Long-Bond Yields Keep Making New Highs. What Can the US Actually Do?

Either the nation must destroy public credit, or public credit will destroy the nation.
— David Hume, Of Public Credit (1752)

What This Episode Is About

MacroMicro’s After Meeting Podcast, EP 213 (published 2026-09-06, recorded September 3). Host Roger, with their US analyst Ralice as guest. The topic: long-bond yields keep making new highs, so what can Washington actually do about it?

The setting: the 10-year Treasury yield at 4.8%, the 30-year at 5.2%. Both the 30-year and the 20-year are at ten-year highs. The previous episode called it “the gates of the underworld opening” on Treasuries, and Roger opens this one by saying the gate shows no sign of closing — not even after the Treasury Secretary rolled out a buyback program the week before. Add the just-finished central bank symposium, where the market read Warsh as hawkish, and even short-end yields are climbing.

What I liked is that the show spends almost no time guessing where yields go next. It does something else: it picks up every single thing the US government is currently doing, drops each one into a single equation, and asks which term it moves and by how much.

There’s also a side bit at the open that made me laugh. Roger tests whether Ralice is a cat person or a dog person using three questions about how she uses AI: do you come in with an answer looking for supporting evidence, or do you just talk openly? When the AI ignores your instruction, do you scold it or coax it? When it forgets earlier context, do you paste it again or switch providers? Her answers: come in with a view and ask for data to support it; start polite, then get sharp; switch straight to another platform. Verdict: cat person.

I kept that in mind, because her answer to the first question ties directly into what the rest of the episode is about.

The Main Points

1. The short end and the long end are different animals. The short end tracks rate expectations, which is straightforward. The long end mixes three things: real rates, inflation expectations, and term premium. Ralice’s line is that with long bonds you can’t just look at whether the yield went up or down — you have to know what pushed it. Since July, the driver has been term premium, which is what investors demand as compensation for “will you actually pay me back?”

2. The market’s read on Warsh may be something the market added itself. Ralice lines this speech up against the July press conference and finds most of it — inflation too high, price stability is the first job — was already said in July. She counts exactly one piece of new information: for the first time he stated plainly how he personally sees the economy. Her read is that the market took “not more dovish” and turned it straight into “more hawkish,” skipping the neutral middle. Her own call: a September hold, with 70% confidence.

3. The counterintuitive bit: a hike could be good news for long bonds. This is the part I most wanted to write down. If the Fed hikes while under public pressure from the President, that proves its independence isn’t for sale — and the term premium the long end demands could fall. The same action is pressure on the short end and medicine for the long end, because the two ends are pricing different things.

4. The equation: change in debt ratio = fiscal deficit + (R − G) × current debt ratio. R is the financing rate, G is the growth rate, everything expressed as a share of GDP. Three variables, three possible roads. This equation is the skeleton of the episode; every later段 hangs off it.

5. The deficit road is closed. The 2025 deficit came in at $1.78 trillion, below 2024’s $1.83 trillion, thanks to tariffs: the effective rate went from around 2% to over 10%, and tariff revenue grew more than 50%, adding close to $200 billion. But Ralice points at the order of magnitude — revenue is in the hundreds of billions, the deficit is in the trillions. And 2026 is only counted through July, with the cumulative deficit already at $1.8 trillion, past all of last year. Tariff rates can’t climb forever, and nobody looks willing to cut spending.

6. The rates road is plumbing work. This is the detailed part. Issuance shifts toward the short end (money market funds are approaching $8 trillion, the Fed rolls maturing mortgage-backed securities into T-bills, and short paper never lacks buyers). On regulation, eligible Treasury trades move to central clearing so a borrow and a lend net against each other, freeing up balance sheet. And since last June the Fed’s repo window has had an added morning session, because funding stress often shows up in the morning. Ralice is clear about what these are: unlike stablecoins, they don’t create new demand. They remove friction, so the same capital can absorb more Treasuries.

7. One word. Ralice mentions scrolling Threads during her commute and seeing someone dissect the Treasury’s quarterly refunding statement. The wording on coupon auction sizes had gone from increases to changes. She went back and checked — it really had. One word that might hint the next adjustment isn’t necessarily upward. She’s cautious about how fast that happens, but agrees the direction is both supply and demand leaning short.

8. The only real answer is growth. As long as G keeps up with or exceeds R, a high rate still gets the debt ratio diluted away. The show points to the 1990s internet era: Treasury yields sat at 5–6% for a long stretch while the debt ratio fell, because productivity was driving growth. They also drop in one aside — Taiwan currently has the lowest R minus G among major economies.

Going Further

”I don’t own bonds. Why does this concern me?”

The show asks this on the listener’s behalf right at the start, and it’s the one worth settling first. Ralice’s answer is short: rising rates are a tightening of the funding environment, and tightening pressures equities.

Unpacked, though, it becomes useful. The same event wears two faces. One is compressed valuation: the risk-free rate is the floor under every asset, so when the floor rises, the same future cash flows discount back to less today and the multiple contracts — while nothing at all happened to the business. The other is that rates get high enough to genuinely shrink demand and cut capital spending. That one is fundamental.

