# US Treasury Yields: 2002 vs 2026

*Date: 1 October 2026*
*Source context: CNBC chart "10-year and 30-year Treasury yield" (1980 to 2026), with the 30-year at 5.6702%, its highest level since July 2002.*

---

## Prompt

> Back to where it was when we got to Australia may 2002. And then thinking this is as low as rates could go. And the next move higher. How wrong that decision was.
>
> Now. 2026. What do you think?
>
> How could I have known then 2002 that rates were to go toward 0%. And how can we learn from that where they may go now?

---

## Response

The chart shows 2002 as one stretch of a 40-year downtrend. The 30-year at 5.67% is now back to where it was in July 2002, which is a nice symmetry for the story.

### Could you have known in 2002?

Mostly not, but there were clues:

- **Anchoring on levels.** "5% is low" was judged against the 1960s-90s, when rates were mostly higher. The chart shows lower highs and lower lows since 1981. Levels tell you little; the drivers matter more.
- **Disinflation was still running.** Inflation was falling, and by 2003 the Fed was openly worried about deflation. China's entry into world trade was pushing goods prices down, and Asian savers were piling into Treasuries (Bernanke's "savings glut" idea, 2005).
- **Japan was the preview.** Its 10-year was already around 1.3% in 2002 and went toward zero. An aging society with heavy debt and weak demand showed that "rates can't go lower" is false. That was the best available signal, and few people applied it to the US and Europe.
- **Later shocks were unforeseeable.** The 2008 crisis, QE and the 2020 pandemic took rates to the floor. No one could have forecast those.

The lesson isn't that you should have predicted zero. A forecast based on mean reversion carried too much confidence, and a structural regime that held for decades defeated it.

### 2026: what I think

I'd frame it as a regime question rather than a number. The forces behind the 2002-2020 decline have partly reversed.

**Pushing yields up:**

- Large deficits even at full employment (US interest costs now exceed $1 trillion)
- A rebuilt term premium
- Tariffs and deglobalization
- Defense and energy-transition spending
- AI capex
- Aging populations that spend down savings
- Rising Japanese long yields, which means the old anchor has loosened

**Pulling yields down:**

- Recession or financial accidents
- A possible AI productivity boom
- Slowing growth
- The fact that governments can't sustain very high rates. If yields spike, central banks and treasuries tend to respond with QE, shorter issuance, or financial repression.

My read is that a return to near-zero rates needs a crisis, so I don't see it as the base case. But the 2002 lesson cuts both ways: I'd be equally wary of confidently saying "rates only go up from here."

### Three scenarios

1. **Higher for longer (most likely):** the 10-year at 4-5.5% and the 30-year at 5-6.5%, with a persistently positive term premium.
2. **Fiscal stress episode:** disorderly moves higher, like UK gilts in 2022, then an official response that caps them.
3. **Growth or financial shock:** a flight to safety that drops yields quickly, even within the higher regime.

### Signposts to watch

- Term premium versus expected policy rates
- Real yields (these separate inflation worries from supply worries)
- Deficit path and auction demand
- Japan's long end
- Inflation expectations

### Separate the view from the position

Separating the view from the position matters more than the view itself. In 2002 the costly part was probably how much was riding on a single forecast. With uncertainty this wide, think in terms of what you can survive being wrong about in either direction.

*This is analysis, not financial advice, and I'm not a financial advisor.*

---

**Possible next step:** build the scenarios out with the debt-sustainability numbers.