Systemic risks, policy paths, and an investment map.

This note doesn’t try to settle “is China going to crash” as a binary. The lens is sharper: given the current leadership’s objective function and constraints, where will capital be steered, and what gets compressed? The framework is Fact (data) → Style (objective function) → Path (foreign analogies) → Policy (forecast) → Asset (mapping). The conclusion isn’t “long” or “short” — it’s structurally allocated to policy-favored directions, structurally short policy-sacrificed directions, and hedged for tail risk.

Marker conventions: [Official] = Chinese official data (NBS / MoF / PBoC / MOFCOM); [External] = third-party / overseas estimates (Rhodium / IMF / World Bank / academic / think-tanks); [Inference] = logic-driven extrapolation; [Reference] = foreign historical analogy. External and official sources diverge meaningfully on GDP, population, local debt, and capital outflow — flagged in each section below.

Three core takeaways

1. Structural deflation + balance-sheet repair is the base case for the next 3–5 years, not a V-shaped rebound. The closest analog is Japan 1990–2010, but China’s per-capita GDP is far below Japan’s at its peak — the middle-class buffer is thinner.
2. “Hard tech + national security” is the policy-dividend mainline; property, platforms, consumer finance are sacrificed. SOE high-dividend (“China Special Valuation”) stocks function as a transition-era “policy-guaranteed yield” vehicle.
3. Three tail risks — Taiwan (2027–2032 window), policy lurching further left (anti-market, anti-wealth), and US escalation — are not adequately priced into asset markets.

Official vs. external sources — quick gaps (detailed below): Rhodium estimates 2025 real GDP grew 2.5–3.0%, about half the official 5.2%; IMF puts broad local-government liabilities at 84% of GDP (no official broad-aggregate figure); a DZH-database compilation estimates LGFV interest-bearing debt at $12.1T vs the IMF $9.04T; on a SAFE BoP basis 2024 net FDI was −$168B, the largest capital flight in the data series since 1990; Brad Setser estimates the official current account understates the true surplus by ~$500B; Yi Fuxian estimates China’s actual population is overstated by 130M+. The gap itself is part of why foreign-investor valuation discounts, gold strength, and Hong Kong capital outflows persist.

§01 · Systemic Risks — a six-dimension data scan

1. Demographics — collapsing faster than expected

[Official] Total population peaked at 1.412B in 2022. 2023 fell 2.08M, 2024 fell 1.39M, 2025 fell 3.39M — the slope is accelerating (NBS, Jan 2026). 2025 births: 7.92M (down 1.62M YoY); birth rate 5.63‰; death rate 8.04‰; natural growth rate −2.41‰. Working-age population (16–59) lost 6.62M in 2025 alone. Population 60+: 323M (23.0%); 65+: 224M (15.9%). Official TFR estimated ~1.05 (SWUFE 2025).

[External] Yi Fuxian (University of Wisconsin-Madison) has long argued China’s population is overstated by ~130M+ (more than one-third of the US population), based primarily on cross-checks against newborn BCG vaccine doses — e.g. 2018 reported 6.21M doses imply ≤9.9M actual births vs. the official 15.23M. Other overseas demographers estimate 2025 actual TFR closer to 0.9 (rather than 1.05) — sharing the global bottom tier with Korea’s 0.7. Even setting the Yi estimate aside, 2025’s 7.92M births versus the 14.33M projected when the one-child policy was repealed in 2016 — less than 56% of forecast — confirms birth policy has structurally failed.

[Inference] Demographics is the most irreversible 30-year variable — the question isn’t whether, but how fast and what offsets it. The 2025 print matters because the slope is accelerating: 1.62M fewer births in a single year and 6.62M working-age loss collapse the “China can buy time like Japan” timeline from theory into reality. Direct implications: ① long-run aggregate demand pressure, especially durables, new homes, education; ② savings rate elevated by aging, not released by consumption; ③ the “long-termism” narrative is forced forward (pensions, healthcare, social security).

