The Walled Garden of Debt

China borrows at 1.7% while the model says it should pay 13.4%. That 11.7-percentage-point gap — the largest structural distortion in global sovereign credit — is not a market signal. It is a wall. And walls, in the end, are what autocracies build best.

312%
Total non-financial debt
to GDP (2024)
124%
Augmented govt debt/GDP
incl. LGFV (IMF est.)
−11.7%
Yield gap: model vs market
largest structural outlier
58T
RMB
LGFV debt (IMF est.)
PBoC claims 14.8T

The Debt Architecture

China's debt is not one problem. It is four problems stacked inside a black box. The official general government debt — 69% of GDP — would be unremarkable for a major economy. But that figure is a fiction maintained by accounting convention. The real figure, once you include the off-balance-sheet liabilities of Local Government Financing Vehicles, is the IMF's "augmented" estimate of 124% of GDP. And even that understates total exposure: non-financial sector debt stands at 312% of GDP, a ratio exceeded only by Japan among major economies — and Japan is three times wealthier per capita.

The core pathology is the LGFV system: approximately 10,000 quasi-commercial entities created by local governments to circumvent borrowing limits. They funded the infrastructure boom. Now the boom is over, land revenues have collapsed, and LGFVs cannot service their debts. The IMF estimates LGFV liabilities at 58 trillion RMB (48% of GDP). The PBoC governor claims 14.8 trillion. The 4× divergence between China's central bank and the IMF on the same metric is itself a symptom of the governance topology: in an L=5 system, there is no independent audit, no free press to investigate, and no opposition to demand transparency. The debt's true scale is unknowable — and that unknowability is the risk.

China's Debt Stack
Multiple layers of obligation, each obscuring the next. Official figures capture only the visible portion.
Official govt debtMoF reported+ LGFV debtIMF augmented+ Corporate & HHTotal non-financial+ Shadow bankingTotal social financing69%of GDP124%of GDP — "the real number"312%~345%TSF: ¥430T (Jun 2025) — growing at 8.9% vs nominal GDP 4.1%0%100%200%~300%+← Official visibility ends here
IMF Article IV (2025); World Bank China Economic Update (Dec 2025); PBoC; Total Social Financing via NBS. LGFV estimates vary 4× between official and IMF figures.
"The problem is not the debt itself, but the large-scale misallocation of investment funded by the debt. China has 'capitalised' an expense: converting what should have been recognised as losses into fictitious assets."
— Michael Pettis, Carnegie Endowment, 2025

The critical dynamic: debt is growing at twice the rate of nominal GDP. Total social financing expanded 8.9% year-on-year through mid-2025 while nominal GDP grew just 4.1%. Each unit of new credit produces less than half a unit of output. This is the textbook definition of a debt trap — and it has been worsening since 2008. Prior to that year, China's rapid debt accumulation was matched by rapid GDP growth, keeping ratios stable. The divergence since is the strongest evidence that a rising share of investment is non-productive.

The Four Roads — Applied to China

The sovereign credit model identifies four historical paths out of unsustainable debt. For China, each path is constrained by the governance topology in ways that make resolution uniquely difficult.

Road 1: Financial Repression
CURRENT PATH
Force savers to accept below-market returns. Use capital controls to trap domestic capital. Direct state banks to roll over LGFV debt at below-cost rates. Suppress yields via closed capital account. Transfer wealth from households to state balance sheet.
Who pays: Chinese households — via 1.7% deposit rates during near-deflation, destroyed property wealth, suppressed consumption. Liberty impact: −5 to −15 pts (requires tighter controls to prevent capital flight). Duration: Decades. Japan has done this since 1990.
Road 2: Restructure
COSMETIC ONLY
Beijing's ¥10T debt swap (Nov 2024) converts LGFV hidden debt to on-budget bonds — extending maturities, lowering rates. But swaps don't reduce debt, they relabel it. Fitch estimates only 25% of hidden debt is covered. IMF calls for actual restructuring with insolvency frameworks.
Who pays: Nobody yet — that's the problem. True writedowns would expose losses in state banks, require recapitalisation, and admit the growth model failed. Politically: Admission of failure is incompatible with an L=5 system where the Party's legitimacy rests on economic competence.
Road 3: Austerity
BLOCKED
Cut spending, allow defaults, let unproductive firms fail. Would require accepting GDP growth well below the 5% target that anchors CCP legitimacy. Already causing strain: 12 provinces restricted from new investment; local government hiring freezes; contractor arrears of ¥10T.
Why blocked: The social contract is prosperity for quiescence. Breaking the growth promise risks the one thing L=5 systems cannot tolerate: mass unrest. Youth unemployment already politically sensitive (data collection was suspended in 2023 and recalibrated).
Road 4: Inflate / Devalue
BLOCKED
Destroy debt through inflation or currency devaluation. But China is fighting the opposite problem: CPI near zero, PPI deflationary for 2+ years. Massive monetary easing has failed to generate inflation. The liquidity trap is real.
Why blocked: You cannot inflate away debt when the economy won't inflate. Devaluation would trigger trade retaliation, capital outflows, and destabilise the RMB — the one macro variable Beijing controls most tightly. The demographic headwind (shrinking workforce) compounds deflationary pressure.
"China has one road open: financial repression. It is the autocrat's preferred tool because it requires no democratic consent — you simply pay households less than their savings are worth. But it guarantees a decade of suppressed consumption, demographic decline, and slow growth. Japan's lost decades, with Chinese characteristics."

