US Treasury Yield Repricing — Episode 1: The 5.00% Breach

What changed since episode 0

In episode 0 we registered two separable parts — (a) structural repricing at the long end, (b) forced revaluation in equities and credit — at medium confidence, and named the central open question as regime floor versus mean reversion. The level has now arrived: the 10-year closed at 5.00% on 2026-09-15, the first close at the registered target (intraday 5.03%, the highest since July 2007; FRED DGS10, through 2026-09-15), and the 30-year sits at 5.36% with its ≥5.00% foothold streak doubled from 25 sessions to 50 (FRED DGS30).

The September FOMC delivered a unanimous 25bp hike to a 3.75–4.00% range, the first since 2023: “Inflation remains elevated… The Committee will deliver price stability” (Federal Reserve decision, 2026-09-16; CNBC). Zero of eighteen dots show cuts and twelve see at least one more hike this year; the YE2026 median is 4.1% (Federal Reserve SEP, 2026-09-16; Wolf Street). Chair Warsh attributed part of the long-end rise to “the competition for capital… the so-called hyperscalers are out in the market raising funding” (Wolf Street, 2026-09-16): our AI-capex leg is now the Fed’s own stated explanation.

The ambiguity moved rather than resolved: from floor versus mean reversion to attribution — how much of the 5.00% is durable repricing versus a reversible oil-shock premium.

The evidence both ways

September’s auctions answered episode 0’s named discriminator — regime floor. Demand improved at higher yields: the 10-year reopening cleared at a 4.834% high yield, bid-to-cover 2.71 (+0.18 vs prior), indirect 79.2%; the 30-year at 5.308%, 2.61 (+0.22), and 79.5% (Treasury auction results, 2026-09-09/10). No tails widened. The capital-competition leg is now measurable in issuance: the top five hyperscalers took 12.9% of USD investment-grade issuance year-to-date 2026 against 1.7% in 2024 (InvestmentNews, 2026). August core CPI printed +0.3% m/m vs +0.2% consensus (BLS, 2026-09-11), and the labor escape hatch closed: +162K against +53K consensus, July revised from −23K to +21K (BLS, 2026-09-04). Part (b) shows first leakage: a 20-session DGS10/SP500 correlation of −0.384 (window 2026-08-17→09-15, computed from FRED series), and mortgages at 6.76% (Freddie Mac PMMS, week of 2026-09-16).

The counter-case, at full strength: the 5.00% close is a war artifact atop a fully-priced hike — the least durable components available. The 10-year crossed 5.00% the same week Brent settled at $105.68 after US–Iran strikes and Hormuz tanker attacks (Al Jazeera / Trading Economics, Sep 2026). Oil premia unwind in days, not quarters. Expectations don’t corroborate an inflation regime: the 10-year breakeven fell 5bp on hike day to 2.33% (FRED T10YIE, 2026-09-16); core CPI is 2.4% y/y, down from 2.5% (BLS, 2026-09-11). Yields initially fell on delivery — buy-the-fact (Wolf Street, 2026-09-16). Buyers feast rather than strike: ~79% indirect and improving coverage is mean reversion seen from the inside; a fiscal-distrust repricing would show concession. The move is global, not US-specific — JGB 10-year +96bp YTD to ~3.3% (StoneX, 2026-09-16). Part (b) is absent broadly: the S&P 500 sits 2.73% below its August high, spreads near tights (FRED SP500/BAMLC0A0CM/BAMLH0A0HYM2, 2026-09-15/16). Stated honestly: the counter explains the timing of the print better than the thesis does, and the auction-demand fact is two-sided — only the 20-year is soft (52.5% indirect on 2026-09-15 after 55.1% in August; TreasuryDirect). It no longer explains the policy leg — a unanimous hike with zero cut dots is a regime fact, not an oil artifact — or the capital-competition leg, where the 12.9% issuance share is structural and growing.

Where we stand

Part (a): strengthened — all three registered drivers gained evidence, and the series records its first executed mechanism, the Fed-hike leg. Part (b): from absent to nascent, still unconfirmed; a −0.384 correlation and a 7bp mortgage move are friction, not revaluation. On the ladder we move part (a) from medium toward medium-to-high; part (b) carries no confirmation, so the honest single episode value is medium. No numeric probabilities — no calibrated basis exists.

The crux is attribution: the 10-year’s +28bp since registration splits +19bp real and +6bp breakeven, roughly 68%/22% of the observed move (FRED DFII10 through 2026-09-15, T10YIE through 2026-09-16; as-of dates differ). Caveat: the real-yield rise conflates expected policy path with term premium — some is rate-path repricing, structural but Fed-determined. The 10Y−2Y curve at +0.27% (FRED T10Y2Y, 2026-09-16), flattened from +0.53% a month ago, fits either reading.

What would falsify it now

The full ledger carries forward, unrevised.

Not one is close. F1’s distance widened from +42bp to +70bp (5.00% close vs the 4.30% line; zero sessions below), F2’s from +45bp to +56bp. F3 moved further from both legs: BEI at 2.33% sits 67bp below its line, and the cut leg is effectively dead for 2026. F4 saw no flight-to-quality (the 10-year rose 32bp over 21 sessions); F5 has no enacted consolidation, with war costs above $33.4B pushing fiscal pressure the other way (Pentagon, via CBS, Sep 2026). The nearest live risk is not a ledger condition firing; it is a de-escalation unwinding the level before the “sustained” qualifier is earned — a 10-year holding ≥4.80% through an oil retracement would be the cleanest structural-floor evidence yet. Catalysts: August PCE (~2026-09-25); 10Y reopening (2026-10-01); 20Y/30Y auctions (~2026-10-15); debt-management statement (2026-10-16); October FOMC (2026-10-27/28); November 4 refunding, resetting the coupon path; December FOMC/SEP (2026-12-08/09); Hormuz headlines and Brent.