Vol.193 Macro Talk 96 | A Look at China’s October Economic Data (Recorded Nov. 20)
Summary
October’s economic data did not show a “consumption collapse,” but a mixed signal: a weak PMI, CPI/PPI slightly beating expectations, and retail sales supported by the long holiday and subsidies. 李丰 had expected October PMI to keep improving after September’s bottoming and turn, but PMI first poured cold water on that view, while the price data pulled it back to neutral: “Some things were worse than expected, while others were about as expected.” If the historical lag between capital-market improvement and price transmission still holds, PPI had reached its roughly two-quarter recovery window by October, counting from May.
The sharp drop in household short-term lending cannot be explained solely by weak consumption appetite; the supply contraction caused by tighter caps on internet-assisted lending rates may be the bigger variable. A central-bank industry guideline issued in April required assisted-lending rates to be strictly capped at an annualized 24% from October. That eliminated the room to charge 24%–36% through membership cards, points and similar mechanisms, abruptly interrupting borrowing to repay old debt. 李丰 put it sharply: “People with money aren’t consuming; people without money want to consume, but can no longer get loans.”
The contraction in assisted lending may amount to RMB1T-plus of deleveraging and will continue to weigh on related platforms’ fourth-quarter results. 李丰 estimates that the largest integrated internet platforms originate more than RMB3T and nearly RMB4T in loans annually; including standalone platforms, the industry may exceed RMB4T–RMB5T. He initially estimated policy-driven annual contraction at least RMB1T, possibly RMB1T–RMB2T, and said the contraction realized since July—especially during August–October—could be at least RMB1T, or even RMB2T–RMB3T. Third-quarter reports already showed several U.S.-listed platforms with no scale growth and some with declines in both revenue and profit. Quarter-on-quarter performance in the fourth quarter will be “at best flat,” though the business may return to a normal track after one or two quarters of cleanup.
China’s social financing has already shifted structurally, making bank-loan data alone an increasingly unreliable gauge of the strength of real-economy financing. Direct financing accounted for about 44% of social financing in the first 9 months, while bank loans made up less than 50% in January–October. Bond and equity financing now represent “nearly half,” although the current increase is mainly being driven by central and local government bonds. If market stability continues to improve, corporate bonds, equities and REITs could take over; risk would move from bank balance sheets to subscribers and be priced directly by the market.
China’s consecutive foreign-currency sovereign-bond offerings have sent a rare strong signal to global investors allocating to Chinese assets and established a pricing benchmark. The $4B Hong Kong sovereign bond drew about $118B in orders, nearly 30x the offering, with roughly two-thirds coming from insurers, banks and sovereign funds. The 5-year tranche priced only about 2bp above comparable U.S. Treasuries. China then issued €4B of bonds on Nov. 18. For investors, this reflects both the willingness of “long money” to increase exposure and a world awash in dollar liquidity but short of assets.
The core risk in AI infrastructure is expanding from highly valued stocks to rapidly accumulating debt, some of it moved off balance sheet. 李丰 says major U.S.-listed AI companies have issued nearly or more than $200B in bonds this year to fund data centers and related infrastructure, with the market expecting the figure to reach more than $1T next year. The positive feedback loop depends on GPU utilization, depreciation, technology paths and cash flow continuing to cover financing costs of roughly 5%–12%. “The more AI stories they tell” can lift valuations and make bond issuance easier, creating a “loop with themselves.”
The A-share market’s rise from roughly 2,500 to 4,000 has not produced broad-based gains because the rally has concentrated in two crowded asset groups, while the coming rotation from high to low could further pressure the index. One group is technology and AI chips favored by public-fund crowds; the other is banks, power, infrastructure and parts of the upstream sector favored by insurers. Balanced portfolios have often barely kept up or lagged. Changes to public-fund benchmarks, together with improving CPI/PPI, are giving investors a reason to rotate from high-valuation crowded sectors into cheaper industries: “Balanced allocation does not make it easy to earn money, while aggressive allocation makes it relatively easy.”
Deep dive
1. October Was Not a One-Way Weakening: PMI and Prices Diverged Again
September PMI had already shown a bottoming rebound, so 李丰 had expected October to extend the improvement. Instead, he first saw an “unexpectedly bad” PMI, followed by CPI and PPI that slightly beat expectations, pulling his assessment from weak back to neutral.
His review preserved the shift in real time: after seeing PMI, he “didn’t think it would be very good,” but the price data could still be considered “not too bad.” In the end, “some things were worse than expected, while others were about as expected.”
