Join our community of smart investors
OPINION

Experiments in markets

Experiments in markets
April 12, 2013
Experiments in markets

The advantage of such experiments is that researchers can set a dividend over particular trading periods and so know for sure an asset's fundamental value. This allows them to measure how far prices differ from value. We cannot do this so easily in the real world, because asset values are much more subjective and cannot be measured precisely; we can never really be sure whether a high price means an asset is overvalued, or whether investors are attaching a reasonable (if wrong ex post) probability to a shift to happier times.

A big finding of these experiments, say Noussair and Tucker, is that markets "have a strong tendency to generate bubbles and crashes".

One reason for this has been shown in an experiment by Colin Camerer of the California Institute of Technology and Keith Weigelt of the University of Pennsylvania. They found that bubbles occurred in roughly one in 12 trading sessions, mostly early in the markets when traders were less familiar with pricing behaviour. This tells us two things. One is that while bubbles might not be as common as pundits sometimes claim, they are more common than the rational markets hypothesis predicts. Secondly, they are more likely to occur in unfamiliar assets - such as technology stocks were in the late 1990s; this is a reason to suspect that government bonds, the most familiar and easily priced of assets, are not in a bubble.

Such bubbles occur, they say, because of "information mirages". Traders sometimes interpret a price rise as a sign that others know something they don't, so they buy at the higher price, believing the asset to be worth more than they previously thought. Sometimes, though, this belief is wrong and so such buying merely propels the asset further above its value.

This might not be merely a laboratory artefact. Some believe such behaviour contributed to the 2008 financial crisis. Uninformed investors thought low spreads on credit default swaps were a sign that default risk was low, and so they sold insurance against it. When the risk turned out to be high, they lost billions.

Some other experiments have shown how investors' moods can generate bubbles. Yaron Lahav and Shireen Meer, two Israeli researchers, showed some subjects film of a Jerry Seinfeld comedy routine before they started trading. Such subjects traded assets at higher prices than fundamental value to a greater extent than did those not shown the film. Terrance Odean and colleagues at the University of California Berekeley performed a similar experiment, showing some subjects film of an exciting car chase. They generated bubbles more than subjects who were not so aroused.

Both experiments show that trading has an emotional as well as rational element, and that this emotion is easily influenced by outside forces.

Again, this is no laboratory curiosity. It's consistent with one of the most important and well-established facts about stock markets - that they do better in winter than summer. One reason for this is that as the weather improves in the spring - yes, it usually does - we become more cheerful and more excited; there's a reason why May Day is associated with fertility rituals. This mood causes share prices to rise too far, with the result that they often do badly during the summer.

You might think this evidence that assets can be often mispriced represents a case for being an active stock-picker, so we can profit from such deviations of price from value.

Not so fast. An experiment led by Matthias Sutter of Innsbruck University gives us reason for caution. He got subjects to trade an asset that had an equal probability of paying a dividend or not in each of several trading periods. Some traders were told in advance whether a dividend would be paid or not; some were told it for the first period, some for the first two periods, some for the first three and so on. He found that middlingly informed traders actually lost money. "There is a wide range of information levels where additional information does not yield higher returns." A little learning is indeed a dang'rous thing. This is because while semi-informed traders can make money by trading against less informed ones, they lose by trading against better-informed ones. And they can't tell which is which.

Other experiments by Richard Deaves of McMaster University and colleagues have found another reason for half-informed traders to lose. People who are overconfident about their knowledge - in that they believe it to be more precise than it is - trade too much and lose money. Shallow draughts of the Pierian spring intoxicate the brain.

Now, one common complaint about experimental economics is that what we see in the laboratory can be unrepresentative of the real world. I don't think this is valid; it often amounts to a rejection of evidence in favour of the prejudices of one's closed mind. Yes, the stakes in experimental markets are usually (though not always) lower than in real life. But so what? The financial crisis has taught us what psychology and other experiments told us before then - that big incentives do not produce better results than smaller ones.

Perhaps, then, we have much to learn from experimental economics. Not least of these lessons is that financial market participants don't always know what they're doing. But then, this lesson might explain why proper economics is so often ignored.