The Structural Outlier

In the Governance Topology sovereign credit model, China is flagged as a "structural outlier" — meaning its yield cannot be explained by governance fundamentals and the mispricing is not tradeable. The model predicts that a country with Liberty=5, Tyranny=87, and debt/GDP of 124% should pay approximately 13.4% to borrow for 10 years. China pays 1.7%. The −11.7 percentage point gap is sustained entirely by the closed capital account: domestic savers have nowhere else to put their money.

Yield Mispricing: China vs. Model Prediction
Where China borrows vs. where its governance quality says it should. The gap is maintained by capital controls, not creditworthiness.
Market yield(what China pays)1.7%Model yield(what L=5 should pay)13.4%11.7pp GAPsustained by capital controls + state-directed bankingImplied CDS: 1,090bp (model) vs 0bp (market). China's CDS doesn't trade because the capital account prevents price discovery.The wall doesn't make the risk disappear. It makes the risk invisible.
Model: Governance Labs 4-factor (Liberty, Debt/GDP, ΔL, GDP/capita). Market yield: PBoC 10yr CGB (Feb 2026). Regression: Y = 33.05 − 0.35×Liberty, R²=0.37.

This is what makes China simultaneously safe and dangerous. It is safe in the narrow sense that Beijing can suppress yields indefinitely as long as capital controls hold. It is dangerous in the profound sense that the price signal that normally forces governments to address debt problems does not exist. In democracies, bond vigilantes eventually impose discipline. In China, there are no vigilantes — just a walled garden where the rules of financial gravity are suspended by political decree.

Sensitivity Analysis: What Breaks the Wall

The model identifies six shock vectors that could breach the capital account wall and force market pricing onto Chinese debt. Each has a probability estimate, a transmission mechanism, and a severity assessment.

ScenarioP (5yr)MechanismImpact on YieldsLiberty Impact
Property cascade35–45%Unfinished housing stock (est. 20M units) triggers bank NPL wave → state bank recapitalisation → fiscal cost 15–30% GDP → confidence collapse → capital flight pressure+200–500bp if controls loosen; contained if notNeutral to −2 (tighter controls)
LGFV default chain25–35%Weaker provinces (Guizhou, Yunnan, Gansu) cannot roll over ¥10T+ in maturing LGFV bonds → contagion to wealth management products → retail investor panic+100–300bp onshore; WMP losses trigger social instability−2 to −5 (crackdown on dissent)
Trade war escalation20–30%US/EU tariff escalation → export shock → RMB depreciation pressure → capital controls tighten → but reduced FX earnings weaken reserve buffer+50–150bp (manageable unless reserves depleted)−1 to −3 (nationalism + controls)
Taiwan contingency5–15%Military action → comprehensive sanctions (Russia-scale) → capital account forcibly opened by asset freezes → yields go to market clearing+800–1,200bp (sanctions regime)−3 to −5 (wartime autocracy)
Demographic cliff~100%Working-age population shrinking ~0.5%/yr → pension liabilities mount → savings rate declines → less captive capital to absorb government debt+50–100bp by 2035 (slow, structural)Neutral (but erodes fiscal base)
Capital account opening3–5%Reform-driven or crisis-forced liberalisation → domestic capital seeks global returns → yield must rise to competitive levels → re-pricing of entire sovereign curve+500–1,100bp (full re-pricing to model)+5 to +15 (if reform) or −5 (if crisis)

The Autocratic Trap: Vulnerability Through the Governance Topology Lens

The five transmission channels through which democratic erosion degrades sovereign credit — legal risk, information degradation, policy unpredictability, capital flight, and repudiation risk — operate differently in a system that was never democratic. China doesn't face erosion. It faces the structural vulnerabilities of consolidated autocracy: the features that make the system stable also make it fragile.

