Anna Wong — Fed Walking Into a "Mistake"

0:00 The setup — August CPI, a 4.93% 10-year, $107 Brent

• Host: headline CPI +0.4% m/m, 3.4% y/y (in line); core +0.3%, hotter than expected; gasoline +3.9% was over a third of the monthly gain.
• 10-year near 4.93%, highest since 2023, after Bessent lifted Treasury buybacks to $6bn per operation; oil fell 3% on news of Iran–Gulf talks.

2:14 Decomposing the core surprise — wireless plans and hotels

• "When you dig into the details it is really due to a couple outliers": wireless phone services (Bloomberg Price Project data traces it to Verizon) and a "discrete jump in hotel prices."
• Both are "irregular, one-off, temporary, volatile" categories.

3:01 The disinflation hiding underneath

• Primary rents and owners' equivalent rent low and still moderating.
• Food mean-reverting after lettuce and tomato supply shocks; prescription and non-prescription drug prices declining.
• "More mixed and nuanced than the hotter-than-expected headline and core readings suggest."

4:02 How Warsh boxed the Fed in

• His hawkish Jackson Hole speech told markets to price a hike if CPI didn't show falling y/y inflation; Governor Waller echoed it — "if the CPI report shows inflation is hot, we move."
• Warsh cut forward guidance in favour of "the unfiltered market signal to shape economic reality" — and the signal after CPI was ~90% for a hike. "As a chairman you have to be true to your words."
• Result: no room to discuss the disinflationary detail.

7:22 Diesel at $6 — headline now, pass-through next

• September gasoline up more than 5% so far adds ~0.2 pp to headline — September headline "probably at least 0.4%."
• Diesel hits farmers, freight and trucking, so pass-through to groceries and heavy appliances; kitchen appliances already popped in August — "not due to tariffs… the tariff pass-through is over."
• Food prices are flexible both ways (thin margins — "just like eggs").

10:08 If oil de-escalates, headline CPI could drop below 2%

• With crude back to "a more regular 70 or 60," headline CPI "could possibly fall below even 2%" in Q1 next year.
• "The faster it rises, the faster it falls" — why the Fed "should not and does not respond to headline gasoline price increases."

11:20 A 5 bp surprise moved hike odds 20 points

• Core came in at 0.29 vs 0.24 expected — "entirely due to wireless telephones and Airbnb shifting their policy."
• The rounding did the damage: 0.2 to 0.3 on the terminal read as "a hot CPI," checking Warsh's box; Kalshi 60%→81%, CME FedWatch 66%→88–89%.

13:13 "This hike is a mistake" — payrolls were a seasonal quirk

• Seasonally adjusted payrolls (+162k) surprised high, but the non-seasonally-adjusted gain was lower than last August's and below her forecast.
• Last August a stronger NSA print produced a mere +20k SA (later revised); this year a weaker NSA print produced 160k+. "Not much substance in it."
• "Markets are not always right… groupthink, markets herd" — she expects next week's FOMC to be a closer call than priced, and "only one hike this year."

15:58 Trump's threat and Warsh's political mastery

• Trump's Truth Social threat was aimed at "the Fed," not Warsh; Hassett and Trump say Warsh "has to do what he has to do."
• Warsh can tell Trump he was outvoted, the market demanded it, and the long end would blow out otherwise — "a masterful job at managing the politics… he basically deflected all the blame from himself," though he created the situation.

18:17 Early-to-mid cycle, fragile — the Fed hikes too early

• Longer average workweek (+0.6%) lifting real weekly earnings shows improved momentum — Bloomberg Economics called the cycle bottom last fall.
• But "a pretty fragile recovery" that still needs Fed support; "the Fed tends to hike late cycle, not in the early to middle cycle" — a hike squeezes the labour-market recovery's runway.

21:18 Bessent's buybacks — the long end is illiquid

• "One of the more sympathetic figures": buying back at the long end and issuing short is "a natural thing to do."
• The yardstick: Operation Twist (2011) was $400bn, ~100x the $4bn announcement, yet the one-day yield effect was far more than 1/100th as large — so "the long-end market is very illiquid… and that… is when you want to intervene." Why policymakers like August.

23:45 Intervention rules — the $6bn is too small

• From her FX-intervention research: effective when the market is disorderly, with a surprise element, proof of abundant resources and readiness "to go all in" — the same rule as the yen, the Fed in 2008 and 2020.
• The $6bn "was not as large as what markets had hoped for," so yields went back up; "Treasury still has a lot of firepower" — she's optimistic it works ultimately.

25:36 Oil and yields moving together — "a bit concerning"

• Not supply-shock causation: the correlation reflects "global risk aversion, a pullback from bonds."
• In past uncertainty spikes long yields fell on Treasuries' safe-haven status; "now we are seeing it increase."

27:00 Bessent vs Treasury convention — issuance for cost, not predictability

• Two sacred Treasury conventions: strong-dollar rhetoric (since Rubin) and regular, predictable issuance across the curve to lower the risk premium.
• Bessent challenges the second: deficit near 6% of GDP, over half interest, above CBO's projection because of yields; issuing short at ~4.2% instead of ~5% long "could make hundreds of billions of difference" — crisis-type flexible policymaking.

31:03 A 5% 10-year restrains the economy — and the long end has already hiked

• October 2023's 5% brought buyers in (she recalls Bill Gross or Bill Ackman buying), but also the lagged labour slowdown into 2024 and the August 2024 flash crash.
• Housing is already cooling fast with the 10-year above 4.5%; at 5% expect clear restraint on activity, small businesses and eventually jobs.
• The 10-year is up at least 0.5 pp since March — "equivalent to about 100 basis points of rate hikes."

32:56 Hiking won't tame bond vigilantes — growth and tariffs will

• Evidence: after Warsh's Jackson Hole speech the long end dipped for a day, then rose again.
• The long-end rise is fiscal; the Supreme Court's IEEPA ruling costs ~$1tn of revenue over 10 years. A Fed-induced slowdown with a 6% deficit leaves no fiscal space — "a downward spiral."
• Tariffs "have been generating revenues without tanking the economy" and have bipartisan support; slowing growth would force zero rates or QE, "which Kevin Warsh doesn't like."

36:13 AI capex — the second derivative has peaked

• GDPNow 4.4% for Q3 — but "right before 2008, growth was amazing too"; GDP is contemporaneous.
• It is the growth rate of capex that adds to GDP growth; AI capex contributed ~1 pp in H1 2026 and could fall to ~0.5 pp or lower next year even with capex still rising — "we are at peak AI capex boom contribution to GDP."