To The Moon

Everyone's Waiting For QE. Warsh Won't Print. Here's The Lever I'd Watch Instead.

Mathematical Money | September 25, 2026

The Treasury went into the market on Wednesday and bought back its own long bonds. Up to $6 billion of 20- to 30-year paper. Second time in a fortnight.

The 10-year closed at 5.11%, up 15 basis points, the highest it's been since 2007. The 30-year did the same.

Think about what that means for a second. The single largest issuer of long-duration debt on the planet showed up as a buyer of its own paper, twice, and the yield went up. That's not a liquidity problem you fix with six billion dollars. That's the market telling you something about who's willing to own thirty-year duration at these levels, and the answer is fewer people than there used to be.

Scott Bessent said back on 19 August that Treasury would at least double its normal buyback size. They've since tripled the long-end operations from a $2 billion cap to $6 billion, and said future ones will run at least $4 billion. The official reason is liquidity support — giving holders of older, less-traded bonds a way out.

I think the official reason is true and also not the whole story.

So what happens next?

Almost everyone I read lands in the same place: eventually the Fed has to step in and buy bonds. Restart QE, cap the long end, done.

I don't think that happens, and the reason is the man sitting in the chair.

Warsh is the wrong chair for QE

Kevin Warsh was sworn in on 22 May this year, confirmed by the Senate 54 to 45 — the narrowest vote for a Fed chair on record. And his entire public identity is built on opposing exactly the thing people expect him to do.

He resigned as a Fed governor in 2011, in the middle of QE2, over that programme. He's spent the fourteen years since arguing the balance sheet is too big. He called it "bloated" in a Wall Street Journal piece last year.

At his confirmation hearing he said something that I think tells you where he'll land. Roughly: as the Fed grew its balance sheet, the people who owned financial assets benefited — and if you cut rates instead, a broader group benefits, because QE "tends to move through financial assets first."

That's not a technocrat weighing tools. That's a man who thinks QE is a distributional choice that favours the rich, saying so under oath, six days before taking the job.

For him to restart large-scale bond buying now, he'd have to repudiate the single position he's most publicly associated with. Not a policy reversal — a personal one, in front of everyone who confirmed him by nine votes.

So I put the probability of outright QE — the Fed announcing open-ended purchases of long-dated Treasuries — at somewhere around 10-15% over the next twelve months. Low, not zero. If something breaks badly enough, everyone's principles get flexible.

Except the printing already started

Here's the part I think most people have missed, and it's the bit that reframes the whole debate.

After the December 2025 meeting, the Fed began what it calls "reserve management purchases." Since January, it has bought roughly $160 billion of Treasury bills under that programme. Add MBS principal reinvestment and total SOMA buying is closer to $250 billion. The balance sheet has grown about $150 billion to around $6.75 trillion — $4,558 billion of Treasuries, $1,910 billion of mortgages, and change. Bank reserves are back up near $3.1 trillion.

The Fed is adamant this isn't QE. Their position is that it's a technical operation to maintain ample reserves, not an attempt to lower long-term borrowing costs.

And here's the thing — that distinction is real, and it's the entire reason your 30-year is at a nineteen-year high.

They're buying bills. Short-dated paper. QE, properly understood, is about buying duration to push down long-term rates. Reserve management purchases deliberately avoid doing that.

So the Fed is expanding its balance sheet and simultaneously not helping the long end, on purpose. Stealth QE isn't something I'm forecasting. It's running right now, at the wrong end of the curve, and the curve is telling you it noticed.

I'd put the probability that this continues and quietly scales through 2027 at around 75-80%. It's the path that requires nobody to announce anything.

The lever I'd actually watch

If you can't print and you won't buy duration, but you still need someone to absorb a lot of long bonds, there's a third option that almost nobody talks about.

You get the banks to do it.

Before 2008, there was no capital rule that specifically penalised a bank for holding Treasuries on its balance sheet. The supplementary leverage ratio — the thing that does that now — is a post-crisis invention. It treats a Treasury bond and a corporate loan identically for leverage purposes, which is a strange thing to do to the world's safest asset and has been criticised for years on exactly that basis.

Relax that, and bank demand for long bonds goes up without the Fed buying a single one.

Now here's why this isn't speculation. It's already half done.

Regulators finalised a rule in November 2025 modifying the enhanced supplementary leverage ratio for the largest US banks. It took effect on 1 April this year, with early adoption allowed from January. It recalibrated the GSIB leverage buffer from a flat 2% to half of each bank's risk-based surcharge, which frees up real balance sheet capacity.

But — and this is the important bit — the final rule kept Treasuries and central bank reserves inside the leverage exposure calculation. The proposal specifically asked for comment on excluding the Treasury holdings of banks' broker-dealer arms from the denominator. That exclusion was consulted on, considered, and left out of the final rule.

