What 2s10s Actually Measures
The number, and what it is made of
The 2s10s is a subtraction. Take the yield on the 10yr Treasury and subtract the yield on the 2yr Treasury. Both are par yields — the rate at which a new bond of that maturity would sell at face value — so the answer is in basis points, where 1 basis point is one hundredth of a percentage point.
On 21 September 2026 the 10yr closed at 4.96% and the 2yr at 4.76%. The spread is therefore 20 basis points. That is the whole calculation. There is no model in it and nothing is forecast by it.
What makes it worth watching is not the number. It is that you have subtracted two yields that are set by two different things.
Two yields, two different questions
The 2yr is short enough that its yield is mostly an average of what the market expects the Federal Reserve's policy rate to be over the next 2 years, plus a small premium. It barely cares about the deficit. It cares about what the central bank does next.
The 10yr is long enough that expectations over 10 years stop doing most of the work. The extra compensation investors demand for holding a long bond rather than rolling short ones — the term premium — becomes a large part of the yield, and that premium is where fiscal risk, inflation uncertainty and supply all get priced.
So the spread is not one opinion. It is the gap between the market's view of the policy path and the market's view of the long run. When the two disagree, the spread moves, and which one moved tells you what the market is actually saying.
Four ways the spread can move, and only one question that matters
Yields can rise or fall, and the two ends can move together or separately. Four combinations follow, and they mean opposite things.
Bear steepener — long yields rise faster than short. Typically fiscal or inflation risk: investors demanding more to hold duration. Bull flattener — short yields fall faster than long. Typically an easing cycle arriving. Bull steepener — short yields fall faster, but the long end refuses to follow, which usually means the market thinks cuts will be met with more supply than growth. Bear flattener — short yields rise faster than long, which is a policy tightening being priced at the front with the long end unconvinced.
The question that sorts them is always the same: which end moved? A spread that narrows because the front end climbed is a policy story. A spread that narrows because the long end fell is a growth or a risk story. Same number, different world.
A worked example from our own board
This is not hypothetical. On 4 September 2026 the spread was 41 basis points: 10yr 4.78%, 2yr 4.37%. By 21 September it was 20 basis points on a 10yr of 4.96% and a 2yr of 4.76%.
The spread narrowed 21 basis points in under 3 weeks, and almost two-thirds of it came from the front. The 2yr rose 39 basis points; the 10yr rose 18. That is a bear flattener: the policy path was repriced upward while the long end moved less.
Read it as a forecast and you would say the bond market is warning about growth. Read it as what it is — the front end repricing the path, the long end pricing a term premium — and you get the opposite conclusion about which end carries the risk. The spread alone cannot tell you which reading is right. The decomposition can.
Inversion: what it predicts and what it does not
When the 2yr yields more than the 10yr the spread is negative — an inverted curve — and this has preceded every US recession in the post-war sample, typically by six to 24 months. That record is real and it is why the number is watched.
It is also weaker than the folklore. The inversion does not cause anything; it records that the market expects policy to be tight now and easier later, and that expectation has usually been followed by a slowdown. The lead time is not a timetable: the 2022 to 2024 inversion ran longer than any in the sample without the recession the model points to arriving on the historical clock. A signal that has been early and is still waiting is not yet wrong, but it is not a forecast you can lever.
The honest use is this: a sustained inversion tells you the market thinks policy is restrictive relative to the long run. It does not tell you when, and it does not tell you where the growth comes from.
The two errors
The first is treating the spread as a single fact rather than a difference of two. Whenever you see 2s10s quoted, the only useful follow-up is: which end moved? Without that, the number has no position attached to it.
The second is confusing it with the funding market. The 2s10s says nothing about whether cash is scarce or abundant — that is the business of the overnight complex, the spread between the policy rate and the administered rate, and the balances that sit underneath them. A curve can flatten smoothly while the plumbing underneath it is going dry. Those are two different pages of the same book.