Caltech bubble experiment

The Caltech bubble experiment, conducted at the Experimental Economics Laboratory at Caltech, showed that speculative trading could inflate prices even when the shares’ value was known in advance and the experiment had a fixed end. Designed by economist Colin F. Camerer, it made the gap between an asset’s calculable worth and its market price visible—and showed how traders’ expectations of one another could keep that gap open.

Experimental Design

Students traded shares in an imaginary company over fifteen periods, each lasting five minutes. Every share paid a dividend of 24 cents at the end of each period. A share held for all fifteen periods therefore yielded $3.60, making that amount the maximum rational price before trading began. After the first dividend, its remaining value fell to $3.36; after the second, to $3.12, declining by 24 cents with each period.

The experiment gave participants two shares to start and some money with which to buy more. Unlike an actual stock market, the share’s total dividend and the time at which trading would end were known. The benchmark for value was therefore not a matter of estimating a company’s uncertain future performance: it could be calculated from the remaining dividends.

The Bubble in Prices

Despite that clarity, the market price jumped almost immediately to $3.50 and stayed near that level until almost the end. It remained there even as the shares’ remaining dividend value fell below $3, then below $2, and eventually below $1. Participants continued paying $3.50 for shares whose scheduled returns had become far smaller.

Camerer asked students why they had bought at prices they knew were excessive. They explained that they had seen others trading at high prices and expected to buy, collect one or two dividends, and then resell at the same price to someone else. Their reasoning depended on the prospect of a later buyer willing to pay too much—the greater fool. The overpricing was not simply a failure to know the dividend schedule; participants could recognize the discrepancy and still trade on the expectation that someone else would sustain it.

What the Experiment Isolates

The setup removed several complications of real-world bubbles. Participants knew when the experiment would end, limiting the time available to resell. They also did not communicate with one another except through their buy and sell orders, so the bubble did not require direct encouragement or discussion. Even with clear information and a finite horizon, prices departed sharply from the shares’ calculable value.

The experiment thus illustrates the risks of dependent decision-making: a trader’s judgment can turn not on the asset’s returns but on the anticipated actions of other traders. Its contrast with a real market is important, however. In actual bubbles, the belief that prices will keep rising can coexist with, or become entangled in, a belief that the company itself may be worth the high valuation. The Caltech experiment makes one mechanism unusually plain, but it does not reduce every real bubble to conscious resale speculation.

Significance and Limits

Because the experiment supplied a fixed value, a known endpoint, and a restricted channel of interaction, it showed that the ingredients for a bubble need not include uncertainty about what an asset will pay or open verbal persuasion among traders. The decisive uncertainty lay in what other participants would do. A price could remain at $3.50 while the remaining dividends dwindled because buyers expected to sell before the market closed.

That result gives the experiment a specific role in the book’s account of markets: it demonstrates how individually intelligible expectations about other people can produce a collectively inflated price. It is an experimental illustration of bubble dynamics, not proof that all bubbles arise in the same way. Real bubbles can become harder to recognize as they persist, as traders’ expectations about resale mingle with beliefs about an asset’s underlying worth.