E014. East Asia Analysis I: China at the Inflection

1. Executive Summary

The dominant Anglophone narratives about China — both the "imminent domination" school and the "imminent collapse" school — fail under empirical scrutiny. The data from 2025 and the first months of 2026 supports neither extreme. What it supports instead is a more analytically demanding finding: China has entered a phase of structural deceleration, in which the country retains substantial industrial scale and selective technological capability, but loses the macroeconomic foundations required for indefinite ascent. Headline GDP growth of 5.0 percent in 2025 (IMF 2025) was sustained by export front-loading, consumption subsidies, and a fiscal impulse of approximately 1.4 percent of GDP (Rhodium Group 2026a); the underlying growth model is not.

The convergent assessment of the IMF (2025), Rhodium Group (2026a, 2026b), Peterson Institute scholars, and CFR's Brad Setser (2025) is that China is approaching a Japan-style structural stagnation, but entering it at a lower per-capita income, with weaker demographic foundations, and under tightening external constraints. This report consolidates the 2025–2026 evidence base, identifies the systematic errors in popular discourse, and applies a four-type error taxonomy (order mismatch, unit mismatch, extrapolation, discontinuity misreading) to dismantle five specific claims that circulate widely but fail empirical and methodological tests. The objective is the removal of hype in both directions and the establishment of an evidentiary floor for discussion.

2. Key Findings

• Headline growth conceals decelerating trend: The IMF (2025) projects 5.0 percent for 2025 and 4.5 percent for 2026, with 2027 at 4.0 percent. China's own 2026 target of 4.5–5.0 percent is the lowest since 1991 (Trading Economics 2026a). The GDP deflator continued to decline in 2025, indicating persistent deflation (IMF 2025).
• Demographic contraction is now severe and structurally irreversible: Births fell 17 percent year-on-year to 7.92 million in 2025, the lowest absolute number since 1949 (National Bureau of Statistics 2026; PBS 2026). The total fertility rate is approximately 0.93 (Wikipedia/UN data 2025). Rhodium Group (2026d) projects a population decline of approximately 60 million over the next decade.
• Capital flows have effectively reversed:  Balance-of-payments inward FDI fell to $4.5 billion in 2024, or 1.3 percent of the 2021 peak of $344.1 billion (Mitsui Global Strategic Studies Institute 2025). Net FDI showed a record outflow of $168 billion in 2024 (Bloomberg 2025). The downward trend continued through 2025 (–9.5 percent) and into early 2026 (–5.7 percent YoY) (Trading Economics 2026b).
• The trade surplus is unprecedented and politically untenable: Setser (2025) estimates the current account surplus at over $650 billion through Q3 2025, nearly double IMF projections, with passenger car exports projected to reach 8 million in 2026 — exceeding Japan's historical peak. This is producing tariff responses and protectionist coalitions across the EU, US, India, and Latin America (East Asia Forum 2025).
• Semiconductor catch-up is real but economically marginal: SMIC's 5nm production via DUV multi-patterning operates at ~20 percent yield with ~50 percent cost premium versus TSMC (AEI 2026; Reuters via SemiWiki 2025). CSIS (2026a) projects domestic share of China's AI chip market reaching 50 percent in 2026 — but this is import-substitution under sanctions, not global commercial competitiveness.
• Industrial policy cannot offset macro contraction: Rhodium Group (2026a) finds that new industries (AI, robotics, EVs) added only 0.8 percentage points to GDP from 2023–2025, while real estate and traditional sectors contributed –6 percentage points. The "high-tech offset" thesis is quantitatively insufficient.

3. In-Depth Analysis

3.1 Macroeconomy: Surface Stability Over Structural Weakness

China's 5.0 percent growth in 2025 (IMF 2025) was the product of three identifiable factors: export front-loading ahead of US tariffs, targeted consumption subsidies (home appliances, automobiles, electronics), and a fiscal impulse of approximately 1.4 percent of GDP (Rhodium Group 2026a). None of these factors are structural; all are policy interventions with diminishing marginal returns.

