The Repricing Nobody Scheduled
Bond yields can move while the policy rate stands still. Separating the observed move from its possible causes changes the balance-sheet lesson.

Editor's note, September 20, 2026: Revised September 20, 2026: Rebuilt the explanation throughout, removing the unsupported claim that rising term premium alone caused the move. Distinguished yields from model estimates and corrected the discussion of bank accounting and liability duration.
A central bank can leave its overnight rate unchanged while the market reprices borrowing for the next thirty years. That distinction is the useful lesson of the spring yield movement. It does not, by itself, identify the cause of that movement.
The Treasury’s daily par-yield series records the 30-year yield at 4.91 percent on April 1 and 5.18 percent on May 19, 2026. It subsequently retraced. The Fed’s June 17 statement maintained the federal-funds target at 3.50–3.75 percent. These are observations about two different prices: the current overnight policy setting and a market yield extending far beyond the next meeting.
Our earlier account treated the long-yield move as a clean increase in term premium. That conclusion was stronger than the evidence. The corrected interpretation is narrower: a fixed policy rate does not fix the yield curve, and institutions must manage exposure to that curve even when the policy announcement appears uneventful.
Three layers of evidence
The first layer is the observed yield. It is a price-derived measure available on each trading day. Comparing two endpoints describes a change, but choosing the May high instead of the June close tells a different story about its persistence. A claim about a lasting repricing needs a stated window, not just a conspicuous peak.
The second layer is a model decomposition. A long yield can be understood as expected future short rates plus compensation for holding duration, with additional details depending on the model and instrument. The New York Fed’s ACM estimates are estimates, not separately traded prices. Its published series run through ten years; a ten-year estimate cannot simply be assigned to a thirty-year bond. Estimates may also change with the sample and model specification.
The third layer is interpretation. Issuance, inflation uncertainty, hedging demand and expectations for future policy may all affect yields. An unchanged current policy setting does not eliminate changing expectations for future policy. Nor does evidence of heavy issuance establish how many basis points it contributed. We have not identified those effects separately, so the article does not attribute the spring move to one of them.
The balance-sheet mechanism survives the correction
For a fixed-rate bond, a higher required yield generally means a lower price. Modified duration approximates the percentage price response to a small yield change; convexity refines that approximation. As an illustration, a bond with duration of 15 years would lose roughly 4.05 percent for a 27-basis-point yield increase before the convexity adjustment. That is our arithmetic, not the measured return of the thirty-year Treasury between the dates above. Coupons, the instrument chosen and changes across the curve matter.
Duration can exist at the two-year maturity too. The front end is particularly sensitive to expected policy, but describing its term premium as nonexistent would be incorrect. Likewise, a change in a Treasury discount rate does not mechanically produce the same percentage loss in a stock or private fund. Their cash flows and risk premia can change at the same time.
Assets must be compared with their funding
The Federal Reserve’s review of Silicon Valley Bank shows why asset duration, liquidity and runnable deposits belong in the same discussion. Holding a security to maturity does not eliminate its economic sensitivity to rates or guarantee that an institution can fund it until maturity.
Accounting classification matters, however. Unrealized fair-value changes on held-to-maturity securities do not automatically flow through reported tangible book value as though the securities were marked through equity. A sale under funding pressure can make an economic loss immediate. Our previous formulation blurred these distinct mechanisms.
For an insurer or pension plan, a higher discount rate can reduce the present value of liabilities as well as assets. Whether funded status improves or worsens depends on their relative duration, valuation rules and hedges. It is therefore wrong to assume that assets necessarily become cheaper while liabilities remain unchanged.
What to watch next
A useful review starts with three questions: which cash flows are fixed, which funding can leave or reprice, and which valuations move with which curve? That produces a testable exposure map. A general prediction of delayed losses across private markets does not.
Use the Treasury comparison tool to separate yield changes from changes in debt composition. The debt-interest scenario then isolates the arithmetic of refinancing a selected share at a different rate. Neither tool establishes why yields moved. They help show which consequences follow from a stated assumption and which still require evidence.