Term premium repricing drives the widest 2s10s curve in four years

The Treasury curve has done something it has not done since the autumn of 2022. Between last Friday and this one, the ten-year yield climbed 34 basis points to 4.53 per cent while the two-year fell 9 basis points to 3.35 per cent. The spread between them closed at 118 basis points, the widest in four years, and it got there without a single data release that would justify it. Growth trackers are flat. The labour panels are soft. Nothing in the week's economic calendar argues for higher long yields, which is precisely why rates desks spent Friday afternoon rewriting their client notes.

Steepening curves usually carry a story, and the story depends on which end moves. A bull steepener, where the front end rallies harder than the back, says the market expects policy rates to fall. A bear steepener, where long yields do the moving, says something colder: investors want more compensation to hold duration and are prepared to keep selling until they get it. This week delivered both halves at once. That combination is rare, it is uncomfortable to hedge, and it tends to persist longer than desks expect because the two legs are driven by different buyers.

Where the move came from

The proximate trigger was Wednesday's policy decision, which ended balance sheet runoff and directed all reinvestment of maturing proceeds into Treasury bills rather than coupons. Read narrowly, that is friendly to the front end. The portfolio stops shrinking, reserve balances stabilise near $2.9tn, and the money market pressure that pushed secured funding above the floor rate in late August should ease. Read across the whole curve, the same decision removes a standing buyer from the coupon sector at the exact moment net issuance in the ten-year and thirty-year points is rising. Two audiences heard two different announcements.

Two-year yields fell on the decision and kept falling. Ten-year yields rose on it and kept rising. That divergence repeated on Thursday and again on Friday with no fresh catalyst, which is the tell that this is a flow story rather than a forecast story. Positioning surveys from Harrow Lane Capital had the buy side net long duration into the meeting, sized for the front end rally that arrived and the long end rally that did not.

The curve by tenor

A thirty-year yield of 5.21 per cent is not a crisis level by any historical standard. It is, however, 94 basis points higher than where the year began, against a policy rate that has come down 75 basis points over the same stretch. Those two facts sit awkwardly together in any framework that treats the long bond as an average of expected short rates plus a small constant. The pivot point of the curve has migrated out to roughly the five-year, and everything beyond it is now trading on supply and demand rather than on the policy path.

Clients keep asking what the long end knows about growth. The answer is nothing at all. It is a supply and compensation question, and the compensation has been too low for two years.

Ines Marchetti, head of rates strategy at Harrow Lane Capital

How much of this is compensation

Bellweather Macro publishes a ten-year term premium estimate that has swung 73 basis points since June. Its June reading was minus 12 basis points, meaning investors were accepting less yield than the expected path of short rates alone would justify, a condition that persisted for most of the past three years and quietly subsidised every levered duration trade in the market. Friday's reading was plus 61. The subsidy is gone.

If that estimate is close to right, about two thirds of the ten-year move since June is compensation and the remaining third is a genuine upgrade to the expected path of policy beyond 2028. The split matters more than the total. Term premium shocks and growth shocks move equities in opposite directions, so a portfolio that treats the ten-year as a single risk factor will misprice its own hedge ratio by a wide margin.

What to check before Monday

• Curve hedges structured for a bull steepener lose on both legs in a bear steepener, because the long leg moves against you while the equity book falls alongside it.
• Duration overlays sized on trailing volatility are now undersized, since ten-year realised volatility has risen from 4.8 to 7.1 annualised since June.
• Carry trades funded at the front end and invested at the long end have seen more than a year of expected roll down erased in four sessions.
• Levered mortgage basis positions face the same supply arithmetic, and the mortgage index has widened 16 basis points against Treasuries this month.
• Euro based buyers are the marginal bid for long coupons, and their hedged yield on the ten-year turned negative this week for the first time since January.

We stopped treating duration as the hedge for the equity book in July. The correlation flipped, it has not flipped back, and cash and gold are doing that job for us now.

Theodora Vance, portfolio manager at Kestrel Asset Management

The week ahead offers two clean tests. A thirty-year auction on Thursday will show whether real money will step in at 5.2 per cent, and the Marlowe Institute services nowcast on Wednesday will show whether the front end rally has any inflation cover. A weak nowcast steepens the curve further from the front. A soft auction steepens it from the back. Only a firm nowcast paired with strong auction demand flattens the thing, and almost nobody is positioned for that combination.