Stagflation is not simply high inflation arriving at an awkward moment. It is a specific, diagnosable combination: prices accelerating while output stalls and joblessness climbs, with the wage-setting process and long-run expectations both giving way. Run the September 2026 data against that definition, condition by condition, and the United States clears two of seven. The oil shock is real and the inflation print is ugly, yet unemployment sat at 4.1% in August, real GDP has not printed a negative quarter since early 2025, and the 10-year breakeven inflation rate implied by the Treasury’s own curves closed Thursday at 2.33% — a hair below where it averaged a year ago, before Brent went anywhere near $104.
That last number is the one the stagflation commentary keeps skipping. On 17 September the Treasury published a 10-year nominal par yield of 4.94% and a 10-year real par yield of 2.61%; the difference is what bondholders will accept as compensation for a decade of CPI. It is 2.33%. In September 2025 the same calculation averaged 2.37%. Brent crude is up about 77% from its 52-week low of $58.72 over that span, the Fed has swung from expected cuts to a delivered hike, and the market’s decade-long inflation price has gone precisely nowhere. Stagflation is, above all, an expectations failure. The expectations have not failed.
Key facts
- Headline CPI rose 3.4% in the 12 months to August 2026, up from 2.4% in January — BLS series CUUR0000SA0, released 11 September 2026
- Core CPI, all items less food and energy, rose 2.45% y/y in August, down from 2.85% in May — BLS series CUUR0000SA0L1E
- The gasoline index is up 27.4% y/y, against minus 7.5% in January — BLS series CUUR0000SETB01
- Unemployment was 4.1% in August, down from 4.4% in February — BLS series LNS14000000
- Real GDP grew at a 1.5% annual rate in Q2 2026 after 2.1% in Q1 — BEA second estimate, 26 August 2026
- The FOMC raised its target range 25bp to 3.75–4.00% on a 12–0 vote — Federal Reserve statement, 16 September 2026
- The Fed’s September projections lifted 2026 growth to 2.3% and cut 2026 unemployment to 4.1%, from 2.2% and 4.3% in June — SEP, 16 September 2026
What the hike actually landed on
The Federal Open Market Committee lifted the federal funds target range by a quarter point to 3.75–4.00% on Wednesday, unanimously, the first increase since 2023. The statement ran to a few sentences and contained no hedging about growth: “Economic activity is expanding at a solid pace… Job gains have kept pace with the workforce, and the unemployment rate has changed little. Inflation remains elevated.”
Read that as a diagnosis rather than a press release. A central bank staring at stagflation writes about weakening demand and a deteriorating labour market, and then agonises publicly about which half of the mandate to sacrifice. This Committee did neither. It described a strong economy with an inflation problem and treated the two as separable.
The projections behind the vote say the same thing more precisely. In June the median participant saw 2026 growth of 2.2% and unemployment of 4.3%. In September those became 2.3% and 4.1%. Growth was marked up, joblessness marked down, and the 2027 unemployment median moved from 4.3% to 4.1% as well. Inflation was the only variable revised the wrong way: headline PCE for 2026 went from 3.6% to 3.7%, core from 3.3% to 3.4%. Our morning piece on the dot plot and the September SEP covers the rate path in detail; the point here is the real economy underneath it, which the Committee upgraded.
You cannot forecast stagflation and simultaneously revise growth up and unemployment down. Whatever the Fed thinks it is fighting, it is not a demand collapse.
Michael Pearce, chief US economist at Oxford Economics, put the constraint plainly in comments to CNN on 17 September: “The Fed cannot control energy prices. The economy is solid and can withstand a few rate hikes, but the risk is higher interest rates begin to weaken the labor market.” Note the sequencing in that sentence. The weakness is a risk created by the policy response, not a condition already present in the data.
Scoring the seven conditions
Here is the checklist, with the reading that decides each one.