October also included the long holiday and the sudden escalation and subsequent deal in trade frictions, among other special factors. A single month therefore cannot be mechanically extrapolated; the better approach is to separate the contributions of cyclical recovery and one-off disturbances.
2. Gold Explains Only Part of CPI; Holiday Services Consumption Was Another Clue
The market attributed the CPI upside surprise to “people buying gold and pushing prices up.” 李丰 estimates that explanation is “probably 40% to 50% right.” Precious metals did make a positive contribution, but meat, poultry, eggs and vegetables actually declined.
Travel, accommodation and transportation prices also rose, showing that holiday services consumption mattered as well. Subsidized categories such as communications equipment, office equipment and smart wearables also contributed positively to retail sales.
李翔 noted that per-capita daily spending during the National Day holiday fell. 李丰 stressed that “total spending rose, but spending per person per day fell” is not contradictory: the 8-day holiday pulled long-distance travel forward and concentrated short-distance trips in the middle of the break, while transportation costs were spread across more days. Many people also kept 1 day for rest.
After checking the historical data, 李丰 said 8-day National Day holidays occurred around 2017, 2020, 2023 and 2025. Except in 2023, per-capita daily spending typically declined year on year. Meanwhile, Beijing’s October retail sales shifted to growth in the teens, pulling the January–October figure back to about -3%; Shanghai’s first-10-month growth also returned to near 5%.
3. PPI Recovery Is Not Just About Gold; It Is Moving from Mines into the Midstream
Looking at commodities and third-quarter reports from listed companies, 李丰 observed that “it’s not just gold” that has improved over the past 1–2 months. Some mined products and upstream segments have strengthened, and industries closest to the upstream are beginning to benefit.
He also viewed Central Huijin’s purchases of midstream companies since the end of the second quarter alongside the improvement in midstream companies’ third-quarter results. His conclusion remained qualified: this could only be called a “possible recovery, or restoration process, starting from the bottom of the industrial chain.”
Past research shows that PPI tends to improve about 2 quarters after capital markets turn higher, with CPI turning after 2–3 quarters. That is a statistical average, not a clockwork rule. Counting from May, after the tariff shock, October marked roughly 6 months; the long holiday’s reduction in working days also created some negative pressure on PPI.
4. The Drop in Household Short-Term Lending Includes an Overlooked Regulatory Supply Shock
The common interpretation is that long-term household loans reflect property demand while short-term loans reflect consumer confidence: long-term loans were still growing in January–October, while short-term loans turned negative. 李丰 does not dismiss those factors, but says “people aren’t consuming” cannot explain the split between still-acceptable retail sales and a sharp drop in short-term lending.
The central bank issued industry guidance on internet-assisted lending in April, requiring strict enforcement from October of a loan-rate cap no higher than 24% annualized. The previous 24%–36% gray zone was often implemented through charges for membership cards, points cards or membership status.
Small-ticket credit offered by traffic platforms such as WeChat, Ant, Didi, Meituan, Ctrip and Douyin is funded primarily by banks. Assisted lending adds an intermediary layer and allows interest to accumulate to elevated levels. Borrowers paying above 24% are often already overleveraged or under cash-flow pressure.
李丰’s rebuttal best explains the data gap: “People with money aren’t consuming; people without money want to consume, but can no longer get loans.” Banning high-interest lending is necessary in itself, but it also means that funding chains dependent on repeated refinancing can break immediately.
5. The Assisted-Lending Industry Is Undergoing a RMB1T-Plus Contraction
On 李丰’s rough estimate, the largest integrated internet platforms originate more than RMB3T and nearly RMB4T in loans annually. Including standalone assisted-lending companies, he guesses the industry is at least RMB4T–RMB5T, possibly higher, while acknowledging that complete regulatory data is unavailable.
Platforms began cutting volumes within the third quarter after the guidance was published. From July and August, “borrowing new money to repay old money” became increasingly difficult, with deeper cuts at the end of the third quarter and in October. 李丰 estimates that the high-interest portion removed may account for about 20% of total volume. Policy-driven annual contraction could be at least RMB1T, possibly RMB1T–RMB2T; the contraction realized since July, especially during August–October, could be at least RMB1T, or even RMB2T–RMB3T.
He is not minimizing the impact: “I hope it won’t have much impact, but I know it actually has a significant impact.” The increase in collection texts offers a social window into old loans no longer being rolled over, with delinquent assets sold or outsourced for recovery.
