TechniqueFrameworkMacro and ratesMarkets
A yield is expected short rates plus a term premium, and the premium is modelled, not seen
Treasury Term Premia: 1961-Present · 2014
The idea
When markets say the term premium rose, they usually mean the New York Fed's ACM series, and knowing how that is built separates a prepared candidate. A long yield is decomposed into the average short rate expected over its life plus a premium for the risk that short rates turn out differently. Neither part is observed. ACM estimates them with a five-factor model fitted to yields alone back to 1961; survey-based measures subtract forecasters' expected short rate instead. The two disagree, and a third model sits roughly halfway between, which tells you how much the answer depends on the method.
Term premium is an estimate from a model; say which model, and treat the level as less certain than the direction.
When it works
Any rates conversation that invokes the term premium, and any question about why long yields moved while policy did not.
When it may fail
Yields-only models are sensitive to their sample period, as the 2023 note shows, and survey measures exist only twice a year. There is no measurement, only estimates that disagree.
How to use it
Quote the term premium as a range across ACM and a survey-based estimate rather than one number, and say which direction both agree on. That is the honest version of the claim.
Where it came from
Research Treasury Term Premia: 1961-Present · Tobias Adrian, Richard K. Crump, Benjamin Mills and Emanuel Moench, Federal Reserve Bank of New York (Liberty Street Economics)
What the source says. ACM is a five-factor, no-arbitrage affine model in which yields are linear functions of pricing factors, estimated daily back to June 1961, and Kim-Wright closely matches an average of ACM and a survey-based estimate.
Connected ideas
Takes it further
Split a move in the ten-year into expected policy and term premium with one rule of thumbA steepening is mostly term premium; a move in the two-year is half expectations, half premium. Use the rule, then say which model you trust.
Takes it further
The near end of the curve carries the recession signal, not the long endAn inverted curve is the market expecting cuts; read the near-term forward spread, because the long spread is diluted by the term premium.
Says something similar
Mortgage bonds shorten exactly when you least want them toA prepayable bond gets shorter when rates fall and longer when they rise; the spread you are paid is for that asymmetry.
Concepts
Put it to work on L3VLUP
More like this
- FrameworkCrises and market historyBook
A bubble needs marketability, money and credit, and speculation together
Check all three legs, marketability, cheap money and a population buying for resale, before calling something a bubble.
From Boom and Bust · William Quinn and John D. Turner
- FrameworkCrises and market historyBook
Every bubble runs displacement, boom, euphoria, distress, revulsion
Ask where a market is in the sequence rather than whether it is a bubble; the stages are more recognisable than the verdict.
From Manias, Panics, and Crashes · Charles P. Kindleberger
- Company insightMarketsBook
Physical traders make money on logistics and access, not on a price view
In physical commodities the money is in moving, financing and accessing the cargo; the price direction is usually hedged.
From The World for Sale · Javier Blas and Jack Farchy
- TechniqueMarketsBook
A commodity index does not return what the spot price did
Returns on a rolled futures position include the roll, so the shape of the curve matters as much as where spot goes.
From The Little Book of Commodity Investing · John Stephenson