1. Headline inflation well above target and rising. MET. CPI was 3.4% y/y in August against 2.4% in January and 2.9% a year earlier. On the Fed’s preferred gauge, PCE prices rose 3.7% in the year to July, per BEA’s Personal Income and Outlays release. Five and a half years above target is not a rounding error.
2. An adverse energy supply shock. MET. Brent settled near $103.99 and WTI near $101.22 on Thursday afternoon, per CNBC quote data read at 20:00 UTC. The gasoline index has swung 35 percentage points, from minus 7.5% y/y in January to plus 27.4% in August. FinanceFeeds has tracked the physical cause through the shut Saudi pipeline and the Hormuz disruption.
3. Underlying inflation broadening beyond energy. PARTLY. This is the honest ambiguity in the data, and it depends entirely on which index you use. Core CPI is 2.45% and falling, down from 2.85% in May. Core PCE was 3.3% in July. That 85 basis-point gap runs opposite to the historical relationship, in which CPI typically prints above PCE. Warsh cited both in his press conference: “Core PCE and CPI prices are running at about 3.2 percent and 2.4 percent respectively. Too many categories are still posting increases above 3 percent.” Half a mark, not a full one.
4. Rising unemployment. NOT MET. 4.1% in August, against 4.4% in February. Payrolls added 162,000 in August. Unemployment is nearer a cycle low than a cycle turn.
5. Stagnant or contracting output. NOT MET. Q2 real GDP grew at a 1.5% annual rate, slower than Q1’s 2.1% but positive, and the Fed’s own median has 2026 finishing at 2.3%. Advance retail sales hit $773.9 billion in August, up 1.2% on the month and 6.0% on the year, per Census release CB26-153 on 16 September. Deflate that 6.0% by the 3.4% CPI and real retail spending is still growing about 2.5% a year. Neither agency published that comparison; it falls out of putting the two releases side by side.
6. A wage-price spiral. NOT MET. Average hourly earnings rose 3.09% y/y in August, down from 3.38% in June and below headline CPI. Real wages are being squeezed by energy, which is the opposite of the mechanism that turns a price shock into an inflation regime. Workers are absorbing the oil shock, not passing it on.
7. Unanchored long-run expectations. NOT MET. The 10-year breakeven closed at 2.33% on 17 September, against a September 2025 average of 2.37% and a May 2026 average of 2.45%, calculated from the Treasury’s daily nominal and real par yield curves. The oil shock has not moved the decade.
Two met, one partial, four not met.
Why core matters more than the pump
The chart above is the whole argument in one frame. Headline CPI, in red, breaks upward from March 2026, the month the gasoline index flips positive year on year. Core CPI, in blue, does not follow. It drifts down. Unemployment, in green, sits flat near 4.1%. The GDP bars keep printing positive.
Three years of data, and the only series responding to $104 crude is the one that contains $104 crude.
Relative price changes and inflation are different phenomena, and the distinction is not academic. When petrol takes a larger share of a household budget, something else takes a smaller one, unless the money supply and wage expectations validate the increase. The 1970s validated it. Wage indexation clauses covered a large share of union contracts, the Fed’s reaction function was ambiguous, and a crude shock propagated straight into everything else.
Our coverage of Waller’s test for the August CPI print anticipated exactly this split, and the Governor’s own framing has held up. Speaking on 3 September, Christopher J. Waller, Governor at the Federal Reserve, said: “we haven’t seen a significant increase in longer-term inflation expectations, but it must be acknowledged as a risk — one that the FOMC should be prepared to act on if longer-term expectations rise and progress on inflation reverses.”
The 1970s comparison, with the real figures
Analogies to 1974 and 1979 get made loosely. The numbers make the break obvious.
| Measure | Dec 1974 | Mar–Jun 1980 | Aug 2026 |
|---|---|---|---|
| Headline CPI y/y | 12.34% | 14.76% (March peak) | 3.40% |
| Core CPI y/y | 11.11% | 13.59% (June peak) | 2.45% |
| Unemployment rate | 7.2%, rising to 9.0% by May 1975 | 7.8% by July 1980 | 4.1%, falling |
| Real GDP, calendar year | −0.5% (1974), −0.2% (1975) | −0.3% (1980) | +2.1% (2025), Fed median +2.3% for 2026 |
Sources: BLS series CUUR0000SA0, CUUR0000SA0L1E and LNS14000000; BEA NIPA real GDP, percent change from preceding period.
