Economists at the University of Ulm measured overconfidence by getting people to answer general knowledge questions such as "how long is the river Nile?" and give 90 per cent confidence intervals around their answers; overconfident people tend to give too small intervals. They then got people trading in an experimental asset market. And they found that while trading activity varied a lot, the number of trades individuals made were unrelated to their degree of overconfidence. Bumptious knowalls are no more likely to trade a lot than modest folk.
But this is not the only type of overconfidence. The economists also got their subjects to predict numbers drawn from a random normal distribution based upon signals, which they were told were themselves unreliable. They found that subjects who overestimated the reliability of the signal were significantly more likely to trade than those who were wise to the signal's unreliability.
This implies that trading is motivated not by overconfidence about our ability, but by overconfidence about signals. Some people mistake noise for signal and trade on the erroneous belief that noise conveys information about future returns.
And lots of traders do this. In recent days over 1.1bn UK shares have changed hands every day. These trades cannot all have been based upon reliable signals about future returns; if they were, there wouldn't have been a seller for every buyer. This implies there's a lot of noise trading going on.
But if investors believe that others are trading on noise, they'll be more inclined to trade themselves in the belief that they can profit from the ignorant activity of the noise traders. In this sense, trading begets trading. And no, the smart trading does not drive the noise traders out of the market, not least because sometimes the smart money actually trades on the same side as the noise traders.