2. Property — fifth year of clearing, new starts still collapsing

[Official] Rogoff & Yang (2020, NBER) estimate property’s full-stack contribution to Chinese GDP peaked around 25–29% (incl. up- and down-stream). PBoC’s 2019 urban household survey found ~70% of household assets are real estate. Full-year 2025 (NBS, 2026-01-19): residential floor sales 881M sqm (−8.7%, residential −9.2%); sales value ¥8.39T (−12.6%); new starts 588M sqm (−20.4%); completions (−18.1%) — new starts are now under one-third of the 2021 peak. 70-city new-home prices YoY remained negative through 2025; official Q2 print −6.4%.

[External] Beike/KE Holdings data (Caixin, Dec 2025): secondary listings have reached 6.5M units, +60% from four years ago; the average buyer now views 17 properties before purchase (vs. 10 in 2021) — supply oversaturated, pricing power firmly with buyers. Tier-1 four-city secondhand prices fell for a 7th consecutive month through November 2025, down 1.1% MoM — a clear gap from the official narrative of “stabilization.” A Reuters poll forecasts new-home prices −3.8% in 2025 and −0.5% in 2026 — more bearish than Beijing’s own implicit 2026 stabilization assumption. S&P Global Ratings (2025-01) judged that surging secondary sales “would stabilize 2025 property” — but conditional on developers halting the new-starts collapse. With actual 2025 new starts at −20.4%, that condition didn’t hold.

[Inference] The trajectory isn’t “hard landing” — it’s “controlled deflation + state advance, private retreat” in slow motion: private developers are allowed to wind down (floor: “deliver pre-sold homes”); state-owned developers (Poly, China Resources, China Overseas) consolidate share; property is reframed from “growth engine” to “social infrastructure”. Key read: sales fell −9% but new starts fell −20%, meaning developers themselves are deeply pessimistic about the next 2–3 years. This active supply contraction implies property’s drag on GDP continues into 2026–2027. The cost: property-related employment (agents, white-collar staff, mid-managers) and downstream sectors (building materials, new-home appliance channels) are compressed for 5–10 years.

3. Local government debt — halfway through the resolution program

[Official] November 2024 NPC-passed ¥12T “6+4+2” package: ¥6T one-time additional local-debt quota for swap + ¥4T (5 × ¥800B special bonds) + ¥2T post-2029 shantytown implicit debt. By end-2024, hidden local debt slated for resolution fell from ¥14.3T to ¥10.5T (MoF). By end-2025, ¥5T+ of replacement bonds had been issued, >50% of the program; projection has implicit debt at ¥2.3T by 2028. 2025 net LGFV bond issuance was just ~¥36B — less than 20% of the 2024 level (Yicai/CLS); over 60% of financing platforms have exited.

[External] Estimates diverge meaningfully. IMF estimate (end-2024): LGFV debt at $9.04T (¥60T); combined with explicit local debt, broad local government liabilities reached 84% of GDP — sharply up from 62% in 2019. A DZH-database compilation across ~4,000 LGFVs put end-2024 interest-bearing liabilities at $12.10T — about one-third higher than the IMF figure. Atlantic Council headline framing: “Beijing extends and pretends” — the core critique is that the program swaps creditors and lengthens maturity but does not reduce total economy-wide leverage. Carnegie/Pettis (Aug 2025) warned that “using the central balance sheet to clean up local debt is a bad idea” — because it converts local credit risk into sovereign credit risk. Read: “halfway through resolution” is true; but the underlying size of the debt stock differs materially between IMF/third-party and official estimates — this is the root of the long-running valuation gap on Chinese banks and LGFV-related high-yield credit between domestic and offshore investors.

[Inference] Debt resolution doesn’t reduce debt — it extends maturity, lowers coupons, and changes the creditor mix, shifting from city commercial banks / trusts / shadow banking to the central government and big four banks. This is central credit absorbing local credit. The cost: central government leverage rises, and local fiscal latitude is materially compressed for 5–10 years — slower infrastructure, civil-servant pay cuts, retreat of public services in lower-tier cities. The 2025 collapse in LGFV bond issuance has already validated this path.

