Treasury's Buyback Schedule Adds Up to Exactly $38 Billion. Two of the Operations Printed on It Don't Count.

by Jonathan Mercer · August 17, 2026

Chief Markets Correspondent, US Market Current

At 11:00 a.m. Eastern on Monday, the Treasury Department is scheduled to publish a preliminary list of eligible CUSIPs for an operation that will run for twenty minutes on Tuesday afternoon. Between 1:40 p.m. and 2:00 p.m. ET on Aug. 18, Treasury will take offers from primary dealers to sell back nominal coupon securities maturing between Aug. 19, 2046 and Aug. 18, 2056. The most it will buy is $2 billion of par. The least it will buy, according to its own schedule, is nothing at all.

The document that sets those terms is the Tentative Schedule of Treasury Buyback Operations, released Aug. 5 alongside the August 2026 quarterly refunding statement. It is a single page of tabular data, nine columns running from Announcement Date through Maximum Purchase Amount, and it is one of the few places where Treasury commits in advance to a calendar of trades on the other side of its own issuance machine. Read the columns rather than the headline and it does something fiscal documents rarely do: it reconciles exactly.

The schedule prints 20 rows. Two carry an asterisk, and the note at the foot of the page explains it: "Operation is scheduled within the May 2026 refunding quarter." Those two are the Aug. 6 operation in the 1-month to 2-year nominal bucket and the Aug. 11 operation in the 10- to 20-year bucket. Both also appear on the tentative schedule Treasury published on May 6. They are printed for continuity, not counted against the new quarter's budget.

Strip those two rows out and the arithmetic closes. The remaining 18 rows split into 16 operations that Treasury's own Operation Type column labels "Liquidity Support" and two it labels "Cash Management." Add the maximum purchase amounts on the 16 liquidity-support lines and the total is $38 billion. Add the two cash-management lines, each capped at $12.5 billion, and the total is $25 billion. Those are the two figures the refunding statement names.

The statement, issued Aug. 5 by Deputy Assistant Secretary for Federal Finance Brian Smith, put it this way: "Treasury is releasing a tentative buyback schedule for the upcoming refunding quarter. Treasury anticipates that, over the course of the upcoming quarter, it will purchase up to $38 billion in off-the-run securities across buckets for liquidity support and up to $25 billion in the 1-month to 2-year maturity bucket for cash management purposes." The statement does not break the $38 billion down by bucket. The schedule does, and the breakdown is where the program's shape becomes visible.

Five nominal buckets below ten years — 1-month to 2-year, 2- to 3-year, 3- to 5-year, 5- to 7-year and 7- to 10-year — get one operation apiece in the quarter, each capped at $4 billion, for $20 billion. The 10- to 20-year bucket gets four operations at $2 billion each, and so does the 20- to 30-year bucket: $8 billion apiece, $16 billion together. Three TIPS operations account for the remaining $2 billion.

The long end gets the most frequent attention and the smallest tickets. Eight of the 16 liquidity-support operations fall in nominal maturities of ten years and longer, but each is capped at half the size of a single operation in any nominal bucket shorter than ten years.

Set that against the issuance side of the same statement. The August refunding sold $58 billion of 3-year notes, $42 billion of 10-year notes and $25 billion of 30-year bonds, a $125 billion package scheduled to settle Monday. One 30-year auction is more than three times the entire 20- to 30-year buyback ceiling for the quarter. Treasury also repeated its forward guidance: "Based on current projected borrowing needs, Treasury anticipates maintaining nominal coupon and FRN auction sizes for at least the next several quarters."

The two cash-management operations are a different animal, and their size makes the point. Each is capped at $12.5 billion in the 1-month to 2-year bucket, on Sept. 3 and Sept. 9, settling the following day in each case. Either one of those afternoons carries a ceiling larger than everything Treasury plans to buy in the 20- to 30-year sector across the whole quarter. Treasury's schedule labels them separately from liquidity support, and they sit at the front end, where the department's cash position lives.

What did not change is the plan's size. The May 6 quarterly refunding statement, covering the quarter now ending, named the same two numbers: up to $38 billion for liquidity support and up to $25 billion in the 1-month to 2-year bucket for cash management. Two consecutive quarters, the same ceilings.

That stability sits against a question Treasury itself put on the table. In a charge dated July 29, 2025, the department asked the Treasury Borrowing Advisory Committee for input, recalling that "In the May 2025 quarterly refunding statement, Treasury announced that it is evaluating a broad range of possible enhancements to the buyback program, such as: changes to maximum purchase amounts, buyback operation scheduling and frequency, security eligibility, maturity bucket composition, execution process, and counterparty eligibility." That was a question posed to an advisory committee, not a decision, and more than a year later the published maximums stand where they were.

The column nobody quotes sits second from the right. Every row lists a Minimum Purchase Amount of $0. The plan is a ceiling, not a forecast, and dealer participation decides what clears. Treasury's Office of Debt Management, reviewing results through October 2024, put the ratio of offers received to the maximum on liquidity-support operations at anywhere from 0.77 to 9.24, meaning at least one drew fewer offers than Treasury stood ready to buy. Cash-management operations clustered far more tightly, between 2.6 and 5.2.

Treasury publishes the par amount offered and the par amount accepted for each operation separately, after it closes. Tuesday's release will show both for the 20- to 30-year bucket. The number to look at is not whether Treasury spends the full $2 billion, but whether dealers offer more than that in the first place.