Japan logged 10,425 corporate failures in the year to March 2026, the second year running above ten thousand. Read the headline on its own and the story tells itself: the small-firm sector is finally cracking. The same dataset says something stranger underneath. Total liabilities came to ¥1.55trn, down for a second year from ¥2.25trn, and failures of firms with under ¥50m of liabilities made up 62.1 per cent of the count, a record in data back to fiscal 2000 (Teikoku Databank). The number of failures is climbing while the money involved shrinks. The two point opposite ways.
They diverge because what is vanishing is small. No large bankruptcy is shaking the system. Tiny firms kept alive for years are now surfacing in the tally. The reflex that reads more failures as a weaker economy is worth stopping on, then. This is not a downturn. It is a change in the character of the cull.
Two forces drive that change. One is the Bank of Japan’s return to positive rates. The other is a reform of trade-credit practice – the phasing out of promissory notes and a tightening of payment terms – that runs on its own track and owes nothing to the central bank. The conclusion, stated first: rate normalisation is a single coin. One face is wider bank margins, the other is the cull of borrowers who cannot pass the cost on. And the trade-credit reform is sealing off, in the same months, the working-capital routes those borrowers would have used to get clear.
Pricier, not scarcer
The BOJ’s Tankan survey takes the temperature of roughly 10,000 firms each quarter. Line up three of its small-firm series and the picture resolves at once (March 2026 survey).
Start with the judgement on borrowing rates, the share reporting “rising” less the share reporting “falling”. The small-firm reading climbed to 64 from 46 in December, a jump of 18 points. The BOJ raised its policy rate to 0.75 per cent from 0.5 in December 2025 and banks followed by lifting their short-term prime rate. That jump is the pass-through reaching borrowers as lived experience.
The judgement on banks’ lending stance, “accommodative” less “severe”, barely moved: 13 a year ago, 12 now. Banks are not pulling back from small firms. Funding conditions, “easy” less “tight”, are the most static of the three, holding around plus 8 for over a year and easing only to plus 7 in March 2026. The average small firm still sits on the comfortable side.
Put the three together and the route of the cull comes into view. Price went up. Quantity did not, and neither did the comfort of day-to-day funding. What is squeezing borrowers is the rate itself, not a credit freeze. The “banks have stopped lending” account is contradicted by the official data before it gets going. The money confirms the pass-through: interest paid across all industries accelerated from a 14.3 per cent year-on-year rise in the first quarter of 2025 through 19.4, 20.2 and 29.2 to 31.2 per cent in the first quarter of 2026 (Ministry of Finance).
The unmeasured tail
Higher interest bills do not sink the sector. The SME white paper itself reckons that even after the rise in interest payments, small-firm recurring profits could end up higher, not lower. The split is not in the average. It is in the tail.
Break recurring profit down by capital size and the dispersion is plain. In the first quarter of 2026, firms capitalised at ¥10bn or more grew profits 24.7 per cent year on year, those between ¥1bn and ¥10bn grew 11.2 per cent, and those between ¥10m and ¥1bn managed just 1.5. The largest tier races ahead while the smallest barely moves. It is this layer that is caught between higher rates, costs it cannot pass on and the wage round.
There is a catch built into the figures. The MOF survey does not cover firms with under ¥10m of capital at all. The genuinely fragile, the smallest of the small, sit outside the frame by definition. The average looks fine partly because the layer that is actually breaking was never in the sample.
On zombie firms the counts disagree. Teikoku has the 2024 share at 14.3 per cent (about 210,000 firms) and falling for a second year, while Tokyo Shoko Research, on the Bank for International Settlements (BIS) definition, has it at 15.20 per cent and worsening by 0.63 of a point. One reads the same population as improving, the other as deteriorating, capturing firms newly slipping into the zombie bracket as rates rise faster than their earning power. The point is not which is right. In a world with rates, the same set of companies can read better or worse depending on how you count – the same shape as the gap between the rising failure count and the falling liabilities. And the more the count diverges, the harder it becomes to trace where the genuinely weak have gone, except outside the statistics.
Into the vacuum
Their destination was prepared by a force other than the rate rise. What moves it is not the BOJ but the Civil Code and the Japan Fair Trade Commission.
The 2020 revision of the Civil Code laid the groundwork. Receivables carrying a no-assignment clause became assignable in principle (article 466), and the assignment of future receivables was written explicitly into law (article 466-6). That gave a legal footing both to factoring, which turns receivables into cash early, and to revenue-based finance (RBF), which advances money against future sales. The first spread across small-firm working capital generally, the second mainly among start-ups. They share one thing: neither has a conduct law of its own. Japan’s Financial Services Agency states plainly that no statute governs factoring and that it falls outside its remit. RBF sits in the same unregulated zone, and a registration regime, though discussed, was still not in place as of 2026.