4. Deflation — CPI flat in 2025; PPI negative for 40+ months

[Official] CPI 2025 full-year: 0.0% (flat, NBS, Jan 2026); food −1.5%, energy −3.3%. PPI 2025 full-year: −2.6%; YoY negative for 40+ consecutive months. The GDP deflator turned negative in Q2 2023 — negative for 11+ quarters running — unprecedented since reform-era data begins. M2/GDP exceeds 230% (BIS). Q1 2026 GDP +5.0% YoY (NBS, 2026-04-16); industry +6.1%; retail +2.4%; FAI +1.8%.

[External] Rhodium Group (Dec 2025) estimates 2025 real GDP grew only 2.5–3.0% — about half the official 5.2% pace through Q3. The biggest divergence is on investment: official FAI is down ~−11% nominal YoY July-Nov 2025 — a collapse — yet GCF (the GDP-side investment series) still contributed +0.9pp to Q3 GDP growth in official data; the inconsistency is Rhodium’s main objection. Rhodium also notes the ~$1T 2025 trade surplus accounts for 57–68% of official GDP growth — without exports, domestic demand on its own would be in negative territory. Atlantic Council: “Same overstatement” echoes the critique. Read: even if you discount the Rhodium estimate, the Q1 2026 industry +6.1% / retail +2.4% scissor itself confirms “supply pulled by state narrative; demand independently weak” — and that structure self-deepens deflation. Foreign investors’ valuation discount on Chinese assets partly reflects skepticism of the official GDP print, not pure pessimism.

[Inference] This is balance-sheet recession (Richard Koo’s framework) — households and firms are voluntarily deleveraging. PBoC rate cuts can’t restore confidence because investment-return expectations are anchored on falling property prices. Despite multiple structural rate cuts in 2024–25, the real rate (nominal minus deflator) is rising because deflation is widening the wedge. That’s why monetary policy looks loose but doesn’t feel loose to households or firms. Q1 2026’s industry +6.1% / retail +2.4% scissor confirms the binary structure: supply pulled by state narrative; demand independently weak — a structure that itself deepens deflation.

5. External pressure — trade cooling, tech containment continues

[Official] MOFCOM print: 2024 actual FDI −27.1% YoY; 2025 11M run-rate −7.5% YoY — narrowed but still negative. Newly registered foreign-invested enterprises in 2025 4M: 18,832 (+12.1% YoY), but actual capital inflow −10.9% — “more new entities, less capital” reflects foreign capital shifting from strategic bets to exploratory entries.

[External] More bearish: on a SAFE balance-of-payments basis, 2024 net FDI inflow was −$168B — the largest capital flight in the data series back to 1990 (Bloomberg, 2025-02). The World Bank’s China Economic Update (June 2025) notes that net capital outflows (incl. errors and omissions) in early 2025 exceeded the current account surplus, producing a $31B (−0.7% of GDP) decline in FX reserves; Mainland-HK Stock Connect Q1 2025 net equity outflow was $57.3B. Brad Setser (former NY Fed) further argues China switched in 2024 to an “internal payments dataset” for current-account reporting that understates the real surplus by roughly half — implying perhaps $500B in concealed surplus and capital flight. He calls it “a fudge factor used to mask hot money leaving Xi’s China.” Read: external sources broadly believe official capital-outflow data is systematically understated, putting more pressure on the renminbi and the HKD peg than headline numbers suggest — this is part of why “grey-market” demand for gold, USD, and crypto inside China has stayed strong through 2025.

[Official — US-China tariffs] In 2025, five rounds of negotiation (Geneva → London → Stockholm → Korea) yielded a “one-year pause” framework: US revoked 91% of tariffs, suspended 24% tariffs for 1 year, suspended port fees and high-tech penetration rules for 1 year; China revoked 84% of its tariffs and suspended new rare-earth export controls for 1 year. But on 2025-12-22 the US implemented its Section 301 finding, imposing additional tariffs of up to 50% on Chinese semiconductor products (with carve-outs for Taiwan-fab capacity) — tech containment is still escalating. China’s share of US bilateral trade fell from ~21% (2017) to ~13% (2024).