Beneath the headline figure, Pettis (2025) identifies the operative pathology as the persistence of China's underlying imbalance: household consumption remains at approximately 39 percent of GDP, compared to 55–70 percent in most developed economies. Investment remains at 42 percent of GDP, well above the 25–30 percent threshold considered sustainable. The result is what Pettis (2025) terms the "involution trap": investment displaces consumption, generating overcapacity, which is then exported as deflation to trading partners. The 2025 data confirms the mechanism — China's GDP deflator continued to decline, even as nominal export volumes surged (Setser 2025).

The IMF (2025) explicitly recommended that China reduce industrial-policy subsidies from approximately 4 percent of GDP to 2 percent and restructure the debt of unsustainable Local Government Financing Vehicles (LGFVs). LGFV debt reached $9.2 trillion in 2024, with the central government providing approximately $1.4 trillion in refinancing over the past year (Atlantic Council 2026). The financial system is, in Rhodium Group's (2026b) formulation, in "fiscal and financial decay" — increasingly unable to function as a transmission mechanism for domestic demand stimulus.

Rogoff and Yang (2026), in a Brookings Papers contribution, formalize the comparison with Japan's lost decade. Their conclusion: China enters the analogous phase earlier in its development trajectory, with weaker fundamentals on demographics, household balance sheets, and external openness — implying a deeper and longer adjustment than Japan's.

3.2 Demographics: The Binding Constraint

The 2025 demographic data constitutes the single most important development. Births collapsed 17 percent to 7.92 million, a level last observed in 1738, when China's population was approximately 150 million (Yi 2026). The total fertility rate is estimated at 0.93 — among the lowest globally and roughly half the replacement level. The population declined by 3.39 million in 2025, the fourth consecutive year of contraction (PBS 2026).

The Rhodium Group's demographic projection (2026d) estimates population losses of approximately 60 million over the next decade, with annual declines reaching 7.6 million by 2035. The cohort of women aged 20–34 — responsible for 85 percent of births — is projected to decline from 105 million in 2025 to 58 million by 2050. Approximately 23 percent of the population is now over 60, and 16 percent over 65 (PBS 2026).

This is the variable that places a hard ceiling on long-horizon Chinese economic trajectories. Unlike fiscal or industrial policy variables, demographic trends are essentially irreversible on policy-relevant timescales. Singapore, which has pursued the most aggressive pro-natalist regime in the world, recorded a TFR of 0.87 in 2025 — demonstrating empirically that mid-cycle policy interventions do not reverse fertility decline once it has crossed certain thresholds (Rhodium Group 2026d).

The RAND Corporation (Pollard et al. 2025) frames this as a national-security-relevant constraint: declining cohorts limit military recruitment, fiscal capacity for entitlement spending, and the labor force on which industrial policy depends. The window for China to achieve specific developmental objectives is closing measurably each year.

3.3 Capital and Trade Flows: Two-Way Decoupling

The most striking 2025 data concerns capital flows. The Mitsui Global Strategic Studies Institute (2025), using State Administration of Foreign Exchange data, documents that balance-of-payments inward FDI collapsed from $344.1 billion in 2021 to $4.5 billion in 2024 — a reduction to 1.3 percent of peak. Net FDI outflows of $168 billion in 2024 represented the largest capital flight in records extending to 1990 (Bloomberg 2025).

The Asian Macroeconomic Research Office (AMRO 2025) attributes a portion of this to interest rate differentials and debt repayment dynamics rather than political decoupling — foreign affiliates repaying intercompany loans in a higher-dollar-rate environment. The decomposition matters analytically, but the macroeconomic implication is the same: foreign capital is no longer reinvesting in Chinese productive capacity at scale. Greenfield and capacity expansion investment have collapsed.

Trade flows present the inverse asymmetry. Setser (2025), in his closely-tracked CFR analysis, estimates China's current account surplus at over $650 billion through Q3 2025, compared with IMF spring projections of $375 billion for the full year. Setser further argues that recent revisions to Chinese current account methodology have understated the true surplus by approximately $500 billion (Yahoo Finance/Benzinga 2025). Passenger car exports reached approximately 6 million in 2025 and are projected to exceed 8 million in 2026 — surpassing Japan's historical peak (Setser 2025).