Third-quarter reports from several U.S.-listed assisted-lending platforms already showed little year-on-year or quarter-on-quarter growth. At some companies, revenue and profit both declined, with profit falling especially sharply. Many related stocks doubled in 2024 and have since pulled back roughly 50%.
6. Credit Repair and Loan Extensions Are the Other Side of Cutting High-Interest Debt
Recent regulatory signals have two layers. Individuals who already have credit problems may enter the credit-system repair process; those temporarily unable to repay may receive a one-time deferral, extension or renegotiation rather than continuing to roll debt on high-interest platforms.
李丰 gave an example: a loan originally repaid monthly over 1 year could be extended to 3 years or longer, reducing monthly payments while lowering the rate cap. Assets more than 6 months overdue that are classified as bad debt and transferred to collection agencies need to be addressed through credit repair.
Taken together, the policy approach is to cap rates “at the top,” ease interim cash flow through extensions, and address the “bottom line” through credit repair. Deliberate multiple borrowing still exists, but borrowers facing genuine income shocks should not remain trapped in a high-interest cycle indefinitely.
October was the first month of strict enforcement. 李丰 expects platform business in the fourth quarter to be flat at best sequentially versus the third quarter, and possibly to decline further. Once all institutions realign below 24%, the old cycle should largely clear after 1–2 quarters, allowing domestic business to return to a normal track.
7. The 2024 Assisted-Lending Boom Explains the 2025 Sudden Brake
After the cleanup of internet finance eliminated many institutions, the surviving platforms became scarce. Banks were also encouraged after the pandemic to shift lending from property, manufacturing expansion and infrastructure toward personal consumption, but lacked customer acquisition, consumption data and risk-assessment capabilities. The intermediaries therefore became “scarce and important, and impossible to replace.”
Supply and demand both strengthened in 2024, making assisted lending a high-margin link that “both sides needed,” and producing “super-good” results at several platforms. 李翔 added that Douyin had accumulated transaction data after developing e-commerce from 2021; although it obtained its license relatively late, it was considered capable of entering the business. 李丰 said its related business once grew extremely quickly, roughly doubling.
This year’s regulatory objective is to address excessive rates, asset quality and bad-debt pressure, so the most profitable intermediary layer of last year has been sharply compressed. It is a “wave-like cycle”: cleanup creates concentration, policies encouraging consumption produce another boom, and accumulating risk eventually brings another tightening.
Payday-loan rates in the U.S. are higher and have long faced moral controversy. 李翔 asked whether the annualized rate could exceed 100%; 李丰 did not confirm that figure, saying only that “the interest is higher.” In China, the 24%–36% gray zone also involved high rates and revolving borrowing.
李翔 noted that revolving borrowers often “have no cash flow and need cash flow.” 李丰 distinguished between excessive forward consumption and people facing genuine pressure in daily life.
8. Flexible Employment Has Absorbed Workers While Amplifying Consumer-Finance Fragility
李丰 cited a claim that flexible employment now accounts for about 40% of China’s employment, but immediately noted that he had “not checked it” and could not confirm whether it was reliable statistics or an emotional estimate. If broadly accurate, it would show both a sharp increase in the services sector’s employment share over the past decade and continued pressure on stable incomes; the services sector has also served as a buffer for employment pressure.
He contrasted this with his experience researching WeWork in 2015. At the time, self-employed and part-time labor accounted for about half of the U.S. workforce, versus only the low teens in China. His team therefore concluded that China lacked an equally large base of individual users for shared offices and did not invest in the model.
Subsequent developments confirmed the distinction. U.S. demand came from communities and business collaboration among individual operators in design, advertising, law and other fields; Chinese shared offices more often became small offices serving companies with 3–10 employees. The growth of digital-service jobs in food delivery, courier work, ride-hailing and livestreaming has since absorbed some of the employment pressure.
9. Platform Wars Compressed Operating Profit and Helped Financial Businesses Emerge
李丰 says it is not mysterious that large platforms eventually develop financial businesses. Execution and operations businesses face fierce competition and low gross margins—“you get involved in mine, I get involved in yours”—while platforms already have traffic and transaction data, making finance an easier source of supplemental profit.
The food-delivery war from July to September hit two types of companies at once. Subsidies reduced the prices of freshly made tea drinks, hurting the profits of companies such as Luckin; after consumers could buy freshly made drinks for about RMB10, the share purchasing bottled RTD drinks priced around RMB5 also declined.