[Inference] The dominant logic of external pressure has shifted from trade to tech and capital — structural and irreversible. The 2025 tariff “pause” is tactical de-escalation; the December 2025 50% semiconductor tariff is strategic escalation evidence — markets shouldn’t read the pause as reversal. Two implications: ① exports lose pull on low-end labor-intensive manufacturing but reverse-feed forced indigenization in hard tech; ② the risk premium on Chinese assets has stepped up systemically — A-share / H-share multiples cannot return to 2020 levels.

6. Social mood — not unrest; low desire and “downshifting”

[Official] Youth unemployment (16–24, urban survey, students excluded): rose to 18.9% in August 2025 — a record high under the post-2023 methodology — then fell to 16.5% in December 2025, 16.1% in February 2026, before rebounding to 16.9% in March 2026 (NBS). The 2026 college graduate cohort is projected at 12.22M (Ministry of Education). Household savings rate ~38% in 2023 (vs. ~30% pre-pandemic) — precautionary savings. “Lying flat” (躺平), “let it rot” (摆烂), and “full-time children” (全职儿女) have become cultural memes. Marriage registrations in 2024 hit 6.10M, the lowest since 1980 (Ministry of Civil Affairs).

[External] Yao Lu (Columbia University, sociology) has written extensively in 2024–25 that China’s “college credential bubble is bursting”, and that the official measure systematically excludes “exam preparation, flexible employment, and discouraged workers” — meaning actual youth unemployment + under-employment is materially higher than the published 16–19%. The rapid rise in PhD/master’s “flexible employment” she documents is essentially disguised unemployment. Multiple Chinese academics (incl. Yao Yang at Peking University) have raised similar concerns about the methodology revisions. The June 2023 redefinition (which dropped the headline from 21.3% to a lower base) followed a record print — externally widely read as politicization of statistics. The fact that the post-revision measure still hit 18.9% in August 2025 suggests the underlying trend is harsher than the headline implies.

[Inference] This isn’t “pre-revolution” — it’s the seed of a Japan-style low-desire society. Young people don’t consume and don’t protest; they opt out of the system. For the leadership this is double-edged: short-term stability, but long-term consumption engine fading, irreversible birthrate decline, tax base shrinking. The current response — birth incentives + consumption stimulus + information control — is weakly effective on the first two and counterproductive on the third.

§02 · Policy Style — three objectives and their priority

[Inference] The current leadership’s objective function, ranked by priority:

1. Regime security (paramount) — primacy of party leadership, internal power concentration, prevention of any structural challenge. The prerequisite for everything else.
2. Social stability (binding constraint) — avoid mass incidents, control information flow, manage expectations. Slow deflation is acceptable; rapid collapse is not.
3. Long-term great-power competition (strategic goal) — tech indigenization, military modernization, supply-chain resilience. Lower priority than #1 and #2 in any given quarter, but the steering wheel for long-term resource allocation.

Style observations:

Core inference: This style resembles late-Soviet upgraded + MITI-era Japanese industrial policy, not the 2000–2015 “market-friendly technocrat” mode. Many investment frames from the 2000–2020 era — platform economy, consumer upgrade, perpetual property bull — are no longer valid. This is structural, not cyclical.

§03 · Reference Paths — who they’re studying, who they’re avoiding

Primary: Japan 1990–2010

[Reference] Closest historical sample — surviving an asset-bubble burst via controlled deflation, industrial upgrade, and social stability. Japan’s path: ① house prices took 25 years to bottom (some tier-2 still haven’t recovered); ② the banking system worked through bad debt for a decade and consolidated (SMBC, Mizuho, MUFG); ③ BOJ ZIRP from 1999, QE from 2001; ④ manufacturing climbed up the curve (robotics, semiconductor equipment, automotive, precision machinery); ⑤ society settled into a low-desire, long-life, low-growth equilibrium.