The political economy of this asymmetry is unstable. The EU has imposed tariffs of up to 35 percent on Chinese EVs, with discussions in early 2026 of replacing tariffs with price floors and voluntary export restraints (Autoblog 2026). The Trump administration's 100 percent EV tariff was incorporated into the Phase 1 truce of November 2025 (IMF 2025), but the underlying structural pressure remains. China's own "anti-involution" (反内卷) policies — phasing out EV purchase tax exemptions, removing solar VAT rebates — represent an implicit recognition that the bleed-and-dump industrial model is internally unsustainable (Firstlinks 2026).

3.4 Industrial Policy and Technology: Selective Capability, Marginal Macroeconomic Impact

The Rhodium Group (2026a) provides the most analytically important finding on industrial policy: from 2023 to 2025, the new high-technology industries (AI, robotics, electric vehicles) added approximately 0.8 percentage points to economic output, while real estate and traditional sectors contributed approximately –6 percentage points. The high-tech sector cannot, by orders of magnitude, offset the contraction in legacy sectors.

This does not imply that industrial policy lacks effect — Naughton's (2021) foundational analysis documents Chinese industrial policy spending at 1.5–2 percent of GDP annually, the largest in modern history. The 2026 national S&T budget is approximately RMB 1.3 trillion, up 7.1 percent (Rhodium Group 2026c). The effect is real and visible in specific sectors. But its macroeconomic impact is insufficient to substitute for the collapse of the property-driven growth model.

In semiconductors specifically, the evidence supports selective catch-up but undermines claims of global competitiveness. SMIC has achieved 7nm production at yields of approximately 20–40 percent (vs. TSMC's ~90 percent) and is attempting 5nm via DUV multi-patterning at yields of approximately 20 percent and cost premiums of approximately 50 percent over TSMC (AEI 2026; SemiWiki 2025). Huawei targets 1.6 million high-end AI logic dies in 2026 for its Ascend product line (AEI 2026). CSIS (2026a) projects that domestic chips will reach 50 percent of China's AI chip market in 2026 — a substantial increase, but driven by import substitution under export controls rather than by global commercial competition.

This is the critical analytical point. Chinese semiconductor advancement is a sanctions-response capability, not a global competitive frontier. The chips are more expensive, lower-yielding, and produced at a scale that requires sustained state subsidization. The structural ceiling at sub-5nm nodes remains binding without EUV access, and China's indigenous EUV programs (LDP-based systems) remain behind schedule (DigiTimes 2025).

3.5 Political Economy: The Contradiction at the Center

Rhodium Group (2026c) and Naughton (2021) converge on the structural contradiction of the Xi-era political economy: the regime simultaneously demands maximum private-sector innovation in strategic industries and maximum political control over private actors. The crackdowns on Alibaba, Ant, Didi, the education sector, and consumer internet platforms from 2020–2022 demonstrated to all potential entrepreneurs that crossing political lines, even on technical or economic grounds, carries existential consequences. The rational response — reduced risk-taking, capital flight, talent emigration — is empirically observable in declining venture investment, slowing unicorn formation, and a documented exodus of high-net-worth individuals to Singapore, Japan, and the UAE.

This contradiction explains why headline indicators of strategic ambition (semiconductor self-sufficiency targets, AI capability claims) coexist with declining indicators of underlying innovation infrastructure (FDI, R&D productivity, originator biotech). The regime is attempting to substitute state direction for the conditions that historically generated frontier innovation. The historical track record of this substitution is poor.

4. Five Common Claims, Debunked Through Diagnostic Framework

Applying the four-type error taxonomy and the M1–M6 protocol, the following widely-circulated claims fail systematic scrutiny:

Claim 1: "China will dominate all advanced industries within a generation."

Error type: Unit mismatch (Type 2) compounded with extrapolation (Type 3). The claim aggregates batteries, semiconductors, biotech, quantum, aviation, software, and pharmaceuticals into a single dominance vector. Disaggregation under M3 reveals these are eight functions with divergent trajectories: Chinese leadership in batteries and solar coexists with structural lag in originator pharmaceuticals (zero global blockbusters), aviation engines (no commercial CJ-1000A), advanced lithography (no domestic EUV), and software platforms (no global cloud or OS leader). The aggregation is empirically false. Extrapolation from EV and battery dominance to all advanced sectors fails M4 (scope consistency) — a five-to-ten year tangent in two industries cannot project to thirty-year dominance across all industries (Pettis 2025; CSIS 2026a).

