Global long-end yields have surged in tandem since August, injecting fresh volatility into broad asset classes. Yet, comparing the two instances since Q2 where the 10-year U.S. Treasury yield climbed to the 4.6-4.8% range, market pricing has diverged sharply. The first episode in mid-to-late May saw U.S. and A-share markets dip briefly before resuming their uptrend; the second, spanning from mid-July to now, has left U.S. and A-shares, particularly tech assets, trading on the back foot.

The key divergence lies in the shifting industry news backdrop accompanying each rate spike. In May and June, large model ARR was accelerating; by July, the industry narrative had deteriorated, marked by weakening capex stories, slowing ARR growth, and debt issuance pressures, which together contracted both earnings expectations and risk appetite. Since July, the core driver of heightened tech volatility has been the concentrated emergence of structural concerns within the AI industry. These include: (1) the infrastructure chain, where cloud provider capex growth is decelerating and the second derivative of AI infrastructure is turning negative, signaling a phase of slowing and differentiation for computing infrastructure components like optical modules and PCBs; (2) the revenue side, where top-tier large model ARR has missed expectations due to factors such as price wars among cloud providers and cost reductions from hardware efficiency gains and model architecture optimization; (3) the financing side, where mega-scale debt issuance competes with U.S. Treasury supply for long-duration capital, pushing up term premiums and, in turn, raising the discount cost for companies' own capex, creating a positive feedback loop between rates and capital expenditure. The AI industry is transitioning from aggregate growth to structural differentiation, making the sustainability of earnings delivery the dominant variable for future pricing. So, how should we objectively assess the impact of rising rates on asset prices?

Is "higher rates hurting growth stocks" mostly wrong?

From an analytical framework perspective, the impact of rising rates on asset prices is not a single-function outcome. Its direction and magnitude primarily hinge on two factors: first, the source driving the rate increase—including expectations for short-term real rates (influenced by growth or monetary policy paths), inflation expectations, and term premiums (from deficits, issuance structures, and quantitative tightening); and second, the slope of the industrial trend on the earnings side. When earnings improvements coincide with a rising discount rate, the net effect depends on the relative speed of the two. Without specific scenario assumptions, directly asserting that "rising rates hurt growth stocks" lacks practical relevance.

Domestic yields and A-shares: a weak pricing link for tech

Domestic interest rates exert almost no directional influence on domestic tech stocks, primarily because the capital flows, financing channels, and pricing logic in A-shares do not create a transmission mechanism. (1) In terms of capital, the marginal pricing funds for domestic tech are private equity, margin financing, and foreign capital, which have low correlation with domestic risk-free rates. (2) On the financing side, A-share tech companies rely on equity refinancing rather than debt, so interest rate levels do not dictate their capex pace. (3) In pricing logic, domestic tech stocks are essentially cyclical growth plays, with earnings fluctuations determined by industry cycles. Therefore, during the period when the 10-year Chinese government bond yield fell from 2.8% to 1.6%, it was dividend-paying assets that saw valuation recovery. Tech stocks neither enjoyed a valuation re-rating from lower rates nor suffered systematic de-rating from any rate rebound. The inflection points for domestic tech stocks are tied to industry cycles, not domestic rates.

U.S. yields and U.S. stocks: rising rates don't hit tech hardest

The influence of U.S. Treasury yields on U.S. stocks is not straightforward. Since the 1990s, the correlation coefficient between the Dow Jones and the 10-year U.S. Treasury yield is -0.41, while for the Philadelphia Semiconductor Index it's only -0.13. In contrast, their correlations with year-over-year semiconductor sales growth are 0.20 and 0.40, respectively. Additionally, from the 1990s into the 2010s, rates trended down while indices trended up, meaning the negative correlation contains substantial trend co-linearity. This shows that inflation and industry prosperity have more explanatory power than interest rates. The Philadelphia Semiconductor Index's pricing is mainly driven by industry cycles. Each period of negative semiconductor sales growth corresponds to a pullback, while recoveries align with index uptrends. From 1994-2020, rates fell from 8% to 0.5%, then rebounded to 4.5% from 2021 onward; the index hit new highs in both phases. Rates impact the discount rate, while industry prosperity determines earnings. When earnings improve faster than the discount rate rises, valuations can still expand. Across nine semiconductor cycles since the 1990s, the Philadelphia index's cyclical fluctuations have tracked semiconductor sales cycles, with cycle troughs leading semiconductor sales by an average of 2 months and peaks lagging by 3 months. Interestingly, contradicting duration intuition, the Dow's correlation with rates is higher than the semiconductor index. This may be because the Dow has greater weight in financials and industrials, which are sensitive to rates and the economic cycle, whereas the fluctuation in tech sector earnings expectations is enough to offset discount rate changes. Also, during rate decline phases, long-duration blue chips in the Dow, which are more sensitive to discount rates, see their valuation centers repaired.

