Stagflation is a word until you write down the test, and then it is eight episodes

Stagflation is the most rhetorically useful word in macroeconomics and the least defined. No statistical agency publishes it, no committee dates it, and no threshold for it exists in any document read for this page. So this page does the only honest thing available: it writes down an explicit, arithmetical test, applies it to every month of American data since 1948, and reports what comes out — 86 months in eight episodes, all but three of them before 1992, and not one of them during the largest inflation burst in forty years.

Why the definition has to be written down first

There is no official stagflation. The National Bureau of Economic Research dates recessions and nothing else; the Bureau of Labor Statistics publishes price and labour series and no combination of them; no document opened for this page contains a threshold, a duration or a test. Every published claim that a country “is in stagflation” therefore rests on a definition the writer has chosen and, in almost every case, has not stated.

This page states one. A month qualifies when the consumer price index is 5% or more above its level twelve months earlier and the unemployment rate is higher than it was twelve months earlier. That is inflation plus a deteriorating labour market, in the two most widely published series, with no smoothing and no judgement. It is not the only reasonable test and it is not endorsed by anyone: it is simply written down, so that the count below can be reproduced or disputed on its own terms.

Two design choices are worth defending in the open. The 5% threshold is arbitrary; it is high enough to exclude ordinary years and low enough to admit 197 months since 1948, roughly a fifth of the record. The direction test on unemployment is used rather than a level, because a level threshold would make the answer depend on the structural rate of unemployment, which has moved over eighty years; a year-on-year rise means the same thing in 1974 and in 2026.

What the test finds, over 938 months

Applied to every month from January 1949 — the first with a twelve-month comparison available on both series — to July 2026, the test is met in 86 months. Grouped into runs, with gaps of two months or less merged, they form eight episodes.

The distribution is the finding. Five of the eight episodes fall between 1969 and 1982, and they account for 75 of the 86 months. Two more are in 1990 and 1991 and one is in 2008. The most recent qualifying month in the entire record is August 2008— eighteen years before this page was written.

Every one of the eight overlaps or immediately precedes a recession the NBER dated. That is not a coincidence and it is not a discovery either: the second half of the test requires unemployment to be rising, which is what happens in and around recessions. The point of stating it is that the test is not picking out a mysterious third state of the economy. It is picking out recessions that happened to arrive with high inflation, which is exactly what the word was coined to describe.

The 2021 to 2023 inflation burst does not qualify, and that is the useful part

The most recent period of high American inflation was large and long. Yet no month of it meets this test, and understanding why is more informative than the count itself.

The inflation half of the test was satisfied for a long stretch ending in February 2023. The unemployment half was not: over that period the unemployment rate was generally lower than it had been twelve months earlier, not higher. Whatever else that episode was, it did not combine high inflation with a deteriorating labour market, which is the specific combination the word names.

The current position, on the same test, is clear and dull. The twelve-month change in the consumer price index for July 2026 is +3.36%, below the 5% threshold. The unemployment rate for July 2026 is 4.1% against 4.2% a year earlier, so it is lower, not higher. Both halves of the test fail. On this definition, and it is only this definition, the answer is no.

A different threshold gives a different answer, and that is worth conceding plainly rather than hiding. Lower the inflation bar to 3% and July 2026 still fails, because unemployment is not rising. Replace the direction test with a level test at 5% and July 2026 fails again, because 4.1% is below it. The result is not fragile to the thresholds in this instance, but it would be in others, and any page reporting a verdict without its test is reporting an opinion.

The 1970s, at the size they actually were

The episodes of the 1970s and early 1980s are invoked constantly and their magnitude is usually understated. The highest twelve-month change in the consumer price index since 1948 is +14.76%, in March 1980. The unemployment rate reached 10.8%in November 1982, its post-war maximum until April 2020. The Federal Reserve’s policy response is visible in a different series entirely: the ten-year Treasury yield reached 15.84% on 30 September 1981, the highest value in a daily file that runs from January 1962 to August 2026.

The two longest episodes in the table — 24 months from January 1974 and 18 months from September 1969 — each contain an entire NBER recession. And the two episodes of 1979–81 and 1981–82 are separated by a five-month gap and correspond to the two separate recessions the NBER dates seventeen months apart, which is why they appear as two rows rather than one. The merge rule is two months, and it is stated for the same reason the inversion page states its own: a different rule gives a different episode count from identical data.

For scale at the other end of the price series, and because it is rarely shown: the whole file from January 1913 contains a twelve-month change of +23.67% in June 1920 and −15.79%in June 1921 — a swing of nearly forty percentage points inside twelve months, before the unemployment series existed to pair it with.

