One number, published every afternoon by the US Treasury, that tells you what the bond market thinks about the next few years. It is the most-quoted recession indicator in finance and also the most consistently misread — because almost everybody watches the wrong half of it.
The US government borrows money for different lengths of time and pays a different interest rate for each. The 2-year yield is what it pays to borrow for two years; the 10-year yield is what it pays to borrow for ten. Subtract one from the other and you get the spread — usually written 2s10s.
Curve spread = 10-year yield − 2-year yield
Normally the 10-year pays more. Lending money for a decade is riskier than lending it for two years, so you demand extra for the privilege. That is a positive or normal curve, and it is the resting state of a healthy economy.
When the number goes negative, something strange has happened: the market is being paid more to lend for two years than for ten. That only makes sense if investors expect interest rates to be materially lower in the future than they are now — which is another way of saying they expect the economy to weaken enough that the Federal Reserve has to cut. This is an inverted curve.
The reason the spread carries information is that its two halves are driven by completely different things.
| Yield | Mostly reflects | Moves when |
|---|---|---|
| 2-year | Where the market thinks the Fed's policy rate will be over the next couple of years | Rate expectations shift — a hot inflation print, a dovish speech, a weak jobs report |
| 10-year | Long-run growth and inflation, plus the extra compensation demanded for locking money up (term premium) | The structural outlook changes — deficits, supply of new bonds, long-term inflation views |
So the spread is really a comparison between what the Fed is expected to do soon and what the economy is expected to look like over a decade. When those two disagree sharply, the curve moves.
The headline version of this indicator is: the curve inverts, a recession follows. That is true in the loosest sense — every US recession since the 1960s has been preceded by an inversion — but it is close to useless as a timing tool, because the gap between the inversion and the recession has ranged from about six months to well over two years.
The more useful observation is about where in the sequence the damage actually shows up. Look at what happens around each of the last several recessions and the pattern is consistent:
Here is the actual Treasury data through the last downturn. Nothing is smoothed or selected; these are the published 2-year and 10-year par yields on the first business day of each month.
| Date | 2Y | 10Y | Spread | What the curve was doing |
|---|---|---|---|---|
| Aug 2019 | 1.50 | 1.50 | 0.00 | Inverts. The warning arrives. |
| Jan 2020 | 1.53 | 1.80 | +0.27 | Back above zero. Markets at record highs. |
| Mar 2020 | 0.59 | 0.92 | +0.33 | Widening as yields collapse. The crash. |
| Apr 2020 | 0.23 | 0.62 | +0.39 | Still widening. Recession under way. |
| Jun 2020 | 0.19 | 0.82 | +0.63 | Widening as yields rise. Recovery. |
| Dec 2020 | 0.16 | 0.97 | +0.81 | Same pattern. Rally continues. |
The spread widened the whole way through. Read on its own it would have told you the same thing in March 2020 as in December 2020, which is worse than useless — those were the two extremes of the year. What separates them is the 10-year: collapsing in the crash, climbing in the recovery.
This is also why a single reading of the spread tells you less than the reading plus its direction. Which brings us to the four ways the curve can move.
Two questions, and the answers combine into four states. First: is the gap widening (steepening) or narrowing (flattening)? Second: are yields overall rising (bear, because rising yields mean falling bond prices) or falling (bull)?
| State | Mechanics | What drives it | What it means |
|---|---|---|---|
| Bull steepener | 2-year falls faster than the 10-year, so the gap widens | The central bank cutting hard into a slowing economy | Late-cycle stress turning into recession pricing. Historically the state the economy is in when a downturn actually begins. |
| Bull flattener | 10-year falls faster than the 2-year, so the gap narrows | Money rushing into long bonds on a growth scare, while short rates are still held high | The bond market signalling a sharp slowdown before any cuts have arrived. |
| Bear steepener | 10-year rises faster than the 2-year, so the gap widens | Heavy government bond supply, worsening deficits, or investors demanding more compensation to lend long | Lenders wanting more to hold long-dated debt. Raises the cost of capital across the board with no growth to justify it. |
| Bear flattener | 2-year rises faster than the 10-year, so the gap narrows or inverts | The central bank hiking aggressively to kill inflation | Tight policy squeezing the system. This is the move that walks a curve into inversion. |
Notice that steepening on its own tells you nothing at all. Three of the states above involve a widening gap, and they mean completely different things: the Fed cutting into trouble, growth being priced in, and the bond market losing patience with government borrowing. Track only the spread and all three look identical.
The homepage card checks all three legs: the gap, the 10-year, and the 2-year.
The card at the top of the home page pulls the official 2-year and 10-year par yields straight from the US Treasury's daily publication, works out the spread and its 20-trading-day change, and colours itself accordingly.
| Colour | Card reads | Shown when |
|---|---|---|
| Green | Stable | Positive curve that is not making a meaningful move |
| Amber | Bear steepening | Gap widening as yields rise — the long end being sold |
| Amber | Bull steepening | Gap widening as yields fall — cuts being priced |
| Amber | Bear flattening | Gap narrowing as yields rise — tightening |
| Amber | Bull flattening | Gap narrowing as yields fall — growth scare |
| Amber | Nearly flat | Spread under 0.25 points |
| Red | Inverted | Spread below zero |
| Red | Re-steepening | Back above zero on falling yields, after inverting within the year |
One honest limitation: the Treasury publishes once a day, after the close. This is a daily indicator and nothing more. It is not designed to be traded intraday, and treating it as though it were is a good way to manufacture noise out of a signal that only speaks in months.
The same logic as an indicator you can put on a chart. It plots the spread, shades it green above zero and red below, marks the four movement states in a corner table, and fires two alerts:
Inputs let you swap the two symbols (the defaults are TVC:US10Y and TVC:US02Y), change the lookback used to judge direction, fade the corner read-out so the chart shows through, and switch between percent and basis points.
A second script, for the price chart rather than its own pane. It shades red wherever the curve was inverted and grey over each official NBER recession, labelled with its length — 18 months for 2008, 16 for 1973–75 and again for 1981–82, 2 for COVID.
Put it on the S&P and the sequence described above stops being an assertion and becomes something you can see: the red arrives first, and the grey opens as the red closes.
Both scripts are published invite-only while they are finalised. Access will open here first.
On its own the curve is a slow, blunt instrument — it has been early, it has been early for a very long time, and it has spent stretches looking broken. It earns its place as one input among several rather than as a trigger. Read it alongside the leading indicator hierarchy, where housing permits and ISM new orders turn first, and the monthly regime score, which folds the curve in with everything else rather than letting a single series drive the call.
Primary source: US Department of the Treasury — Daily Treasury Par Yield Curve Rates. Longer history: FRED series T10Y2Y.
For educational purposes only. This explains how one indicator is constructed and read; it is not investment advice, and the curve has been wrong on timing often enough that nobody should trade it alone.