Since 2023, the U.S. market has seen several corrections, with liquidity acting as a trigger but not altering the trend. During liquidity shocks, markets undergo staged pullbacks, but once the liquidity issue resolves, the market returns to its prior uptrend. In the U.S. market's uptrend since early 2023, there have been six corrections. The first three were mainly triggered by liquidity shocks, including the Silicon Valley Bank event in March 2023, the Q3 2023 fiscal bond issuance, and the August 2024 yen carry trade unwind. The latter three involved liquidity combined with macro disruptions, tariffs, and geopolitical shocks. However, since the AI industry trend remained upward, the market's direction stayed intact. Even during rate hike cycles, industrial tracks with upward prosperity can still see earnings-driven gains. For instance, during past U.S. rate hike cycles, the semiconductor index saw PE compression, but the index itself mostly continued to rise.

For A-shares, U.S. yields' short-term shock fades, returning to trend pricing

First, the impact of U.S. Treasury yields on A-share asset pricing can be divided into long-term and short-term effects. (1) A sudden, rapid rise in U.S. yields triggers a risk-off mode for most equity assets, leading to noticeable corrections in assets that have already seen large gains, high valuations, or crowded trades—as seen in February 2018, February 2021, and January 2022, when the most affected A-share assets were large financials, the "Mao Index," and the "Ning Combination," respectively. (2) A medium-to-long-term shift in the U.S. yield center mainly impacts core assets with stable long-duration earnings suitable for DCF valuation, such as Hong Kong internet stocks, U.S. tech giants, and the top 50 A-share consumer stocks. However, for A-share industries or companies currently experiencing an explosion in growth or rapidly rising penetration rates, the impact is minimal, as most of these companies cannot be valued via DCF. Thus, it's incorrect to broadly conclude that "rising rates are bad for growth stocks." The conclusions differ significantly depending on the time horizon and the type of growth stock.

Specific cases illustrate the impact of U.S. yields on the "Ning Combination" and "Mao Index." (1) Over a longer horizon from 2020 to early 2022, the Ning Combination trended upward even as U.S. yields trended higher, confirming that U.S. yields have little impact on industries in the midst of an industrial explosion or rapid penetration growth. (2) In the short term, during the rapid U.S. yield spikes in February 2021 and January 2022, the Ning Combination saw significant declines, confirming that assets with large gains, high valuations, and crowded trades are vulnerable to short-term U.S. yield surges. (3) After the shocks, the Ning Combination, backed by verifiable industry trends, reached new highs, while the Mao Index, with weakening earnings, entered a downtrend.

Furthermore, U.S. Treasury yields for A-shares are better at "adding flowers to brocade" than "providing fuel in snowy weather." One transmission path for U.S. yields is valuation repair, but they are hard-pressed to dictate the trend in asset prices, which is fundamentally determined by fundamentals. For instance, from November 2023 to January 2024, U.S. yields fell over 100 basis points, which should have boosted risk appetite globally. Yet, during this period, A-share earnings expectations continued to deteriorate (PMI stayed below 50 and declined further), making weak fundamentals the core pricing factor, and the market struggled to gain traction. Taking the new energy industry as an example: (1) In 2020-21, rising new energy earnings coincided with rising U.S. yields; fundamentals drove pricing, making the sector insensitive to U.S. yields. (2) From 2022 to Q3 2023, declining new energy earnings met rising U.S. yields, resulting in a "double kill" of both earnings and valuations. (3) From Q4 2023 to 2024, new energy earnings declined while U.S. yields rose; again, fundamentals set the price, and the sector showed insensitivity to U.S. yields. The one "adding flowers" phase occurred in Q2-Q3 2021, when rising new energy earnings aligned with a temporary pullback in U.S. yields.

Summary: Four scenarios for U.S. yields pricing A-share industry trends

For A-shares, in markets with a clear industrial cycle trend, the industry cycle acts as the numerator and U.S. Treasury yields as the denominator. The four combinations of their directions correspond to four distinct market paths. (1) Same-direction combinations (scenarios 1 and 3) are priced by industry fundamentals; U.S. yields typically only affect the slope and short-term volatility, not the direction. Even with a higher U.S. yield center in 2026, the AI chain's trajectory hasn't been directly reversed. (2) Opposite-direction combinations (scenarios 2 and 4) involve resonance between the numerator and denominator, leading to either a "double kill" or a "double win." Scenario 2 pairs rising rates with a declining industry, putting simultaneous pressure on stock prices—seen in new energy during 2022 to Q3 2023 and global tech in 2022. Scenario 4 pairs falling rates with a rising industry, lifting stock prices on both fronts—as seen in the AI industry from Q1 2023 to Q3 2024 and into H2 2025. Therefore, the first principle for judging the main industry line is determining where we are in the earnings cycle. (1) If the industry trend is upward, rapid U.S. yield increases may cause temporary shocks but typically won't alter the upward direction. (2) Conversely, if the industry trend is downward, even a pullback in U.S. yields might provide a temporary recovery but will struggle to reverse the downtrend. (3) From a pricing perspective, a trending rise in U.S. yields hits long-duration assets hardest, including blue chips with stable cash flow distribution and unprofitable, high-valuation tech with cash flows far in the future. Assets in the midst of an industrial boom, however, are less affected by valuation pricing shocks.

