Reading the yield curve without guessing

The yield curve is one of the most watched recession signals in markets, and one of the most misquoted. People repeat a headline value they half-remember, or state that the curve “is inverted” without saying which spread, on which day, from which source. This post is about reading it properly: what the spread measures, where the live number lives, and how much weight the history actually supports.

We will not print a current value here. A number stated in a blog post is stale the moment it is published. The point is to send you to the live source and teach you to read it.

What the spread is

The signal is a difference between two Treasury yields of different maturities. Two versions dominate.

• The 10-year minus 2-year spread, published by the St. Louis Fed as T10Y2Y. It is the 10-Year Treasury Constant Maturity minus the 2-Year Treasury Constant Maturity, quoted in percent, updated daily, with the underlying yields sourced from the U.S. Treasury.
• The 10-year minus 3-month spread, published as T10Y3M, same construction with a 3-month bill at the short end.

When the number is positive, longer money yields more than shorter money, the normal state. When it goes negative, the curve is “inverted”: short rates sit above long rates. That is the configuration that draws attention, because it has tended to appear ahead of recessions.

To read the live value, open either FRED series page. The current figure, the full history, and a chart are all there, updated daily. That is the number to quote, with its date, not one from memory.

Why an inversion carries information

The short end of the curve tracks the current policy rate closely. The long end embeds the market’s expectation of average rates over the coming decade, which in turn reflects expected growth and inflation. When the long end falls below the short end, the market is effectively saying it expects policy to be cut from here, which is what happens when growth is expected to weaken.

The idea has a specific origin. In The Yield Curve as a Predictor of U.S. Recessions (Estrella and Mishkin, Federal Reserve Bank of New York, 1996), the authors report that the spread between the ten-year note and the three-month bill “is a valuable forecasting tool,” one that “significantly outperforms other financial and macroeconomic indicators in predicting recessions two to six quarters ahead.” Note the horizon they claim: two to six quarters, not next month. The signal is early and coarse by design.

The Fed operationalizes this with a probit model that converts the term spread into a recession probability. The New York Fed maintains it as a standing indicator: its yield-curve model “uses the slope of the yield curve, or the term spread between long- and short-term interest rates, to calculate the probability of a recession in the United States twelve months ahead,” and it favors the 10-year minus 3-month version, because the 3-month bill hews closely to the current policy rate.

What the history actually supports, and what it does not

This is where careful framing matters, because the popular version of the claim is stronger than the evidence.

A Federal Reserve FEDS note on predicting recession probabilities from the yield curve states that the 10-year minus 3-month term spread “narrowed prior to each of the six most recent recessions,” and that an inversion “preceded five of those” downturns. That is a strong track record. It is not the same as “the curve inverts before every recession, every time,” and the difference is the whole discipline of reading the signal honestly.

The same note is candid about the limits:

• The lead time is variable. An inversion has preceded recessions by anywhere from months to more than a year, so it dates the risk poorly even when it flags it.
• The model can miss. The authors note their basic specification “fell close to zero in early 2008,” on the eve of a severe recession, because the signal lives entirely in the yield curve and ignores everything else.
• The curve can flatten for benign reasons. A falling term premium, rather than an expectation of rate cuts, can compress the spread without carrying the same recessionary meaning.

The St. Louis Fed has made the same point in plain language: different spreads and different model specifications produce different recession probabilities from the same yield-curve data, so the honest answer is a range, not a single certainty.

How to actually read it

Put together, a disciplined reading looks like this.

1. Name the spread. Say “10y minus 2y” or “10y minus 3m,” because they do not always agree, and a claim about “the yield curve” without one of these is unfalsifiable.
2. Quote the live value with its date, from the FRED series page, not a remembered figure.
3. Treat an inversion as elevated risk with an uncertain lead time, not a dated forecast. On the Fed’s own framing the useful horizon is measured in quarters, not weeks. It has preceded most recent recessions; it has also given false or early signals, and it has missed at least once.
4. Cross-check rather than lean on one number. The Fed’s own work pairs the spread with other indicators, and reports a probability rather than a verdict, precisely because the spread alone can mislead.

Read this way, the yield curve is a genuinely useful gauge and a poor oracle. That is the correct amount of weight to give it.

Next time someone tells you the curve inverted, ask which spread, on what date, and from what source. If they cannot answer, they read a headline. If they can, they read the FRED page, which is where you should be reading it too.