The Curve

The yield curve is simply the difference between what the government pays to borrow money for ten years and what it pays to borrow for two. In a normal economy, long-term debt costs more than short-term debt — just as a 30-year mortgage charges a higher rate than a 2-year car loan. The lender bears more uncertainty over a longer horizon: inflation could erode the bond’s value, the borrower’s situation could change, and the lender’s capital is locked up for longer. Investors demand compensation for that extra risk. A positive 10-year-minus-2-year spread, which the FRED series T10Y2Y tracks, is the normal state of affairs. Over the past fifty years, the median spread has been +0.79% and the average +0.85%.

When the curve inverts — when short-term rates exceed long-term rates — it means the bond market is making a collective bet that the future will be worse than the present. Investors are willing to accept a lower yield on ten-year Treasuries because they expect the Federal Reserve to cut short-term rates in response to economic weakness. They’re essentially saying: “I’ll take 4% for ten years because I think rates will be much lower than 4% within the next couple of years.” That expectation of falling rates is, in effect, an expectation of recession — because the Fed almost never cuts rates aggressively unless the economy is faltering.

This is why the yield curve has such a remarkable track record. It isn’t an arbitrary technical indicator like a moving average crossover or a momentum oscillator. It aggregates the forward-looking expectations of millions of bond market participants — pension funds, central banks, insurance companies, sovereign wealth funds — who collectively manage hundreds of trillions of dollars and whose livelihoods depend on correctly forecasting interest rates. When the curve inverts, it means the smartest money in the world is positioning for trouble.

The T10Y2Y series begins in June 1976. In those fifty years, the spread has been negative on 2,049 of 12,460 trading days — about 16.4% of the time. But that 16.4% is not evenly distributed. The 1970s and 1980s, with their extreme interest rate volatility under Paul Volcker, account for the bulk of inverted days. The entire decade of the 2010s saw exactly 3 days of inversion. Then the 2020s arrived, and the curve spent 537 trading days inverted between July 2022 and August 2024 — more than two straight calendar years.

The deepest yield curve inversion in American history didn’t come from a financial crisis or a market panic. It came from a deliberate act of policy. In October 1979, Paul Volcker — newly appointed chairman of the Federal Reserve — announced that the Fed would stop targeting the federal funds rate and instead target the money supply directly. The practical effect was to let short-term interest rates go wherever the market took them. And where the market took them was into the stratosphere.

By March 1980, the 2-year Treasury yield had soared to 15.03%, while the 10-year sat at 12.62%. The spread collapsed to −2.41% on March 20, 1980 — the deepest inversion in the T10Y2Y dataset and a record that still stands forty-six years later. The bond market was screaming: short-term rates were unsustainably high, and the economy would buckle. It was right. The economy entered recession in January 1980 — actually two months before the deepest inversion point. The 1980 recession was brief (six months), but Volcker’s crusade against double-digit inflation wasn’t over.

The curve briefly un-inverted during the summer of 1980, as the recession dragged short-term rates down. But Volcker tightened again. By September 1980, the curve was back in negative territory, and by May 1981 it had plunged to −1.36%. This second inversion lasted 278 trading days. The recession that followed — July 1981 to November 1982 — was the worst since the Great Depression at that time, with unemployment reaching 10.8%. The Volcker era established the yield curve’s reputation: two inversions, two recessions, no ambiguity.

But the Volcker inversions were unusual in an important way. They were caused by the Fed deliberately pushing short-term rates to extreme levels — the fed funds rate hit 20% — to crush inflation that had reached 14.8%. The inversions weren’t the market anticipating recession; they were the market recognizing that the Fed was engineering one. This distinction matters because later inversions would operate through a different mechanism — the market anticipating Fed cuts, rather than reacting to Fed hikes.

Since 1976, there have been six major yield curve inversions — episodes where the T10Y2Y spread stayed negative for more than 20 trading days. Five of those six were followed by recessions. The sixth, which ended in September 2024, remains the great exception. The table below shows each inversion, its depth, its duration, the recession that followed (or didn’t), and the lag between the two events.

Notice the pattern in the post-Volcker era. The 1989 inversion was shallow (−0.45%) and brief (123 trading days), but the recession followed nine months after de-inversion. The 2000 inversion was slightly deeper and lasted 220 days; recession arrived two months after the curve turned positive. The pre-GFC inversion in 2006–2007 was the shallowest of all — the curve barely touched −0.19% — but the recession that followed was the worst since the 1930s. Depth of inversion, it turns out, doesn’t predict severity of recession. It simply signals that something is wrong.

The most remarkable feature of this table is the lag. In the post-Volcker era, recessions typically began 2 to 9 months after de-inversion — not during the inversion itself. This makes intuitive sense. The inversion reflects the market’s expectation of future rate cuts. The de-inversion happens when those cuts actually begin (or when long-term rates rise due to fiscal fears). By the time the curve turns positive, the recession is often already underway or about to start. It’s the un-inversion, not the inversion itself, that should make you nervous.

The 2022–2024 inversion was exceptional by every measure. It was the deepest since Volcker, bottoming at −1.08% on July 3, 2023. It was the longest since 1978, spanning 537 trading days (783 calendar days). And it produced no recession — the first time a major, sustained inversion failed to deliver one in the history of the T10Y2Y series.

