The Yield Curve Has Predicted Every Modern Recession, and Here Is What It Is Saying Now

What the Yield Curve Actually Measures

Line up all the U.S. Treasury bonds by maturity from a 3-month security to a 30-year bond and graph the yield on each. That line is the yield curve. In a normal healthy economy the slope is upward. Short-term Treasuries pay less than long-term bonds because an investor who holds money for ten or thirty years wants to get paid more for not touching it. That extra payment has a name the term premium and I'll come back to it later because it turns out that it matters more than the simple version of this story usually lets on

An inverted curve inverts the relationship. Short-term yields exceed long-term yields. The usual cause is simple: the Fed is aggressively raising short-term rates to combat inflation and at the same time bond investors are betting that the economy will weaken enough that the Fed will eventually have to cut driving down long-term yields even as short-term yields continue to rise. Two forces that bend the same line in opposite directions

The most commonly cited version of this is the 2-year/10-year spread. It inverted in July 2022 reached its most inverted point in October 2023 at negative 108 basis points and did not rise above zero again until late 2024. base point is one-hundredth of a percentage point so negative 108 basis points means that the 2-year yield was about 1.08 percentage points higher than the 10-year

Why 2s10s and Why 3-Month/10-Year

You can compare any two points on the curve and people do. However two spreads dominate the conversation: the 2-year spread versus the 10-year spread called 2s10s and 3 months versus 10 years sometimes written 3m10yThey don't measure exactly the same

The 2-year yield depends primarily on what traders expect the Fed to do in the next few years. It's looking ahead but in a narrow window. The 3-month yield almost exactly tracks the Fed's current policy rate since no one expects much to happen with short-term rates three months from now. So 3m10y really compares where policy is today with where the bond market expects the economy and inflation to be a decade from now. It's a clearer read on whethercurrent policy is too tight for the long term. 2-year and 10-year bonds by contrast can move based on changing expectations about the pace of cuts even when the 10-year market view has barely changed

My own opinion close to what I understand to be the consensus among Fed researchers is that 3m10y has the best historical record of recession forecasts. It is cited less frequently mainly because 2s10s is easier to find and explain not because it is the most reliable version

A Worked Example: Building a Curve Out of Expected Rate Cuts

The clearest way to see why a reversal occurs is to build a small one yourself. Each number below is illustrative chosen to be round and easy to verify by hand not an actual market quote

Let's start with an idea: the yield on a bond maturing N years from now is roughly the average of the short-term rate that investors expect to see in each of those N years ignoring the term premium for a moment to keep the arithmetic simple. Now suppose the Federal Reserve's overnight rate is at 5.00 percent today the market expects it to fall to 4.00 percent next year then to 3.00 percent.the following year and then remain at 3.00 percent as the long-term "neutral" level from year four to year ten

The illustrative 2-year return is the average of the first two years: 5.00 plus 4.00 is 9.00 divided by 2 is 4.50 percent. The illustrative 10-year return is the average of the ten years: 5.00 plus 4.00 plus eight years at 3.00 is 5 plus 4 plus 24 which is 33 divided by 10 is 3.30 percent

Subtract the 2-year term from the 10-year term and you get 3.30 minus 4.50 which equals negative 1.20 percentage points or negative 120 basis points. This toy curve is inverted more than the actual October 2023 peak of negative 108 basis points although nothing dramatic happens after year three. All of the inversion comes from the market expecting cuts in years two and three. That's the miniature mechanism. AHigh current rate plus a credible expectation of future cuts translate on average into a lower long-term yield than the short-term yield paid today

Why the 2022 to 2024 Inversion Did Not Produce a Recession

Based solely on historical records a recession should have appeared in late 2023 or early 2024. It never did at least not on schedule. The most credible explanation I've found is fiscal not monetary: Roughly $6 trillion in federal support flowed between 2020 and 2023 through COVID relief infrastructure legislation the CHIPS Act and the Inflation Reduction Act

