China's financial clean-up enhances quality, U.S. Treasury yields under pressure

On Wednesday (2026-09-09) U.S. large-cap stocks softened slightly, with SPY down 0.46% and QQQ down 0.29%; but looking at the past 5 trading days, SPY is still up 0.08% and QQQ up 1.23%, which indicates the indices haven't collapsed and capital is moving while picking directions. ETF Radar today feels more like a reminder: calm on the surface, but growing divergence underwater.

The thread that ties things together today centers on China’s financial system 'cleaning out' and the capital market 'upgrading quality' moving forward at the same time. Start with the banking side. Multiple listed banks clearly increased write-offs of nonperforming loans in the first half of the year: 42 listed banks cumulatively wrote off and disposed of about 6517 100-million-yuan, at historically high levels. The term 'write-off', in plain terms, means removing bad debts that are very difficult to recover from the books as quickly as possible. Imagine a house full of old junk—unless you clear it out first, there’s no room for new things. Banks are the same: if bad loans aren't cleared, new lending, capital usage, and risk assessment all get held back.

Why does this matter? Because a stable-looking nonperforming loan ratio on the books does not mean the risk has actually disappeared. Increasing write-offs now is equivalent to admitting old problems must be dealt with, and that the processing speed is accelerating. The market transmission is that financial institutions' reports become cleaner, capital maneuvering room expands, and concerns about 'whether existing risks will continue to fester' decline. Put bluntly, if you first move the stalled bad issues aside, subsequent lending, mergers and acquisitions, and asset pricing all get a bit easier.

Next, look at the capital market side. The CSRC mentioned at a press conference today that, since the 2024 'Six M&A Measures' were issued, listed companies have cumulatively disclosed 370 major asset restructurings; at the same time, more than 2000 companies have paid dividends for 5 consecutive years. M&A and restructuring, in plain terms, means companies buying other companies, selling non-core assets, or reorganizing businesses to become more profitable. Consecutive dividends, in plain terms, mean the company is not just telling stories but is also willing to return real cash to shareholders. Putting these two signals together is clear: on one hand, allow poor assets and inefficient businesses to exit; on the other hand, encourage more standardized companies that are willing to reward shareholders to stay.

Thus, the preceding banking 'cleaning out' connects with the capital market 'upgrading quality'. Faster clearing of bank risks makes the financial system’s foundation more stable; more active M&A and dividends in the capital market enhance the perceived quality of listed companies. At the sentiment level, capital is more likely to shift from 'fear of stepping on landmines' to 'willingness to reassign valuations'. Valuation, in plain terms, is how high the market is willing to tag a company's price. With less worry about risk, price tags are easier to revise upward.

But we should also see a reality: today’s main news is mostly happening in the China market, while U.S. large caps fell on Wednesday. That is not contradictory. The story in China is about medium-term repair, while U.S. markets are still digesting a higher interest rate environment in the short term. Higher rates, in plain terms, mean borrowing costs are more expensive. When borrowing is expensive, stocks—especially small companies and long-duration assets—feel more pressure. Duration, in plain terms, is how sensitive an asset is to interest rate changes; assets with lots of distant cash flows are more sensitive to rates. So what we see today is local policy improvements, but global pricing hasn't fully synchronized yet.

There is also a side thread that contrasts with the main line. In U.S. real estate, leading developers are already making budgets for 2027, and the market is focused on pricing pressure next year. Pricing pressure, in plain terms, means houses aren’t as easy to sell at ideal prices. If U.S. developers are preparing for price cuts, promotions, and thinner profits, that indicates the transmission of high rates is still ongoing: mortgage costs are up, buyers are more hesitant, and developers’ bargaining power weakens. That is exactly two different stages compared with China’s financial 'cleaning out' and capital market 'upgrading': one is repairing confidence, the other is still bearing high-cost pressure.

Now look at ETFs related to today’s main line. First FXI. FXI actually fell 2.24% over the past 5 trading days. That is the price that has already happened. Looking forward, the ETF Radar model is mildly positive on FXI for the next 1 to 3 months, but the signal is moderate. Mildly positive here, simply put, means the system thinks there is room for improvement ahead, but the evidence is not one-sided. The main logic supporting it is not whether the stock price dropped today, but China’s financial regulatory thread: the Ministry of Finance plans to issue 3000 100-million-yuan special treasury bonds to replenish core Tier 1 capital for central financial enterprises; the central bank also conducted 5000 100-million-yuan, 89-day outright reverse repos. A reverse repo, in plain terms, is the central bank releasing liquidity first to make the financial system a bit more relaxed. Core Tier 1 capital, in plain terms, is a bank’s hardest 'safety cushion'. Together, these two measures relieve pressure on financial institutions and reduce risk premia. Risk premium, in plain terms, is the extra return investors demand for taking on risk. When required returns fall, asset prices usually have an easier time rising. To be honest, FXI has still been falling over the past 5 trading days, which diverges from the model’s mildly positive view; that’s because the model is looking at the next 1 to 3 months, and short-term sentiment has not yet fully priced in the policies.