[Inference] Decision-makers systematically study Japan (multiple internal-circulation reports from CICC, CITIC, Renmin University, CASS). Lessons taken: industrial upgrade, deleveraging without a banking crisis, social stability. Lesson not transferable: Japan’s per-capita GDP at the bubble’s peak was ~$25,000 (developed-country floor); China’s 2024 print is $13,400 — China is “old before rich” and the starting point is more fragile. That’s why the leadership cannot copy Japan’s “let the market clear” — sovereign credit must absorb the shock.

Secondary 1: Korean developmental state (1997–present)

[Reference] After the 1997 Asian crisis, Korea exited via chaebol concentration + hard-tech indigenization + export orientation: Samsung, LG, Hyundai, SK made long-horizon, capex-heavy bets with state R&D backing, eventually winning global positions in semis, displays, autos, batteries. Costs: high household debt (~100% of GDP, 2024), birthrate collapse (0.7), hyper-competitive social culture.

[Inference] China’s “new productive forces” narrative roughly clones the Korean model — bigger scope, bigger subsidies — across semis, batteries, EVs, AI, robotics, biopharma, commercial aerospace, low-altitude economy. Investment logic: don’t bet winners, bet sectors — the leadership keeps the sector alive; specific firms can be replaced.

Secondary 2: Singapore governance model

[Reference] Singapore’s stack: HDB public housing (95% of citizens live in state-built housing, prices deeply managed), Temasek/GIC sovereign-wealth steering of strategic industries, authoritarian + technocratic + high-quality civil service, a limited but functional safety net.

[Inference] The ceiling for China’s new property model is the Singapore-style public-housing path — affordable rentals + price-controlled commodity housing + market-priced premium housing in three layers. Housing ceases to be a wealth vehicle and becomes utility infrastructure. This is the rationale for the “three major projects” (urban-village renovation, indemnificatory housing, dual-use facilities) pushed since 2024. Implication for property investors: commodity housing as an asset class earns suppressed returns until the system reaches a Singapore-style equilibrium.

Anti-pattern: Soviet collapse (1989–1991)

[Reference] The leadership’s deepest fear is the Soviet model — collapse caused by relaxing ideological control, abandoning party leadership, and being penetrated by Western color revolutions. This is why “political security” and “ideological battlefield” are repeated themes, and why control over information, education, culture, and religion only tightens.

[Inference] Any market expectation of “loosen control to unlock vitality” is mis-directed — political control’s perimeter only expands. Direct implication: private-sector confidence cannot be restored by property-rights documents; it requires private capital becoming politically unthreatening to the leadership. That is exactly what happened to Tencent / Meituan / Alibaba post-rectification — large-scale compliance and political-alignment retooling.

§04 · Policy Forecast — directions across five dimensions

[Inference] Likely policy direction over 3–5 years:

1. Fiscal & monetary — central leverage backstops; welfarism rejected

• Special and ultra-long government bonds become normalized; central debt rises from ~22% to 30%+ of GDP.
• Debt-resolution program continues: LGFV implicit debt is swapped to longer, lower-coupon explicit bonds; banks absorb interest-margin pain.
• Rate-cut headroom is compressed (LPR around 3.0%); more emphasis on structural tools (PSL, re-lending, targeted facilities).
• No US/EU-style direct cash transfers to households — the leadership repeatedly rejects “feeding the lazy”, a path-dependent ideological constraint that overrides economic logic.
• Fiscal deficit drifts toward 4–5% (the historical 3% line is effectively breached).