This is a huge cushion of demand big enough to offset the credit squeeze that an inverted curve typically causes. Households were still spending pandemic savings well into 2023. Business investment had tax credits. The Federal Reserve raised its rates at the fastest pace since the 1980s over this same stretch and the economy absorbed most of the hit because fiscal policy was pushing in the opposite direction at the same time. Two forces clashed and for a time the one wonfiscal.The investment did not fail.It predicted a recession that fiscal policy delayed and quite possibly avoided entirely

Case Study: Silicon Valley Bank and the Spread the Curve Squeezes

The banking mechanism in the above description is not theoretical. It has a recent and very public example: Silicon Valley Bank

During 2020 and 2021 with short-term rates near zero SVB received a wave of deposits from tech startups and cash-filled venture funds and put much of that money into long-duration Treasuries and agency mortgage bonds. That's an ordinary banking strategy in a low-rate world: borrow short-term with deposits earn more on longer-term assets and pocket the spread

Then the Fed made a reversal and broke off trading in both directions at once. Rising rates drove down the market value of SVB's long-duration bond portfolio as bond prices and yields move against each other and the same environment made its deposits much less stable. Startups were burning cash instead of raising new rounds so they were withdrawing deposits instead of adding them just when the bank's assets were underwater. When SVB tried to sellpart of that portfolio to raise cash had to realize the losses instead of just keeping them on paper and the announcement itself triggered a run on depositors concentrated in a small closely interconnected set of venture capital-backed companies. Regulators seized the bank in March 2023. It's the clearest real-time demonstration I know of of the previous announcement. An inverted curve doesn't just predict a credit squeeze somewhere in the economy. It can directly ruin a bank's balance sheet.specific if that bank gets caught holding long-duration assets financed by short-term fleeting money

What the Curve Is Saying Right Now

In June 2026 the 2-year Treasury bond yields about 4.18 percent and the 10-year bond about 4.45 percent. This is a 2-10 shilling spread of about 27 positive basis points. At first glance a normal positive curve

Look closer and the shape is stranger than the headline number suggests. The return above zero is mainly due to a long-term rise due to inflation fears linked to the energy shock of the Iran war not an initial drop due to recession expectations as a healthy re-intensification usually works. Traders have a name for this pattern: a bearish steepening both ends of the curve rising and the long end rising faster. Historically that shape manifests itself morein stagflationary periods than in clean expansions

The Federal Reserve kept its policy rate between 3.50 and 3.75 percent at the June 2026 meeting and the updated dot chart showed that the median monetary authority was expecting a possible increase not a cut by the end of the year. That's not what a central bank does when it's confident that a soft landing is underway. Whether this shape of the curve ends up meaning a soft landing stagflation or simply a longer delay before a recession sets in depends hugely on how quickly the policy frameworkIran peace will actually restore energy flows across the Gulf. I don't think anyone has a sure answer to this question yet and I would treat anyone claiming certainty here with some suspicion

The Counterargument: A Small Sample and a Very Elastic Lag

"Predicted every modern recession" is the kind of phrase that sounds like a law of physics and is actually a description of a very small data set. It's worth thinking about this before taking the indicator too seriously

Count the recessions for which the modern era curve is credited and you'll be working with something like eight or nine data points depending on exactly where you draw the line for "modern" and how you count double-dip recessions as one recession or two. A perfect record in eight or nine tests sounds impressive until you remember how little chance the indicator had of being wrong. A coin that comes up heads nine times in a row is either a very unusual coin or a very small sample size. I don't really know what the modern era curve is.performance and I'm skeptical of anyone who tells you they do know it with complete confidence

The lag problem compounds this. An investment does not come with a countdown clock. Historically the gap between the investment and the start of a recession has ranged from less than a year to more than two years and there is no reliable way to know in advance what type of cycle you are in.The 2022 reversal is the extreme case within the sample itself. It began in July 2022 deepened until October 2023 did not intensify again until late 2024 and at the time of writing in mid-2026 has not yet been followed by a recession by any conventional dating. This is not a short delay and arguably not even a long delay by historical standards. It may be acompletely different animal and expanding the definition of "eventually being right" enough will make almost any indicator seem infallible

Did Quantitative Easing Break the Signal?

