WhitmanTrading

What Is the Yield Curve?

Yield curve is the plot of interest rates against time to maturity, showing what money costs for a month, a year or thirty years all at once. Normally longer money pays more; when it does not, the curve is inverted, and its shape changes independently of its level.

There is no such thing as “the interest rate.” There is a rate for overnight money, one for two-year money, one for thirty-year money, and they move independently. Plot them against time and you have the curve.

How it works

A price series with several maturities and their rates marked.
The curve is every maturity's rate at once. Illustrative chart - not real market data.

Each point is the rate for lending over that period. The horizontal axis is time to maturity, the vertical axis is yield, and the line joining them is the curve.

A steady series with longer maturities paying progressively more.
Normally longer money pays more. Illustrative chart - not real market data.

The normal shape slopes upward. Lending for thirty years carries more uncertainty than lending for three months, so it pays more. That extra is compensation for time, not for credit — government curves slope up too.

A choppy series where short rates exceed long rates.
An inversion means short money pays more. Illustrative chart - not real market data.

When it inverts, short money pays more than long. That is unusual and it means the market expects rates to be lower in future — investors accept less on a long bond because they think the alternative will be worse.

Shape and level are different events

A rising series where the long end moves furthest.
A steepening curve reprices long bonds hardest. Illustrative chart - not real market data.

The curve can move up, or it can change shape, or both. A parallel shift lifts every maturity equally. A steepening raises the long end relative to the short. A flattening compresses them together.

A falling series where the gap between maturities narrows.
A flattening one compresses the difference. Illustrative chart - not real market data.

These have completely different consequences. A parallel rise damages every bond roughly in proportion to its duration. A steepening damages long bonds and barely touches short ones. Reading either as “rates went up” loses the information that mattered.

A calm series with a single rate marked against the whole curve.
One rate changing is not the curve moving. Illustrative chart - not real market data.

And the central bank sets one end, not the whole thing. Policy controls the very short rate directly. The long end is set by what the market expects over decades, which is why a rate cut sometimes moves long yields the wrong way.

A worked example

Take a curve where two-year money pays 4% and ten-year money pays 4.5%. The gap of half a percentage point is what you are paid for accepting eight extra years of uncertainty.

Now the two-year rises to 5% and the ten-year stays at 4.5%. The curve has inverted. Nobody’s headline rate “went up by half a percent” — one point on the curve moved and the relationship reversed.

A slow series where the long end moves most for a given change.
It moves in shape, not just in level. Illustrative chart - not real market data.

What that does to holdings depends on duration. A bond with a duration of 2 falls roughly 2% for a one-point rise; one with a duration of 17 falls roughly 17%. Same event, an eightfold difference in damage, decided by a property of the bond rather than by the size of the move.

A falling price series with a stop level marked.
A long bond falls furthest when rates rise. Illustrative chart - not real market data.

This is why “bonds are safe” is an incomplete sentence. A short-dated government bond is close to what people mean by safe. A thirty-year one moves like an equity on an announcement day.

Why the shape carries information

An upward slope has two possible explanations and they are not the same. One is that the market expects short rates to rise, so long rates average in those higher future levels. The other is that lenders simply demand extra compensation for tying money up — the term premium — regardless of expectations.

Both produce the same picture, and they imply different things. If the slope is expectations, it is a forecast you can disagree with. If it is term premium, it is a price for a service and carries no view at all. Nobody can decompose the two from the curve alone, which is why confident readings of “what the curve is telling us” are usually reading one interpretation into a shape that supports both.

Inversion is the case where the ambiguity narrows. A term premium explanation struggles to produce an inverted curve, since it would mean lenders demanding less to wait longer. So an inversion is closer to a genuine expectation — that rates will be lower later, usually because something has slowed.

It still says nothing about when. The shape is a statement about direction and level, with no timestamp anywhere in it.

The original data

Use this site’s own measurement of what an annual percentage does over a long horizon. Over thirty years, an annual drag of 5 basis points costs 1.5% of the final balance, 20 basis points costs 5.8%, 75 costs 20.2%, and 150 costs 36.5%.

Read those as yield differences rather than fees. The gap between a bond yielding 4.00% and one yielding 4.20% is 20 basis points — worth 5.8% of the final amount across thirty years. The curve’s half-point gaps are not rounding.

A candlestick chart annotated with the cost of a round trip.
And every trade on it still costs a round trip. Illustrative chart - not real market data.

And every position taken on a curve view costs a round trip — 0.0098 on this site’s shared series, about 2% of the median bar range of 0.493. A curve trade held briefly pays that on a difference measured in fractions of a percent.

A price series with heavy volume on one day.
Auction days are volume days. Illustrative chart - not real market data.

When it fails

The characteristic failure is treating an inversion as a dated prediction. Inversions have preceded recessions often enough to be famous, and the lag has ranged from months to years — which makes the signal real and useless for timing. A position taken on the inversion itself carries an unknown holding period, pays costs throughout, and can be correct about the eventual outcome while being ruinous about the path.

A candlestick series with a step change at an announcement.
A decision arrives as a step, not a slope. Illustrative chart - not real market data.

A second failure is reading one maturity as the curve. The ten-year moving is not the curve moving, and the difference between a shift and a twist decides what happens to what you hold.

A third is assuming the central bank controls the long end. It sets the short rate; the rest is expectation, and the two can disagree loudly.

A fourth is ignoring inflation. A 5% yield with 3% inflation is about 2% of purchasing power, which inflation covers in full.

A declining series cut short at a decision point.
The curve just inverted. What does that price in? Illustrative chart - not real market data.

And a fifth is trading the shape without knowing your duration. The curve is the map; duration is how far your particular holding moves on it.

Interest rate covers what a single point on the curve actually is. Bond duration covers how far a holding moves when the curve shifts. And inflation covers the difference between a nominal yield and a real one.

What I actually do

The thing that took me longest to grasp is that ‘rates went up’ is almost never a statement about the curve. Short rates can rise while long rates fall, which is a completely different event from both rising together — and the two have opposite implications for anything you hold.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.