The Sticky Core: Why Services Inflation Became the Fed's Last Remaining Problem

September's consumer price index arrived at 3.7% year over year — well above the 3.6% consensus and the highest headline reading in months. The number made headlines. But a single item drove most of the alarm: gasoline surged 9.1% in the month, accounting for more than half of the total monthly increase. Strip out energy, and a different — and in some ways more consequential — inflation story comes into view.

Core CPI, which excludes food and energy, rose 0.3% in September, nudging the annual pace from 2.4% to 2.6%. That modest beat against the 0.2% consensus expectation is worth examining. It was the third consecutive month of core readings coming in slightly hotter than forecast. Individual misses can be noise; a pattern of them is a signal.

The concern is not the aggregate core figure itself, which at 2.6% remains well below the 3%-plus levels that defined the 2025 inflation surge. It is the composition. Two components of core inflation — shelter and services broadly — are behaving in ways the Federal Reserve has long identified as the hardest to dislodge.

Shelter Reawakens

Shelter is the largest component of core CPI, comprising roughly one-third of the overall index. It had been a welcome source of disinflation for much of 2026. Not anymore. After a prolonged lull — the index rose just 0.1% in July — shelter accelerated to 0.3% in August and then to 0.4% in September. Owners' equivalent rent (OER), a measure of what homeowners would pay to rent their own homes, rose an estimated 0.35% in September, pushing its year-over-year rate to approximately 3.3%.

The mechanics behind the acceleration are well understood, if stubbornly slow to resolve. Official BLS shelter measures lag actual market rent conditions by roughly twelve to eighteen months, because they capture what renters are paying under existing leases rather than the prices new tenants are negotiating today. The rental market surged violently in 2022 and 2023; the statistical catch-up has been working its way through the system ever since. The problem for policymakers is that a second set of forces is now sustaining demand in the rental market even as the original surge fades. With the 30-year mortgage rate at 7.32% as of October 22, the lock-in effect (the phenomenon in which existing homeowners are effectively trapped in their properties by below-market mortgage rates they locked in years ago) remains powerful. Fewer homes list for sale; fewer buyers can afford the ones that do. Rental demand stays elevated, and landlords know it.

The ISM Upstream Signal

Beyond shelter, the service sector as a whole is getting structurally more expensive to operate. The Institute for Supply Management's (ISM) Non-Manufacturing Prices Paid index — a monthly survey of what service businesses are paying for their inputs — rose to 74.0 in September, up 1.4 points from August and the highest reading since July 2022. ISM said fuel costs were cited twice as often as any other factor by survey respondents. That matters because energy costs flow into service prices before they register in the CPI: airlines, ground logistics firms, cleaning services, and food delivery all carry fuel as a major operating expense.

The comparison to mid-2022 requires context. When ISM Services Prices last ran at this level, headline CPI was approaching its 9% peak and the Fed was still behind the curve. Today's demand conditions are materially different: consumer spending has slowed, credit conditions are tighter, and the Fed is actively restraining growth. But the reading does signal that service-sector disinflation — a steady trend for most of 2024 and 2025 — has stalled. Upstream cost pressures that eventually reach consumers are rising again, driven this time by the same energy shock that inflated the September headline.

Supercore's Stubborn Persistence

The Fed's preferred internal barometer for stubborn price pressure is what economists call supercore inflation: core services excluding shelter (CPI-CSXS). This measure strips out both the energy volatility that distorts the headline and the lagged rent dynamics that complicate shelter, leaving a clearer view of wage-driven, discretionary services pricing. Supercore rose 0.34% in September. Over the past three months, it has run at approximately a 4.5% annualized pace — more than double the roughly 2% rate that prevailed before the pandemic. PNC's economics team noted it was the third consecutive month of elevated supercore readings.

The persistence of supercore is the structural dimension of the current inflation problem. Goods prices — the original driver of the post-pandemic surge — have largely normalized and are now exerting little inflationary pressure. Energy prices move with geopolitical conditions; they are volatile but mean-reverting. Services, by contrast, are labor-intensive, locally provided, and difficult to offshore. Once service businesses adjust their pricing upward, they rarely reverse it. The same rigidity that makes services inflation slow to arrive makes it slow to depart.

What the Fed Is Actually Watching

The Federal Reserve meets October 27–28, with futures markets pricing an approximately 83% probability of a hold at the current target range of 3.75%–4.00%. That near-certainty of inaction reflects a deliberate signaling by Chair Kevin Warsh following September's hike — the first rate increase in roughly three years and a unanimous 12-0 vote — as well as a soft September payroll report (+29,000, a significant miss) and a modest easing in Treasury yields heading into the meeting. The Fed is not abandoning its tightening posture; it is pausing to let cumulative policy restraint work through the financial system.

The critical release of the coming week is Thursday's September Core PCE (personal consumption expenditures) deflator, the Fed's preferred inflation gauge, due October 30. Its August reading was 3.0% year over year. Given September's hotter-than-expected core CPI and the continued elevation in services prices, most analysts expect it to move higher. The Fed has made clear, through the dot plot it updated in September, that 16 of 18 participants see at least one additional hike before year-end. A September Core PCE reading that breaks clearly above 3.0% — or even prints at the same elevated level — would harden December hike probability, already priced at roughly 80% in futures markets, into near-certainty.

The arc of this inflation cycle has not been a single wave but a sequence of overlapping ones. Goods prices surged first, then normalized. Energy added a new layer each time supply disruptions intervened. And now services — the slow-moving, labor-intensive, rent-catching, fuel-absorbing portion of the economy — are the holdout. They are the last component that has not yet returned to anything resembling pre-pandemic norms. That is precisely what makes them the hardest problem the Fed has left to solve.