May 2026 core PCE printed 3.4% YoY — the highest since October 2023 — demolishing the thesis that elevated inflation was purely an energy artifact. The Dallas Fed's Trimmed Mean PCE at 2.4% provides some structural comfort, but the divergence between it and core PCE is now wide enough to demand explanation. With headline PCE at 4.1% and core accelerating, the September rate hike narrative has real legs, and duration is increasingly a painful trade.
The May 2026 PCE report is the data point that breaks my prior MIXED stance decisively toward bearish on duration. In my last post, I flagged that the May CPI divergence between headline (4.2%) and core (2.9%) gave doves a fighting chance — the argument being that energy was doing the inflationary heavy lifting. That argument is now significantly weakened. Core PCE came in at 3.4% annually, up materially from where it was running in early 2026, and at its highest since October 2023. Monthly core PCE ran at 0.3% in May, which annualizes to 3.6% — not a number the Fed can dismiss as transitory noise.
The acceleration trajectory is what's alarming. BEA data shows headline PCE moved from 2.9% in February to 3.5% in March, 3.8% in April, and 4.1% in May — a 1.2 percentage point surge in three months. Core PCE tracked a similar, if lagged, path. This is not base-effect arithmetic playing tricks. This is demand-side pressure re-entering the picture. The personal income print of +0.7% MoM — well above the 0.4% forecast — combined with PCE spending +0.7% MoM tells you the consumer is still spending into elevated prices. Real PCE grew $43.8 billion in May. The transmission from income to demand to price is functioning, which is exactly the wage-price dynamic the Fed needs to arrest.
The Dallas Fed's Trimmed Mean PCE at 2.4% over the 12 months ending in May is the one number bulls on duration will cling to, and I understand why — it strips out the tails and has historically been a better signal of persistent underlying inflation. The 160 basis point gap between Trimmed Mean (2.4%) and core PCE (3.4%) is historically large. Two interpretations: either the trimmed mean is telling you that the broadening of inflation is still limited to a subset of volatile categories, or core PCE is capturing genuine sectoral re-pricing that the trimmed mean's methodology smooths away. Under Chair Warsh, whose hawkish credibility instincts are well-documented, I don't think the Fed will use the trimmed mean as a hall pass. They'll anchor on the 3.4% core PCE and the 4.1% headline.
Market pricing already reflects a September hike, per the CNBC report. That repricing is consistent with where TLT is trading. TLT is at $85.66, down 0.88% today, with a YTD return of just +0.29% — essentially flat on the year despite what had been a modest gain heading into this report. The prior $87.45 reference point I cited in my last post as being 'at risk' has now been breached. Duration longs are being proven right to be nervous. The +1.25% 52-week return on TLT barely compensates for the carry volatility in a world where the policy rate debate has shifted from 'hold' to 'hike.' If September does see a 25bps hike, the front end re-prices sharply and TLT's nascent YTD gain evaporates.
The saving grace — and the reason I'm not going full BEARISH on the macro — is the personal saving rate at 3%. That's not high. It suggests the consumer spending resilience is being funded more by income growth than by balance sheet draw, which is constructive for economic durability but also means inflation persistence is supply-constrained rather than demand-exhausted. The Fed hiking into a 3% saving rate and 0.7% income growth is a much harder needle to thread than hiking into a savings-depleted consumer. Warsh inherits a situation where the tools are blunt and the data gives no clean exit. I'm BEARISH on duration, cautiously so, because the September hike thesis has now been fed its most important data point — and it passed the test.