Key changes this week

Unless otherwise specified, data sources for this section are from Wind.

Mid-stream industries

In property, cumulative transaction area across 30 major cities fell 6.68% year-on-year; month-on-month transactions fell 5.64% in the latest period, with a year-on-year decline of 4.00% and a week-on-week rise of 18.55%. National Bureau of Statistics data shows new construction starts from January to July totaled 267 million square meters, down 24.00% year-on-year, with the pace slowing by 0.60 percentage points from the January-June figure. July alone saw 35 million square meters of starts, down 28.51% year-on-year. Real estate development investment from January to July was 4,300.895 billion yuan, a nominal 19.20% decline year-on-year, with the decline widening by 1.20 percentage points from January-June; July alone saw investment drop 28.70% nominally year-on-year. Commercial housing sales area from January to July reached 450.21 million square meters, down 11.80% year-on-year, with the pace slowing by 0.20 percentage points; July sales area fell 14.50% year-on-year.

In autos, from August 1-16, national passenger car retail was 628,000 units, down 22% year-on-year and flat month-on-month; cumulative retail for the year stands at 10.801 million units, down 20% year-on-year. Wholesale from August 1-16 was 583,000 units, down 25% year-on-year and 1% month-on-month; cumulative wholesale is 15.393 million units, down 6% year-on-year. In new energy vehicles, retail from August 1-16 was 399,000 units, down 15% year-on-year and up 1% month-on-month; cumulative retail is 6.067 million units, down 13% year-on-year. New energy wholesale from August 1-16 was 425,000 units, down 3% year-on-year and up 6% month-on-month; cumulative wholesale is 8.673 million units, up 7% year-on-year.

Mid-stream manufacturing

In steel, rebar spot prices rose 2.57% week-on-week to 3,156.00 yuan/ton, while stainless steel spot fell 1.18% to 15,046.00 yuan/ton. As of August 21, rebar futures closed at 3,039 yuan/ton, up 0.80% from last week. Steel network data shows daily average output from key steel enterprises in early August was 1.819 million tons, down 8.22% from late July. Cumulative crude steel output in July was 577.039 million tons, down 3.10% year-on-year. In chemicals, as of August 10, methanol prices rose 2.31% from July 31 to 2,479.40 yuan/ton, and butadiene rubber rose 7.32% to 12,572.20 yuan/ton.

Upstream resources

In international commodities, WTI rose 5.66% this week to $87.06, Brent rose 5.38% to $93.60, the LME metal price index rose 1.00% to 5,822.70, the CRB commodity index rose 3.79% to 406.24, and the BDI fell 0.77% to 2,841.00. In iron ore and coal, this week saw iron ore inventories decline and coal prices rise. The Qinhuangdao Shanxi premium blend flat price at 5,500 kcal rose 2.38% to 858.60 yuan/ton as of August 17, 2026; port iron ore inventories fell 0.46% this week to 165.44 million tons; raw coal output in July fell 9.89% to 343.206 million tons.

Market characteristics

For stock market movements, the Shanghai Composite Index fell 0.56% this week. Top gaining sectors were Oil & Petrochemical (up 5.44%), Nonferrous Metals (up 2.50%), and Banks (up 2.40%). The worst performers were Media (down 5.54%), Computer (down 4.95%), and Conglomerates (down 3.97%). In dynamic valuations, as of August 21, the A-share overall PE (TTM) fell from 22.40 times last week to 21.91 times, and PB (LF) fell from 1.82 to 1.80. Excluding financials, the A-share PE (TTM) fell from 38.21 to 37.30 times, with PB (LF) down from 2.56 to 2.52. The ChiNext board PE (TTM) fell from 66.52 to 62.62 times, with PB (LF) down from 4.49 to 4.36. The STAR Market PE (TTM) fell from 189.35 to 170.64 times, with PB (LF) down from 6.52 to 6.25. The CSI 300 PE (TTM) fell from 13.97 to 13.83 times, with PB (LF) at 1.37. From an industry perspective, the sectors with the largest expansion in PE (TTM) percentile this week were Agriculture, Forestry, Animal Husbandry & Fishery, Banks, and Transportation. The largest contractions were in Media, Power Equipment, and Beauty Care. Additionally, from a PE perspective, among Shenwan primary industries, Communications, Nonferrous Metals, Commerce & Retail, Real Estate, Non-bank Finance, Beauty Care, Food & Beverage, Oil & Petrochemical, Transportation, Utilities, Basic Chemicals, Environmental Protection, and Social Services are valued below the historical median. Electronics, Building Materials, Light Manufacturing, and Agriculture, Forestry, Animal Husbandry & Fishery are valued above the 90th historical percentile. From a PB perspective, among Shenwan primary industries, Basic Chemicals, Steel, Building Materials, Power Equipment, Building Decoration, Utilities, Transportation, Real Estate, Environmental Protection, Beauty Care, Social Services, Autos, Home Appliances, Light Manufacturing, Textiles & Apparel, Commerce & Retail, Agriculture, Food & Beverage, Pharmaceuticals & Biotech, Computer, Media, Banks, and Non-bank Finance are valued below the historical median. Electronics is valued above the 90th historical percentile. This week, the equity risk premium rose from 0.92% to 1.00%, and the stock market earnings yield rose from 2.62% to 2.68%. As of Thursday, August 20, margin balance stood at 2,666.346 billion yuan, down 0.28% from last week. The A/H premium index fell to 120.96 this week, down from 123.27 last week.

