On Wednesday the Treasury told the bond market it would step in. It said it would more than double the maximum size of its liquidity-support buybacks of long-dated debt, from $2 billion to at least $4 billion per operation, targeting the 10-to-20-year and 20-to-30-year buckets. The change runs from 9 September to 4 November, the date of the next quarterly refunding.
The 30-year yield fell almost 9 basis points to 5.196%, off a 19-year high above 5.3% set the session before. Stocks snapped a three-day losing streak. Bitcoin ripped.
On Thursday the whole thing came back. The 30-year rose more than seven basis points to as much as 5.27% — roughly where it sat in the minutes before the Treasury said anything at all. The rescue had a shelf life of about one session.
What it cost the tape
The Dow closed down 703.84 points, or 1.32%, at 52,759.21. The S&P 500 fell 0.87% to 7,641.16 and the Nasdaq Composite lost 1.0% to 26,067.17. Walmart, down about 9% on its comparable-sales miss, did most of the damage to the Dow on its own. But the index was heavy before Walmart, and it stayed heavy after.
Oil did not help. West Texas Intermediate rose 2.83% to $86.78 a barrel and Brent gained 2.58% to $93.98 after President Trump promised a “crushing economic operation” against Iran over the Strait of Hormuz. Higher crude into higher long yields is the least welcome combination this market has.
Our take: A buyback is not quantitative easing and the market spent Thursday remembering it. The Treasury does not print the money it buys bonds with — it funds repurchases by issuing other debt, mostly shorter. So the operation swaps duration rather than removing supply. It can smooth a disorderly auction or a liquidity air pocket. It cannot shrink a deficit that ran $432 billion in the month of July alone. Bloomberg’s reporting called the plan a circuit breaker at best, and one session of price action agreed. When the fix is plumbing and the problem is arithmetic, the fix works exactly as long as everyone is looking at the plumbing.
The part that should worry people
The size of the intervention is the tell. Doubling per-operation capacity to $4 billion sounds substantial until you set it beside the volume the long end trades every day, or the amount of paper the government needs to sell between now and November. The market did the sum overnight.
That leaves the long end priced on the two things buybacks cannot touch: fiscal supply and inflation expectations. The 30-year at a 19-year high was never really about liquidity, and the July FOMC minutes — more hawkish than expected — landed in the same 48 hours. Daniela Hathorn of Capital.com framed the tension as policymakers still worried about inflation while long-term borrowing costs become a problem in their own right.
Rate-sensitive equity is where this shows up first. Chip stocks shed $680 billion in a day earlier this month with the trigger coming from the bond market, not from anything a semiconductor company said. Long-duration cash flows get repriced by the long bond whether or not the earnings are fine.
What to watch
- 9 September. The larger operations do not start until then. Between now and the first one, the buyback is a promise, and Thursday showed what promises are worth at this yield level.
- Warsh at Jackson Hole. The Fed chair speaks next week. The question is whether the hawkishness in the minutes turns up in his own language — and whether he says anything about the long end at all.
- The 4 November refunding. That is when the current window closes and the Treasury has to say how much it plans to borrow and at what maturities. Supply guidance will move the 30-year more than any buyback schedule.
- Whether 5.33% holds as the ceiling. Tuesday’s high is now a marker. A clean break above it, after an intervention has already been spent, would tell you the bid is thinner than anyone wants to admit.
The Treasury bought itself a day. That is not nothing — disorderly is worse than expensive, and Wednesday was calmer than Tuesday. But a market that gives back an entire intervention in a single session has told you what it thinks the intervention was for. Not lower yields. Just slower ones.