The Yield Curve Inverted for 27 Months and No Recession Came

The yield curve inverted in July 2022. It stayed inverted until late 2024, roughly 27 months, the longest sustained inversion in the FRED series.^[1] Every explainer written during it said the same thing: this indicator has never been wrong.

The recession did not come. Not on the timeline, not in the shape people described, not at all in the way the signal promised.

That is worth sitting with, because it is the first significant miss for a measure usually presented as an economic law. And the more interesting conclusion is not “the curve is broken.” It is that the part everyone stares at, the moment of inversion, was never the part that actually lined up with recessions.

What the Curve Actually Is

Plot the yield on US Treasuries against how long until they mature and you get a line. Normally it slopes up: lending the government money for ten years pays more than lending it for two, because you are committing for longer and bearing more uncertainty.

Inversion is when that flips and the short bond pays more. The usual shorthand is the 10-year minus 2-year spread, the FRED series T10Y2Y.^[1] Negative means inverted.

Why would anyone accept less yield to lend for longer? Only if they believe rates will be lower later, and locking in today's rate for a decade beats rolling short-term paper at whatever comes next. So an inverted curve is a statement about expectations: the bond market thinks rates are coming down.

Rates come down when the Fed cuts. The Fed cuts when the economy weakens. That is the whole chain of reasoning, and it is why inversion got its reputation.

The Second Mechanism, Which Is More Concrete

Expectations are a story. There is also a mechanical channel, and it is the one that does real work.

Banks borrow short and lend long. They pay depositors something near short-term rates and earn something near long-term rates on mortgages and business loans. That gap is the margin the business runs on. Invert the curve and you compress it directly, so lending gets less profitable, so banks lend less, so credit contracts, so the economy slows.

This is not a forecast. It is a squeeze applied to the plumbing of credit, which is why the indicator was ever more than superstition. And it is a useful lens on why mortgage rates do not follow the Fed: the long end is a different market with a different buyer, and the spread between the two ends is doing its own thing.

Why It Missed

Nobody has a settled answer, and anyone offering one confidently is overselling. The candidate explanations worth knowing:

• The long end was distorted. Years of central bank bond buying suppressed long-term yields, so the curve may have inverted at a lower level of genuine economic stress than the historical comparison implies.
• The labour market did not crack. The transmission from tighter credit to recession runs through people losing jobs and spending less. Employment stayed resilient, and the chain broke in the middle.
• Fiscal policy pushed the other way. Government spending was expansionary while monetary policy was restrictive, which is not the configuration the historical record was built on.
• The sample is tiny. This is the one people skip. The modern record contains fewer than ten recessions. An indicator that calls eight of eight is not a law of nature, it is eight data points, and the confidence interval around that has always been much wider than the way it gets described.

The Part Nobody Watches

Here is the detail that changes how you read this, and most coverage gets it backwards.

Inversion precedes recession by a long and variable lag, commonly quoted at 12 to 24 months.^[2] A range that wide is close to useless for timing anything. But un-inversion, the curve steepening back to positive, has historically landed closer to the actual onset of recession than the inversion did.^[2]

The mechanism is not mysterious once stated. The curve un-inverts because the Fed cuts short-term rates. The Fed cuts because something is deteriorating. So the return to a normal-looking curve is not the economy recovering, it is the emergency response beginning.

Which means the popular reading, inversion is the warning and normalisation is the all-clear, is close to inverted itself. If you were going to watch one of the two, the steepening is the one that has historically arrived nearer the trouble.

Where It Sits Now

As of mid-July 2026 the curve is positively sloped: the 10-year around 4.55%, the 2-year around 4.18%, a spread near 37 basis points.^[3]The spread has crossed back and forth over zero repeatedly since February 2026, which is worth more than any single day's reading.^[2]

Read that alongside what the Fed is actually doing rather than on its own. This week the committee held for a fifth meeting on a 9-3 vote with all three dissents wanting a hike. A curve pricing modest future cuts, against a committee arguing about raising, is a disagreement rather than a signal, and disagreements between the bond market and the Fed resolve in one direction or the other rather than staying comfortable.

What To Do With It

Not timing. A 12-to-24-month variable lag cannot tell you when to do anything, and an indicator that just missed entirely should not be promoted to a trading rule now.

Read it as a price, not a prophecy. The curve tells you what the bond market currently expects rates to do. That is genuinely useful when you are pricing anything rate-sensitive, from a mortgage to a discount rate in a valuation.

Watch the direction, not the level. A spread that crosses zero repeatedly is a flat curve, which mostly means uncertainty. Sustained steepening while the Fed cuts is the configuration with the better historical track record, and it is not the reassuring one.

Do not rebuild a portfolio around it. The people who went defensive on the 2022 inversion spent two years being early, which in practice is indistinguishable from being wrong. That is the same lesson as every other label people mistake for a forecast.