How it works
A twenty-year bond and a ten-year bond are, in principle, interchangeable if you are willing to roll. Buy the twenty-year and hold it. Or buy the ten-year, get your money back in a decade, and reinvest it in whatever ten-year yield exists then. For both to be fairly priced today, the market must be implying some particular reinvestment rate for that second leg. That implied rate is the 10y10y forward.
The arithmetic is a compounding identity: one plus the twenty-year yield, raised to the twentieth power, equals one plus the ten-year yield to the tenth, times one plus the forward to the tenth. Solve for the forward. A serviceable rule of thumb is that the 10y10y is roughly twice the twenty-year yield minus the ten-year. Purely as illustration: with a ten-year at 4.0% and a twenty-year at 4.5%, the implied 10y10y sits near 5.0%. The steeper the curve out there, the higher the forward, and the relationship is geared — a small widening between the two spot yields produces a large in the forward.
What makes the number useful is what it excludes. Rate cuts, a recession, a tightening cycle, the entire tenure of whoever runs the central bank next: all of it falls inside the first ten years and therefore washes out of the forward. What is left is the market's price on two things it cannot observe. First, where the neutral short rate plus inflation is expected to settle in the long run. Second, the — the extra yield demanded simply for the discomfort of owning you cannot hedge cheaply.
This is a price, not a forecast. Bond markets have a long and undistinguished record of predicting the level of interest rates a decade out. What the forward reliably tells you is what investors are currently charging, which is a different and for our purposes more interesting question.
Why it matters to this crash
The diagnostic value shows up when the forward moves in the wrong direction relative to the news. On 19 August 2026 we wrote up two bad auctions and a global : $25bn of 30-year Treasuries stopping at 5.216%, the highest at auction since 2001, and the 30-year TIPS yield at 3.09%, the highest since 2008 — real yields, not an inflation scare. In the same dispatch we noted Robin Brooks pointing out that 10y10y forwards had risen almost everywhere that week despite soft US inflation data: France up 14bp, the UK up 13bp, Japan up 11bp.
That pattern is the whole reason the metric earns its place. Soft inflation data should pull down expected short rates and therefore pull down forwards. Forwards went up instead, in several countries at once, with no common growth story. is risk premium: investors repricing what it costs to lend to governments for thirty years, and doing it in an orderly way, with nobody forcing them. Japan's ten-year at 2.93% was the highest since 1996, with Deutsche Bank's Shoki Omori noting that 3% is "a critical defence line for fiscal credibility" because it is the rate the government's own budget assumes (FT).
Every long-dated valuation in our coverage discounts against something like this rate. Terminal values in the complex, , thirty-year corporate paper issued to build data centres into the same limited pool of duration buyers that governments are already competing for. A permanently higher 10y10y is not a headline event. It is a quiet upward shift in the denominator of nearly everything.
What would make this dangerous
The specific signature to watch for is 10y10y forwards rising while near-term inflation prints soften. That combination cannot be explained as a cycle call and leaves one interpretation standing: investors demanding more to hold fiscal risk. We saw exactly that in August 2026.
Second, simultaneity without a common shock. France, the UK and Japan moving together in the same week is not three domestic fiscal stories that happened to coincide. It suggests a single global bid for long duration that is thinning out everywhere at once.
Third, forwards grinding higher while rate volatility grinds lower. In the same 19 August dispatch our rate-volatility gauge was down 16% over five days and 31% over twenty, with yields at multi-decade highs. Cheap volatility alongside a repricing long end means nobody is paying to hedge the thing that is actually moving, which is how an orderly repricing becomes a disorderly one when somebody finally has to.
Fourth, a level breach with a named consequence attached — Japan's ten-year through the 3% its own budget assumes being the clearest currently on the board.