The capital-market outcomes for the two platforms differed. Alibaba rose from a lower base and had AI-related narrative support, so its stock performed relatively well; Meituan came under more pressure. This shows that the same subsidy war can have different marginal effects on an aggressor and an incumbent with an existing advantage.
李翔 was surprised that Alibaba had about RMB600B in cash and Tencent about RMB400B, while Alibaba’s leverage also appeared lower. 李丰 did not reach a firm conclusion, but guessed that Alibaba may have used different methods to issue corporate bonds directly after 蔡崇信 returned, potentially lifting its cash balance. He also cited Bilibili’s roughly $700M bond issuance as an example of direct financing.
10. Three-Year Time Deposits Maturing Are Helping Reactivate Deposits
Household deposits rose substantially in September, while deposits at non-bank financial institutions unexpectedly fell. 李丰 said he could not be “100% sure,” but one possibility was that public funds bought in size in 2021 returned to cost levels during August and September, prompting some holders to redeem.
A clearer chain began in 2022. Volatility in interest rates and bond prices pushed wealth-management products below net asset value, sending large amounts of money into 3-year time deposits whose yields were then attractive and sometimes close to or above those of 5-year deposits. Those deposits are maturing in concentrated volumes during the third and fourth quarters of this year.
The return on a new 3-year deposit is now far below what it was then, forcing the money to be reallocated. The maturity wave, combined with fund redemptions, may explain the acceleration in M1 growth in September and October and the clear narrowing of the M2–M1 “scissors spread,” reflecting cash reactivation.
By October, the month-on-month net decline in household deposits broadly matched an increase in non-bank financial deposits, restoring the familiar pattern of “deposits moving into wealth management.” That wealth management may include bonds, fixed income, equities, funds or insurance; it does not mean all the money entered stocks.
11. Direct Financing Has Reached Parity with Bank Lending in China’s Social Financing
October social financing appeared to show only moderate growth. Even including RMB500B of new government bonds, the main contribution still came from government debt. Household short-term loans fell, long-term loans grew only slightly, and both household and corporate loan categories faced year-on-year and sequential pressure.
After checking the figures, 李丰 found that direct financing accounted for about 44% of social financing in the first 9 months, while bank loans represented less than 50% in January–October. Bond and equity financing now account for “nearly half,” with a small amount of bill financing making up the remainder—hard to imagine when indirect bank financing dominated.
Direct financing means buyers subscribe directly to bonds or equities at market prices, rather than banks first taking deposits and then extending loans. Corporate default losses therefore fall on investors, and assets can be traded. China’s property dollar bonds, which rose from RMB1 at par to about RMB1.3 before falling to roughly RMB0.04, are an extreme example.
The current shift is being driven mainly by central and local government bonds. 李丰 guesses that government-related direct bond issuance could eventually reach more than RMB10T, with 80%–90% still consisting of government-related bonds; he also estimates total government bonds this year could reach about RMB13T. If capital-market efficiency and stability improve, corporate bonds, equities and REITs could gain share. “Looking only at loans” can no longer fully capture the scale or efficiency of funding entering the real economy.
12. Repeated Foreign-Currency Sovereign Issuance Has Established a Rare Baseline for Chinese Assets
China issued $4B of sovereign dollar bonds in Hong Kong, split between 3-year and 5-year tranches, drawing about $118B in orders—nearly 30x the offering and, by its own description, a record subscription multiple. Based on 李丰’s recollection, the two issuance yields were about 3.64% and 3.78%.
On Nov. 18, China issued another €4B of sovereign bonds in Europe, split evenly between 4-year and 7-year maturities. The yields were about 0.3 percentage points above the average level for comparable euro-denominated bonds in Europe. Including the roughly $2B issued earlier in the Middle East, the three offerings total about $10B.
For a country with ample foreign-exchange reserves and a history of relying mainly on domestic debt, this amount does not mean China needs foreign debt to solve an FX-funding problem. What matters is the issuance frequency, geographic coverage and pricing function: Hong Kong, the Middle East and Europe are becoming reference benchmarks for future offshore financing by Chinese companies.
13. Near-30x Subscription Reflects Global Long Money, Not Offshore Retail Deposits
Some attributed the Hong Kong demand to dollar deposits held by offshore residents. 李丰 directly rejected that view: roughly two-thirds of subscriptions came from insurers, banks and sovereign funds, while asset managers accounted for less than one-third. Investors were roughly half Asian and half non-Asian by geography.
