When was the last yield curve inversion? It ran from 2022 to 2024. The gap between the 10-year and 2-year Treasury yields first dipped below zero for two days in April 2022, then went negative for good in July. It sank as far as 1.08% below zero in July 2023 and didn’t turn positive for good until early September 2024. Counting from that first dip, the episode lasted 888 days, the longest since the data begins in 1976.
An inversion means short-term Treasuries pay higher yields than long-term ones, which usually happens when the Fed has pushed rates up and investors expect it to cut later. It has come before every recession in that record. So where’s the recession this time? It has now been 55 months since the curve first inverted, and no recession has started.
How long after an inversion does a recession start?
We found every inversion since 1976 and measured how long it took for the next recession to begin.
Every earlier inversion was followed by a recession; 2022 is still waiting.
The typical wait was about a year and a half, 17.5 months at the median. But the range is wide, from six months in 2019 (with a pandemic in between) to 34 months after a short dip in 1998. The 2022 episode has now run well past the longest of them.
Seven cases is a small sample. The curve has a strong record, but it’s not a timer that goes off on a set date.
Is the danger over once the curve turns positive?
This is where the history surprises people. In past cycles, the curve turning positive again didn’t mean the danger had passed. It often meant the recession was close.
Recessions usually started within months of the curve turning positive again.
For the inversions that began in 1978 and 1980, the curve was still inverted when the recession started. After the 1988, 2000, 2005 and 2019 inversions, the recession began within seven months of the curve turning positive. Why? The curve usually turns positive because short-term rates fall, and they fall when the Fed cuts as the economy weakens.
Only the 1998 dip broke the pattern: the next recession came 32 months later, after the curve had inverted again in 2000. This time the curve has been positive for 26 months.
So was the signal wrong this time?
It’s too early to say for sure. Recessions are dated by the National Bureau of Economic Research, a private research group. It names the starting month long after the fact. In theory, a downturn could already have begun. But the jobs data show no sign of one yet. A separate warning sign tied to unemployment, the Sahm rule, also went off in mid-2024, and no recession followed.
Where is the curve now? Only slightly positive. The Fed cut rates through 2025, then raised its target range again in September:
The curve is positive again, but only just.
| Measure (Nov. 6) | Value |
|---|---|
| 10-year minus 2-year Treasury yield | +0.35% |
| 10-year minus 3-month Treasury yield | +0.67% |
| Fed target rate, upper limit | 4.00% |
| Chance of recession within a year, our model | 9% |
The New York Fed’s own recession model uses the 10-year minus 3-month gap. We ran that gap through the same kind of formula, using every month since 1982. It puts the chance of a recession within a year at about 9%. That’s low, but the formula learned from the same handful of recessions.
I don’t think the signal was simply wrong. But I wouldn’t call the recession “late,” either. An inverted curve shows that money is tight. It doesn’t cause the recession. This time, many households and companies had borrowed at long fixed rates, so the squeeze reached them more slowly than in past cycles.
What I’d take from it
When the curve inverted in 2022, I mostly saw it in headlines. They kept warning of a recession while paychecks and jobs around me looked the same. I bought a home in 2023 on a fixed rate, and the curve played no part in that choice. I picked a payment I could keep making if my income fell. I’ve never made a trade because of the curve.
If you take one thing from it, don’t try to guess the month. Check whether your loans are fixed or variable, and how many months you could cover your bills without a paycheck. That answer helps whether a recession comes next year or not at all.
What I’m watching: the monthly jobs reports, whether the 10-year minus 2-year gap dips below zero again, and the Fed’s December 9 decision.
Sources
Method: We took the daily 10-year minus 2-year Treasury spread from FRED (since June 1976) and marked every day it closed at or below zero. Inverted days less than a year apart count as one episode, unless a recession started in between, and days during a recession don't count. For each episode we measured the months from its first inverted day, and from its last, to the next recession start in the NBER monthly dates (FRED's USREC series). Data: US Treasury and NBER via FRED; calculated by MKViz. Code and data
- Federal Reserve Bank of St. Louis (FRED), 10-Year Treasury Constant Maturity Minus 2-Year Treasury Constant Maturity
- Federal Reserve Bank of St. Louis (FRED), 10-Year Treasury Constant Maturity Minus 3-Month Treasury Constant Maturity
- Federal Reserve Bank of St. Louis (FRED), NBER based Recession Indicators for the United States
- National Bureau of Economic Research, US Business Cycle Expansions and Contractions
- Federal Reserve Bank of New York, The Yield Curve as a Leading Indicator