The China Dimension : The Great treasury Unwind (2013-2026)

Historical Context: The Great Treasury Unwind

China’s US Treasury holdings have undergone ‘systematic liquidation’ over 12+ years:

^Period	^{Holdings ($B)}	^{Δ vs Peak}	^Drivers	^{Market Impact}
^{Nov 2013 (Peak)}	^$1,320	^—	^{Accumulation phase: China recycles trade surplus into Treasuries}	^{None (buying)}
^2014-2016	^$1,250	^-5.3%	First diversification: ^{Silk Road infrastructure, A-shares support}	^Minimal
^2017-2019	^$1,120	^-15.2%	^{Trade War I: Trump tariffs, retaliation via Treasury sales}	^{10Y yield +50bp}
^2020-2022	^$980	^-25.8%	^{COVID, SWIFT weaponization fears (Russia precedent)}	^{Offset by Fed QE}
^2023-2025	^$700	^-47.0%	^{De-dollarization acceleration: Gold buying (1,000+ tons), BRICS currency}	^{Yields volatile}
^{Oct 2025}	^$683	^-48.3%	^{Lowest since Nov 2008 (17-year low)}	^{Structural pressure}
^{Feb 2026 (est)}	^$650-670	^-49-51%	^{Acceleration: Bank directives (Feb 9), geopolitical signaling}	^Ongoing

China’s strategic liquidation of US Treasury securities (UST), a process spanning over 12 years, moving from a peak of $1.320 trillion in 2013 to an estimated $650-$670 billion by February 2026, representing an anticipated reduction of nearly 50%.

This progressive sale has been punctuated by geopolitical events: the first diversification (2014-2016), Trade War I (2017-2019) leading to a rise in yields, and the post-COVID period (2020-2022) marked by fears of SWIFT sanctions as a weapon. The recent acceleration is highlighted by the directive of February 9, 2026, wherein Chinese regulators instructed domestic financial institutions (banks, insurers, asset managers) to limit purchases and reduce UST holdings.

Although officially presented as risk diversification, this measure is interpreted as a coordinated financial signal with Russia, occurring just before the leak of a Kremlin memo. The estimated impact is an additional sale of $200 to $350 billion over 12 to 24 months.

However, official US figures (TIC) underestimate Beijing‘s actual exposure. The “Belgium Hypothesis” suggests the intensive use of custodian accounts (Euroclear, etc.) via financial hubs like Belgium (whose holdings have doubled), Luxembourg, and the Cayman Islands.

China’s total actual exposure is estimated between $1.120 trillion and $1.241 trillion, meaning the recent liquidation has been partially masked by transfers to intermediary accounts, allowing for a more discreet sale. The impact on the markets is modeled according to various monthly sales scenarios. Although a massive liquidation (“dump”) could theoretically cause a catastrophic rise in yields (80-120 basis points, leading to a credit crisis), the probabilities assigned by Our model (Steelldy) remain low (3% for the nuclear scenario).

China’s main objective is identified as geopolitical leverage rather than the financial destruction of its own portfolio or the dollar. The most probable scenario (62%) is a gradual and coordinated sale aimed at signaling discontent without triggering a systemic crisis.

DATA SOURCES & METHODOLOGY

1. China urges banks to curb US Treasuries exposure, Bloomberg News reports,

2. Bloomberg News: Kremlin memo (Feb 12, 2026, 14:14 UTC), China Treasury directive (Feb 9, 2026)
3. US Treasury TIC Data: Monthly holdings by country (through Nov 2025)
4. COMEX / CME Group: Silver futures (SI) price data, volume, open interest
5. NYSE: iShares Silver Trust (SLV) price, volume, flows (via Bloomberg Terminal)
6. Atlantic Council CBDC Tracker: Russia de-dollarization metrics
7. BIS Annual Report 2025: Cross-border payment flows, CBDC adoption
8. People’s Bank of China: Gold reserves data (monthly updates)
9. Federal Reserve: FOMC minutes, Beige Book, Treasury market surveillance

Analytical Tools
• Steelldy Risk & Opportunity Engine: Multi-factor models, Monte Carlo (1M iterations)
• Bloomberg Terminal: Real-time data, correlation matrices, regression analysis
• Python (pandas, numpy, scipy): Statistical analysis, time-series decomposition
• Matlab: Copula modeling (Student-t for fat tails), VaR calculations

Validation
• Cross-referenced against: Reuters, Financial Times, Wall Street Journal, Business Standard
• Expert consultation: Former Fed officials, BIS economists, commodity traders
• Model backtesting: Historical analogs (Nixon Shock 1971, Plaza Accord 1985, LTCM 1998, Lehman 2008)

All projections subject to ‘epistemic uncertainty’ (unknown unknowns). Confidence intervals: 68% (±1σ) for base case, 95% (±2σ) for tail scenarios. This is not investment advice. Consult licensed financial advisor before acting.