The Yield Curve: What an Inversion Signals, and What It Does Not

Few indicators carry the reputation of the yield curve. It has preceded most modern US recessions, it is free to check, and it updates every day.

It is also over-interpreted more often than almost anything else in markets, largely because people treat a signal with a two-year error bar as if it were a calendar.

What the curve is

Plot the yield on government bonds against how long until they mature - three months, two years, ten years, thirty years - and join the points. That line is the yield curve.

It normally slopes upward. Lending money for ten years exposes you to more inflation risk, more interest rate risk and more uncertainty than lending for three months, and investors expect compensation for that. The extra compensation is called the term premium.

An upward-sloping curve is the resting state of a functioning bond market.

What inversion means

Inversion is when the line slopes downward - short-dated bonds yielding more than long-dated ones.

To see why that is unusual, consider what a long yield is made of. A ten-year yield is approximately the average of expected short-term rates over the next ten years, plus a term premium.

For the ten-year yield to fall below the two-year, the market has to expect short-term rates to be substantially lower in future than they are now. Central banks cut rates when growth is weakening or inflation is falling. So an inverted curve is the bond market saying, collectively, that it expects conditions to deteriorate enough to force cuts.

The curve is not a prediction engine. It is an aggregation of what a very large number of participants are already expecting.

Why the track record is good and the signal is weak

Both things are true and they sit awkwardly together.

The correlation is real. Inversions of the 10-year/2-year spread have preceded most US recessions of the modern era, and the NBER's recession dating makes that comparison straightforward to check.

But the lag has ranged from months to roughly two years. That range makes it nearly worthless for timing. Someone who moved entirely to cash on the first day of an inversion has, in some episodes, then sat out a substantial further advance in equities before any downturn began. Being early is not the same as being right when you have to live through the gap.

There have also been false signals - inversions that were not followed by a recession - and the sample size is small. Modern recessions number a handful. A pattern with a handful of observations, a two-year error bar and no controlled test is a piece of evidence, not a rule.

The complication nobody wanted

The classic interpretation assumes long yields move mainly on expectations of future policy. Large-scale central bank bond buying broke that assumption for an extended period.

When a central bank buys enormous quantities of long-dated government bonds, it depresses long yields directly, independently of what anyone expects rates to do. That compresses the term premium and makes the curve flatter than expectations alone would produce.

This means an inversion during or after a period of heavy bond buying may partly reflect the central bank's balance sheet rather than a recession forecast. Which part is which is genuinely contested, and reasonable economists disagree.

Which spread to look at

There is no single yield curve. Different pairs are watched, and they do not always agree.

10-year minus 2-year is the most commonly quoted, and the version with the longest record behind it.

10-year minus 3-month is preferred by a good deal of academic work, on the argument that the 3-month rate tracks actual policy more closely than the 2-year does.

5-year minus 30-year describes the long end and says more about inflation expectations than about the near-term cycle.

They can send different signals at the same time. When commentary refers to "the yield curve" without saying which spread, that is worth noticing.

Where the curve sits now

As of early September 2026, the 10-year minus 2-year Treasury spread was positive at roughly 0.4 percentage points. The curve was upward sloping.

This is checkable rather than arguable. The US Treasury publishes the full curve daily, and the Federal Reserve Bank of St Louis publishes the 10-year/2-year spread as a single series updated each business day. Anyone can look at the number rather than take a description of it.

A curve that has recently un-inverted is its own discussion. Historically, some downturns have begun after the curve steepened back, not while it was inverted - because the steepening was itself caused by the central bank starting to cut. Using inversion as the alarm and treating normalisation as the all-clear gets that sequence backwards.

How to use it sensibly

Treat the curve as one input among several, describing what the bond market currently expects rather than what will happen.

It is most useful in combination. An inverted curve alongside deteriorating employment data, tightening credit conditions and falling new orders is a coherent picture. An inverted curve on its own, with everything else steady, is a question rather than an answer.

And check the actual number before accepting any characterisation of it. The data is published daily, free, by the issuer.

The bottom line

The yield curve is the bond market's aggregated expectation of the future path of interest rates. Inversion means enough participants expect enough cuts to overwhelm the normal term premium, which usually implies they expect trouble.

It has a good hit rate, a terrible sense of timing, a small sample, and a known distortion from a decade of central bank bond buying. That combination makes it worth watching closely and worth acting on cautiously.

This article is educational and is not financial advice. Bond prices and yields move inversely and are subject to interest rate risk.