Duration Dashboard: Treasury Yields After the August 2026 CPI Report

Prior-week reference is the Friday, September 4 close; markets were closed Monday, September 7 for Labor Day. The 10-year's 99th percentile reading sits approximately 2 basis points below its five-year closing high of 4.99%, set October 19, 2023. The 30-year's five-year closing high is now 5.37%, set Thursday, September 10, 2026; Friday's close is 1 basis point below it.

The week's move was concentrated in a single session, and it was not the CPI session. Tuesday and Wednesday were orderly: the 2-year added 3 and 4 basis points, the 10-year 1 and 5, as crude extended its climb and Treasury announced an expansion of its buyback program for longer-dated debt. Thursday delivered the repricing. Headline producer prices for August rose 0.4% on the month and 5.4% over the year, above the 5.3% consensus and up from 4.8% in July, on a 24.1% monthly increase in diesel; West Texas Intermediate settled above $100; and the European Central Bank raised rates for a second time. The 2-year rose 16 basis points that day, the 5-year 15, the 10-year 12 and the 30-year 8, which took the long bond to a five-year closing high of 5.37%. Friday's August 2026 CPI print then confirmed rather than extended the move: the 2-year added 4 basis points, the 10-year was unchanged and the 30-year declined 1.

The composition of the flattening is what distinguishes it from the July episode covered in our July 26 report on the oil-driven bear flattener. In July, crude lifted the front end because the market expected the Committee to respond to energy pass-through. This week the front end rose because the response is now scheduled: with the FOMC four days away and pricing above 80% for a quarter-point increase, the 2-year is discounting a policy rate of 3.75%–4.00% rather than the possibility of one. The long end's smaller move reflects a market that has already priced a substantial energy and fiscal premium at the 99th percentile, and that is now asking whether a tightening Fed shortens the horizon over which that premium needs to be paid.

Treasury Auctions and Buybacks: Long End Clears at 5.31%

Three coupon auctions cleared into a rising-yield tape ahead of the August 2026 CPI release, and the results split by tenor in the pattern that has held all summer. The 10-year reopening drew a 2.71x cover with indirect participation at 79.2%, a strong result that was nonetheless priced 13 basis points below where the tenor closed on Friday. The 30-year sale, which followed the 5.22% stop recorded in our August 16 report, cleared at 5.31%, the highest stop of the current cycle, and tailed. Demand for duration inside ten years continues to arrive without concession; demand beyond twenty continues to require it, and this week required more of it than any sale since the tenor's five-year high was first set in August.

The buyback operation Treasury expanded in August, examined in our August 23 report, was tested for the first time under stress and delivered less than the market had assumed. Treasury said Wednesday it would raise the size of its longer-dated purchase to $6 billion, roughly triple the prior operation; the execution repurchased $5.2 billion against approximately $10.5 billion offered. The 30-year rose 4 basis points on the announcement day and 8 the day after. A backstop that accepts half of what is tendered is a liquidity facility, not a price-support facility, and the 30-year's five-year closing high arrived in the same week that distinction became clear.

Credit Pulse

Credit widened in the week of the August 2026 CPI, but by an amount that describes a rates event rather than a credit event. Investment grade finished 1 basis point wider at 80, the 33rd percentile of its five-year range, after trading in a 78–80 basis point band all week and touching its tightest level on Thursday, the day Treasury yields moved most. High yield widened 6 basis points to 265, the 17th percentile, with the entire move arriving Wednesday and Thursday before a 1 basis point retracement Friday. Implied equity volatility rose to 17.84 on Thursday, its high for the week, and finished at 15.84 after Friday's equity rally, 1.31 points above the prior Friday and at the 30th percentile. The S&P 500 declined four consecutive sessions through Thursday and recovered 0.86% on Friday to finish the week down 0.8%; the Russell 2000 declined 2.4%, the rate-sensitive underperformance one would expect.

Primary supply supplied the week's more informative credit signal. The post-Labor Day investment grade calendar, which dealers had expected to open a month of $175 billion to $250 billion in issuance, instead produced the lightest post-holiday week since 2020, with 18 issuers marketing deals on Tuesday and volume subdued thereafter as rate volatility kept borrowers on the sidelines. That follows an August in which high-grade supply set a monthly record above $145 billion. Borrowers who front-loaded into a stable summer tape are now electing to wait through the FOMC, which removes the supply test the market had anticipated and leaves secondary spreads supported by scarcity rather than demand.

