The U.S. stock market is up 14% so far this year. How should we interpret this fact? On the one hand, 14% sounds pretty normal; a solid number but not flashy. On the other hand, you could argue that 14% is extremely surprising,and the market should be down 50%, not up 14%, due to the rise in interest rates.
Here’s the math. Using a simplistic measure, the duration of the U.S. stock market is 50 years.[1] Year to date, both nominal and real long rates have increased by 1%, with the 10-year TIPS rising from about 1.9% to 2.9%. Now, there’s a lot we could say about why interest rates went up, but let’s just accept that real rates have risen for some reason. Rule of thumb, an asset with duration of 50 years should go down 50% when the interest rate goes up 100 basis points.[2] According to this rule, the stock market should have crashed in 2026, along with long-duration bonds, but has not. Why not?
One explanation is that while the rate at which we discount future corporate profits has risen, expectations about those profits have risen even more, so that the higher rates have been more than offset by higher growth expectations. Under this scenario, today’s stock market might be a good deal for investors, because AI will bring about a new era of economic growth. Of course, this rosy scenario assumes that a substantial share of the AI benefits accrue to today’s listed firms rather than to consumers, workers, or currently private or nonexistent firms. That’s a big assumption.
The advent of magnificently productive AI could itself cause interest rates to go up. As firms borrow more for AI-related capex, they drive up interest rates. As savers anticipate a future of AI-generated abundance, they save less today, also driving up interest rates.[3]
So, one story is that the stock market is fine today, perhaps better than fine if the expected returns on both stocks and bonds have risen thanks to AI. An alternative explanation is that the stock market is making a mistake and has failed to react properly to the rise in rates. This latter explanation seems plausible to me: we have millions of news reports about every minor development in the halls of the Federal Reserve, but not much discussion of the far more consequential rise in long-term real rates.
As an empirical matter, stock and bond prices do not move in lockstep.[4] From roughly the late-1990s to the early 2020s, nominal Treasury prices and stock prices were negatively correlated, a phenomenon often called “risk-on/risk-off.” Seen from that perspective, 2026 is an example of risk-on: risky assets such as stocks rose and (allegedly) riskless assets such as Treasuries fell.
Setting aside the comovement of stock and bond prices, let’s consider forward-looking expectations. As I’ve previously discussed, as of January 2026, the expected equity premium was around 1.6% as measured by Shiller’s “excess CAPE yield,” which compares a measure of real 10-year rates with trailing earnings yields. An expected premium of 1.6% is low, but maybe not crazy low. But as of October 2026, the excess CAPE yield has fallen to 0.6%. That’s getting uncomfortably close to negative.
How does today’s situation compare to the tech stock bubble? The excess CAPE yield fell to 0.6% in November 1994, then bounced between 1% and 2% for a few years before falling to the seemingly insane level of -1.5% in January 2000 near the bubble’s peak. You could interpret this history in two ways:
- Pessimist: All is lost! We are getting close to a negative yield, just like 2000!
- Optimist: Cool! Today is similar to November 1994, which means we’ll have a gigantic bull market over the next five years.
I’m not sure which view is correct. One thing is clear, however. If we are in a bubble today, it is totally unlike the COVID bubble of 2021 and more like the tech stock bubble of 1999/2000. Back in January 2000, it seemed like a no-brainer to overweight bonds and underweight stocks, because bonds had attractive yields, with long-term real rates north of 4%. In contrast, 2021 presented a more perplexing choice, since bonds and stocks both looked expensive, with long-term real rates negative or zero. Today looks more like 2000 than 2021, with long-term real rates at around 3%. All we need is another 100-basis-point increase in rates, and we might have another no-brainer on our hands.
Endnotes
[1] The current dividend yield on U.S. stocks is about 1%. Suppose net repurchases add another 1%, so that total shareholder yield is 2%. Under simplifying assumptions, the duration of the U.S. equity market is the inverse of the yield, getting you a duration of 50 years. Using more sophisticated calculations, Polk and Vuolteenaho (2026) find the U.S. stock market has an average duration of 55 years.
[2] This calculation ignores convexity. A more precise statement is that if the interest rate goes from 1.9% to 2.9%, the price of a zero-coupon bond with a duration of 50 years should fall by 39%.
[4] See Campbell, Pflueger, and Viceira (2026), Gormsen and Lazarus (2026), and Polk and Vuolteenaho (2026).
References
Campbell, John Y., Carolin Pflueger, and Luis M. Viceira. "Bond-stock comovements." Annual Review of Financial Economics 18 (2026).
Gormsen, Niels Joachim, and Eben Lazarus. Interest rates and equity valuations. No. w34814. National Bureau of Economic Research, 2026.
Polk, Christopher, and Tuomo Vuolteenaho. "Purifying the Equity Premium," 2026.
Rachel, Lukasz. "What Next for r*? A Capital Market Equilibrium Perspective on the Natural Rate of Interest." Brookings Papers on Economic Activity 2025, no. 1 (2025): 131-213.
