A bubble exists when people are paying prices that make no sense. "No sense" as in, "There's no way in hell that this asset will ever generate enough income to pay them back."

Here's a nice illustration.

The authors of this paper conducted an experiment with college students. They gave each student some money and some "stock." Trading was divided into 10 periods. During each period, students were allowed to trade stock. At the end of each period, each share of stock earned a divident of 20 cents with a 50% probability or 0 cents with a 50% probability.

The expected value of each share of stock was $1 before the first period (10 periods with an expected value of 10 cents after each period). At the end of the second period, each share had an expected value of exactly 90 cents. And so on. At the end of the 9th period, each share had an expected value of 10 cents.

The students were told before the experiment how to value the shares of stock. But once trading began, the students quickly bid the price of shares well above their expected value.

This was not simply an instance of risk-preferring behavior, as the chart below shows. Beginning with period 7, the average share price exceeded the maximumpossible total return. Students paid an average of $1/share during period 7 even though the maximum return -- assuming a share hit the 20-cent dividend four times in a row -- was only 80 cents. During period 10, shares traded at 90 cents even though the most each share could earn was 20 cents.

This was not a freak event. Nine out of ten experiments produced similar results.

If people can react this irratonally in a controlled experiment with clearly defined financial returns, the bubbles in the housing and credit markets are easy to understand. We should probably be asking ourselves why we don't see more bubbles.