Decision day: first hike since 2023 priced at 92–95%; the dot plot and Warsh's presser carry the risk

0. Weekly Arc

The past week completed the hawkish consolidation that began with Chair Warsh’s Aug 28 Jackson Hole keynote. The Sept 4 payrolls beat, the Sept 10 PPI and the Sept 11 core CPI beat moved September hike pricing from a post-Jackson Hole ~70% to an 85–90% band and now to a 92–95% near-certainty, pulling the 10-year above 5% for the first time since 2007 and the 30-year above 5.35%. The ECB hiked on Sept 10; the BOJ and BOE decide next. The tradeable question is no longer the decision but how Warsh frames it.

1. Policy Narrative & Expectations

The past ~24h have removed the last meaningful uncertainty about the decision and concentrated it in the communication. With the FOMC statement, the updated Summary of Economic Projections and Warsh’s press conference all due inside a two-hour window, hike pricing has firmed into a 92–95% band across vendors rather than extended materially ^[1][2][3][4] — market-implied odds were already ~87–90% in the Sept 14 briefing — while OIS pricing shows roughly 23bp of tightening discounted for this meeting and a ~95% implied probability on Deutsche Bank’s read ^[5]. Attention has therefore shifted, in Bloomberg’s framing, from whether the Fed raises borrowing costs to how many hikes follow ^[6]. The consensus sell-side map for the dot plot has settled: Goldman Sachs expects a narrow 10-to-8 majority for only one further 2026 hike, with Waller at zero, 2027 and 2028 each showing one cut at 3.625% and 3.375%, and the 2026 core PCE median nudged down from 3.3% to 3.2% ^[7][8]; Citi expects the move to be framed as a “calibration” or “fine-tuning” rather than the start of an aggressive cycle, with the median dot showing one more hike this year and cuts resuming in 2027 ^[8][9]. The counterweight is that Warsh is widely expected not to submit his own dot at all, leaving the chart with 18 points rather than 19 and reinforcing his stated resistance to forward guidance ^[10][7][11].

1.1 FOMC Officials’ Remarks

• [NEW] Dovish: Governor Stephen Miran, on CNBC on Tuesday (Sept 15), argued that a rate hike would be a mistake — the lone sitting FOMC voice in this batch against the move, per Business Insider’s relay ^[12].
• [ONGOING] Neutral/swing: Governor Christopher Waller — the operative line remains that “sternly staring at inflation until it melts before our withering gaze is not an option,” per CNBC ^[13]; his earlier dovish framing (holding if disinflation continues, considering a hike if inflation runs hot) is unchanged in substance ^[7][8].
• [ONGOING] Chair Warsh (listed separately): no fresh public remark; the Jackson Hole frame still governs — he said recent inflation reports “do not tell me that underlying trends have improved,” that 12-month PCE inflation is 3.7% and the six-month annualized rate 4.1%, and that absent clear progress toward 2% “we have work to do” ^{[10][7][8][14]}. Goldman relayed his Jackson Hole warning against the “hall of mirrors” of officials adopting market views that merely reflect the Fed back at them ^[10].

