For most of the decade after the financial crisis, the term premium was close to zero and occasionally negative. Investors were accepting no additional compensation — sometimes a penalty — for lending to the government for ten years rather than rolling over short-term bills.
That is no longer the case, and the change has consequences well beyond the bond market. Mortgage rates, corporate borrowing costs and equity valuations all take their cue from long-term yields. When the term premium moves, it moves them all.
What the number actually is
A ten-year government bond yield can be decomposed into two parts.
- Expectations. The average short-term interest rate investors expect to prevail over the next ten years. If you could roll a series of one-year bills instead, this is what you would expect to earn.
- The term premium. Everything else — the extra yield demanded for committing capital for a decade rather than staying short.
The critical point is that the term premium is not observed. It is inferred. Nobody publishes a market price for it; it is what remains after a model estimates the expectations component and subtracts it from the actual yield. Different models produce different estimates, and they disagree most precisely when the number matters most.
This is worth holding onto whenever a commentator states the term premium to two decimal places. They are quoting a model output, not a market quote.
Why it should normally be positive
Lending for ten years carries risks that lending for three months does not.
The obvious one is inflation. A fixed coupon paid in 2036 is worth whatever 2036 prices allow it to buy, and nobody knows what that is. The second is interest rate risk: if yields rise after you buy, the market value of your bond falls, and the longer the maturity the more violently it falls. A ten-year bond loses roughly eight or nine times as much value per unit of yield increase as a one-year bill.
Rational investors demand compensation for bearing both. A positive term premium is the normal state of affairs. The two decades in which it was not are the anomaly requiring explanation.
Why it vanished
Three forces compressed it, and all three have weakened.
Central bank purchases. Quantitative easing programmes bought long-dated government bonds in enormous size. The stated mechanism was precisely this: remove duration from the market, force private investors into other assets, and compress the premium. It worked. As balance sheets have run down, that compression has reversed.
Low and stable inflation. When inflation has been near target for twenty years, the risk of holding a fixed coupon for a decade feels theoretical. Once inflation has been above target for five, it does not.
The bond–equity correlation. This is the least discussed and arguably the most important. For most of the 2000s and 2010s, bonds rose when equities fell — a bond portfolio was insurance against a bad equity outcome, and insurance is worth paying for. That relationship depends on the shock being a growth shock. In an inflation shock, bonds and equities fall together, and the hedge stops working. An asset that no longer diversifies your portfolio has to compensate you in yield instead.
Supply matters more than it used to
Textbook treatments often assume government bond supply has little effect on yields, because the market is deep and buyers are plentiful. That assumption held better when a price-insensitive buyer — the central bank — was absorbing a large share of issuance.
Without that buyer, the marginal purchaser of long-dated debt is a pension fund, an insurer or a foreign reserve manager, all of whom are price-sensitive and have alternatives. Heavy issuance concentrated at long maturities then has to clear at a higher yield. This is why quarterly refunding announcements — which specify the maturity mix of upcoming issuance, not just the total — now move markets in a way they did not a decade ago.
What it means in practice
Three consequences follow.
- Central bank control over long rates is weaker than assumed. A policy rate cut lowers the expectations component. It does nothing directly to the term premium, and can raise it if the cut is read as tolerating higher inflation. Long-term borrowing costs can rise while the central bank is easing.
- Long-duration equities are exposed twice. Companies whose value sits in distant cash flows are discounted at a rate that includes the term premium. A rising premium compresses their valuations independently of anything happening to their business.
- The yield curve becomes harder to read. A steepening curve driven by rising growth expectations means something quite different from one driven by a rising term premium. The shape looks identical.
That last point matters for anyone using the curve as a recession indicator, and it is a caveat we set out at greater length in what an inverted yield curve actually signals. The signal was calibrated in a period when the term premium was stable. It is not stable now.