Four things to insist on whenever the word appears

• The test, in numbers. Which inflation measure, over which period, at what threshold; which labour measure, and level or direction. Without those five, a stagflation claim cannot be checked by anybody.
• The count. How many months since 1948 met that test, and when the last one was. On the test used here the answer is 86 and August 2008.
• The merge rule. Eighty-six qualifying months become eight episodes at a two-month merge and a different number at any other.
• The absent month. Neither series has an October 2025, so no test of any kind can be run on it.

What these sources do not say

• No agency defines or measures stagflation.The test on this page is this page’s own, stated in full so it can be reproduced or rejected.
• The thresholds are choices. A different inflation bar or a level test on unemployment gives a different count, and the effect of two alternatives is stated above.
• No cause is identified. Nothing read for this page explains why any of the eight episodes occurred, and no explanation is offered.
• October 2025 cannot be tested. Neither series has a value for it.
• Nothing outside the United States. No foreign price or labour series was read, and definitions differ between statistical agencies.
• No claim about what any asset does in these episodes. One yield series is quoted for scale and nothing is inferred from it.

How to verify this yourself

1. Download CPIAUCNS and UNRATE and align them by month. Both are monthly and neither has an October 2025.
2. Compute the twelve-month change in the price index for every month and count those at or above 5% since 1948. You should get 197.
3. Add the second condition, that unemployment exceeds its level twelve months earlier. The count should fall to 86.
4. Group those 86 months into runs, merging gaps of two months or less. You should get eight episodes, the longest 24 months from January 1974.
5. Check the most recent qualifying month. It should be August 2008.
6. Apply the same test to every month of 2021, 2022 and 2023. None passes, because the labour condition fails.

Where to go next on this site

• Inflation and deflation — the price index behind half this test, and the four questions any inflation figure has to answer.
• Unemployment and the labour market — the other half, and why one survey has no October 2025.
• Economic cycles — the chronology every episode above is placed against.
• All macroeconomics guides — the area index, one click away.

Sources opened for this page

Each source below was opened and read on 13 August 2026. Verified on 13 August 2026.

• Federal Reserve Bank of St. Louis — FRED series CPIAUCNS, consumer price index for all urban consumers, not seasonally adjustedUsed for: all 1,362 monthly index values from January 1913 to July 2026, every twelve-month change computed from them, the 197 months at or above 5% since 1948, the maximum of plus 14.76% in March 1980, the most recent month at or above 5% in February 2023 at plus 6.04%, the July 2026 reading of plus 3.36%, and the 1920 and 1921 extremes. Read 13 August 2026.
• Federal Reserve Bank of St. Louis — FRED series UNRATE, civilian unemployment rateUsed for: all 942 monthly values from January 1948 to July 2026, the year-on-year direction test in every month, the post-war maximum of 10.8% in November 1982, the July 2026 reading of 4.1% against 4.2% in July 2025, and the absence of any value for October 2025. Read 13 August 2026.
• National Bureau of Economic Research — US Business Cycle Expansions and ContractionsUsed for: the peak and trough months used to relate each of the eight episodes to the chronology, including the separate January 1980 and July 1981 recessions that produce two episodes rather than one, and the note that the business cycle data were last updated on 14 March 2023. Read 13 August 2026.
• Federal Reserve Bank of St. Louis — FRED series DGS10, 10-year Treasury constant maturityUsed for: the daily series from 2 January 1962 to 11 August 2026 and its maximum of 15.84% on 30 September 1981, used to give the scale of the 1981 episode in a series independent of the two used in the test. Read 13 August 2026.
• US Bureau of Labor Statistics — Public Data API v2, series CUUR0000SA0 and LNS14000000Used for: confirmation at the primary publisher that both the consumer price index and the unemployment rate return no value for October 2025, so no test can be applied to that month. Read 13 August 2026.

Change log

• 13 August 2026— page rewritten from scratch against the full FRED consumer price index and unemployment files, the NBER chronology, the ten-year Treasury series and the BLS public API, all opened that day. Removed from the previous version: stagflation described as a recognised economic state with no definition given; a claim that the 2021 to 2023 period was stagflationary, which fails on the labour condition in every month; references to the 1970s with no figures attached; a set of policy prescriptions no source supports; and any suggestion that an official body identifies or dates such episodes.

Vextor Capital is not authorised under MiFID II as an investment firm, is not registered with the SEC, FINRA, the CFTC or the NFA, is not an investment adviser, holds no position in any instrument named on this page, and receives no payment from any bank, broker, exchange, index provider or data vendor. Every outbound link on this page points to a statistical agency, a central bank or a public data series.