1.2 Policy Signals & Institutional Communication

• [NEW] The decision mechanics are the event: the statement and updated projections land at 2 p.m. EDT (1800 GMT) with Warsh’s press conference at 2:30 p.m., i.e. 02:00 and 02:30 Beijing time on Sept 17 ^{[10][11][1][3][15][16]}.
• [NEW] Warsh is expected to withhold his own dot: he has publicly criticized the dot plot and declined to submit a projection at his first meeting as chair in June, so this chart may again carry only 18 points ^[10][7][11].
• [ESCALATED] The dot-plot consensus has hardened around “one more, then done”: Goldman expects a 10–8 majority for a single additional 2026 hike with the neutral rate drifting up to about 3.25%–3.5% over the coming year ^[7][8]; TD Securities expects no forward guidance but a hawkish dot ^[3]; Citi’s base case is a “dovish hike” in which guidance no longer points to further increases and the core PCE forecast is revised down on methodology changes ^[8][9].
• [ESCALATED] The institutional forecasters are now clustered on a shallow cycle: HSBC flipped its long-held “on hold” view to a 25bp hike today plus another 25bp by year-end 2026, then unchanged through 2027 — explicitly a small cycle versus a post-1990s norm of at least 200bp ^[17]; ING and Nomura both frame the move as a “recalibration” rather than the start of a consecutive hiking cycle ^[3][18]; BofA expects 75bp completed within 2026 — smaller in total but faster than the ~100bp markets price over a year ^[19].
• [ESCALATED] Deutsche Bank frames a hold as near-impossible: its FOMC preview argues a hold would be the largest dovish surprise at a scheduled meeting since the Fed began announcing policy actions in 1994, and that the committee would be “very uncomfortable” engineering one ^[5].
• [ONGOING] A dissenting hold camp persists: Standard Chartered judges the Fed should hold this week and calls a hike the “wrong policy choice,” arguing it has a very low cost to wait ^[8][20]; Oxford Economics argues three months of relatively mild inflation could just about assemble a slim majority to stand pat ^[7].
• [ESCALATED] Political pressure continues into the meeting: President Trump reiterated on Sept 13 that the US “should be paying the lowest interest rate in the world,” following his threat this month to cut off trade with surplus countries if the Fed does not cut ^{[12][11][21][14]}; Council of Economic Advisers chairman Christopher Phelan told CNBC on Tuesday it “doesn’t make sense to raise rates now” ^[22].
• [ONGOING] Market-implied hike odds sit in a 92–95% band across vendors, up from the 85–90% band of the Sept 14 briefing ^[1][2][3][4].

2. Key Data & Market Read

• [ONGOING] August CPI (released 9/11): headline in line with expectations at 3.4% year-on-year and 0.4% month-on-month; core rose 0.3% m/m, above expectations, with the year-on-year rate easing slightly ^{[22][16][23][24]}.
• [ESCALATED] The composition read still says “narrow, not broad”: the core overshoot was concentrated in volatile items — wireless communications services up 5.9% in the month, airfares and lodging — while rents and owners’ equivalent rent rose only about 0.2% and medical services prices fell, leaving no evidence yet of broad-based, sustained pressure ^[24]; energy was the headline driver, with the energy CPI sub-index rising year-on-year and month-on-month on the oil move ^[23].
• [ONGOING] July PCE (the Fed’s target measure) ran at 3.7% year-on-year ^[10][1].
• [NEW] A weak auction and a weak regional print complicate the hike: the recent 20-year Treasury auction tailed by 2bp with end-user demand falling to its lowest since February 2026 ^[25], and the September Empire State manufacturing index fell to 7.6 against a 15.0 consensus ^[25][19].
• [NEW] Household energy costs are the transmission channel into headline inflation: diesel prices hit a record $6.27 a gallon on Tuesday and gasoline jumped 18 cents over the week to $4.33, per AAA ^[16][26].
• [NEW] Activity tracking was revised up: BofA raised its third-quarter 2026 GDP tracking estimate to 2.6% quarter-on-quarter annualized on stronger-than-expected services revenue and consumption ^[19].
• [NEW] Today’s calendar is Fed-light and consumer-heavy: US August retail sales and import prices, the September NAHB housing index, July business inventories and EIA crude inventories, alongside UK August CPI, euro-area July industrial production and the Bank of Canada minutes ^[27][15].
• [NEW] Narrative impact: the data continue to justify the hawkish read without settling the path. The hawkish case rests on a core PCE conversion that several houses read as stalling above Waller’s threshold ^[8][5]; the dovish case rests on a single noisy category doing most of the work and a September 30 BEA methodology revision that Citi expects to revise the core PCE forecast lower ^[24][9]. That asymmetry is precisely why the dot plot and the press conference outweigh the decision itself today.