Claim 2: "If China dominates semiconductors, global hegemony shifts to China."

Error type: Unit mismatch (Type 2) — reduction of hegemony to a single industrial input. Hegemony in international relations is conventionally measured across at least nine dimensions: military force projection, reserve currency status, alliance density, technological standard-setting, higher-education attraction, soft power, original technology generation, food and energy security, and geographic safety (relevant compositions in Brad Setser 2025; CFR institutional analyses). The US retains overwhelming leads in eight of nine. Semiconductor leadership is a critical input but not the sole determinant. The claim collapses the multidimensional vector to one variable, then applies a counterfactual reasoning that does not survive disaggregation.

Claim 3: "China's industrial dominance gives it control over global innovation."

Error type: Order mismatch (Type 1) — conflation of market share (state) with innovation trajectory (derivative). Producing 80 percent of solar panels does not confer control over next-generation perovskite research at Oxford, MIT, KAIST, or EPFL. Producing 50 percent of batteries does not confer control over solid-state, sodium-ion, or lithium-sulfur research at Toyota, Samsung SDI, QuantumScape, or academic laboratories worldwide. Manufacturing market share and frontier-research trajectory are different functions with different dynamics. Dominance in commodified production is a form of dumping leverage; it is not innovation governance (Pettis 2025; Rhodium Group 2026a).

Claim 4: "China's economy is on the verge of collapse."

Error type: Discontinuity over-detection (Type 4) compounded with extrapolation (Type 3). The data does not support collapse. It supports stagnation with high tail risk. The IMF (2025) baseline is 4.5 percent growth in 2026. Rhodium Group (2026a) finds the system functionally inefficient but not failing in the near term. The "collapse" school (e.g., elements of Gordon Chang's longstanding thesis) has been falsified across a quarter century by the system's continued operation. The correct base case is Japanification (Rogoff and Yang 2026): a prolonged, painful adjustment over decades, not a sudden discontinuity. Discontinuity over-detection produces strategic miscalibration in the opposite direction from domination claims.

Claim 5: "China cannot innovate; the system is too rigid."

Error type: Unit mismatch (Type 2) — failure to distinguish between innovation domains. China has demonstrated genuine innovation in process engineering and applied scaling (CATL battery chemistry industrialization, BYD vertical integration, DJI consumer drones, DeepSeek model architecture efficiency). It has not demonstrated parallel capability in originator biotech, foundational software platforms, or frontier theoretical research at scale. Treating "innovation" as monolithic produces an under-detection of real capabilities in specific domains. The accurate finding is bounded innovation: strong in domains where state direction and market scale align, weak in domains requiring patient capital, intellectual freedom, and global integration (Naughton 2021; CSIS 2026a).

5. Conclusion

The data assembled here supports neither the maximalist threat assessment that has dominated Anglophone discourse since approximately 2018 nor the maximalist collapse assessment that periodically resurges. It supports a more demanding intermediate position: China is a structurally decelerating but still globally significant economy, with selective high-technology capabilities, severe and accumulating macroeconomic constraints, and a binding demographic ceiling. The base case for the 2030s is approximately Japan in 1995–2005, but at lower per-capita income, with greater external pressure, under more authoritarian political conditions, and across a population roughly ten times larger.

The strategic implications of this base case differ materially from either of the popular alternatives. The principal external risk to the world economy is not Chinese global domination of innovation chains; it is the disinflationary export of Chinese overcapacity into markets that no longer have the political tolerance to absorb it (Pettis 2025; Setser 2025). The principal internal risk to China is not regime collapse; it is the political consequences of prolonged stagnation under a leadership that has staked its legitimacy on continued ascent.

Analytical rigor in this domain requires three disciplines: disaggregation of composite variables (Chinese power is at least eight functions, not one); explicit calibration of confidence by derivative order (state observations are more reliable than rate-of-change claims, which are more reliable than acceleration claims); and resistance to the institutional incentive structures that produce systematic bias in both directions. The hype, in either direction, is a luxury that policy and investment decisions cannot afford.

REGNIS publishes occasional essays at the intersection of historical analysis and institutional intelligence. The views expressed are those of the author.