Information Opacity
Critical
No free press, no independent audit, no opposition scrutiny. The 4× divergence between PBoC and IMF on LGFV debt is not a rounding error — it is the cost of L=5 information architecture. Investors cannot know what they don't know. Beijing can hide losses until they become uncontainable.
Error Correction Failure
Critical
No mechanism to change course when policy fails. The 5% GDP target is politically sacred — so investment continues into unproductive assets because admitting lower growth means admitting the model is broken. Youth unemployment data was suspended rather than published. Bad news is suppressed, not addressed.
Succession Risk
Critical
Xi abolished term limits (2018). No successor identified. No institutional mechanism for power transfer. Bondholders pricing 30-year CGBs must assess: will the next regime honour these obligations? Historical base rate for post-strongman debt repudiation: 23%.
Demographic Erosion
Structural
Population peaked 2022. Working-age cohort shrinks ~5M/yr. Old-age dependency ratio doubles by 2050. Pension liabilities unfunded. The captive savings pool that funds government debt will shrink — slowly, then suddenly.
Geopolitical Isolation
Rising
Decoupling from Western tech supply chains. FDI declining. "De-risking" policies reducing economic integration. Each step of isolation reduces the external anchors that constrained policy excess. Sanctions risk (Taiwan scenario) is a fat-tail event with existential credit implications.
Reserve Buffer
$3.2T
The single strongest defence. Foreign exchange reserves of $3.2T provide crisis buffer. But: reserves are static while liabilities grow. The ratio of reserves to total debt has deteriorated from 15% (2010) to ~6% (2025). The buffer is large in absolute terms, diminishing in relative terms.
The Autocratic Credit Paradox
How the same features that suppress yields also create systemic vulnerability. The wall works — until it doesn't.
WHAT SUPPRESSES YIELDSWHAT CREATES FRAGILITYClosed capital accountNo price discovery → invisible risk accumulationState-directed bankingNPLs hidden via evergreening → zombie economyNo free press / auditFiscal data unreliable → 4× LGFV divergencePolitical control of judiciaryNo rule of law → contract enforcement arbitraryGDP target enforcementInvestment forced into unproductive assets → losses capitalisedNo opposition / dissentNo error correction → bad policy persists indefinitelyEach strength is its own weakness
CLINICAL ASSESSMENT
China's debt crisis is unlike any other in the sovereign credit database because the normal correction mechanisms do not operate. Bond vigilantes cannot discipline a closed capital account. Free markets cannot price risk when the state directs all capital allocation. Independent analysis cannot size the exposure when the data is politically managed.

The result is a system that is stable but not sound. The wall holds — and behind it, losses accumulate in balance sheets that no outsider can fully audit. The IMF's best estimate is that restructuring the LGFV system alone could cost 15–30% of GDP. Property sector losses may add another 10–20%. State bank recapitalisation could require a further 5–15%. These are not additive — there is overlap — but the aggregate fiscal cost of true resolution likely exceeds 30% of GDP, or roughly $5 trillion.

Beijing's chosen path — financial repression — is the autocrat's classic resolution: force the losses onto households through suppressed deposit rates, destroyed property values, and reduced social spending. It works. Japan has sustained a version of this for 35 years. But it guarantees a generation of suppressed consumption, demographic decline, and slow growth. The 5% target becomes 3%, then 2%, then the number that dare not speak its name.

The governance topology model assigns China a <2% probability of regime-threatening credit event in the next five years — because the wall is strong. But it assigns a 35–45% probability of a chronic growth downshift that erodes the social contract without breaking it. The danger for China is not explosion. It is slow suffocation — a system that cannot reform because reform would expose the scale of prior losses, and cannot grow because the debt overhang absorbs all new capital.

For sovereign credit investors, the conclusion is simple: China is not tradeable. The yield reflects political control, not credit quality. The day the wall comes down — whether through reform, crisis, or geopolitical shock — yields reprice by 500–1,100 basis points overnight. Until then, the 1.7% yield is the sound of a system talking to itself.