In November 1974 core CPI reached 11.16%. Energy had leaked into haircuts, rents, insurance and school fees. That is what “broadening out” looks like as a number, and it is the state today’s core reading of 2.45% is roughly nine percentage points away from. Unemployment in that episode climbed from 4.6% in October 1973 to 9.0% in May 1975, and the 1982 disinflation eventually cost 10.8%. Today’s labour market is doing nothing of the sort.
Where the analogy survives: both episodes begin with a physical supply disruption a central bank cannot drill its way out of, and in both, credibility is the variable that determines whether the shock is a one-off or a regime. Where it breaks: the transmission machinery of the 1970s — indexation, an untested mandate, unanchored expectations — has been dismantled, and the 2.33% breakeven is the measurement that proves it.
What a rate hike can and cannot do to $104 oil
A quarter point does not reopen a pipeline. Asked precisely that at Wednesday’s press conference, Kevin Warsh, Chairman of the Federal Reserve, gave the clearest statement of the central bank’s actual job in a supply shock: “We cannot affect any individual price, whether it be oil prices, whether it be food stuffs at the grocery store. But what we can do, and will do, is ensure that any change in relative prices don’t broaden out. Don’t have second and third order effects in the economy.” (Transcript, 16 September 2026.)
Split the ledger honestly.
What the hike cannot do: lower the price of crude, restore Hormuz transit, rebuild refining capacity, or shorten the gasoline index’s 27.4% annual increase by a single basis point. Our reporting on the tanker incident and the East-West pipeline sets out how physical the constraint is.
What it can do: keep the breakeven at 2.33%, keep wage growth at 3.1% rather than 6%, and make it expensive for firms to treat a fuel surcharge as permission for a general price rise. The hike is not aimed at oil. It is aimed at the second-round effects, and the second-round effects are exactly what the seven-point scorecard finds absent.
Not everyone accepts the trade. Selma Hepp, chief economist at Cotality, told NPR on 16 September: “The bigger question is whether the Fed risks fighting the wrong inflation battle. A rate hike is unlikely to lower gasoline prices, reduce tariff-related costs, or accelerate homebuilding, but it will further dampen housing demand and delay a broader market recovery.” That critique is correct about the incidence. Housing and rate-sensitive capex bear the cost of a policy aimed at a price the policy cannot reach. The counterargument is that the alternative bill, denominated in credibility, is the one the 1970s ran up.
The institutional tension nobody resolved
The Federal Reserve Act gives the FOMC maximum employment and stable prices. It gives no instruction for the case where a foreign supply shock pushes one mandate away from target while the other sits comfortably at it. The statutory language assumes demand-driven fluctuations, where the two objectives move together.
That gap is why the framework debate matters more than any single meeting. The Committee’s revealed choice on Wednesday was to treat the 2% goal as the binding constraint and the labour market as the slack variable, which is a defensible reading of the Act and also a change from the 2020 framework’s emphasis on shortfalls in employment. Warsh has been explicit that the shift is deliberate, telling reporters in the same press conference that he does not believe the Committee needs “to do harm to the labor markets to achieve our objective.”
There is a political dimension the Fed will not discuss on the record. The Chairman declined to engage with questions about presidential pressure for lower rates, saying only that the decision was the Committee’s. Independence is easiest to assert when the data cooperate; a genuine stagflation, with unemployment rising into a hiking cycle, would test it in a way that 4.1% joblessness does not.
Bond markets applied their own pressure. Karen Manna, fixed income strategist at Federated Hermes, told CNN: “The Fed raised rates today, but the bond market got there first. In many ways, the bond market has been leading the Fed rather than the other way around.” The 10-year closed Thursday at 4.93%, down from 5.004% the day before, while the S&P 500 gained 1.14% to 7,637.71. Equities and duration both rallied on a hawkish central bank, which is the price action of a market that believes the inflation fight is winnable.