Federal Reserve Policy Outlook: September Hike Pricing After the August 2026 CPI

The Committee entered its pre-meeting blackout with the policy rate at 3.50%–3.75% and the market having moved further than any single Fed communication moved it. Following Chair Warsh's Jackson Hole keynote, futures priced roughly a 56% probability of a quarter-point increase at the September 15–16 meeting. Heading into Friday's release the figure was near 70%, and after it, above 80%. Prediction markets placed the probability of at least one increase during 2026 near 88%. The decision is now the least uncertain element of Wednesday's meeting.

The projections are the more consequential release. The June Summary of Economic Projections showed nine of eighteen participants penciling in at least one increase this year, with a median 2026 year-end rate of 3.8% and a 2027 median of 3.6%, and the Chair again declined to submit a projection of his own. A September increase to 3.75%–4.00% is consistent with that June median. What the market has not priced is a revised median implying a second move — a 2026 dot near 4.1% would do it — and the balance of this week's data makes that revision plausible: core PCE has been stalled near 3.3%, the August CPI core rate exceeded consensus, producer prices are at 5.4%, second-quarter private domestic demand grew 4.2% and claims are at 206,000. The three dissents recorded at the July meeting, covered in our August 5 report, established that a tightening faction existed before the energy shock re-intensified; the data since have given it a majority's worth of reasons.

The counterargument the Committee will weigh is the one this week's sentiment data supplied alongside the August 2026 CPI. A 47.8 reading on consumer sentiment alongside a year-ahead inflation expectation of 4.6% is the stagflationary configuration the Chair would prefer not to tighten into, and the annual core CPI rate at 2.4% is the lowest of the cycle. The most probable resolution is an increase paired with language that conditions further moves on energy pass-through rather than on a preset path. That would leave the 2-year roughly where it closed Friday and put the burden of the next repricing on the long end.

Week Ahead: The September FOMC Decision

• Retail Sales (September 15): August data follows a 0.4% decline in July. Consensus sits near a 0.4% gain, with some forecasts as high as 0.6% on gasoline-station receipts. The control group will show whether real spending held up as fuel costs rose.
• Industrial Production and Import Prices (September 15): Import prices carry the direct tariff and energy pass-through and will be read against the 5.4% PPI print.
• FOMC Decision, Projections and Press Conference (September 16): The decisive event of the week. A quarter-point increase to 3.75%–4.00% is priced above 80%; the updated projections are not. The Chair's handling of the sentiment data and the August 2026 CPI core trend will set the tone for the long end.
• Housing Starts and Building Permits (September 16): The first construction read with mortgage rates re-approaching the levels that stalled activity in late 2023, released the same afternoon as the decision.
• Initial Jobless Claims (September 17): Claims follow a 206,000 print and a four-week average near 206,000. A move above 215,000 the day after a rate increase would carry disproportionate weight.
• Bank of England and Bank of Japan Decisions (September 17–18): A third developed-market central bank tightening in the same week as the Fed would complete the pattern the ECB began; a Bank of Japan increase to 1.25% is the more relevant outcome for Treasury demand.

US Economic Positioning and Global Context

Energy reversed the relief it had provided in late August and set the terms on which the August 2026 CPI was read. West Texas Intermediate settled at $100.05 on Friday and Brent at $104.61, each roughly 8% higher on the week, with Brent trading near $110 intraweek before Friday's pullback; US diesel reached a record above $6 a gallon. That more than reversed the roughly 5% decline recorded during the week of Jackson Hole and restored the transmission channel from the Strait of Hormuz to the long-end inflation premium. The dollar index was little changed on the week after a rally of roughly 4% since mid-August, consistent with a market in which US and European rate expectations moved together. Gold settled near $4,409, lower for a third consecutive week as real yields rose.

The belly's case has weakened, but not enough to abandon it. For most of the summer these reports have argued that the 5- to 10-year segment captured the curve's yield without its policy or fiscal exposure. This week the belly repriced 24 and 19 basis points against 26 at the front end and 11 at the long end, so it did not escape the move; it simply did not lead it. The 5-year now sits at its highest close since October 2023 and the 99th percentile, which is compensation for a policy rate that is about to rise, not for one that might. With the decision priced and the projections the open variable, the 2-year carries the risk of a hawkish dot plot and the 30-year the risk of a foreign-demand shortfall into a tailing auction cycle. The belly still sits between those exposures; it is now being paid roughly 4.8% to do so, and the argument this week is that the compensation improved more than the risk did.