Core Viewpoints
◾ September Institutional Behavior Pattern: Fund Heavy Position Bond Sell-off Accompanied Market Development, Most Bond Varieties Declined in September: 1) Long-end government bonds retreated more than short-end, yield curve steepened, long-end spreads widened, 5Y-3Y spread narrowed. 2) Tier-2 capital bonds and perpetual bonds led the decline in long-end, 5-7Y credit spreads rapidly widened. These maturities and varieties are basically fund-preferred heavy positions. From spot trading perspective, except for short-term commercial paper, funds preferred net purchases of 7-10Y government bonds, 20-30Y treasury bonds, 1-5Y medium-term notes, and 7-10Y tier-2 capital and perpetual bonds since 2024. Compared to Q4 2022 adjustment, all showed tier-2 capital and perpetual bonds leading decline characteristics, but in this round of "fund heavy position bond sell-off," short-end decline was limited, with curve steepening more significant. Trading behavior confirms September fund selling concentrated on tier-2 capital bonds, perpetual bonds, 10Y policy bank old bonds, and ultra-long treasury old bonds. Currently, fund selling has accelerated, but overall still has considerable clearing space, with fund presence still visible in ultra-long active bonds.
◾ Despite Weakening Commodity Demand, Inflation Expectations Remain Strong: If PPI transmits to core CPI, based on historical elasticity data, core CPI year-over-year may reach 1.6% by March next year. Even if 10Y rates rise to 2%, real interest rates would be 0.4%, further breaking current historical lows (0.8%). If nominal rates remain unchanged, it means real rates would decline 69.4BP over the next six months; even if 10Y rates rise to 2%, real rates would still decline 44BP. Re-inflation effects approach 2024 rate cuts: 2024 full-year rate cuts of 60BP drove 10Y rate center and real rates down 70.6BP. Travel and consumption continue improving, scar effects keep healing, RBC business cycle recovering. Ministry of Culture and Tourism data shows this year's National Day and Mid-Autumn Festival 8-day holiday saw tourism trips up 16.1% year-over-year, domestic tourism spending up 13.5%. Daily average trips up 1.6%, consumption up 1%. Every holiday this year has shown recovery. Tourism trips and revenue have already surpassed historical highs. Through October, 2025 tourism trips (+10.4%) and revenue (+3.1%) have both exceeded 2019 levels.
◾ Q4 Bond Market Focus: Valuation, Institutional Behavior, and Tariff Re-trading: EVA valuation perspective shows 30Y treasury bonds regain mortgage rate comparison advantage, 10Y relatively neutral. From bank EVA calculation perspective: As of end-September, 30Y treasury after-tax EVA level was 2.15%, even considering VAT recovery on interest income, compared to existing mortgage rates of 1.71%, already has high comparison advantage, with spread widening to 79% historical percentile since 2015; but 10Y treasury EVA spread versus general loans only recovered to 24% historical percentile level, comparison advantage relatively neutral. Even with weak fundamentals, from real interest rate perspective, rate cut necessity may not be high. Core CPI has shown continuous recovery this year, reaching 0.9% year-over-year growth in August. Considering core CPI and current treasury rates, compared to last year's real rate levels may not be low. Rate cuts based on other considerations may not necessarily benefit bonds. Current IRS implied rate cut expectations are low. Bank behavior shows current market risk appetite is high, insurance bond allocation growth may be weaker than usual. Insurance traditionally has advance bond allocation demand in Q4 for "opening red" considerations, but this year's liability-side premium growth is limited, asset-side bond cost-effectiveness is low. This year's insurance seasonal bond allocation peak was in Q3, Q4 incremental may be limited. Banks may advance reduction of bond outsourcing exposure. Based on year-end bank assessment needs and next year's bond fund redemption fees, Q4 banks may advance fund redemptions, unfavorable to bond market supply-demand structure.
Tariff turbulence resurges - what's different this time? Reviewing April 3 this year, 2406 declined 7.75BP full day, April 7 opened down 10bp to 1.8%, then fluctuated up to 1.8325%, two-day decline about 16BP, then April 8 opened low at 1.82% but rallied throughout session to close at 1.855%. Overall, tariff-based bond trading basically ended within three trading days, then entire April returned to "garbage time." This Saturday's interbank market opening saw significantly reduced transaction volume, 30Y active bond 2502 opened down 4BP, rebounded intraday then continued declining to close at 2.08%, full-day decline 5.1bp, overall magnitude smaller than previous round's April 3 volatility. One reason may be lack of concrete confirmation from equity market risk appetite pricing; another reason is fewer trading institutions, also causing smaller volatility than previous round. Monday's A-share opening adjustment space may not be large. Compared to April, A-share incremental factors are: 1) Strong AI industry trends like US stocks, domestic computing power + overseas computing power; 2) Greater familiarity with TACO investment model; 3) Risk is large gains, expensive static valuations, major index approaching key psychological level of 4000 points. Overall, if Monday opens, it's already the second trading day of bond market tariff reaction, and magnitude would be weaker than April, so further operation space may be limited.
