spreads and rising filings are both true

Here is the apparent contradiction, and it is the most analytically interesting thing in this scenario. Corporate credit spreads are at or near record tights. The high-yield option-adjusted spread was 2.73 percent on September 23, 2026. On April 7, 2025 it was 4.61 percent. Investment-grade spreads stood at 0.77 percent on the same September date, and BB spreads at 1.59 percent. Those are observed facts with dates, taken from the same public data series the market itself watches. The bond market is pricing calm.

At the same time, the bankruptcy counts in section 2 are rising, and Subchapter V elections rose 63 percent year over year. If spreads measure credit risk, why are they tight while filings rise? The answer is that the two measures are not measuring the same thing, and in fact they are barely measuring the same population.

A credit spread is a price. Specifically, it is the compensation investors demand to hold a tradable bond rather than a risk-free instrument of comparable maturity. It reflects the expected loss on a diversified portfolio of liquid, index-eligible securities, and it is set by the buyers and sellers of those securities. The issuers in that market are overwhelmingly large: publicly listed corporations, well-known private companies with institutional followings, and issuers large enough to justify the cost of a public or broadly syndicated deal. To be in a high-yield index at all, an issuer has to be big enough to be indexed.

A bankruptcy filing is a realised outcome. It is not a price and it is not a portfolio expectation. It is what happened, to a particular firm, on a particular date. The firms that file for Subchapter V protection are by statutory design small businesses. They do not issue bonds. Their obligations are bank lines, equipment finance contracts, trade credit and private loans. None of those instruments trade in an index, and none of their terms appear in any published spread series.

So the correct reading is not that the bond market is wrong or that the bankruptcy data is misleading. It is that a tradable spread and a small-business filing count measure different populations, and there is no accounting identity that requires them to move together. Record-tight spreads tell us that the marginal dollar of institutional credit risk is priced expensively relative to history - that is, investors are demanding very little extra yield. Rising small-business filings tell us that a specific group of non-tradable borrowers is running into trouble. Both can be true in the same quarter, and in this scenario both are.

There is a second reason the two diverge, and it is structural rather than statistical. Small-business credit terms are not publicly indexed. When a bank changes the rate on a revolving line, or an equipment finance company shortens a term, or a private lender withdraws from a sector, none of that prints anywhere. It is visible to the borrower and to the lender, and to almost nobody else until a filing appears. The absence of a public price for small-business credit is not evidence that small-business credit conditions are unchanged. It is evidence that we do not have a ticker for them.

6 From fuel squeeze to refinancing failure

Pull the chain together, with each link labelled. The first link is observed fact: short rates sit well above where they were in late 2024, with the two-year at 4.85 percent against 3.56 percent on September 16, 2024. The second link is structural: a share of small-operator equipment debt is floating-rate, so its cost tracks short rates and lender spreads. The third link is the bridge from Scenario A - a fuel-cost squeeze that reduces operating cash flow for freight, farm and construction operators, described in that article as a cash-flow compression rather than a single failure. The fourth link is a lending decision, and it is the one that turns pressure into an event.

Why is Subchapter V the right lens on that decision? Because it isolates the population precisely. Subchapter V of Chapter 11 exists to make reorganisation cheaper and faster for small businesses. It carries a debt-limit threshold that keeps it to smaller debtors, an elected trustee rather than the full apparatus of a conventional Chapter 11 case, and, in typical cases, no creditors committee. The design intent is survivability: a small firm should be able to restructure without the cost and procedural weight that a large corporate case carries.

The second reason is more subtle and more useful. A Subchapter V election is chosen. A firm that files under it has made a decision, usually on advice, to reorganise rather than to liquidate, and it has elected into a specific subchapter of the code. That makes the count a behavioural signal as well as an economic one. A rise from 185 to 302 elections in a year is not just more distress; it is more distress being routed into a reorganisation channel that the firm believes it can survive. Behavioural data of that kind is more informative about small-business expectations than any aggregate filing total, because the aggregate is dominated by consumer cases that have nothing to do with this mechanism.

Put the chain in order and the scenario reads as follows. In this scenario, lenders reprice at reset, add covenants, or decline renewal. Borrowers that can absorb repricing continue at a higher cost. Borrowers that cannot refinance on acceptable terms must restructure or file. Those that file appear as Subchapter V elections, and each one is a dated, identifiable event in a public court record. The result is not a diffuse malaise but a countable series, which is exactly why the scenario is described as discrete.

7 What would deepen it and what caps it

Honesty requires stating the present condition clearly, because it does not match the scenario. Bank lending standards are not broadly tight right now. The Federal Reserve's survey of senior loan officers reports the net percentage of banks tightening standards for commercial and industrial loans to large and middle-market firms. That series shows 33.9 percent in the third quarter of 2023, which was the tightening peak of the recent cycle; 0.0 percent in the fourth quarter of 2024; 18.5 percent in the second quarter of 2025; 5.3 percent in the first quarter of 2026; 8.1 percent in the second quarter of 2026; and 0.0 percent in the third quarter of 2026. A net zero reading means banks tightening and banks easing roughly cancel out. It does not mean credit is easy in an absolute sense, but it plainly does not describe restrictive conditions.

