How it actually works
The US Treasury borrows across a range of maturities, and each maturity has its own yield. Plot them all and you get the yield curve. Nobody wants to discuss an entire curve in a morning note, so traders talk about the gap between two points on it, and the two points that won the popularity contest are the two-year note and the ten-year note. Subtract the former from the latter and you have the 2s10s, expressed in basis points, a basis point being one hundredth of a percentage point.
Round numbers, invented for illustration: if the two-year yields 4.00 per cent and the ten-year 4.50 per cent, the 2s10s is +50bp. If the ten-year fell to 3.80 per cent while the two-year stayed put, the spread would be −20bp and the curve would be inverted.
The two ends measure different things, which is the whole point of comparing them. The two-year yield is mostly a bet on what the Federal Reserve will do with overnight rates over the next couple of years. The ten-year is that same expectation stretched further out, plus a : the extra compensation investors demand for lending for a decade to a borrower whose fiscal path, inflation record and future auction sizes they cannot see. So the 2s10s is really a comparison between what the market thinks will do and what the market wants to be paid for not knowing.
What people get wrong
The same in the spread can mean opposite things. There are four flavours. Bull steepening: short yields fall faster than long ones, usually because the Fed is cutting or about to. Bear steepening: long yields rise while short ones sit still, which is a term-premium event and generally a vote of no confidence in the long-run fiscal picture. Bull flattening: long yields fall, typically a growth scare or a flight to safety. Bear flattening: short yields rise, which is the market pricing hikes. A number on its own tells you nothing. You have to know which end moved.
Inversion — a negative 2s10s — has an unusually good record as a recession signal in the post-war US, because it is the market saying rates are high enough now that they will have to come down a lot later. The lag is long and variable, and the signal has been wrong at least once. Treat it as a temperature reading rather than a diagnosis.
Why it matters to this crash
In this cycle the curve is not being set by the market. It is being set by two arms of the state pulling in opposite directions, which makes the shape almost uninterpretable by the usual rules.
We wrote about this on 19 August. The Fed held at 3.5–3.75 per cent on a 9–3 vote, with three members dissenting towards higher rates after five straight years of inflation above target — pressure on the short end. The same day, Treasury doubled long-end buyback operations from $2bn to at least $4bn to support the 10-to-30-year sector after the 30-year yield hit 5.34 per cent, its highest since 2007 — pressure on , in the other direction. The 2s10s was 52bp that day.
A curve flattened by hawkish dissent at one end and official buying at the other is not telling you anything about growth. It is telling you about . And it is doing so at exactly the moment the labour market started deteriorating: payrolls fell 23,000 in July against expectations of roughly 80,000 added. The usual signal has been disconnected from its wiring, which is a problem, because a great many people size their risk off it.
What would make this dangerous
The move to watch is not the level but the composition. Sharp bear steepening — the ten- and thirty-year selling off while the two-year is anchored or falling because the Fed is cutting into a weak labour market — is the signature of a term-premium shock rather than a growth story, and the single clearest sign that buyers have started charging the United States for fiscal risk. That would show up in internals before it showed up in the spread: at the long end, a rising , buyback operations getting bigger again without the 30-year yield coming down.
Re-inversion is the other danger, and a different one. It would mean the market has concluded the hawkish dissenters will win the argument and hike into a labour market that is already shedding jobs.
Either way, speed matters more than direction. A curve that moves 40 or 50 basis points in a week forces mechanical position-cutting in every that is long the curve with borrowed money, and there is a great deal of that about.