The mechanics were straightforward. In March 2022, the Federal Reserve began the most aggressive rate-hiking cycle in four decades, lifting the fed funds rate from near zero to 5.33% by July 2023. The 2-year Treasury yield, which closely tracks expected Fed policy over the next two years, surged in lockstep — from 0.77% at the start of 2022 to 4.94% by July 2023. The 10-year yield rose too, but more slowly: from 1.63% to 3.86% over the same period. The gap between them produced the inversion.

What made this inversion different was what the economy did while the curve was screaming recession. Real GDP grew at an annualized rate of 2.5% in 2023 and 2.8% in 2024. The unemployment rate never rose above 4.3%. The S&P 500 gained 24% in 2023 and another 23% in 2024. Consumer spending held up. Corporate earnings grew. The labor market, as we’ll explore in Episode 6, remained historically tight. Something in the transmission mechanism from “inverted curve” to “recession” had broken — or at least been delayed.

Several explanations have been offered. The most compelling is that the post-pandemic economy was structurally different. Households had accumulated $2.3 trillion in excess savings during COVID-era lockdowns and stimulus. Corporations had refinanced their debt at ultralow rates during 2020–2021, locking in cheap capital for years. The labor market was so tight — at one point there were two job openings for every unemployed person — that even as the Fed tightened, employers couldn’t find enough workers to lay off. The usual channels through which higher rates cause recessions (credit contraction, job losses, debt defaults) were clogged with pandemic-era buffers.

Another explanation focuses on the Fed itself. In previous cycles, the Fed tended to hike until something broke — a bank failure, a market crash, a surge in unemployment — and then scrambled to cut rates. This time, the Fed paused at 5.33% in July 2023 and held for over a year, allowing the economy to adjust gradually rather than slamming into a wall. When it finally cut in September 2024, it did so preemptively, before any crisis. The curve de-inverted on cue, the economy absorbed the rate shock, and the anticipated recession never materialized.

Wall Street tracks two versions of the yield curve: the 10-year-minus-2-year spread (T10Y2Y) and the 10-year-minus-3-month spread (T10Y3M). They often diverge, and the divergence reveals something important about the nature of the inversion. The 2-year yield reflects market expectations of where the Fed will be over the next two years. The 3-month yield reflects where the Fed is right now. When the two curves disagree, it’s because the market distinguishes between what the Fed has done and what the Fed will do.

The table tells a fascinating story. The T10Y2Y inverted first — in July 2022 — because the 2-year yield, driven by expectations of rapid rate hikes, jumped ahead of the 10-year. The T10Y3M didn’t follow until November 2022, because the 3-month yield (which tracks the current fed funds rate) was still catching up to where the market expected rates to go. By mid-2023, both curves were deeply inverted, but the T10Y3M was far more negative (−1.59% vs. −0.93%) because the gap between what the Fed had already done (5.33% fed funds) and what the market expected long-term (low 4s) was wider than the gap between the market’s 2-year and 10-year expectations.

The divergence was most dramatic in September 2024. The T10Y2Y had already un-inverted (+0.10%), because the 2-year yield was falling in anticipation of imminent rate cuts. But the T10Y3M was still deeply inverted (−1.20%), because the 3-month yield — still pegged to the current fed funds rate of 5.33% — hadn’t budged yet. In other words, the bond market had already priced in the cuts, but the Fed hadn’t delivered them. This lag is why relying on a single curve measure can be misleading. The T10Y2Y gave the “all clear” months before the T10Y3M did.

The chart makes the causal mechanism unmistakable. The yield curve inverts because the Fed hikes rates. As the fed funds rate rose from 0.08% in January 2022 to 5.33% by August 2023, the T10Y2Y spread fell from +0.78% to −0.93%. The correlation isn’t perfect — bond market expectations, term premium fluctuations, and global capital flows all introduce noise — but the direction is clear. Every major Fed hiking cycle in the past fifty years has compressed or inverted the yield curve, because the Fed directly controls the short end of the curve but only indirectly influences the long end.

The long end of the curve is where things get interesting. The 10-year yield reflects not just expected future short-term rates but also the term premium — the extra compensation investors demand for holding long-duration bonds. In theory, the term premium should always be positive (you should get paid more for tying up your money longer). But from roughly 2010 to 2021, the term premium was compressed — sometimes to near zero or even negative — by quantitative easing, foreign central bank purchases of Treasuries, and a global savings glut. When the term premium is abnormally low, the curve inverts more easily, and the inversion may be sending a different signal than it traditionally does.

This is the “term premium” theory for why the 2022–24 inversion didn’t produce a recession. The argument goes: the curve wasn’t inverted because the bond market expected recession; it was inverted because structural forces had compressed the term premium so much that even a modest Fed hiking cycle was enough to push the spread negative. The inversion was real but the recession signal was distorted — like a smoke alarm going off because someone burnt toast, not because the house is on fire.