Later in the episode, answering a listener question about Taiwan, they use exactly this split. The listener notes that M1B growth has flattened at a high level and the central bank faces pressure to hike, so the second half looks bleak. Ralice agrees the worry is reasonable, then adds the most practical sentence in the whole show: separate whether the correction is a funding-side one or a fundamental one. If the fundamentals haven’t been blocked by the hike, then it’s a funding-driven valuation correction — which, for a long-term holder, is an opportunity rather than a warning.

I’ve fallen into this hole myself. During the last leg of rising rates I attributed a software holding’s entire drawdown to “the industry is finished,” sold near the bottom, and then watched the revenue keep growing quarter after quarter. I hadn’t separated the two kinds of falling; I’d only seen the red on my screen. Now I make myself answer one question: did this quarter’s revenue and margin get worse because of rates? If I can’t say yes, what I’m looking at belongs to the denominator, not the numerator.

”The news says they’re fixing it. Is anything actually fixed?”

Once the episode lays out everything the government is doing, the items fall into two categories that headlines routinely blend together.

One kind creates demand. Nearly half of stablecoin reserves have to sit in short-term T-bills — that’s a genuine new buyer. Shifting issuance from the long end to the short end is also moving supply toward where demand already lives.

The other kind removes friction. Central clearing makes everyone’s counterparty the same institution, so a borrow and a lend net out and both the balance sheet used and the capital set aside fall. If money market funds get their counterparty restrictions loosened, cash no longer has to pass through the primary dealer layer. The Fed’s morning repo session simply aligns the window with when the need appears. None of these produce a single new buyer. They let existing buyers buy more.

Separating the two changes the question you ask when reading the news. Instead of “is this bullish or bearish,” you ask “does it move demand, or does it move friction?” Demand shows up in volume. Friction shows up in volatility and liquidity, and it takes time. The show is honest about that: these reforms can indirectly cap the upside risk in rates, but they need time, and time is what’s short.

One more detail worth stealing. Ralice notes the Fed’s repo facility still isn’t inside central clearing, so it can’t net against other repo trades — one reason institutions resist using it. A tool can be fully built and still sit unused because of something that technical. Announcing a facility and having a facility actually used are separated by exactly this kind of gap. When I look at systems I’ve built that nobody uses, the blockage usually looks like this too.

”Rates are this high — why not just lock in 5%?”

The first listener question in the second half is precisely this: at 4.25% you said bonds were fine for income, so at 4.7% you’re suddenly less keen?

I like Ralice’s answer because she doesn’t answer “should you buy.” She answers “what kind of money is this?” If it’s idle cash you have no plans for and can hold to maturity, then locking that yield in is locking it in, and it’s a good one. If it’s money you’ll want to move around, she’d say no — price swings are getting larger and the odds of a capital loss aren’t small.

The same instrument at the same yield is two different products for two different people. The difference isn’t in the bond; it’s in the horizon of your money. Listening to that, I recognised a habit of my own: asking “is this worth buying” without first saying how long I intend to hold it and whether I’ll need it back. Without that sentence, any answer is the wrong answer.

And to close the loop on the cat-person bit. Ralice said that when she uses AI, she usually has a view already and asks it to find supporting data. That’s how most of us use it, me included. But this episode demonstrates the opposite move — the market already had a view (Warsh turned hawkish), and what she did was line the speech up against the July press conference paragraph by paragraph and then say “I’m a little puzzled by that reading.” Going in with an answer to look for support, and going in with an answer to look for the counter-evidence, look like the same motion and point in opposite directions.

Sources and Further Reading

• The episode: MacroMicro After Meeting Podcast EP 213, “Long-bond yields keep making new highs — how does the US fix this?” (published 2026-09-06, recorded 2026-09-03)
• The equation used: change in debt-to-GDP = deficit ratio + (R − G) × current debt ratio, from the government debt chapter of standard macroeconomics texts
• The show mentions they’ve built a cross-country R-minus-G dataset (under the country data centre, in the credit and debt section) if you want to compare debt sustainability across economies yourself
• All figures here are as quoted in the episode, as-of the 2026-09-03 recording: 10-year at 4.8%, 30-year at 5.2%, three-month T-bill at 3.9%, money market funds approaching $8 trillion, 2025 deficit of $1.78 trillion, 2026 cumulative deficit through July of $1.8 trillion

One Thing Worth Taking With You

The single thing that stayed with me: a blended number tells you nothing by going up or down. You only learn something once you split it. A 5.2% long-bond yield can be real rates moving, inflation expectations moving, or “I’m afraid you won’t pay me back” moving. Three causes, three completely different next steps. Without splitting it, all you know is that it got bigger, and then you get anxious.

Here’s a small thing I’ve tried that has nothing to do with investing, if you want it. Take a sentence you’ve been saying to yourself lately — “I’m so tired,” “we’ve drifted apart,” “this job has gone flat” — and treat it as a blended indicator. Write down three components you could observe separately. Tired might be not enough sleep, or the same task repeating too many times, or something unresolved hanging over you. Then for one week, track only one of them: which days it showed up, which days it didn’t. After seven days you’ll know how much of that sentence was that one component moving.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.