3. Financial-Conditions Signals

• [ESCALATED] The long end is the live transmission channel: the 10-year stands at 5.004% with the 20-year at 5.409% and the 30-year at 5.372% ^[1], after Tuesday’s 5.04% print — the highest since 2007 ^[28][2][29] — and the 2-year also touched a 2024 peak ^[30]. Asian and European curves are moving in sympathy, with Japanese 30-year yields up and Australian 10-year yields sharply higher ^[31].
• [ESCALATED] The driver narrative has shifted decisively to real rates and fiscal/term-premium risk: SPDB’s decomposition attributes the year-to-date move in the 10-year overwhelmingly to real rates rather than breakevens, with expected short-end rates up roughly 61bp and the term premium roughly flat — i.e. a “higher for longer” policy-rate repricing rather than an inflation-expectations shock ^[32]. Jefferies adds a structural note: gold’s correlation with 30-year TIPS real yields has flipped to positive (+0.6 versus a -0.4 historical average), which it attributes to rising fiscal pressure and currency-debasement hedging ^[33]. Bloomberg Economics’ Anna Wong frames the reflexive risk explicitly — if a hike fails to lower long yields, that would confirm fiscal policy, not monetary policy, as the driver, since a hike raises the government’s marginal funding cost and hence future issuance ^[34].
• [ESCALATED] Positioning is at extremes and leans one way: Citi calls Treasury short positioning tactically extreme after a week of chasing higher yields ^[30]; BofA reports shorts accumulated across the entire curve with asset managers cutting longs or adding shorts and almost no money buying duration at lows ^[30]; JPMorgan’s client survey shows shorts jumping 10 percentage points in the week to Sept 14, the fastest build since early 2025, with net longs at a four-month low ^[30]; the market is described as “extremely” short with almost no dip-buying ^[4].
• [NEW] The investor-survey read confirms the crowding: BofA’s September Global Fund Manager Survey shows 33% naming a “disorderly rise in bond yields” as the top tail risk, overtaking the AI bubble, with a net 48% underweight in bonds and cash at 3.9% triggering a sell signal ^[35][36][37].
• [NEW] Dollar and rates plumbing: the dollar index traded near a two-week high around 99.656 ^[2] with USD/JPY in the mid-155s ^[38][39]; Treasury Secretary Bessent, asked about the yield rise, attributed it to “global issues” and called the upsized buybacks a success ^[40][41].
• [ONGOING] The buyback backstop has not worked: the expanded long-dated operation has failed to push yields down, and the market read it as ineffective ^[42][43].
• [ONGOING] Household credit transmission remains intact: card rates above 20% are expected to rise further and the average 30-year mortgage sits near 7.4% ^[21][44].

4. Global Central-Bank Linkages

• [ESCALATED] BOJ: BofA expects a 25bp hike to 1.25% at the Sept 17–18 meeting, almost fully priced at roughly 97%, followed by roughly quarterly hikes to a 2% terminal by July 2027, with the implied neutral range at 1.1%–2.5% and Ueda expected to validate a faster pace without committing to timing ^[45]; Nikkei frames the BOJ as eyeing hikes to fight inflation ^[46], and Dow Jones relays a shift from cautious normalization to a nimbler, data-responsive stance ^[47]. BofA recommends selling 6m1y payer 1x2 options and stays short USD/JPY ^[45].
• [ESCALATED] ECB: swaps now imply four more quarter-point increases from the ECB as the energy-driven inflation revival turns money markets more hawkish than central bankers ^[48]; Nomura expects hikes in December 2026 and March 2027 taking the deposit rate to 3.00% ^[18].
• [NEW] BOE: UBS expects two “adjustment” hikes over the next 12 months via the rate-expectation curve, with the front end showing a “hump” for the first time since the conflict began and the MPC tone turning more cautious on energy ^[49]; BofA expects a hold at 3.75% on a 6–3 vote with a more hawkish tone than July, sees market pricing of ~110bp of hikes over the next year as too aggressive, and expects QT slowed to £50bn a year with a possible halt to long-dated gilt sales ^[50].
• [NEW] PBoC / China: August financial data showed aggregate financing and loan growth continuing to slow, with direct financing (corporate bonds, equity and government bonds) exceeding the share of RMB loans for the first time on record ^[51]; the PBoC’s Q2 report de-emphasised loans as a single financing channel ^[52]. HSBC reads the fast pace of RMB fixing adjustment as a deliberate policy bias toward appreciation, with August net FX settlement rebounding sharply and importer purchase demand falling to its lowest since January 2013 ^[53].
• [ONGOING] BoC: minutes are due today, with next week’s UK and Japan decisions following in succession ^[15][54].

