Of bubbles and crashes

Risk AI

There are plenty of warnings about a pending stock market crash, alongside a few hopeful voices. The only real uncertainty is when it will happen.

While there can be many reasons for market bubbles, momentum plays a key role. Investors see prices going up, which makes them buy — and because they buy, prices go up further, all in a happy feedback loop.

This dynamic also affects how we perceive risk. Along the bubble, volatility often declines, for the simple mechanical reason that when prices are steadily ticking upward, volatility is measured as low.

But volatility does not really capture the risk of crisis — it is a fair-weather risk measure.

We could call that the money-for-nothing trade, after the 1980s pop song.

However, prices go up the escalator and down the lift.

So what can go wrong with that money-for-nothing trade? I like to think about it in terms of endogenous risk: the risk we measure is usually not the risk we care about. The actual underlying risk rides the escalator up with prices, building invisibly during the calm phase of a boom.

Meanwhile, measured perceived risk — volatility, if you will — goes down as we ride the bubble and only shoots up after the crisis hits, when it is too late to trade out.

Perceived risk is negatively correlated with actual risk.

So what punctures the bubble?

In my book Illusion of Control I used the analogy of the little boy yelling that the Emperor has no clothes, like in H. C. Andersen’s fairytale.

Or, we could take the cue from Joseph P. Kennedy, the father of President John F. Kennedy, who famously said:

“When the shoeshine boys are giving stock tips, it is time to get out of the market.”

What about the fundamentals?

As reported in the Financial Times this week, Harvard economist Jason Furman estimates that if you exclude investment in data centers and information-processing technology, US GDP growth in the first half of 2025 would have been only 0.1%. In other words, almost all growth came from tech investment.

Today, Jamie Dimon warns in a BBC interview that he feels there are increased risks that US stock markets are overheated. “I am far more worried about that than others,” he said.

So even if I am confident that AI will transform society, and that some investors in AI will make an absolute fortune — it seems likely that many of those who are late to the party, investing at the top of the bubble, face significant losses. Periods of genuine innovation often coincide with speculative overreach and spectacular losses. Good for society. Fantastic for some investors. Terrible for others.

Look at history.

The best example of dot-com-era excess was Yahoo!, which had a market capitalization of $128 billion on 3 January 2000. It had about 180–200 million users worldwide at that time, which suggests the market was valuing each user at about $680. Back then, Yahoo was the AI of its era — a must-own name everyone believed would dominate the digital future. The details change, but the pattern of extrapolating early success into infinite growth is familiar.

At the same time, Amazon reached a peak valuation of about $39 billion in late 1999. Anyone buying then would have seen the share price collapse by more than 90 percent within two years. Yet an investor who held on until today would have seen that position rise almost fifty-fold, an annualised return of about 15 percent over twenty-five years. Even a purchase at the top of the bubble turned out to be remarkably profitable in the very long run.

So when I think about the market today through the lens of endogenous risk and the discussion in my book Illusion of Control, yes — it does look like a bubble, and a crash seems likely.

And I am very appreciative of all the investors who are now providing the money that fueles the AI boom that will make our lives so much more comfortable in the years to come. The challenge, as always, is telling progress from euphoria before the lift starts its descent.


Models and risk | Financial Regulation, Systemic Risk, Stability and AI

Blogs and appendices on artificial intelligence, financial crises, systemic risk, financial risk, models, regulations, financial policy, cryptocurrencies and related topics

© Jon Danielsson