Thursday, June 18, 2026
Economy

Treasury’s Trillion Refunding Is the Quietest Liquidity Story of the Second Half

· · 3 min read

Treasury’s $7 Trillion Refunding Is the Quietest Liquidity Story of the Second Half

The U.S. Treasury’s Borrowing Advisory Committee meets Tuesday with one number on the table: roughly $7 trillion in total marketable debt outstanding is on track to roll over between now and the September 30 fiscal-year close, and dealers are already pricing the bid-to-cover window as the tightest liquidity event since the 2014-2015 supply glut. The 2-year, 5-year, and 7-year auctions will absorb the bulk of the issuance, but the marginal buyer has shifted: primary dealer takedowns now account for 24% of all auctions, the highest share since the debt-ceiling standoff of 2023, and that share is rising week over week. That is the tell. When dealers are forced to absorb paper they cannot syndicate, the cost of carry rises in the when-issued market, and the long bond responds first.

Why this matters for the curve: the Treasury’s Quarterly Refunding Announcement, due Wednesday, will set the coupon auction sizes for August and September. Front-end supply relief is unlikely. The 4-week and 8-week bill issuance has already been cut by $40 billion over the past two months as the Treasury tries to avoid bumping up against the debt ceiling, but coupon supply is structurally rising because net new issuance must absorb a $1.9 trillion fiscal-year deficit. That means the 5-year and 7-year sectors will be the marginal supply point, and the 10-year and 30-year will price off the 5-year auction tail. If Wednesday’s announcement raises coupon sizes by more than $1 billion per tenor, expect a 4-6 basis point backup in the long bond within 24 hours, and a steeper 2s10s curve by the end of the week.

The desk positioning is the second-order story. Levered funds have flipped net long the long bond for the first time since February, and the CFTC Commitment of Traders report released Friday shows asset managers holding the largest short position in 10-year note futures since November 2023. That is a one-sided trade, and one-sided trades unwind violently when the marginal auction tail surprises. The July 10-year reopening auction is the next major data point, but the September refunding is the structural event. Anyone trading the long bond through year-end is essentially trading the Treasury’s auction-calendar decisions plus the debt-ceiling timeline, not the Fed’s policy path.

The technical level to watch is the 4.45% area on the 10-year yield. That was the high from May, and it acted as resistance twice in June. A clean break above 4.45% on heavy volume would signal that the supply story is overwhelming the Fed-dovish narrative, and would likely push the 30-year yield through 4.75% within a week. A failure to break 4.45% on a $44 billion 10-year reopening would suggest primary dealer demand is strong enough to absorb the coupon supply, and the curve would steepen modestly while the long end holds. The trade is not whether the Fed cuts in September; the trade is whether the Treasury’s auction calendar allows the long bond to hold its current range, and that question gets answered Wednesday.

The liquidity backstop is the most underpriced risk. The Fed’s overnight reverse repo facility has drained from $2.5 trillion in 2022 to under $50 billion today, and the Standing Repo Facility has not been meaningfully tested since its 2021 launch. If a coupon auction tails by more than 2 basis points and primary dealers hit their takedown limits, the SRF becomes the de facto backstop, and the Fed’s balance sheet mechanically expands by the size of the dealer credit. That is a tail risk, not a base case, but it is the kind of tail risk that the long bond market has not had to price since March 2023. The marginal investor in the 7-year sector should be aware that the bid-to-cover is not just a number on a Bloomberg page; it is a leading indicator of whether the Federal Reserve will be forced back into the market as a buyer of last resort, and that outcome would reprice the entire back end of the curve in a single session.