Second, KWEB. KWEB actually fell 3.54% over the past 5 trading days. ETF Radar is mildly positive on KWEB for the future, but the signal is moderate. This line has real transmission from today’s main thread because internet platforms and growth sectors are very sensitive to changes in financial conditions. Financial conditions, in plain terms, mean whether money in the market is tight or financing is difficult. The accelerated bank write-offs and capital market quality improvements in today’s main thread, combined with the aforementioned capital replenishment and liquidity injections, will improve the market’s overall view of Chinese risky assets. In KWEB’s systematic basis, the price has already partly reflected this benefit, so the signal is not very strong. In other words, policy is moving toward repair, but stock prices are still under short-term pressure; that divergence itself is information: the news is new and capital is still confirming.

Third, LQD. This one is not directly related to today’s China-focused main line; it is the system’s independent medium-term stance. LQD actually rose 0.09% over the past 5 trading days. But for the outlook, the model reading points to LQD being mildly negative, with multiple pieces of evidence aligned. LQD is U.S. investment-grade corporate bonds, in plain terms, bonds issued by relatively creditworthy large companies. Why is the system biased negative? First, since 2026-09-02 there were net redemptions of -$2,518M, equivalent to about -8.52% AUM. AUM, in plain terms, is the total assets this fund manages; net redemptions this large mean capital is continuously leaving. Second, 30-day cumulative outflows are -9.79% AUM, indicating this is not a one-off sentiment but sustained outflows. Third, U.S. September nonfarm payrolls added 162k versus expectations of 55k. Nonfarm payrolls, in plain terms, are U.S. employment data. Employment that is too strong makes the market think: the Fed may not cut rates quickly, and rates will stay higher for longer. High rates usually pressure existing bond prices, and corporate bonds are not immune. There is divergence here too: LQD rose slightly over the past 5 trading days, but the model is negative. The reason is simple—price has not yet fully reflected the outflows and high-rate pressure in the short term.

Fourth, IEF. This is also unrelated to today’s main line and is ETF Radar’s independent rate view. IEF actually fell 0.22% over the past 5 trading days. For the outlook, the ETF Radar model is negative on IEF, and the signal is fairly consistent. IEF is a medium-term U.S. Treasury fund. The quantitative evidence has three points. First, the U.S. 10-year Treasury yield is about 4.8%. Yield, in plain terms, is the return a bond gives; when yield goes up, bond prices usually go down. Second, the term premium is 0.8892%. Term premium, in plain terms, is the extra compensation investors demand for lending longer-term money to the government. This value is relatively high, indicating the market demands more compensation for the long end of the curve. Third, CFTC positioning shows 10-year Treasury net longs at -2,062,502 contracts, and this sits at the 100th percentile of its own range. CFTC positioning, in plain terms, looks at which side large players are betting on in the futures market; this extreme net short implies many funds are betting on yields rising, i.e., bond prices weakening. IEF has already been falling over the past 5 days, which aligns with the model’s negative view and the short-term move somewhat validates the system reading.

Fifth, IWM. IWM actually only rose 0.02% over the past 5 trading days, essentially flat. Looking forward, data points to IWM being mildly negative, and the signal is fairly consistent. IWM represents U.S. small-cap stocks, which can be understood as smaller companies with weaker resilience. Evidence shows CFTC Russell 2000 net positioning at -109,499 contracts, indicating large players in futures markets are biased toward betting on declines; IWM has had net flows of -$663M since 2026-09-02, i.e., -0.83% AUM, meaning capital is withdrawing; plus WALCL−TGA−RRP four-week change is -$72B. This indicator is a bit awkward—plainly, it’s a thermometer the system uses to roughly look at total market liquidity, and it’s marginally tightening. Marginal tightening, in plain terms, means the faucet isn’t fully shut but the flow has decreased. Small companies fear a tightening financing environment the most, so this line is consistent with the 'high rates pressuring small caps' logic.

Putting today’s news and these ETFs together, the picture becomes clear: the new messages from China emphasize risk clearing and capital market quality improvement; on the U.S. side, the system’s medium-term concern remains that high rates have not fully eased. One side is repairing confidence, the other is still enduring funding cost pressure. Short-term prices and medium-term judgments diverge, which is not surprising—instead, it reminds us the market is in a transition stage.

That’s it for today: the main news thread points to China’s financial system 'cleaning out' plus capital market 'upgrading quality', but global asset pricing is still being pulled by high rates. A reminder: the above is the system quantitative model’s directional forecast for general markets and does not constitute investment advice tailored to you; please consult a licensed investment advisor before investing.