The Treasury Doubled Its Bond Buybacks. It Wasn't Enough.

Things started getting a bit ridiculous toward the end of last week, and then today the Treasury decided to grab something — anything — out of the toolkit.

Here's what happened. The 30-year Treasury yield had been climbing toward its highest level since the 2007-2008 financial crisis, hitting 5.31% on Tuesday. The 20-year was at 5.28%. The recent 30-year auction had priced at 5.216%, the highest since 2001. Nobody was sleeping. So on Wednesday morning, Treasury Secretary Scott Bessent announced he would double the size of the Treasury's bond buyback program for longer-dated securities, from $2 billion to at least $4 billion per operation, effective September 9 through November 4. The market rallied. The 30-year yield fell almost 10 basis points to roughly 5.187%. Stocks followed higher. TLT, the long Treasury bond ETF, surged 1.7% on the day. The S&P 500 was up a modest 0.2%, while the equal-weight RSP ETF actually outperformed at +1.0%.

So was this a genuine liquidity intervention or a political gesture timed three months before the midterm elections? The answer matters because it tells you whether the long-end bond selloff is actually solved or just briefly distracted.

Let me walk through the plumbing.

Treasury buybacks work like this: primary dealers sell off-the-run (less liquid, older) Treasury securities back to the government. The Treasury pays from its Treasury General Account, the government's checking account at the Fed. When that money leaves the TGA and lands in a dealer's bank account, it increases bank reserves in the broader system. So on the face of it, buying bonds should add liquidity, not drain it. The dealer gets cash instead of an illiquid bond sitting on their balance sheet. It's supposed to free up capacity.

But here's the part most commentary misses. The overall quarterly liquidity-support buyback allocation — the total envelope — hasn't changed. It remains $38 billion per quarter, exactly what it was set at in the most recent refunding statement. What Bessent did was reallocate and intensify within that existing budget. He bumped the per-operation ceiling from $2 billion to a $4 billion floor and raised the number of long-end operations from two to four per quarter. That's a scheduling change, not a funding change. Same $38 billion. Different pace.

And $38 billion per quarter, even at full deployment, is what exactly in a market with roughly $28 trillion of outstanding debt? As of this announcement, the buyback program — relaunched in May 2024 after the March 2020 Treasury market shock — has cumulatively repurchased $239 billion. That's real money, sure. But the IMF's own analysis of the program found that individual buyback amounts represent about 0.1 percent of the targeted off-the-run market. Yes, they account for roughly 10 percent of average daily trading volume in those buckets, which is why dealers care. But 0.1 percent of the total market is not going to reset the supply-demand balance at the long end.

And even the liquidity benefit the IMF documented is concentrated at the short end. The study found the most significant bid-ask spread narrowing — up to 0.8 basis points — occurred in the 1- to 3-year maturity bucket. For longer maturities beyond three years, where the current selloff is concentrated, the spread reduction effects were statistically insignificant. Bessent's expansion specifically targets the 10-to-20-year and 20-to-30-year sectors. Those are the places where the evidence says buybacks have the least effect on dealer spreads.

Now, let me concede something. Yes, the market reacted. Yields fell. The rally was genuine. Jeremy Stretch at CIBC put it directly: the Treasury Secretary acted to prevent the bond market selloff from dragging down other asset classes. And for a day, it worked. But here's the mechanism question: why was the selloff happening in the first place?

Three forces. First, the national debt keeps surging — the CBO is projecting an annual deficit of $2.1 trillion. Second, CPI is running at 3.4% year over year, which means the Fed isn't cutting its way out of this and real rates stay elevated. Third, the supply overhang at the long end is structural: the government has been aggressively issuing 30-year bonds while simultaneously pushing to replace longer-term debt with short-term bills. That works fine until the market starts pricing in the rollover risk — the continuous need to refinance massive amounts of maturing debt. At that point, the 30-year bond stops being a safe haven and starts being a liability that nobody wants to hold for the duration.

Bessent's move addresses none of those three forces. It briefly adds demand for off-the-run bonds within a tiny envelope, but it doesn't reduce the deficit, it doesn't change the inflation trajectory, and it doesn't meaningfully alter the supply schedule. The recent refunding statement kept quarterly debt sales steady while announcing these buybacks. So the government is simultaneously selling and buying, just at different maturities. It's not a net reduction in supply. It's a reshuffle.

And there's the timing. A DZ Bank analyst noted that the Treasury feared the economic pain of yields rising above 5%, and pointed out the political calendar: three months until the midterms. When you look at the sequence — yields spike, equities get nervous, announcement drops Wednesday morning, markets rally — it's hard to separate genuine liquidity support from the kind of late-summer intervention that looks good on TV.

Let me give you the historical calibration. This is what the 2000-2001 liquidity regime looked like from the inside: a series of increasingly obvious interventions that each produced a short-term rally and then were swallowed by the underlying supply problem. The Treasury can't print money to fund these buybacks — they're constrained by cash on hand and the ability to raise new funds. So when the deficit keeps growing and the debt keeps expanding, the buyback envelope either gets absorbed or has to come out of something else.

So here's what I think about the setup going forward. Understanding what I understand about spreads and economics, the 10-basis-point rally was the market's relief reflex, not a repricing of fundamentals. The underlying pressure points — $2.1 trillion annual deficit, 3.4% inflation, structural long-end supply — are all still there. If Bessent's expanded buybacks run four times through the quarter at $4 billion each, that's $16 billion hitting the long end. Divided across the 10-to-30-year and 20-to-30-year buckets, that's maybe $8 billion per bucket over seven weeks. Against daily Treasury secondary market turnover in those maturities that runs in the hundreds of billions, it's a rounding error.

But here's the conditional chain. If the buyback program actually does narrow bid-ask spreads for longer-dated securities — and the IMF evidence is thin on that — it could reduce dealer inventory risk enough to improve market functioning during the next bout of stress. If inflation data starts printing lower in the September-through-November window, and the Fed shows it's willing to cut, the yield decline will look like Bessent's idea even though the Fed would have done the heavy lifting. If neither of those happens, and the deficit trajectory stays on track, yields will resume their climb and the market will price this intervention for what it was: a speed bump, not a roadblock.

The equal-weight index outperforming the cap-weight today is interesting. RSP up 1.0% versus SPY up 0.2% is the kind of broadening move that normally signals the market isn't just being carried by a handful of mega-caps. But don't confuse breadth with conviction. A bond-buyback announcement lifts everything mechanically because duration risk is a common denominator across asset classes. The rally is real. The relief is genuine. The supply problem is unchanged.

What to watch: the September 9 buyback start date. The first actual operation will tell you whether dealers are genuinely sitting on excess long-end inventory they want to offload, or whether this is mostly a listing effect — the mere announcement of eligible securities doing the work while actual purchases barely register. The IMF found the listing effect is where most of the liquidity improvement lives. If the first operation comes in at the $4 billion floor and clears at modest bid-to-cover ratios, you'll know the market was already priced for this. If it gets oversubscribed 8-to-1, like the front-end buybacks historically have, maybe there's something real here.

Same deficit. Same bills. Different impact — because this time the Treasury is trying to buy its way out of a supply problem it created with its own issuance schedule. It's not going to work. But the market is going to trade it like it does until the data says otherwise.

The views expressed here are personal and do not constitute investment advice.