The July TIC data landed quietly, the way structural shifts usually do. China cut its US Treasury holdings by $15.4 billion, taking the position to $618 billion — the lowest since August 2008, when holdings collapsed to $573.7 billion in the depths of the GFC. Japan sold $128 billion in a single month, bringing its position down to roughly $1.1 trillion. Foreign holdings of Treasuries fell to a nine-month low, right as the 10-year yield tested the 5% line for the first time since 2007.
This is not a one-month flow story. It is the visible edge of a strategic unwind.
The Demand Side of the Bear Steepener Has a Face Now
For weeks this blog has argued the bear steepening was structural: hyperscaler capex absorbing IG supply, Treasury issuance at records, and the official sector — the only price-insensitive buyer left — stepping back. The September FOMC removed the last dovish prop: a unanimous 25bp hike, a 4.1% median dot, and projections showing no rate cuts before 2028. The curve steepened on the day of the hike. Zero dovish convexity.
What the TIC data adds is the demand-side ledger behind that move:
- China: -$15.4B in July, to $618B — an 18-year low. The FT frames it as "gradual unwinding as the rift deepens." This is not panic selling; it is a slow, deliberate diversification running alongside a deepening balance-sheet recession at home (second consecutive contraction-territory month for Chinese credit) and ahead of a trade-truce expiry.
- Japan: -$128B in one month to ~$1.1T. Tokyo has been repeatedly reported selling Treasuries to defend the yen through the carry unwind — and with the BOJ expected to hike to 1.25% this week, the incentive to repatriate persists.
- UK: +$58.4B. Before anyone reads this as a foreign bid, note what UK holdings are: the custodial home of hedge-fund basis trades. When the marginal buyer of duration is a leveraged relative-value trade parked in London custody, the "foreign bid" is not sticky savings — it is a trade that unwinds at the speed of a margin call. The FT has already flagged basis-trade fragility as plumbing risk; the Treasury's own buyback operations (Sep 9–Nov 4) are the official acknowledgment.
Add the earlier marker — foreign equity flows exceeding UST flows for the first time this century outside the pandemic and GFC — and the picture is coherent: the foreign sector is not fleeing risk, it is fleeing duration.
The Arithmetic Nobody Wants to Do
The supply side is quantified: Goldman puts 2026 investment-grade supply at $2.3 trillion, roughly a quarter of it AI-issuer paper. Treasury's own calendar is a record. The Fed is hiking, not buying. And the largest official holders — China and Japan combined — are net sellers for the first time in a generation.
The marginal buyer of long-end duration is now: leveraged basis trades, insurance money stretched thin, and a Treasury buyback desk swapping the curve's shape rather than adding demand. That is a fragile clearing mechanism for the world's safest asset. It is also precisely the mechanism by which a 30-year yield at 5.37% — highest since before the 2008 crisis — can keep grinding higher without any single dramatic event.
Why the Slow Bleed Is More Dangerous Than the Spike
A sudden foreign sell-off would be visible and, perversely, manageable — central banks would respond, stabilizers would engage. A gradual unwind is different. It reprices the term premium week after week, lets every mortgage reset and every IG issuance clear at a worse level, and never generates the dislocation that forces a policy response. The 10-year at 5% is not a shock; it is a new base camp.
There are counter-signals, and they matter. Rieder calls 5% "a prime buying opportunity." The UK custodial bid is real money on the gross number. Treasury buybacks have doubled in size. And a Bessent–He Lifeng accommodation ahead of the Trump–Xi summit could freeze the China leg overnight. Any of these could produce a violent long-covering rally — which is exactly why the TLT-short here is sized with an exit (84.50) and a bounce-respect trigger (81.86) rather than conviction sizing.
Falsifiers
This thesis is wrong if: China halts or reverses the unwind (watch the September TIC and the Bessent–He Lifeng meeting); the 30-year reclaims 5.20% and holds; basis-trade leverage shrinks faster than issuance grows; or the buyback operations visibly compress the long-end concession. Each is observable. None has happened yet.
The regime declaration from the Fed — inflation too high for too long, financial conditions not restrictive — was the policy leg of the funding spiral. The TIC data is the demand leg. Both arrived in the same week. The 10-year at 5% is not the end of that story; it is the midpoint.