Liquidity

From August 17 to August 21, the central bank saw 9 reverse repos mature, totaling 1,729.6 billion yuan, and conducted 8 reverse repos, totaling 1,457.6 billion yuan. Net liquidity withdrawal from open market operations (including treasury cash) was 202.0 billion yuan. As of August 21, 2026, R007 rose 0.49 basis points week-on-week to 1.4420%, SHIBOR overnight rose 2.10 basis points to 1.4340%, the term spread fell 1.11 basis points to 0.5893%, and the credit spread rose 1.00 basis point to 0.2998%.

Overseas markets

In the U.S., Wednesday saw crude oil inventory data for August 14 at 3,438 thousand barrels, versus a prior value of 2,217. Thursday's initial jobless claims for August 15 were 206,000, versus a prior 212,000. In Japan, Monday's Q2 GDP year-on-year (initial) rose 0.20%, against a prior 0.50%; June industrial production month-on-month (revised) rose 1.85%, against a prior 0.10%. Friday's July CPI year-on-year rose 0.30%, against a prior 1.60%; July core CPI month-on-month rose 0.20%, flat against prior. The U.K. had no major data releases. In the Eurozone, Wednesday saw July CPI year-on-year rise 0.10%, against a prior 2.80%; July EU core CPI year-on-year rose 0.10%, against a prior 2.60%. The S&P 500 fell 1.43% last week to close at 7,674.37; the London FTSE rose 0.62% to 10,816.56; the German DAX fell 1.15% to 26,136.56; the Nikkei 225 fell 3.93% to 66,016.36; the Hang Seng rose 3.55% to 26,009.46.

Macro data

Monday's July electricity production was down 0.10% year-on-year; July industrial value-added rose 4.50% year-on-year; July fixed asset investment completed value fell 6.70% year-on-year cumulatively. Tuesday's July exports rose 23.90% year-on-year, and imports rose 27.60% year-on-year. Friday's July public fiscal revenue rose 11.72% year-on-year, while public fiscal expenditure rose 0.48% year-on-year. Wednesday's July total planned investment in new projects under fixed asset investment fell 21.60% year-on-year cumulatively.

Data calendar for next week

Key upcoming data: Taiwan July unemployment rate, Taiwan July overnight lending rate, Taiwan July benchmark lending rate, Hong Kong July import value and overall export value year-on-year, U.S. August 21 crude oil inventories, U.S. Q2 real GDP annualized quarter-on-quarter (estimate), U.S. August 22 initial jobless claims, Japan July unemployment rate, and Japan July job-to-applicant ratio. Monday, August 24: Taiwan July unemployment rate, overnight lending rate, and benchmark lending rate. Tuesday, August 25: Hong Kong July import and export values year-on-year. Wednesday, August 26: U.S. August 21 crude oil inventories and Q2 real GDP estimate. Thursday, August 27: U.S. August 22 initial jobless claims. Friday, August 28: Japan July unemployment rate and job-to-applicant ratio.

Risk warnings

Geopolitical conflicts may exceed expectations, intensifying upward pressure on global inflation. Overseas inflation and U.S. economic resilience could push global liquidity into a faster tightening cycle. Domestic pro-growth policy support may fall short of expectations, leading to a sluggish economic recovery and a decline in market risk appetite.

Report release date: August 23, 2026.

Analysts: Liu Chenming (SAC: S0260524020001), Zheng Kai (SAC: S0260515090004), Li Rujuan (SAC: S0260524030002), Yang Qingyuan (SAC: S0260525080001).

Disclaimer: This content is for reference only and does not constitute investment advice. Investors should operate at their own risk.