2. Industrial — subsidies funnel into “new productive forces”

• Semis (mature-node indigenization first, advanced-node slow under sanctions): NAURA, AMEC, Piotech, SMIC, CXMT, YMTC.
• AI / compute: Huawei Ascend ecosystem, Cambricon, Hygon, SenseTime; data centers and power infrastructure (UHV, energy storage).
• EV + batteries + storage: BYD, CATL, Sungrow already global leaders.
• Robotics + factory automation: Inovance, Estun, the humanoid supply chain (Unitree, Galaxy, Agibot).
• Biopharma: license-out and out-bound biotech (BeiGene, Innovent, Akeso, Keymed).
• Commercial aerospace + low-altitude economy: new 14th-FYP priorities.
• Defense: AVIC family, China Shipbuilding.
• Vehicles: government guidance funds, the Big Fund Phase III ¥344B (2024), supply-chain subsidies, tax breaks.

3. Finance & assets — manage deflation, manage prices, prop SOEs

• Property: “deliver pre-sold homes” floor, controlled clearing of private developers, commodity-housing prices grind down for another 2–3 years before bottoming, indemnificatory supply expands; tier-1 core may stabilize first.
• A-shares: Central Huijin national team backstops at key levels (verified multiple times in 2024) — but the goal is stability, not boom.
• “China Special Valuation” SOE high-dividend acts as a policy-dividend vehicle, offering 5–7% yields — a quasi-bond substitute under deflation.
• H-shares: long suppressed by valuation anchors, but AI export + bottoming platform regulation create structural opportunities.
• Capital controls: not loosened; likely tightened around the edges (QDII, offshore insurance, crypto).

4. Social — modest patching, no relaxation of control

• Delayed retirement is law (2024) — male 63, female cadre 58, female worker 55.
• Hukou: liberalized in tier-3/4 cities, marginally relaxed in tier-1; the core control remains points / talent quotas.
• Birth incentives: provincial subsidies (Hubei, Sichuan, Hainan), longer parental leave, expanded daycare — expected effect: weak, since the underlying issue is economic insecurity.
• Social safety net: marginal expansion of rural pensions, urban-rural medical insurance — far short of what’s needed to release consumption.
• Information space: tighter (personal VPN use, self-media registration, AI-generated content labeling).

5. External — strategic competition, tactical de-escalation (2025 cycle complete)

• US — one cycle done: After 5 rounds of negotiation in 2025 (Geneva → London → Stockholm → Korea), a “one-year pause” framework was agreed: US revoked 91% of tariffs, suspended 24% tariffs and high-tech penetration rules for 1 year; China revoked 84% of tariffs and suspended new rare-earth export controls for 1 year. But on 2025-12-22 the US implemented Section 301, imposing up to 50% additional tariffs on Chinese semis — strategic tech containment continues. Read: tactical trade truce + strategic tech escalation — long-run direction is “non-decoupled but de-risked”, and 2026 can re-trigger frictions on any political event.
• EU: courted as part of the not-US axis, but trade frictions persist (EV anti-subsidy, medical-device procurement); the core question is whether EU industry can absorb Chinese mature-capacity exports.
• Russia: deeper strategic binding (energy, food, security), but careful not to be dragged into secondary sanctions.
• Global South: BRI 2.0 — more selective, more financial instruments (RMB settlement experiments).
• Taiwan: gray-zone pressure escalates (drills, economic squeeze, info ops); probability of initiating conflict during the 2027–2032 window remains less than 50% — military modernization is incomplete, the economy is constrained, and decision-makers are risk-averse. This is the base case, not a certainty.

§05 · Sacrificed vs. Favored — who’s pressed, who’s lifted

[Inference] The direction is in the data. Three dimensions — groups, regions, industries:

Industries

Regions

Groups

Bottom line: Policy-favored ≈ “industries that fit the national narrative.” Sacrificed ≈ “industries that ride the previous growth cycle’s logic but don’t fit the new narrative.” It’s not absolute good or bad — it’s relative resource allocation.

§06 · Investment Mapping — four buckets

[Inference] Four buckets across instrument-sector-time:

1. A-share core (policy-positive beta)

2. H-shares (valuation gap + platform recovery)

3. Indirect overseas beneficiaries (hedging China constraints / shifts)

4. Hedges and tail risk

A suggested framework (principle, not advice)

• Core (50–60%): China Special Valuation high-dividend + hard tech + new energy + innovative pharma — anchored to policy-favored direction.
• Satellite (20–30%): H-share platform recovery + indirect overseas beneficiaries — alpha sources.
• Hedge (10–20%): gold + USD assets + tail hedges — to dampen volatility.
• Cash / short-duration (5–10%): rebalancing reserve and dry powder for tail events.

