The Yield Curve

What US Treasuries pay across maturities, whether the curve is inverted today, and every inversion since 1976 — with the recession that followed each one.

Right now

+0.33 The 10-year yields 0.33 percentage points more than the 2-year, so the curve is not inverted — and has been upward-sloping since 2024-09-06.

Federal Reserve H.15 · as of 2026-09-11
MaturityYieldRelative to the 10-yearWhere it sits in its own history
3 monthsDGS3MO 4.00% -0.95 pts 51th percentile
2 yearsDGS2 4.56% -0.39 pts 49th percentile
10 yearsDGS10 4.95% 46th percentile

Percentiles are against the full history of each series, which begins 1962-01-02 for the 10-year. The 10-year minus 3-month spread, the other curve measure people watch, is currently +0.89.

The 10-year minus 2-year spread since 1976

upward-sloping inverted 1976-06-01 – 2026-09-11

Every inversion, and what followed

Inverted fromUntilLengthNext recession beganLead time
2022-07-06 2024-09-05 792 days none yet
2005-12-27 2007-06-05 525 days December 2007 23 months
2000-02-02 2000-12-28 330 days March 2001 13 months
1998-05-26 1998-07-27 62 days March 2001 33 months
1990-03-08 1990-03-29 21 days July 1990 4 months
1988-12-13 1989-11-06 328 days July 1990 19 months
1980-09-12 1982-07-16 672 days July 1981 10 months
1978-08-18 1980-05-01 622 days January 1980 16 months

Episodes separated by less than 90 days are merged, because the spread crossing zero for a week does not end an inversion in any meaningful sense. Recession dates are the NBER's business-cycle peaks. Lead time is measured from the first day of the inversion.

What the curve is

The US Treasury issues debt across maturities from four weeks to thirty years, and each trades at its own yield. Plotting yield against maturity gives the curve. Normally it slopes upward: lending money for ten years carries more uncertainty than lending it for three months, so it pays more.

An inversion is when it does not — when short-dated debt pays more than long-dated. That is unusual, and it says something specific: that the market expects short-term rates to be lower in future than they are now, which generally means it expects the central bank to be cutting, which generally means it expects the economy to weaken.

Why anyone cares

Every US recession since the 1970s has been preceded by an inversion of the 10-year minus 2-year spread. That is a genuinely striking record and it is the reason the indicator gets the attention it does.

Three things are worth holding alongside it. The lead time varies enormously — the table above shows the spread. It has produced signals that were not followed by a recession, so "never wrong" is not the claim. And the sample is small: a handful of episodes over fifty years is not many observations to build confidence on, however clean the pattern looks.

The un-inversion is the part people miss

Recessions have historically begun after the curve returns to normal, not while it is inverted. The steepening happens because the short end falls as the central bank starts cutting — which it does once the weakness it was anticipating begins to show up.

So reading a return to an upward slope as an all-clear inverts the historical pattern. It has usually been the later signal, not the cancellation of the earlier one.

2s10s or 3m10s

Two versions are in common use. The 10-year minus 2-year is the one most widely quoted. The 10-year minus 3-month has some academic support as the better predictor, and the two do not always agree — there have been stretches where one was inverted and the other was not. Both are shown above for that reason.

Where this comes from

Federal Reserve H.15 constant-maturity Treasury yields, distributed through FRED. A work of the US government, free to use, and published here as actual values with full history rather than as summary statistics. The rest of the Markets section is on the Markets overview, and every ticker's sensitivity to these rates is on its page in the correlation finder.