So the lever exists. It's been drafted. It's been through public comment. And it's sitting on a shelf, unused.

Probability it gets pulled in some form within twelve months: I'd say 50-60%. Higher than most people would guess, because of what it costs politically, which is almost nothing.

Why that's the path of least resistance

Think about how each option looks on a front page.

QE is a monetary decision with a press conference. It shows up on the Fed's balance sheet. It has a name people recognise and a politics people have already formed opinions about. It makes Warsh a hypocrite.

A capital rule adjustment is a technical amendment issued jointly by banking regulators. It doesn't expand the Fed's balance sheet by a dollar. It has no announcement effect to speak of. Roughly nobody outside finance will read about it.

And the mechanism is the same in the direction that matters: more demand for long-duration government paper. The difference is whose balance sheet absorbs it — the central bank's, in public, or the commercial banks', quietly.

If I'm designing the least painful route out of a long-end problem under a chair who's spent his career attacking QE, I'm not printing. I'm making it cheaper for JPMorgan to own thirty-year bonds.

The uncomfortable part, and I'll say it plainly because it cuts against my own conclusion: this is exactly the kind of thinking that produced the last crisis. Making it capital-efficient for banks to warehouse duration is how you get concentrated interest-rate risk sitting in the banking system. We watched a version of that in 2023 when a few regional banks discovered what happens when you hold long bonds through a hiking cycle. Doing it deliberately, at GSIB scale, is a trade-off rather than a free lunch.

One thing that doesn't fit the debasement story

I hold the view that this is fundamentally a debasement trade — too much duration supply, a currency being managed rather than defended, and a long end that wants a higher real return to fund it.

But there's a fact that complicates it and I'd rather raise it myself than have it raised for me.

The Fed is currently hiking. The target range is 3.75-4%, raised in a unanimous vote, and the market puts October odds of another hike at roughly two-thirds to seventy percent. That's not a central bank debasing anything at the short end. That's one fighting inflation.

So the honest version of my thesis is narrower than the usual debasement story. It isn't that the Fed is printing money and the currency is being destroyed. It's that the short end and the long end have come apart. The Fed controls one with the policy rate, and it's tightening. The other is a supply and demand problem in a market that has to absorb enormous issuance, and the tools for that are fiscal and regulatory, not monetary.

Two different problems. Two different sets of levers. Most commentary treats them as one.

So — at what yield do I redo the math?

This was the actual question, and my honest answer is that I don't have a number. Not a coy one, not one I'm keeping to myself. I genuinely don't run a level that triggers anything.

That's deliberate, and here's the reasoning.

Pick any threshold and then look at what it would have done to you this year. Say you'd decided at the start of 2026 that 4.25% on the ten-year was where equities stop making sense. You'd have sold. Then it went to 4.5% and you'd have felt clever. Then 4.75%, then 5%, then 5.11% on Wednesday, and somewhere in there the index went up anyway and you spent nine months out of a market that didn't care about your line.

A threshold on a macro variable sounds like discipline. In practice it's a prediction with a number attached, and I'm no better at predicting the ten-year than anyone else posting about it today.

What I do instead is structural rather than tactical. My long-dated positions run to 2027 and 2028. I bought that much time specifically so that no single yield print forces a decision out of me. That isn't cleverness — it's an admission that I can't time this, so I've tried to buy my way out of needing to.

The thing that would actually change my behaviour isn't a level at all. It's persistence. A 5.11% that fades back under 5% in a few weeks is a scare, and scares tend to be where you add rather than where you sell. A ten-year that sits above 5% for a full quarter while the Fed is still hiking is a different animal — that would tell me the market is pricing supply rather than policy, and supply doesn't resolve when the Fed pivots.

I'll be honest about the cost of working this way. Not having a threshold means I'll be slow. If this really is a regime change rather than a scare, someone with a hard rule gets out ahead of me and I eat the first chunk of it. I've decided that's a price worth paying versus the alternative, which is being whipsawed out of positions by a number I made up in advance.

For what it's worth, I haven't sold anything this week.

Genuinely curious where people land on this, because I might be wrong in an interesting way.

Does anyone think Warsh actually restarts QE? I've argued he can't without eating his own record, but chairs have surprised before and I'd like to hear the case.

And has anyone else been watching the leverage ratio file? I get the feeling it's the most consequential thing in this whole story and it's being discussed by about forty people.

Stop guessing. Start calculating.

Live to fight another day. 🤙

Nothing here is financial advice. The probabilities above are my own estimates, not anyone's forecast, and I've tried to separate what's documented from what I'm guessing at. Do your own work.