◾ Risk Warnings: 1) Monetary policy tightening beyond expectations; 2) Large-scale wealth management product return causing market volatility; 3) Credit risk events; 4) Inaccurate statistical standards; 5) Untimely research report information updates.
1. September Institutional Behavior Pattern: Fund Heavy Position Bond Sell-off
Accompanied by market development, most bond varieties declined in September: 1) Long-end government bonds retreated more than short-end, yield curve steepened, long-end spreads widened, 5Y-3Y spread narrowed. As of October 10, 30Y treasury yield adjusted 14.5BP from end-August, 1Y treasury yield basically unchanged, 30Y-10Y treasury term spread returned to 45% historical percentile since 2019, spread widening to around 44BP. 2) Tier-2 capital and perpetual bonds led long-end decline, 5-7Y credit spreads rapidly widened. 10Y tier-2 capital bonds (AAA-) and 10Y perpetual bonds (AAA-) both retreated 34BP from end-August, especially since late August, 5Y credit spreads (versus same-term China Development Bank bonds) historical percentile rapidly rose from 7% low to 50%.
These maturities and varieties are basically fund-preferred heavy positions. From spot trading perspective, except short-term commercial paper, funds preferred net purchases of 7-10Y government bonds, 20-30Y treasury bonds, 1-5Y medium-term notes, and 7-10Y tier-2 capital and perpetual bonds since 2024.
Compared to Q4 2022 adjustment, all showed tier-2 capital and perpetual bonds leading decline characteristics, but in this round of "fund heavy position bond sell-off," short-end decline was limited, with curve steepening more significant. Especially since September, key maturity tier-2 capital and perpetual bonds retreated significantly, 7Y and 10Y tier-2 capital bonds led decline around 30BP, 3Y and 5Y perpetual bonds also adjusted around 20BP; in government bond adjustment, ultra-long maturity varieties led adjustment, 10Y China Development Bank fell 16BP since September leading performance, 30Y treasury fell 11BP.
Trading behavior confirms September fund selling concentrated on tier-2 capital bonds, perpetual bonds, 10Y policy bank old bonds, and ultra-long treasury old bonds. September funds cumulatively net sold 16 billion yuan spot bonds, including other categories (including tier-2 capital and perpetual bonds) net sold 101.5 billion yuan, mainly selling 7-10Y and 30Y+, followed by net selling 20-30Y treasury old bonds 55.2 billion yuan, but simultaneously buying same-maturity treasury new bonds.
Currently, fund selling has accelerated, but overall still has considerable clearing space, with fund presence still visible in ultra-long active bonds. September funds cumulatively sold 87.3 billion yuan 7Y+ other categories (including tier-2 capital and perpetual bonds), significantly up from August's 19.1 billion yuan, 20Y+ treasury old bond selling scale also increased from August; but in pre-holiday last day's 25T6 and 25T2 bond switching, still saw fund active participation in 30Y treasury active bonds, net buying 1.1 billion yuan 25T6 spot bonds, net selling 1.9 billion yuan 25T2 spot bonds that day.
2. Despite Weakening Commodity Demand, Inflation Expectations Remain Strong
Commodities torn between "supply-demand narrative" and "anti-involution": "Golden September, Silver October" below expectations, pre-holiday restocking narrative ended, commodities fell back to post-Politburo meeting lows. Upstream-downstream divergence, downstream "supply-demand logic" dominated, upstream "anti-involution" dominated: rebar fell to early July levels, coking coal maintained early September prices.
Inflation expectations remain strong: Based on "anti-involution" leader coking coal monthly spread calculation, September 30 pricing shows 2026 PPI year-over-year at 1.2%, March PPI year-over-year, April PPI year-over-year even exceeding 2%. If PPI transmits to core CPI, based on historical elasticity data, next March core CPI year-over-year may reach 1.6%.
Even if 10Y rates rise to 2%, real interest rates would be 0.4%, further breaking current historical lows (0.8%). If nominal rates remain unchanged, it means real rates would decline 69.4BP over next six months; even if 10Y rates rise to 2%, real rates would still decline 44BP. Re-inflation effects approach 2024 rate cuts: 2024 full-year rate cuts of 60BP drove 10Y rate center and real rates down 70.6BP.
Travel and consumption continue improving: Ministry of Culture and Tourism data shows this year's National Day and Mid-Autumn Festival 8-day holiday saw tourism trips up 16.1% year-over-year, domestic tourism spending up 13.5%. Daily average trips up 1.6%, consumption up 1%. Every holiday this year has shown recovery.