This is the version of the counterargument that I find hardest to dismiss. Let's go back to the term premium mentioned above the extra yield that investors demand for holding a long-term bond rather than rolling over short-term bonds over and over again. A normal upward-sloping curve is partly a story about expected future rates and partly a story about that premium. An inverted curve is supposed to mean that the portion of expected future rates has become strongly negative enough to completely swamp the term premium.term

But the term premium itself is not fixed. Quantitative easing which means the Federal Reserve and other big central banks directly buy huge amounts of long-duration bonds mechanically pushes down long-term yields by increasing demand for those bonds regardless of what anyone expects short-term rates to happen in the future. Between the 2008 financial crisis and the mid-2020s central bank balance sheets grew enormously and a significant set ofResearch argues that this compression of term premiums was structural and not temporary. If that's true part of what looks like "an inverted curve predicting a recession" over the past fifteen years could actually be a term premium that the central bank's bond purchases artificially set at a low level. That's a different signal wearing the same clothes

I want to be careful here because this is exactly the kind of argument that is easy to overstate. The compression of forward premiums does not make the yield curve meaningless. It means that the threshold for what is considered a "real" investment may have changed and that a given number of negative basis points may contain less recessionary information today than the same number decades ago. No one including me has a clear way to separate the two effects in real time. That is a real limitation of the model not a footnote topage

How I Actually Use the Yield Curve

My read and I want to clearly point out that this is my own framework and not a forecast is that the yield curve is much more useful as a description of what the bond market currently believes than as a countdown timer to a recession

The way I actually use this: I check both 2s10s and 3m10y rather than relying on either of them alone as they may disagree and when they do the disagreement itself is informative. I pay more attention to the shape of a new deepening than to the raw depth of the reversal. There is a pattern not a law in which recessions have often started closer to the time the curve reverses than when it is at its most negativewhich is the opposite of the intuitive story and the part I think people most often overlook. I also strive not to treat the fiscal and monetary context as background noise. The 2020 to 2023 period taught me that a given tax authority can override what the curve is "supposed" to mean for years in a row

Where I land as it stands whatever the value of my reading: a bearish rally fueled by an energy shock coupled with a Fed talking about a hike rather than a cut doesn't strike me as the setup for a clean soft landing. Nor does it resemble the classic pre-recession reversal because the front is not the falling part. Honestly my best guess is that this is a regime that the old playbook was not designed to read clearly closer in spirit to an oil crisis.from the 1970s than from 2007 or 2019. I could be wrong and I say this honestly and not as a hedge to cover myself. This specific curve shape is one of the most difficult I have tried to model. None of this is a call to buy or sell anything. It is a framework for reading one entry among many

The Bottom Line

The mechanism is real. An inverted curve reflects a bond market valuation of future rate cuts and directly reduces bank lending spreads giving the indicator a causal channel and not just a correlation. The 2- to 10-year spread inverted in July 2022 bottomed at negative 108 basis points in October 2023 and did not rise again until late 2024 the longest reversal on record and has so far not been followed bya recession largely because roughly $6 trillion of fiscal stimulus from 2020 to 2023 offset credit tightening. In June 2026 the curve sits at about 27 basis points positive a bearish tilt driven by an Iran-linked energy shock rather than a clear new rise with the Fed holding between 3.50 and 3.75 percent and signaling a possible rise. The "predicted every recession" framework overstates a track record built on ahandful of data points and a lag that has ranged from less than a year to several years and the effect of quantitative easing on the term premium may have changed what a given level of investment even means. My own approach is to read the curve as a snapshot of market expectations rather than a clock weigh both the re-steepening and the investment itself and keep the fiscal context in view at all times. None of that is investment advice. It's the lens I use and I prefer to beBe honest about your limits than pretend the performance curve is a crystal ball