There’s empirical support for this view. The New York Fed’s estimate of the 10-year term premium was negative for most of 2022–2023, and several academic papers published in 2024 showed that adjusting for the term premium would have eliminated much of the inversion signal. Others disagree, arguing that adjusting away the signal is just hindsight bias dressed up as methodology. The debate remains unresolved, which is itself a useful thing to know: the yield curve’s recession-prediction track record, while extraordinary, rests on a sample of only five or six events, and the underlying economic structure has changed significantly since the Volcker era.

The table below shows the annual average, minimum, and maximum T10Y2Y spread for every year in the dataset. Shaded rows mark years with notable inversions. The range column — the difference between the year’s high and low — captures volatility. In calm years like 1985 or 2016, the curve barely moves. In crisis years like 1980 or 2022, it swings by more than two full percentage points.

The T10Y2Y spread sits at +0.51% as of April 9, 2026 — the 40th percentile of its historical distribution. The 10-year Treasury yields 4.29% and the 2-year yields 3.79%, producing a healthy positive slope. The curve has been positive since late August 2024 and has been remarkably stable, oscillating between +0.46% and +0.74% for the past three months. By the standards of this indicator, the signal is unambiguously green.

But green doesn’t mean safe. The curve was also positive in early 2008, when the spread averaged +1.65% for the year — and the worst financial crisis since the 1930s was already underway. The spread was positive at +0.39% through most of 1990, even as the economy was tipping into the Gulf War recession. The yield curve is a leading indicator, not a coincident one. By the time the curve is telling you a recession is happening, it’s already re-steepened because the Fed has cut rates and the 2-year has plunged. The danger zone is not during inversion — it’s in the months immediately after de-inversion.

We are now 19 months past the September 2024 de-inversion. Historically, post-1982 recessions have arrived within 2 to 9 months of de-inversion. We are well past that window. This is the strongest evidence yet that the 2022–24 inversion was, in fact, a false alarm — or at minimum, that whatever it was signaling hasn’t materialized on the traditional timeline. The fed funds rate has come down from 5.33% to 3.64%, the labor market remains strong (initial claims at 219K, as covered in the dashboard), and corporate credit spreads are near historic lows.

The yield curve’s threshold levels, based on the historical distribution, help put the current reading in perspective. Below −0.27% (the 10th percentile) signals deep stress — the market is pricing aggressive rate cuts. Between −0.27% and +0.19% (10th to 25th percentile) is the caution zone: the curve is flat or barely inverted, and the Fed is likely at or near its peak rate. From +0.19% to +0.79% (25th to 50th percentile) is where we sit now — normal territory. And above +1.48% (75th percentile), the curve is steep, typically seen in the early stages of recovery when the Fed has just slashed rates and the economy is beginning to heal. The steepest the curve has ever been was +2.91% on February 4, 2011, coming out of the GFC, when the Fed held rates near zero while the 10-year had risen on recovery hopes.

Five decades of yield curve data offer several lessons for anyone who uses this indicator to assess financial stress.

First, the signal is real but imprecise. An inversion of the 10Y–2Y spread has preceded every recession since the data begins in 1976 (and every recession since 1969 using older data sources). But the lead time varies wildly — from two months to nearly two years. An inverted curve tells you that a recession is more likely, not when it will arrive. Trading on the curve alone has historically been unprofitable because the lag is so variable that you can be right about the direction and still lose money on the timing.

Second, depth doesn’t predict severity. The 2006–07 inversion was the shallowest in the dataset (−0.19%), yet it preceded the Global Financial Crisis. The 1989 inversion was also shallow (−0.45%), and the subsequent recession was mild. The 2022–24 inversion was the deepest since 1981 and has produced no recession at all. If you’re looking for a correlation between how inverted the curve gets and how bad the downturn will be, the data says there isn’t one.

Third, watch the de-inversion, not the inversion. Recessions have historically started in the months after the curve turns positive, not while it’s inverted. The de-inversion typically happens because the Fed starts cutting rates (pulling down the 2-year yield) or because long-term inflation expectations rise (pushing up the 10-year yield). Either way, the de-inversion is the moment when the theoretical risk becomes an imminent threat. In the Volcker era, recessions actually overlapped with the inversion period itself, but since 1989, the pattern has been consistently: invert, wait, de-invert, recession.

Fourth, context matters as much as the number. The yield curve doesn’t exist in isolation. It’s one node in a web of interconnected indicators. An inverted curve paired with widening credit spreads (Episode 2), rising claims (Episode 6), and a triggered Sahm Rule (Episode 5) is a far more alarming combination than an inverted curve with tight credit, low VIX, and a booming labor market — which is roughly what we had during 2022–2023. The dashboard approach from Episode 1 exists precisely because no single indicator, however impressive its historical record, should be read in isolation.

Fifth, structural changes can bend the rules. Quantitative easing, compressed term premiums, pandemic-era excess savings, and the globalization of bond markets have all changed the plumbing of the yield curve in ways that the pre-2008 data doesn’t capture. The curve’s track record was built in an era of higher rates, larger term premiums, less central bank intervention, and fewer global buyers of U.S. Treasuries. Whether that track record will hold in the new regime is an open question — and the 2022–24 inversion may be the first evidence that it won’t.