US Headline Inflation Hit a Three-Year High — But Core Barely Moved

Year-on-year % change, not seasonally adjusted. Headline in terracotta, core in black. The October 2025 reading is unavailable (release gap) and the line bridges it.

US headline inflation reached 4.2% in the year to May 2026, its highest reading since April 2023 and the third consecutive monthly acceleration, from 3.3% in March to 3.8% in April. On its face, the figure revives the question of whether the post-pandemic inflation that peaked above 9% in 2022 has returned.

The composition says otherwise. According to the U.S. Bureau of Labor Statistics, energy accounted for more than 60% of the May increase, led by a sharp move in gasoline prices tied to the disruption of oil supply through the Strait of Hormuz. Core inflation — prices excluding food and energy — rose just 0.2% on the month and held at 2.9% year-on-year, slightly below expectations. This page sets the headline reading against the core series since 2019, isolates the energy contribution, and separates what the data establish from what they do not.

01The headline jumped to a three-year high — and it is almost entirely energy

The May 2026 print broke a pattern that had held for nearly two years. After peaking at 9.1% in June 2022, headline CPI fell steadily and spent 2024 and most of 2025 between roughly 2% and 3%, reaching a low of 2.3% in April 2025. The climb back above 4% is the first since the post-pandemic disinflation, and at 4.2% it is the highest reading since April 2023.

The driver is narrow. BLS data attribute more than 60% of the monthly gain to energy, with gasoline rising about 7% on the month as the conflict around the Strait of Hormuz disrupted crude supply. Food prices rose 0.2% on the month. The categories that capture broad domestic price pressure — shelter, services excluding energy — decelerated rather than accelerated: shelter rose 0.3%, half of April’s pace. An energy-led headline rise of this shape is a supply shock passing through the price level, not evidence that demand-driven inflation is rebuilding.

The two series tell one story. They moved together up to the 2022 peak, fell together through the disinflation, and converged near 2.5–3% by late 2025. In 2026 they separate again — but this time only the headline rises, while core stays flat. The widening gap is the signature of an energy shock layered on top of an otherwise contained trend.

Year-on-year percent change, not seasonally adjusted (BLS series CPIAUCNS and CPILFENS). The headline–core gap widened from roughly zero at the start of 2026 to 1.3 points by May as energy prices rose while core decelerated. Figures rounded to one decimal; the gap is the difference of the rounded figures.

• Headline rose from 3.3% (March) to 3.8% (April) to 4.2% (May) — three consecutive monthly increases.
• Over the same stretch, core moved from 2.6% to 2.9%, and the May core gain (+0.2% on the month) came in below expectations.
• The headline–core gap of 1.3 points in May is the widest since late 2022, and is accounted for almost entirely by energy.

02Core inflation barely moved — which is what makes 2026 different from 2022

The contrast with 2022 is the point. In 2022, headline and core rose together: core CPI reached 6.6% in September 2022, confirming that price pressure had spread well beyond energy into rents, services and goods. That breadth is what made the 2022 episode a genuine inflation regime rather than a commodity spike.

The 2026 reading has no such breadth, on the data available so far. Core at 2.9% is close to where it sat through 2024 and 2025; services excluding energy decelerated in May; shelter, which carries roughly a third of the index, cooled. The mechanism that turns an energy shock into persistent inflation — pass-through into wages, rents and broad services, feeding inflation expectations — is not visible in the May figures. The shock is real and the headline is high, but the underlying trend the core series tracks has not changed.

03Why a supply-side energy shock transmits differently from demand-led inflation

An oil-driven rise in the price level is, in the first instance, a transfer: consumers and firms pay more to petroleum producers, some of them abroad. That compresses real incomes — average hourly earnings rose 3.4% in the year to May, below the 4.2% headline, so real wages fell year-on-year for a second consecutive month — but it does not, by itself, create the self-reinforcing wage-price dynamic associated with demand-led inflation.

