How I read the yield curve without treating it as a timing signal

The yield curve compares interest rates across maturities. Its shape can add context to expectations for growth, inflation and central-bank policy, but it is not a standalone timing signal.

I check it most days because it is a cheap summary of what the bond market currently expects. It is also one of the most over-read charts in finance, so the second half of this piece is about what it does not tell you.

What the curve is

Plot the yield on government bonds against how long they have left to run and you have a yield curve. The most watched version is the US Treasury curve, from one-month bills out to 30-year bonds.

Most of the time it slopes upward: lending money for ten years pays more than lending it for three months. The slope reflects a mix of expected future short-term rates and the extra compensation investors want for holding a longer bond, which economists call the term premium.

You can look at the whole curve, but most day-to-day discussion is about one number: the 10-year yield minus the 2-year yield, usually written 2s10s. The Federal Reserve Bank of St Louis publishes it daily as series T10Y2Y on FRED, free, with the full history back to 1976.

An upward-sloping curve

A curve that slopes upward is the usual condition. Short rates sit below long rates, which is consistent with an economy that is expected to keep growing and with a central bank that is not actively restraining it.

Banks are the clearest transmission point. A lender that funds itself short and lends long earns the gap, so a steeper curve widens that margin and a flatter one squeezes it. That is arithmetic about bank funding, not a recommendation about bank shares.

A flattening curve

The curve flattens when short-term yields rise towards long-term yields, which typically happens while a central bank is raising policy rates.

One common reading is that the bond market expects those rate rises to slow the economy, so it is unwilling to demand much extra yield for lending further out. Another reading is that long-term inflation expectations have fallen, or that demand for long-dated bonds from pension funds and insurers is holding long yields down for reasons unrelated to growth. Flattening on its own does not separate those explanations.

An inverted curve

Inversion means short-term yields sit above long-term ones: 2s10s goes negative.

1 June 2021: upward sloping

The 10 year paid 1.62% against 0.16% on the 2 year. The 2s10s spread was +1.46.

3 July 2023: inverted

The 10 year paid 3.86% against 4.94% on the 2 year. The 2s10s spread was minus 1.08.

This is the shape with a research literature behind it. The Federal Reserve Bank of New York publishes a monthly recession probability derived from the spread between the 10-year and 3-month Treasury yields, following work by Arturo Estrella and Frederic Mishkin published in 1996. The New York Fed's own yield curve FAQ explains the model and its caveats.

The record repays reading in dates. The slogan version leaves out the lags, and the lags are most of the story. US recession start and end dates below are the ones set by the NBER's business cycle dating committee.

• 2006 to 2007. 2s10s was negative for much of 2006. The NBER dates the following recession as beginning in December 2007, so the gap between signal and event was well over a year.
• 2019 to 2020. The spread inverted briefly in August 2019. The NBER dates the next recession from February 2020, but that downturn is generally attributed to the pandemic, not to whatever the curve was signalling six months earlier.
• 2022 to 2024. 2s10s went negative in mid-2022 and stayed negative into the second half of 2024, the longest continuous inversion in the FRED series. No NBER-dated recession began during it.

So the honest summary is narrower than the usual one. Inversion has preceded most post-war US recessions, with a variable and sometimes very long lag, and the most recent episode did not produce one at all. It is a piece of context with a decent record, not a countdown clock.

Two things that add to the picture

Real yields. Subtract expected inflation from the quoted nominal yield and you get the real yield. It is the more direct measure of how restrictive policy is, and it moves the discount rate applied to long-duration assets more directly than the nominal number does.

Term premium. This is the part of a long yield that is not explained by expected future short rates. Nobody observes it directly. It has to be modelled, and different models give different answers, so treat any single figure as one estimate. A rising term premium is usually discussed as investors asking for more compensation to hold duration, whether because of fiscal supply or because inflation looks less predictable.

What the curve does not do

It does not date anything. It does not tell you what to hold, and you will find no allocation instructions on this page, because the right answer depends on your circumstances, your time horizon and your tax position rather than on the shape of a line.

Nor is it independent evidence when you already believe something. The shape has several possible explanations at any moment, which is why it works better as one input beside employment data, survey measures such as the PMIs, credit spreads and what central banks are saying.

Where I learned to use it

Macro analysis of this kind, and the business-cycle and inter-market material around it, is the core of ITPM's Professional Trading Masterclass. My PTM review sets out what the course covers and who I think it suits.

The Macro Dashboard on this site tracks the curve and the other series mentioned here. It is free after an email signup, with no payment and no course purchase required.

Nothing on this page is financial advice or a recommendation to buy or sell anything.