What Is the Yield Curve, and What Do Inversion and Steepening Mean?

The yield curve plots the interest rate on government bonds across maturities, and its shape, normal, inverted, or steepening, tells you what markets expect.

The yield curve is a snapshot of the price of time, and its shape, not its level, is what carries the signal.

What the yield curve actually is

The yield curve is a single line. To draw it, you take bonds from one issuer of the same credit quality, almost always U.S. Treasuries, and plot the yield each one pays against how long until it matures. The short end might be a 3-month bill or a 2-year note; the long end, the 10-year or 30-year bond. Each point answers one question: what annual yield do I earn for locking up my money for exactly this long?

Connect the dots and you get the curve. Its usual state is "normal," meaning upward-sloping: the longer you lend, the more yield you collect. That slope exists because lenders want extra compensation, a term premium, for tying money up longer and for bearing more inflation and rate uncertainty over a longer horizon.

The key idea is that the curve is about shape, not height. Two curves can sit at very different yield levels and still have the same slope. Inversion and steepening describe how the gap between long and short yields is changing, not whether rates themselves are high or low.

What inversion means

Inversion is when short-term yields rise above long-term yields, so the curve slopes downward. The 2-year, for example, ends up yielding more than the 10-year. That is unusual: you are being paid more to lend for two years than for ten.

Mechanically, inversion typically appears when two forces meet: current policy rates are held high, which keeps short yields up, while markets expect the central bank to cut rates later, which pulls long yields down. The curve bends below zero as a result.

Inversion draws attention because, historically, a sustained inversion of widely watched spreads such as the 2-year versus 10-year, or the 3-month versus 10-year, has preceded most U.S. recessions. That track record is why it is treated as a recession warning. The caveats matter as much as the headline. It is a probabilistic leading indicator with long and variable lead times, often in the range of 6 to 24 months; it has produced false signals; and it does not cause a recession, it reflects expectations rather than creating the outcome.

What steepening means

Steepening is the opposite motion: the gap between long and short yields widens. The long end pulls further above the short end, or a curve that had been inverted climbs back toward and over zero. Plainly, the curve is opening out from wherever it had been. Two distinct flavours hide under the same word, and they should not be blurred.

When the short end leads

Short yields fall faster than long yields. This often happens when markets begin to price imminent rate cuts or easing. Note the direction: the curve gets steeper because short rates are dropping, not because long rates are climbing. This is why "steepening" should never be read as a shorthand for "rates going up."

When the long end leads

Long yields rise faster than short yields. This tends to show up alongside stronger growth or inflation, or a rising term premium driven by supply concerns. Here the steepening is led by the long end moving higher. The slope alone cannot tell you which flavour you are in; you have to look at which end is moving.

A signal is not a forecast

The curve's shape, and the z-scores below, describe the current state of markets and how unusual it is versus trend. They are historical and statistical signals, not predictions of a recession, of where rates go next, or of asset returns. Inversion has often preceded recessions, but "often preceded" is a statement about the past, and past patterns need not repeat. Kresmion Research publishes these readings as information, not as advice or a directional call.

Keep two objects distinct. There is the curve itself, a real market object made of Treasury yields by maturity. And there is Kresmion's regime factor, a normalized z-score derived from that curve. This explainer teaches the first; the live number below is the second, used only as today's grounding.

How Kresmion measures this

Kresmion does not quote a raw curve spread here. Its cross-asset regime model converts the curve into a z-score: how many standard deviations the curve sits from its own one-year trend. A value near zero means the curve looks normal versus its recent history; a large value means it has moved unusually far from that trend.

As of 2026-06-18, the yield-curve factor reads z = -2.3236, about 2.32 standard deviations from its one-year trend. Under Kresmion's sign convention for this factor, that negative reading reflects a steepening relative to trend: the curve is unusually wide, or less inverted, compared with where it has been over the past year. It is the single most stretched of the 15 factors in the model today. The second most stretched is breakeven inflation, the market-implied inflation expectation, at z = -2.0327.

Two clarifications. First, this is a statement about the curve relative to its own recent trend, not an absolute claim: a z of -2.3236 does not mean the curve is outright steep or inverted in basis-point terms, only that it has moved far from its recent norm. Second, despite two very stretched rate-and-inflation factors, the blended regime reads Neutral, with HIGH conviction at conviction_pct = 75.00 percent and a smoothed regime score of 0.1499 on a scale centered near zero. The individual extremes are offset by the model's other factors, so the cross-asset picture is balanced rather than risk-on or risk-off. The 0.1499 figure is the smoothed regime score, a separate field from the conviction reading. Rows exist through 2026-06-18, and 2026-06-17 and 2026-06-16 also read Neutral with HIGH conviction, so the reading is stable across the latest days rather than a one-day blip.

See the blended regime read in context in our daily research notes, and how the z-scores and the regime are computed in the methodology.

Honest limitations

This grounding rests on a model-internal, normalized signal, and that carries real weaknesses. The z-score is computed against only a one-year trend, so a short lookback can make a curve look "stretched" simply because the recent window was calm; it is not a long-history anchor. The model exposes the curve as a z-score, not a raw basis-point spread, so it cannot tell you on its own whether the curve is inverted or steep in absolute terms. The sign convention is internal to this dataset, where a negative yield-curve z corresponds to steepening versus trend, and it should not be generalized to outside data. And a single day's reading, even one stable across three days, is a thin basis for any broad conclusion; the regime can shift as the underlying factors move.

Key takeaways

Frequently asked questions

Does an inverted yield curve mean a recession is coming?

Not on its own. A sustained inversion has historically preceded most U.S. recessions, which is why it is watched, but the lead time has been long and variable, often 6 to 24 months, and the signal has been wrong before. It is a probabilistic, historical indicator, not a forecast, and it does not cause a recession.

Is steepening the same as rates going up?

No. Steepening only means the gap between long and short yields is widening. When the short end leads, the curve gets steeper because short rates are falling, not rising. When the long end leads, long rates rise faster than short rates. The slope tells you the gap is widening, not which end moved or in which direction rates went.

What does Kresmion's yield-curve z-score of -2.3236 tell me?

It says that on 2026-06-18 the curve sat about 2.32 standard deviations from its own one-year trend, and under this model's sign convention that reflects a steepening relative to trend. It was the most stretched of the 15 factors that day. It measures how unusual the curve is versus its recent history, not a basis-point spread and not a prediction.