This is why a central bank’s textbook response to a supply shock differs from its response to a demand shock. A demand-driven overheating calls for tighter policy to cool activity. An energy supply shock raises inflation and weakens real demand at the same time, and tightening into it can deepen the second effect. The relevant question for policy is whether the shock is bleeding into core and into expectations — which is, again, why the core series rather than the headline is the operative signal. Following the May report, markets priced the Federal Reserve as likely to hold rather than react to the energy-driven headline. Related material: our overview of how inflation forms and transmits across the economy.

04Does an oil shock mean a recession is coming? A separate question

The 2026 oil move has revived a second consensus reflex — that a major oil shock precedes a US recession. That question turns on the magnitude of the shock relative to the economy, not on the inflation print, and we treat it separately. The relevant metric is the oil burden: the share of US GDP spent on petroleum at prevailing prices. On that measure the 2026 episode has so far peaked well below the level that has historically preceded recessions, even as headline inflation rose. The full episode-by-episode record, the burden threshold, and the 2026 reading are set out in our study on oil shocks and US recessions since 1970.

05What would change this read

The interpretation on this page is conditional on the shape of the data observed through May 2026, and it is falsifiable. Three developments would shift it.

• Core acceleration. If core CPI and, in particular, services excluding energy turn higher over the coming months, that would indicate the energy shock is passing through into the broad price level rather than staying contained.
• A sustained oil price. If crude stays elevated for roughly six months or more rather than retracing, the cumulative pass-through into transport, food processing and other energy-intensive prices rises materially.
• Expectations drift. A persistent move higher in survey- or market-based inflation expectations would signal that a temporary headline impulse is becoming embedded.

None of these is present in the May 2026 data. The page reports the reading as observed; it will read differently if these series move.

06Methodology & data

Headline inflation is the 12-month change in the Consumer Price Index for All Urban Consumers, not seasonally adjusted (FRED series CPIAUCNS). Core inflation is the 12-month change in CPI for all items less food and energy, not seasonally adjusted (FRED series CPILFENS). The not-seasonally-adjusted series are used because the headline year-on-year figures reported by BLS are computed on that basis. The energy share of the monthly increase and the gasoline and shelter movements are from the BLS Consumer Price Index release for May 2026. All series figures here are computed by Eco3min from the public data.

import pandas as pd

# Headline and core CPI straight from FRED — no API key
h = pd.read_csv("https://fred.stlouisfed.org/graph/fredgraph.csv?id=CPIAUCNS", parse_dates=["observation_date"])
c = pd.read_csv("https://fred.stlouisfed.org/graph/fredgraph.csv?id=CPILFENS", parse_dates=["observation_date"])

# Date-aligned 12-month change (a release gap can break a positional shift)
for df, col in [(h,"CPIAUCNS"), (c,"CPILFENS")]:
    df.set_index("observation_date", inplace=True)
    df = df.reindex(pd.date_range(df.index.min(), df.index.max(), freq="MS"))
    df["yoy"] = (df[col] / df[col].shift(12) - 1) * 100

07Data & reproducibility

The monthly headline and core CPI year-on-year series used in the chart and table are available in open format, updated as new BLS releases arrive.

License: Creative Commons Attribution 4.0 (CC BY 4.0). Free for research, academic and journalistic use with attribution.

08Questions & answers

09Sources & limitations

• PrimaryU.S. Bureau of Labor Statistics — Consumer Price Index, May 2026 release; series CPIAUCNS (headline) and CPILFENS (core), via FRED.
• ContextFRED (Federal Reserve Bank of St. Louis) — monthly CPI series used for the chart and downloadable dataset.

• The headline reading is energy-driven and the energy component is volatile. A retracement in oil prices would lower the headline as quickly as the shock raised it.
• Core is a lagging read on persistence. Pass-through from energy into core, if it comes, would appear over subsequent months; the May data are a snapshot, not a verdict.
• A monthly reading is missing. The October 2025 CPI is unavailable in the series; year-on-year changes are computed by date-aligned matching to avoid distortion from the gap.
• This page covers inflation, not recession risk. The question of whether the 2026 oil shock implies a downturn is treated separately, via the oil-burden record linked above.