This is the single most important honesty check on the whole scenario. The trigger here is a turn in standards, not the present state of standards. Anyone reading the Subchapter V counts and concluding that the banking system has already tightened would be reading the data backwards. What the scenario posits is that standards turn from net zero to net tightening, and that the turn arrives while short rates remain where they are and while the borrowing base is still carrying floating-rate exposure taken on at lower rates. Every clause in that sentence is a condition, and the scenario depends on all of them.

What would deepen it, then, is a specific combination. Standards turning net positive on tightening would do it. Short rates staying high for longer than borrowers' reset schedules assume would do it, because each reset transmits the level rather than the change. Fuel costs persisting at the levels Scenario A describes would do it, because the cash-flow side of the debt-service coverage ratio is what makes a repriced loan unaffordable. And a fall in used equipment values would do it, by worsening lender recovery and reinforcing the incentive to decline at renewal.

What caps it is equally specific. Net-zero standards today are a cap, because the mechanism needs a turn. The yield curve is another: at 4.85 percent on the two-year, 5.11 percent on the ten-year and 5.40 percent on the thirty-year, the curve is upward-sloping rather than inverted, which is normal for an expansion and is not the configuration typically associated with acute credit stress. And the structural design of Subchapter V is itself a cap. The subchapter exists precisely to make restructuring survivable rather than terminal, so a rise in elections is partly a measure of how many small firms are using a tool designed to keep them operating. A filing under Subchapter V is a bad outcome. It is not the same outcome as a liquidation.

Second-order effects deserve a mention because they are where a small-business credit scenario touches things outside itself. Suppliers carrying trade credit to affected operators absorb losses when a customer restructures, and those losses land on other small firms. Equipment remarketing channels absorb surrendered fleets and machines, and if enough arrive at once, values fall for every operator trying to sell, which feeds back into the collateral problem described in section 4. Local employment at affected firms is concentrated, not diversified, so a single refinancing failure can be a meaningful share of a small labour market. And small retail and service operators frequently carry commercial real estate exposure, which links a filing count to a property market that is not priced in the same way a bond is. Each of those is a causal inference about transmission, not a measured result.

8 A catalogued vulnerability, not a black swan

The parent article defined the test and then applied it, and the result for this scenario is unambiguous. A black swan is defined by surprise, not by severity. Something that is written down in a public document, counted in a public series and discussed in advance does not satisfy a surprise condition. Subchapter V elections are counted monthly. Total and business filings are published by the Administrative Office of the U.S. Courts. Credit spreads and lending standards are published continuously. A mechanism assembled from those inputs is, by construction, anticipated.

The Federal Reserve's Financial Stability Report of May 2026 is the clearest illustration of the point. It catalogues vulnerabilities rather than forecasting events. It notes that hedge fund leverage is stable at record-high levels and concentrated in the largest funds, and that margin calls have been met without difficulty. That is what a catalogued vulnerability looks like: a condition that is known, measured and named, with no date attached and no probability assigned. Our credit scenario is the same kind of object. Refinancing exposure among small floating-rate borrowers is a known feature of the credit landscape, not a hidden one.

So the honest close is this. This article has described a mechanism, not predicted an outcome. The counts are facts with dates. The spread levels are facts with dates. The lending standards reading that caps the scenario is a fact with a date, and it happens to be the fact most likely to be skipped by anyone eager to tell a distress story. The transmission from fuel squeeze to coverage pressure to lending decision to filing is a causal inference. The Fed's catalogue of vulnerabilities is expert interpretation. The scenario itself is a scenario, labelled as one throughout, with no probability attached at any point.

What remains genuinely unknown is worth naming. We do not know how much of the rise in business filings is distress versus ordinary churn in a normal business population. We do not know the true size of the floating-rate small-business debt stock, because the relevant lending happens off-index and largely off-database. We do not know private lender terms, because they are not published. And we do not know when the reset and maturity dates that constitute the refinancing wall actually fall, which is the single most consequential unknown in this scenario. The mechanism is documented. The timing and the identity of the affected borrowers are not.

That is the correct status for this material, and it is why it belongs in a scenario series rather than on a list of things nobody saw coming. A catalogued, observable vulnerability fails the surprise condition. The open question here is when the calendar meets the cash flow, and which borrowers are standing on the date. If that turn arrives, the evidence will be a count of elections on a court docket, rising from a base of 302 in August 2026 - observable, dated, and by the parent article's own standard, not a black swan at all.