Does an Inverted Yield Curve Predict a Recession?

An inverted yield curve, when short-term Treasury yields rise above long-term ones, has come before all 6 US recessions since 1976, with the first warning a median 17.5 months ahead. It also raised 2 false alarms in 10 warning episodes, and the lead time ranged from over a year to the edge of two. It signals recessions, not stock market crashes.

What is an inverted yield curve?

The yield curve plots Treasury yields from short to long maturities. Normally longer bonds pay more. An inversion means the yield on short bonds, usually the 2-year or the 3-month, is above the 10-year yield, so the spread (10-year minus 2-year) is below zero. It usually reflects an expectation that the Fed will cut rates because growth is slowing. See yield curve in the glossary.

As of September 2026, the monthly average spread is 0.34 points, so the curve is not inverted.

How well has it predicted recessions?

This site replays its warning rule, either the 10-year minus 2-year or the 10-year minus 3-month spread below zero, against every US recession since the data begins in 1976. On the 10-year minus 2-year spread alone, the first inversion came this many months before each recession began:

The site's own record for the rule: it flagged 6 of 6 recessions with a median lead of 17.5 months, and of 10 warning episodes, 8 were followed by a recession within 24 months and 2 were not.

What about the 2022 to 2025 inversion?

The 10-year minus 2-year spread was below zero in 26 of the 33 months from June 2022 to March 2025, one of the longest inversions in the data. In the data used here (recession dates through September 2026), no recession started afterwards. Either the signal was a false alarm, or the lead time was longer than usual. The record shows the lead can be long: before the 1981 recession the first warning came at least 24 months ahead.

Does an inversion predict a stock market crash?

No. The replay checks whether the S&P 500 fell 20% or more within 12 months of a warning. After 11 warning episodes, that happened 0 times (0%), against 6% for any random month. The curve has pointed at recessions, and markets often fall during one, but not on that schedule.

The limits are real: eight recessions is a small sample, the replay uses today's revised data, and a recession and a crash are different events. Read how the Crash Indicator works for the full method and the Sahm rule for a signal built to confirm a recession that has started.

Yield curve inversion FAQ