More important was the price. The 5-year Chinese dollar bond priced only about 2bp above a comparable U.S. Treasury, while the 3-year tranche was broadly in line with Treasuries. Since Treasuries are the pricing foundation for dollar assets, subscribers were demanding almost no additional sovereign-risk premium.
李丰 used corporate bonds as a reference. Google might need to pay only about 1 percentage point above Treasuries, while a high-spending, pre-profit entity such as xAI/Grok might pay more than 10%. China’s sovereign bond paying only 2bp more shows that major international investors price its creditworthiness very close to that of U.S. Treasuries.
The signal also supports the internationalization of the renminbi and Chinese assets, Hong Kong’s offshore financial-center role and overseas bond issuance by Chinese companies. 李丰 linked it to the global shortage of assets after funds pulled back from some high-level, high-valuation U.S. equities. A small news item that appears to have “nothing to do with individuals” is actually a price signal for global allocation shifts.
14. AI Capex Is Shifting from an Equity Narrative to a Debt Test
Nvidia released earnings above expectations that day, and its after-hours price rose sharply again, approaching $200. 李丰, however, was more interested in the financing structure supporting demand than in the stock’s reaction to the earnings report.
Based on the figures he cited, major U.S.-listed AI companies have issued nearly or more than $200B in bonds this year to finance data centers and related infrastructure. The market expects next year’s figure to reach more than $1T. Google, Meta, Oracle and xAI/Grok are all raising funds, and the total has already exceeded related equity fundraising.
Data centers being built today may be about 90% GPU-based. Most bonds have maturities longer than 3 years, with some at 1 year. The investment case depends on utilization and gross margins over the next 3–5 years covering annualized financing costs of roughly 5%–12%, as well as chip depreciation.
The positive loop is that the more companies “tell the AI story,” the higher their valuations may rise and the easier it becomes to issue bonds, which are then used to buy GPUs and support related results. 李丰’s question is whether this becomes a “loop with themselves,” ultimately still requiring validation by real cash flow.
15. Off-Balance-Sheet Data-Center Financing Hides Risk Without Eliminating It
Continuous issuance by a single entity would lift its leverage, so technology companies are designing structures involving project companies, fund ownership and long-term leases that keep data-center debt from appearing directly on their own balance sheets.
李丰 gave a hypothetical structure involving Meta. A project might begin with a fund of about $3B, with Meta contributing 20% and other investors 80%, followed by roughly $12B of debt issued by the project entity. Meta could sign a 5-year lease or guarantee, leaving the debt at a project company in which it holds a minority stake while retaining the long-term expenditure.
Similar arrangements involve large institutions first buying the asset, followed by actual users such as Amazon signing long-term sale-and-leaseback commitments. 李丰 said many “strange and varied workarounds” have appeared recently. Their essence is to get the debt issued without putting the full scale of the liabilities directly on technology companies’ balance sheets.
The risk has not disappeared. AI efficiency, data-center utilization, GPU depreciation, chip iteration speed, and whether TPUs, inference chips or edge chips can alter the GPU share could all expose debt far larger than the equity financing if assumptions deviate materially.
16. A-Share Rotation from High to Low Explains Both the Cooling Index and the Lack of Broad Gains
New public-fund performance constraints require benchmark assessments to focus on fund type and main holding sectors. A healthcare fund can no longer win rankings by concentrating in technology stocks. 李丰’s “wealth code” is that the constraint may push money from high-valuation crowded sectors back into industries it did not previously need to hold.
The active-management logic is the same. China’s economy cannot rely on technology alone, and manufacturing and technology’s replacement of property and infrastructure is only “halfway there.” The destination is not poor companies or speculative small caps, but bringing sector exposure closer to the structure of China’s real economy.
CPI and PPI slightly beating expectations give investors an internal reason to allocate to low-valuation sectors such as consumption and industry. But “selling something for RMB20 and buying something for RMB2” does not necessarily lift the index; it may instead leave the market volatile or lower during the rotation.
The main beneficiaries of the move from roughly 2,500 to 4,000 were concentrated in two crowded groups: technology and AI chips favored by public funds, and banks, power, infrastructure and parts of the upstream sector favored by insurers. That is why 李丰 answered 李翔’s question about the wealth effect by saying balanced allocation “does not make it easy to earn money,” while aggressive allocation “makes it relatively easy.”