Scar effects keep healing, RBC business cycle recovering. By annual data view, economic cycle has been in recovery phase since 2022. Unknowingly, tourism trips and revenue have already surpassed historical highs. Through October, 2025 tourism trips (+10.4%) and revenue (+3.1%) have both exceeded 2019 levels.
3. Q4 Bond Market Focus: Valuation, Institutional Behavior, and Tariff Re-trading
EVA valuation perspective: 30Y treasury bonds regain mortgage rate comparison advantage, 10Y relatively neutral. From bank EVA calculation perspective: As of end-September, 30Y treasury after-tax EVA level was 2.15%, even considering VAT recovery on interest income, compared to existing mortgage rates of 1.71%, already has high comparison advantage, with spread widening to 79% historical percentile since 2015; but 10Y treasury EVA spread versus general loans only recovered to 24% historical percentile level, comparison advantage relatively neutral.
Fundamental perspective: Economic data indeed weak, traditional economy's real estate components also difficult to be optimistic about. July-August economic data performance relatively flat, investment components showed obvious decline, previously bright social retail also retreated. With equipment update support weakening and "anti-involution" policy effects beginning to show, manufacturing monthly year-over-year growth turned negative, infrastructure investment also showed retreat, mainly affected by utility drag; trade-in subsidies front-loaded, later pulling power insufficient, subsidized goods growth declined, but service consumption provided support; real estate investment decline range continued widening, August monthly year-over-year fell to -19.46%, sales side has not bottomed and recovered. If following this logic, might bring rate cut expectations.
But even with weak fundamental trends, from real interest rate perspective, rate cut necessity may not be high. Core CPI has shown continuous recovery this year, reaching 0.9% year-over-year growth by August. Considering core CPI and current treasury rates, compared to last year's real rate levels may not be low. Rate cuts based on other considerations may not necessarily benefit bonds.
Current IRS implied rate cut expectations are low. Using "IRS (FR007: 1-year) - FR007 (20-day MA)" to measure market expectations for rate cut pricing, beginning of year and April tariff event landing, implied rate cut expectations once expanded to 50BP, rate cut expectations continuously narrowed since April, indicator turned positive after August 18, recently declined, but currently basically no implied rate cut expectations.
From institutional behavior perspective, current market risk appetite is high, insurance bond allocation growth may be weaker than usual. Insurance traditionally has advance bond allocation demand in Q4 for "opening red" considerations, but this year's liability-side premium growth is limited, asset-side bond cost-effectiveness is low. This year's insurance seasonal bond allocation peak in secondary market was Q3, Q4 incremental may be limited.
On the other hand, banks advance reduction of bond outsourcing exposure. Based on year-end bank assessment needs and next year's bond fund redemption fees, Q4 banks may advance fund redemptions, unfavorable to bond market supply-demand structure.
Tariff turbulence resurges - what's different this time? Reviewing April 3 this year, 2406 declined 7.75BP full day, April 7 opened down 10bp to 1.8%, then fluctuated up to 1.8325%, two-day decline about 16BP, then April 8 opened low at 1.82% but rallied throughout session to close at 1.855%. Overall, tariff-based fixed income trading basically ended within three trading days, then entire April returned to "garbage time."
This Saturday's adjusted National Day holiday only had interbank market open, transaction volume significantly reduced, 30Y active bond 2502 opened down 4BP, rebounded intraday then continued declining to close at 2.08%, full-day decline 5.1bp, overall magnitude smaller than previous round's April 3 volatility. One reason may be lack of concrete confirmation from equity market risk appetite pricing; another reason is fewer trading institutions, also causing smaller volatility than previous round.
This round's similarities and differences with last time, we believe can be observed from US and China perspectives: US side, Nasdaq opened slightly down around 1%, closed down 3.56%, magnitude weaker than April. Compared to April, US stock investors believe more focused on stock earnings growth certainty, AI's own large capital expenditure (but also concerned about short-term bubbles from strong expectations), and upcoming Fed rate cuts. Market is very familiar with "TACO" strategy, "Trump Always Chickens Out." This round's decline more likely reflects previous high gains bubble release rather than April's uncertainty concerns. Summary: market panic, but not panicked enough to want to flee.
Monday's A-share opening adjustment space may not be large. Compared to April, A-share incremental factors are: 1) Strong AI industry trends like US stocks, domestic computing power + overseas computing power; 2) Greater familiarity with TACO investment model; 3) Risk is large gains, expensive static valuations, major index approaching key psychological level of 4000 points.
Overall, if Monday opens, it's already the second trading day of bond market tariff reaction, and magnitude would be weaker than April, so further operation space may be limited.
Risk Warnings: 1) Monetary policy tightening beyond expectations 2) Large-scale wealth management product return causing market volatility 3) Credit risk events 4) Inaccurate statistical standards 5) Untimely research report information updates
Comments