Two interventions, two round trips: the long end now sets the discount rate

Buybacks and currency operations work on the bid, not on the supply. An intervention changes the hedged arithmetic for one class of overseas buyer for a while; a buyback swaps one maturity of Treasury paper for another and retires nothing. Neither alters the volume that has to be absorbed, and the volume is what has changed. A quarter that adds $1 trillion of debt annualises to roughly $4 trillion a year, about half again the average pace since January 2020 ^[1], and equal to something like an eighth of the entire publicly traded stock arriving every year ^[3].

Four fifths of the federal debt trades publicly ^[2]. That makes the marginal buyer's required yield the base of the term structure sitting under every corporate discount rate, every property cap rate and every terminal value in a growth model.

For most of the past two decades it did not work that way. In Wolf Richter's telling, QE from late 2008 through early 2022 ^[8] turned a bond market that priced risk into one that went along with anything ^[13]. By the summer of 2020, with the Fed buying about $120 billion a month ^[10], the 10-year yielded 0.5% and the 30-year a little over 1% ^[9], and Richter argues the market had stopped pricing risk, inflation or the supply queued up behind it ^[11]. That was an administered number rather than a cleared one, and operators built capital structures on top of it.

The fiscal side offers nothing to lean the other way. Deficits ran near 6% of GDP for the four years through 2025 ^[4], and the Congressional Budget Office projects 5.8% for fiscal 2026 ^[5], which would make five consecutive years at roughly that level ^[5]. Richter's own prescription is fiscal consolidation, on the argument that deficits and inflation are what the bond market actually wants addressed ^[16].

Two honest limits on this read. It is one publisher's account, and Richter himself calls the episode a small rap on the knuckles rather than anything serious ^[12]. He also reports direction, not level: the decline was erased, but the piece does not say where the long end came to rest ^[3].

Even at that modest reading, the useful information is the hold time. Two attempts to lean on long yields inside about three weeks ^[4], both from the Treasury side rather than the Fed's, produced a couple of days each. Anyone whose model still treats a lower long rate as the thing that arrives when policy permits it is holding a position on the deficit path, and paying for it in the one input that discounts everything else.