China's 2.2 Trillion Yuan Deleveraging Is the Consumption Weakness the Market Is Misreading

The mechanics are now visible. The Shenzhen-based Ruiunion Credit Platform estimates that consumer loans transferred off bank balance sheets grew from about RMB 18.6 billion at end-2021 to roughly RMB 454 billion at end-2025 — more than twentyfold in four years. In the first eight months of 2026 alone, more than a thousand loan blocks covering over RMB 275 billion of unpaid balances came to market, and most buyers were state-owned asset managers.

One number deserves to be separated from the rest: official Chinese banking data put household-loan delinquency below 3%, against 4.7% in the United States. Those two facts are not in conflict — RMB 2.2 trillion on an RMB 82.25 trillion household book is 2.7%. The alarming part is not the ratio; it is that the stock of impaired household credit is larger than the entire bad-debt market that has been built to clear it, and it is concentrated in borrowers in their thirties and forties whose unsecured loans carry capped rates near 20%.

PBOC data show the deleveraging is broad, not confined to unsecured consumer loans. In the first nine months of 2026 household loans fell RMB 1.03 trillion, of which short-term loans fell RMB 1.14 trillion while medium- and long-term loans — mostly mortgages — added only RMB 106 billion, against RMB 1.28 trillion for the whole of 2025. Households have stopped borrowing even for housing, which is not what a confident consumer looks like.

The spending data confirm the story rather than contradicting it. Retail sales grew 1.1% in the first nine months of 2026 to RMB 37.0 trillion, and September alone grew 1.1%; strip out automobiles and September grew 3.9%, implying that the value of car sales dropped about 23% year on year, from RMB 471 billion to RMB 364 billion. The weakness is concentrated in the big-ticket, credit-financed category, which is what household deleveraging looks like in the consumption data.

The counterweight is real and worth stating plainly: household deposits rose RMB 9.39 trillion over the same nine months, and the household loan balance of RMB 82.25 trillion was 2.0% lower than a year earlier. So the deleveraging is partly a shift into savings rather than a collapse in income — which makes the drag on consumption large but not obviously permanent. A savings pool that keeps compounding is the swing factor for the next consumption cycle.

Two regulatory squeezes in 2026 have cut the supply of collection capacity just as the stock of bad debt grew. The China Banking Association's personal consumer loan collection guideline took effect on January 30, banning threatening and humiliating contact and repeated call bombardment; the 2026 revision of the internet finance overdue-debt collection self-discipline convention took effect on August 1, allowing contact only between 08:00 and 20:00 and capping a collector at three contacts a day. Regulators have also told lenders to keep average rates on new online-platform loans below 20%. The rules that protect borrowers are shrinking the recovery machine.

The people doing the collecting are already stretched: 200 to 300 calls a day on entry-level pay near RMB 4,000 a month, rising to RMB 20,000-30,000 for experienced collectors, with monthly turnover reaching 20% in some months, and some resolution firms laying off staff or closing. Recovery rates have slowed with them, which is why clearing prices fell toward 1% of face value and why China Renaissance Holdings paid only about 4% for RMB 7.4 billion of outstanding principal on loans already 439 and 854 days overdue.

The pattern in the table is the important part. Average days overdue rose from about 700 to more than 900 in a year, and almost 80% of packages are now more than a year past due, while a parallel wave of 101-to-150-day packages is appearing — lenders locking in losses on claims that have not yet aged. Weighted-average borrower ages in the distressed pools run from the early thirties to the late forties, with bank-affiliated vintages concentrated around 46 to 48. The bad debt is a wage-cycle problem wearing a consumer-credit costume.

Demand for credit is not the only thing shrinking. The internet loan-facilitation rule effective October 1, 2025 put partner platforms on a whitelist and pushed funding and risk back toward banks, and the financial product online marketing rules issued by the People's Bank of China with seven other departments took effect on September 30, 2026, barring platforms from the core sales step and prohibiting 'instant approval' and 'low rate' sales language. With lenders told to cap average rates on new online-platform loans below 20% and consumer finance companies told to push comprehensive financing cost under 20%, marginal borrowers stopped clearing. Supply of consumer credit is contracting in the same quarters that demand is, which is why 'weak demand' and 'weak supply' print on top of each other.

Qfin Holdings put it plainly in its own second-quarter release: industry contraction, tighter regulatory oversight, a liquidity shock in late June, and a more prudent approach to growth, risk and capital allocation. Origination and facilitation volume fell 25.1% year on year to RMB 63.4 billion while revenue fell 17.8% and net income 76.8%. FinVolution reported mainland transaction volume down 19.3% and net profit down 43.2%, with management noting that institutional funding in China is tightening against an evolving risk environment. Nobody is repricing the risk; they are simply writing less.

The listed consumer finance companies show the same squeeze with a different signature: of the six licensed institutions that disclosed first-half results, only two grew their balance sheets. CMB Finance's total assets fell 11.9% to RMB 147.4 billion in six months, its revenue fell 15.3% and net profit 8.2%, while 马上消费金融 and Ant Consumer Financial's RMB 1.268 billion package (weighted-average overdue 316.5 days) show the same wave arriving at different speeds. Notably, Ant Consumer Financial is the outlier that is still growing: revenue of RMB 10.9 billion, up 8.7%, and net profit of RMB 1.94 billion, up 33%, in the first half.

The stress is landing wherever the paper sits. ICBC reported a credit-card overdraft NPL ratio of 5.37% at June 30, 2026 against 4.61% at end-2025, even as first-half revenue rose 9.1% to RMB 465.9 billion and attributable profit rose 3.3% to RMB 173.7 billion. At the asset-manager end, China Cinda acquired RMB 141.3 billion of new financial non-performing claims in the first half and lifted non-performing debt income 55.4% to RMB 3.56 billion — yet group attributable profit was only RMB 784 million, with core tier-one capital at 9.04%. Volume is being cleared at prices that barely cover the cost of clearing it.

Note what the table does not contain: an explosion. Bank profits are still growing, platform balance sheets are still liquid, and the asset managers are still transacting. That is precisely why the market keeps calling this weak consumption. The losses are being distributed thinly across provisioning, disposal prices and thinner origination rather than crystallised in a headline. A crisis that shows up as a 5.37% card NPL ratio is not a crisis yet — which is exactly the argument for pricing it before it is.

Over the next two quarters, three dated things decide the argument. First, December 31, 2026, when the personal consumer loan interest subsidy extended by the finance regulator, the finance ministry and the central bank expires: it pulls borrowing forward into the fourth quarter and leaves a hole in the first quarter of 2027. Second, third-quarter results from Qfin Holdings and FinVolution, where FinVolution's full-year guidance of RMB 11.5-12.9 billion against a RMB 6.6 billion first half implies a second half below the first. Third, the clearing price of the next batch of consumer NPL packages, where another print near 1% of face value confirms that recovery assumptions have been reset rather than merely delayed.

Over one to three years, the variable is institutional rather than economic: whether China builds a discharge channel for households that have already failed. Shenzhen's five-year personal bankruptcy pilot has accepted 869 cases, concluded 809 and discharged RMB 285 million of debt, with 70% of debtors reporting a fixed job and predictable income, and the Supreme People's Court confirmed in March 2026 that the first liquidation case ended with the remaining debt written off after four years of supervision. A national framework would convert deleveraging from a consumption tax into a solved problem.