Why long bonds pay more — two very different answers

Ten-year Treasury yields have climbed toward 4.1 per cent, and fixed-income desks are arguing about why. Everyone agrees that long-dated yields are rising. Quantitative analysts and macro economists offer explanations that cannot both be right.

A busy fortnight for bonds

The move has had plenty of triggers. On 28 July the Bank of Japan loosened its cap on Japanese ten-year yields, removing an anchor that had held down global long rates. On 1 August Fitch cut America's credit rating to AA+. The next day the Treasury raised the size of its quarterly coupon auctions for the first time since 2021 and signalled more increases to come. Meanwhile American growth data have kept surprising on the upside, and July's consumer price figures showed headline inflation at 3.2 per cent, with core inflation at 4.7 per cent.

Each of these events can be read two ways. Strong data and sticky core inflation support the view that the Federal Reserve will keep rates high. A bigger refunding and a downgrade support the view that investors want more pay for holding government debt. The yield itself cannot say which reading is correct, because it bundles both answers into one number.

Expected rates or term premium

The macro economists say long yields are rising because investors expect policy rates to stay higher for good, the 'higher for longer' view built on resilient growth. Term-structure models disagree. They show the expected path of rates barely changing, and put the whole move down to a rising term premium, the extra return investors want for holding long-dated debt and for bearing the risk of heavy government issuance. The split matters in practice. An investor who blames rate expectations buys the dip; one who blames the term premium sells the long end.

Why the models disagree

A ten-year yield can be split into the average short-term rate investors expect over the next decade and a premium on top. Neither piece is observed directly, so both must be estimated, and the estimate depends on how the model is built. The most widely cited decompositions, the New York Fed's Adrian-Crump-Moench model and the Fed Board's Kim-Wright model, infer expectations from the joint behaviour of yields across maturities.

The technical difficulty is persistence. Interest rates move in long, slow cycles, and a few decades of data contain only a handful of them. Statistical models fitted to short samples tend to conclude that rates revert to their average faster than they really do. If a model believes rates snap back quickly, a rise in long yields cannot be explained by expectations, since those are assumed to fade, and the model assigns the move to the term premium almost by default. Researchers have documented this small-sample bias for years, and corrected versions typically attribute more of any move to expectations.

The Kim-Wright model tries to discipline its estimates with surveys of professional forecasters' rate expectations, which helps anchor the expected path. Surveys have their own weakness: forecasters are slow to revise, and their answers tend to cluster around the central bank's own guidance. A model tied to surveys will therefore tend to show expectations as stable, and it too may push the residual into the premium.

What the past says

This argument has been had before. In 2005 Alan Greenspan called it a conundrum that long yields fell while the Fed raised short rates; the models of the day blamed a shrinking term premium, driven by foreign central banks buying Treasuries. In the 2013 taper tantrum, the ten-year yield jumped after the Fed hinted at slowing its bond purchases, and most decompositions attributed the spike to the term premium. In both cases the model reading was broadly accepted, because there was an identifiable buyer or seller of duration to explain it.

This time there are candidates too. The Fed is running down its bond holdings, the Treasury is selling more long debt, and Japanese investors have more reason to keep money at home. For much of the past several years the New York Fed's estimate of the ten-year term premium was below zero, an anomaly that reflected central-bank buying and a world starved of safe assets. A return to positive territory would be normalisation as much as alarm.

The case for both

The strongest objection to the either-or framing is that the two explanations feed each other. If investors come to believe the neutral rate of interest has risen, they will expect higher policy rates and also become less certain about where rates will settle, which raises the premium for bearing duration risk. Heavy issuance, in turn, keeps financial conditions from easing as fast as the Fed might like, which argues for keeping policy tighter for longer. A pure split between the two is a modelling convenience, and real-world moves are usually a blend.

Even so, the proportions matter for positioning. Inflation-protected Treasuries offer a partial check: when real yields rise while inflation compensation stays steady, the move is about the price of money over time rather than fear of inflation. Over recent weeks much of the increase has shown up in real yields, which is consistent with either higher expected real rates or a higher real term premium, and does not settle the dispute.

What would settle it

Three pieces of evidence would help. The first is Fed communication, including Jerome Powell's speech at the Jackson Hole symposium later this month: if officials push back against rate cuts in next year's pricing and long yields barely move, the expectations story weakens. The second is auction results, especially whether long-dated sales clear only with generous concessions to dealers, which would point to supply. The third is the next refunding announcement in early November, which will show whether the Treasury intends to keep lengthening the stack of long debt the market must absorb.

Whether the rise comes from rate expectations or from term premium will decide how investors allocate money for the next decade, and getting it wrong either way will be expensive. On the current evidence the term-premium camp has the stronger hand for the rest of this year, because supply is rising on a published schedule, but the investors most exposed are those holding long bonds on the assumption that the move is purely about the Fed and will reverse once it stops hiking.