Bottom line: This isn’t a “long China” or “short China” portfolio — it’s “long policy-favored direction, short policy-sacrificed direction.” The leadership’s objective function is a better guide to allocation than the GDP print.

§07 · Three tail risks — flagged honestly

[Inference] Three tail risks that materially shift the framework if they hit:

1. Taiwan (2027–2032 window)

Base case (~70–80%): gray-zone pressure escalates (drills, economic squeeze, info ops, coast-guard “law enforcement” normalization), but no kinetic initiation — military modernization is incomplete, the economy is stressed, and decision-makers are risk-averse. Risk case (~10–20%): misjudgment or accident triggers escalation.

External wargame benchmarks: CSIS The First Battle of the Next War (2023, 24 wargames): in most scenarios the US/Taiwan/Japan coalition repels a Chinese amphibious invasion and preserves Taiwan’s autonomy — but at the cost of “a bloody mess” with terrible losses on all sides. CSIS Lights Out? (July 2025, 26 blockade wargames): any blockade generates escalation pressure that is hard to contain and tends to spiral into large-scale war. Adm. Davidson’s “2027” framing is a capability timeline (when China would be capable) — not necessarily an action timeline; most analysts now flag 2030–2035 as more realistic windows. CFR / RAND broadly agree.

Asset implications: if escalated, Chinese assets, HK, Taiwan, global tech, semis upstream all re-price; gold, USD, select defense names are extreme beneficiaries. Read: external consensus is “low probability, very high cost” — this is the core rationale for holding gold / select defense / USD as tail hedges, not a base-case bet.

Signals to monitor: ① US-Taiwan military contact frequency; ② mainland drill scale and duration; ③ semiconductor export-control escalations; ④ US discussion of financial sanctions (SWIFT / FX); ⑤ normalization of coast-guard / economic-blockade “quasi-war” instruments around Taiwan.

2. Policy lurching further left

Base case: at the margin, the line is loosening (Central Economic Work Conference late-2024 returns to “stabilize growth”, language on private economy warms up). Risk case: a political event or external crisis triggers ideological tightening — anti-market, anti-wealth, anti-foreign intensifies. Common-prosperity 2.0 could target real estate, financial assets, and offshore holdings via progressive taxation or restrictions.

Signals to monitor: ① new income / property tax rules; ② treatment of private-sector entrepreneurs in media or law; ③ new capital-control instruments; ④ offshore asset disclosure and tax piercing.

3. US policy path

Base case: Trump 2.0 tariffs oscillate around 60%, tech containment continues to expand (BIS Entity List, AI chip controls, TikTok-style asset divestitures). Risk case: financial decoupling — SDN sanctions on large Chinese entities, partial SWIFT exclusion, forced ADR delisting — re-prices multiples down another step.

Signals to monitor: ① OFAC SDN list changes; ② SEC audit requirements on Chinese ADRs; ③ secondary sanctions on Chinese banks / energy firms; ④ China-related legislation in Congress.

Risk-management takeaway: Always retain 10–20% in hedge positions (gold + USD + select tail options) — never let the portfolio become a single-country / single-scenario bet. Even if the base case plays out, the levered all-in upside doesn’t compensate for tail-risk drawdowns.

Bottom-line summary:

Structural deflation + state-capacity expansion + hard-tech indigenization is the next-3–5-year mainline; property + platforms + premium consumer + low-end export are the sacrificed directions; Taiwan / leftward policy / US escalation are three under-priced tail risks.

Asset translation: core in hard tech + new energy + SOE high-dividend; satellite in H-share recovery + indirect overseas; tail in gold + USD + hedges.

The leadership’s objective function is a better guide to allocation than the GDP print — read the function, then position.