5. Asset Implications

Quadrant	Current probability tilt	Key asset implication	Anchoring narrative
Growth↑ + Inflation↑	Rising	Energy/commodities and TIPS stay the cleanest expressions — diesel at a record and gasoline up on the week — while nominal long bonds are blocked by a real-rate-driven selloff with the 30-year above 5.35% and a weak 20-year auction	§2 / §3
Growth↑ + Inflation↓	Falling	The “calibration” path: Citi’s base case of a dovish hike with a median dot showing no further increases, plus Goldman’s downward core PCE revision, favours quality and rate-insulated equity beta over duration, and the front end over the long end	§1.2 / §2
Growth↓ + Inflation↑	Rising (tail)	The supply-shock leg remains policy-insensitive: diesel and gasoline records, a weak Empire State print and rising term premium mean nominal bonds hedge neither leg, so gold’s official-sector bid is the hedge rather than duration	§2 / §3 / §4
Growth↓ + Inflation↓	Falling	The hold tail is now a genuine tail and expensive to own: Deutsche Bank frames a hold as the largest dovish surprise since 1994, and a no-hike read would likely steepen the curve and hit both equities and the long end simultaneously	§1.2 / §3

Stock-bond correlation call: the regime remains the inflation/policy-driven positive-correlation configuration hardest for risk parity, but today’s tape carries a nuance worth flagging. The long end is being set by real rates — SPDB’s decomposition puts the entire year-to-date move in real terms with breakevens barely moving ^[32] — which is a policy-path and term-premium signature rather than an inflation-compensation shock; historically that configuration is the one in which duration fails to hedge equity drawdown but does not correlate with equities in a pure inflation-hedge sense either. Citi’s event study adds a tactical layer: equities typically pop on the statement then fade into the press conference and close slightly negative, Treasuries mean-revert after an initial extreme, and hawkish FOMC days see multi-day continuation in yields with gold selling off more than 3% ^[55]. The escape hatch remains narrow and event-specific: if the hike lands with a dot plot that caps the cycle at one more move and Warsh frames it as calibration, the policy-uncertainty component of the term premium can compress and medium duration can partially recover its hedge value. If instead the dot shifts up and the tone is hawkish, both legs of a risk-parity portfolio are repriced by the same variable again. The reflexive tail — if tightening fails to lower long yields, fiscal rather than monetary policy is confirmed as the driver — is the scenario in which the correlation structure itself shifts rather than merely the level.

Risk-budget implication: underwrite the decision as fully priced and the communication as two-sided, and do not treat 92–95% as a signal. Keep commodity and energy risk at or above benchmark: energy is the only sleeve whose driver — the Middle East supply shock and refining constraints — is orthogonal to the Fed’s reaction function, and it is the cheapest hedge against the stagflation quadrant. In rates, stay underweight long-end nominal duration into the decision but pre-commit to rebuilding it on a calibration read; prefer curve and relative-value expressions to outright shorts, given documented extremes in short positioning that make a dovish reading a squeeze risk, and note that the demand base for duration is thinning at the long end while the 20-year auction is tailing. In credit, keep duration short and favour spread products over pure duration; household credit transmission via card rates above 20% and mortgages near 7.4% is now the drag rather than corporate leverage. In equities, hold moderate beta but rotate toward lower rate-sensitivity and cash-flow durability, and treat the equity-risk-premium compression and the 5% threshold on the 10-year as the disciplined stop for adding risk. Hold gold as the structural fiscal and dollar-credit hedge, sized for continued real-rate whipsaw, and recognise that its correlation with 30-year real yields has flipped positive — it is not a substitute for duration in a growth shock.

