〔Overnight Brief〕Core CPI +0.3% m/m Locks September Hike at ~85-90%, TD Flips to Three Hikes, MUFG Warns of 'Policy Error' - Long End Stages 'Credibility Rally' on Warsh's 'Inflation Is an Option' Trade

0. Weekly Arc

[ESCALATED] The dovish case for a September hold collapsed on Friday's CPI: Fed funds futures repriced from ~70% to ~85-90% odds for a 25bp hike at the September 15-16 meeting ^[1][2]. Wall Street's "no hike in 2026" cohort capitulated within hours - TD Securities flipped from "no moves" to three hikes through January 2027 ^[3], while JPMorgan, Citi, and MUFG all added at least one September move ^[4][5][6]. The arc is no longer a debate about whether the Fed acts; it is a debate about how much further it goes - and how the long end, which wants Fed credibility on inflation, takes the news ^[7].

1. The Print

[NEW] Headline CPI +0.4% m/m and +3.4% y/y matched consensus, stepping up from the +0.1% m/m prior ^[8]. Core CPI +0.3% m/m beat the +0.2% consensus - largest single-month gain since April - while core y/y cooled to +2.4%, the lowest since March 2021 ^[8]. Energy did the headline heavy lifting: energy index +2.1% m/m, gasoline +3.9%, contributing over one-third of the monthly headline increase ^[8]. The market read was pre-committed: per JPMorgan's pre-release framing cited by Wall Street CN, +0.3% m/m core = hike; +0.2% = hold ^[8]. [NEW] Pass-through into rates and housing: mortgage rates hit their highest levels since early 2025 by end of day, with multiple lenders raising quotes 0.05% in the last few hours ^[9]. Refi demand was already pulling back - MBA total applications -2.7% w/w for the week ending September 4, purchase apps -0.2% SA (-3.0% unadjusted) ^[10]. Existing-home sales fell 2.0% m/m to a 3.98 million annualized pace in August - first sub-4M since June 2025 - with inventory at the highest level in more than a decade ^[11]. University of Michigan consumer confidence fell, with inflation expectations rising (Xinhua relay) ^[12].

2. Wall Street Forecast Revisions

[NEW] TD Securities - the most aggressive flip: from "no moves in 2026" to three 25bp hikes (September, October, January 2027), per strategists Oscar Munoz and Gennadiy Goldberg, who argue the print shows "no further progress" on inflation; the bank also expects a hawkish dot plot and minimal forward guidance ^[3]. [NEW] JPMorgan - now sees September and December hikes ^[5]; the bank's pre-release framing had made the +0.3% m/m core the trigger ^[8]. [NEW] Citi - expects a September hike followed by cuts through mid-2027 ^[4]. [NEW] MUFG - strategists George Goncalves and Agron Nicaj call September +25bp then October hold, citing Fed Chair Warsh's Jackson Hole hawkishness, "solid" August employment data, and the above-consensus core print; they flag the move could be a "policy error" but argue the Fed "cannot afford to do nothing" when markets price the hike in ^[6]. December hike probability: 55-60% ^[6]. The "policy error" caveat and the Warsh framing are MUFG's own characterization, not direct quotes ^[6]. [ONGOING] Fed Chair Warsh - anchor of the hawkish case; MUFG cites his repeated Jackson Hole emphasis that "inflation is an option" ^[6]. Warsh's July 29 comment that he would "let the bond market do the heavy lifting" is the cited origin of the long-end credibility trade ^[7].

3. The Long-End Paradox

[NEW] Mortgage News Daily frames the move as paradoxical: Fed funds futures and the 2-year are selling off on a ~90% September-hike print, yet the long end of the curve is rallying ^[7]. The thesis: the bond market wants the Fed to "get serious about fighting inflation"; yields first spiked on July 29 when the Fed held and Warsh said he would let the bond market do the heavy lifting ^[7]. Today's back-to-back inflation reports are the first test of that frame ^[7]. [NEW] The 10-year push toward 5% - 4.949% intraday spike per Gelonghui, 4.957% per Yicai ^[13][14] - is consistent with the front-end repricing rather than a clean long-end rally at the surface; the single source flagging the paradox narrative is Mortgage News Daily, and a duration-matched cross-check of 10y vs. 2y performance is needed to verify ^[7]. [NEW] Spot direction conflicted in the same window: Gelonghui reported spot gold -$40 to $4,298/oz and the dollar index +20 pips to 99.15 at 12:34 UTC ^[14]; Wall Street CN reported gold +1.5% and 10-year yields slightly lower at 12:30 UTC ^[8]. Flag the contradiction; the dollar move is more reliably sourced than the gold/yield direction ^[14][8].

4. Plumbing, Fiscal, and What Would Falsify It

[NEW] Fed overnight reverse repo (RRP) usage Friday: $5.255 billion across 3 counterparties, up from $4.736 billion Thursday - a one-day uptick, not a trend ^[15]. [NEW] August federal budget deficit: $166.8 billion ^[16]. FY2026 deficit to date: $1.966 trillion vs. $1.973 trillion at the same point in FY2025 - a marginal improvement, not a fiscal inflection ^[17]. [NEW] S&P Global (Cailian Press relay) is "significantly adjusting" its dollar rating and may cut the bond outlook to negative - single-source, thin on trigger and timeline ^[18]. [NEW] A pre-release research note from Sina Finance had tilted toward a September hold as the base case while flagging the hike as a "serious option to evaluate"; that base case is now overtaken by the print ^[19]. - **What would falsify it:** a Fed hold next week would invalidate the entire forecast cluster (TD, JPMorgan, Citi, MUFG) and unwind the bond paradox ^{[3][4][5][6][7]}. A dovish Warsh press conference would re-rate the long end - the "credibility rally" collapses, 2-year underperforms, dollar gives back the +20 pips ^[6][7][14]. A downside September CPI surprise (release mid-October) would revive the no-hike camp ^[19]. - **Source-quality flags:** S&P dollar-rating item is single-source ^[18]; "inflation is an option" Warsh framing is MUFG's characterization, not a direct quote ^[6]; the paradox narrative rests on a single Mortgage News Daily note ^[7]; gold direction conflicted between Gelonghui and Wall Street CN ^[14][8]; the pre-release hold view in the Sina Finance note has not been updated post-print ^[19]; the Reuters market expert's "not fundamentally necessary" view ^[20] and broader media framing of a hike-on-the-table print ^{[21][22][23][24]} were not corroborated by an institution-grade call in the post-print scramble.