The reshuffle shows up in the data. The balance of discounted notes receivable across all industries shrank from ¥620bn in the first quarter of 2025 to ¥415.8bn a year later, down 32.9 per cent. The old habit of discounting a note to bridge a gap is thinning out. This is not the rate at work. What pushed it is the amended Subcontract Act, renamed the Act on Optimising Transactions with Small and Medium Subcontractors, in force from January 2026. It no longer recognises payment by note, and it bars uses of electronic monetary claims or factoring where the receiver cannot get the full sum, fees included, by the due date. One by one, it is closing the routes through which a buyer shifted its own financing burden onto the supplier. Short-term borrowing, meanwhile, rose 8.6 per cent. Borrowers are swapping notes for loans and for pulling receivables forward.
Beyond the closed routes lies the regulatory vacuum, and into it step the impostors. An arrangement that is lending in substance – where the buyer takes no collection risk and forces a buy-back if the debt goes bad – counts as moneylending whatever it is called. In February 2023 the Supreme Court ruled that buying a wage claim and collecting it through the worker amounts to the moneylending business. The FSA and the Japan Financial Services Association warn repeatedly that such disguised factoring is unregistered loan-sharking. RBF crosses the same line when, dressed as revenue sharing, it is really a loan. The worse a firm’s cash position, the harder it is pulled towards operators sitting on the boundary between legal and illegal.
The size of this zone cannot be put in figures. No reliable primary statistics exist, and most of the numbers in circulation are the operators’ own estimates. What can be said is that it is wide enough for the regulators to keep issuing warnings. And the MOF survey shows recurring profit in the moneylending and related sector up 9.4 per cent in the first quarter of 2026, with capital spending up 16.9. The signal that demand pushed out of the banks is being caught on the nonbank side is indirect, but it is in the numbers.
One coin
Pull it together and two forces are tightening on the same tail. The BOJ’s normalisation is one coin with two faces: banks can now price credit, which widens spreads at the regional lenders among others, and that same rate shaves off the tail that cannot pass costs on. The second force, the reform that scraps notes and tries to clean up payment terms, owes nothing to the BOJ and seals off the tail’s old funding routes. The motives differ; the destination is one and the same, a push into unregulated finance. The more the BOJ moves rates and the more the reform shuts routes, the harder the push.
For now the push has somewhere to land. Short-term borrowing is up, moneylenders’ profits are up, and factoring and RBF take what comes next. So the failure count moves late, behind the real stress. The 10,425 already on the board is a trailing print of what is happening below the surface.
The tail is not choosing this. A layer that could still discount a note a year ago has been forced – by two dated events, the December rate rise and the January subcontracting reform – into the channels beyond the banks, and in one direction only. The compulsion is new this year, not a standing condition.
The trouble is that the landing has a limit. Short-term borrowing and nonbank credit reshuffle debt without creating the means to repay it. Buy-back factoring goes further, deepening next month’s hole each time it bridges this one. Teikoku Databank reads the number of firms that may yet fail as building quietly below the surface and feeding through to failures from the second half of fiscal 2026. When the landing fills, the stress surfaces. And since the unregulated zone is small and the regulators are already leaning on it, where it surfaces is more likely the larger, visible failure count.
So the thing to watch is the gap between two Tankan readings. The borrowing-rate judgement has already climbed to 64 while the funding judgement still sits at plus 7. That gap measures the distance borrowers can still absorb the rate over. When the funding reading starts breaking towards neutral as failures accelerate, the absorption is spent. In the market it shows up in regional banks’ credit costs.
What should an investor take from this? The piece argues a macro frame, not a list of stocks. The direction is still worth stating. The winners in a world with rates are those who can price credit. Regional lenders enter a tug-of-war between wider spreads and rising credit costs; the market is pricing the spread gain now, and if the argument here holds, it is the cost side that is set too low. Buying the weakest regionals on a low price-to-book alone – as consolidation targets due for a re-rating – risks missing that the discount may already reflect the tail’s credit risk; a low multiple is as easily the mouth of a value trap as the start of a re-rating. The nonbanks and moneylenders catching the demand pushed out of the banks ride that tailwind for a while. And when that build-up surfaces, the corporate-turnaround field comes into play. None of this is a case for turning bullish on Japanese consumption. It is a case for telling apart, by firm size, whose margin the normalisation becomes and whose cost.
The cull in the statistics is a shadow that falls late. What moves first is the absorption below the surface, and the distance left before it runs out is the thing worth pricing now.