What to watch between now and December
Three things will settle the argument, and each has a date.
First, core CPI. September data lands in mid-October. If core holds below 2.6% while gasoline stays elevated, condition three resolves decisively to “not met” and the stagflation case collapses to a single energy line item. If core turns up toward 3% with the gasoline index still running above 20%, the broadening has begun and the scorecard moves to three met.
Second, the breakeven. A sustained move above 2.6% on the 10-year, computed off the same Treasury curves, would mark the first genuine expectations signal of this cycle. It has not happened through a 77% rise in crude, which sets a high bar for it happening at all.
Third, payrolls. The Pearce risk is that the hike itself manufactures the missing leg. Watch for the three-month average of job gains dropping below roughly 50,000 while unemployment climbs through 4.4%. That combination, alongside core above 3%, would be the real thing rather than the label. On present trajectory it is a 2027 question, not a 2026 one, and the pricing for the remaining 2026 meetings reflects a market that expects the economy to take the hikes.
The likeliest path from here: one more 25bp increase by December, headline CPI peaking near 3.6% in the autumn as base effects from the spring gasoline surge compound, then a fade through H1 2027 as the energy contribution rolls off, with core PCE reaching the Fed’s projected 2.5% for 2027 roughly a quarter late. That is an uncomfortable year. It is not the 1970s, and the difference is measurable to two decimal places.
FAQ
Is the US in stagflation in 2026?
No, on the standard definition. Stagflation requires high inflation alongside stagnant growth and rising unemployment. US unemployment was 4.1% in August 2026 and falling, real GDP grew at a 1.5% annual rate in Q2, and core CPI is 2.45% and declining. Only headline inflation at 3.4% and the energy shock itself satisfy the criteria, two of seven conditions on the scorecard above.
What is the difference between an oil shock and stagflation?
An oil shock is a change in one relative price. Stagflation is a change in the general price level that persists after the shock fades, because wages and expectations have adjusted upward. The test is whether core inflation, which excludes energy, follows headline inflation higher. In 2026 it has not: core CPI fell from 2.85% in May to 2.45% in August while gasoline rose 27.4% year on year.
Can the Fed lower oil prices by raising interest rates?
No. Chairman Kevin Warsh said on 16 September that the Fed “cannot affect any individual price, whether it be oil prices, whether it be food stuffs at the grocery store.” Higher rates work on demand, not on supply disruptions. What tighter policy can do is prevent a relative price change from producing the second- and third-round effects that turn a shock into sustained inflation.
What does the 10-year breakeven rate say about stagflation risk?
It says the risk is being priced as low. The 10-year breakeven, the gap between nominal and inflation-protected Treasury yields, closed at 2.33% on 17 September 2026, below its September 2025 average of 2.37%. Because stagflation is fundamentally a failure of inflation expectations, a breakeven that has not moved while crude rose 77% off its 52-week low is direct evidence against the diagnosis.
How does 2026 compare with the 1970s on the numbers?
Core CPI peaked at 11.16% in November 1974 and 13.59% in June 1980, against 2.45% in August 2026. Unemployment rose from 4.6% in October 1973 to 9.0% in May 1975, and reached 10.8% in December 1982; it is 4.1% today and falling. Real GDP contracted in 1974, 1975 and 1980. The US economy has not posted a negative quarter since Q1 2025.
What would change the stagflation verdict?
Three markers: core CPI rising above 3% while gasoline stays elevated, the 10-year breakeven sustaining a move above 2.6%, and payroll gains averaging below 50,000 a month with unemployment through 4.4%. Any two together would move the scorecard from two conditions met to four, at which point the label would describe the data rather than the mood.
This article is analysis of published economic data and does not constitute investment advice. All figures were read from the primary releases named and linked above on 17 September 2026.