6. Contrarian & Tail Risks

• Consensus fragility: the 92–95% pricing is a self-referential construction, and the historical regularity that hikes land when pricing exceeds 40% cuts both ways — a miss at this level would be unprecedented in the recent sample and would presumably trigger exactly the term-premium backlash the committee is trying to avoid. The falsifiable pillars are narrow: (1) the entire core exceedance rests on wireless services and volatile travel items, with core PCE conversion only marginally above the threshold Waller has cited; (2) the dot plot is the swing variable, and the camp expecting one further hike, the camp expecting none, and the camp expecting the median shift up to imply multiple hikes cannot all be right; (3) Standard Chartered and Oxford Economics both still argue a hold, and the vote arithmetic is genuinely close; (4) Warsh’s own stated method — refusing to guide, learning from markets — is circular when the market’s expectation of the hike is itself the justification for the hike; (5) the survey arithmetic question raised by the CNBC Fed Survey is that respondents expect both multiple hikes and unchanged growth and equity outcomes, which are not obviously compatible.
• Second-order transmission: three chains are not in market pricing. The fiscal-reflexivity chain is the structural one — a hike raises the government’s marginal funding cost, enlarging future interest deficits and Treasury issuance, and if investors doubt fiscal offset the term premium rises, worsening debt service rather than improving it. The energy pass-through chain is still developing: diesel at a record and gasoline up sharply will land in September and October prints rather than the August survey period, meaning the Fed is tightening into an impulse rather than ahead of one, and the same cost pressure directly compresses household real income. The AI-financing chain remains the equity-transmission vector — large-scale corporate borrowing to fund AI capex adds debt supply while stimulating demand, and a majority of fund managers now see AI hyperscaler capex as the most likely source of a credit event. Layered on top, the political channel is active: a hike two months before the midterms runs against presidential preference, and a no-hike outcome would be read as capitulation.
• Source quality control: hike odds are a band across instruments and snapshot times — 92.4% (CME FedWatch), 92.5% (CNBC/CME), 93% (Reuters and Business Insider), 94% (Jin10), 95% (Gelonghui/CME), 90% (NYT and CME in separate citations), 88% (predict.fun), and “more than 90%” (Bloomberg and a dozen others) — so sequencing and instrument matter more than any single reading, and the ~95% figure should not be treated as a precision instrument. Single-source or unverified items include the Bloomberg “History Shows Fed Will Deliver Rate Hike” headline relayed via a social post, the Anil Vohra 5% cross-over thesis, the deerpointmacro “hike as clearing event” argument, Dario Perkins’ claim that fear of the bond-market reaction is itself why the Fed hikes, TimmerFidelity’s correlation and valuation observations, Bob Elliott’s hedge-fund positioning correction, El-Erian’s pre-meeting note, and any commentary attributed to unnamed strategists. Two internal conflicts deserve side-by-side reading: the long-end driver is variously attributed to real rates and policy-path repricing, to a restored term premium from fiscal supply, and to energy-driven inflation — three causal stories for the same yield; and the dot-plot expectations of Goldman and Citi sit against TD Securities’ hawkish-dot call. Finally, several of today’s “remarks” are restatements of speeches from earlier weeks rather than fresh guidance, and the batch contains items with no extractable content.

---

Appendix: Additional Sources

• ^[56] Jin10 — the tradeable element is whether Warsh regains control of the policy narrative
• ^[57] Jin10 — hike fully priced; the “stress test” lies ahead; Treasury yields at 19-year high and gold stalling near $4,300
• ^[58] Jin10 — nominal and real yields surging together have not crushed gold; dollar weakness and sovereign-debt risk are supports
• ^[59] Jin10 — gold trading a “bigger thing”; watch whether long-end yields stabilise or keep climbing
• ^[60] Jin10 — bond shorts crowded to an extreme; real volatility may come after the script lands
• ^[61] EconBerger (LinkedIn, social) — livestream announcement only, no policy content
• ^[62] Peifengke (Chen Dapeng) — long-term logic versus entry point; gold and copper caution
• ^[23] Wanlian Securities — August CPI/PPI detail; hike cannot be ruled out; caution advised
• ^[63] Datong Securities — metals weekly; September hike probability and dollar strength drove the copper slide
• ^[64] Nick Timiraos relay — central-bank watchers expect a hike this week and another within the year

---

This report is a macro-mechanism analysis, not investment advice.