3.3.3.2 Information Cascades

To illustrate the effect of information cascades, Shiller tells the story of two restaurants that open next door to each other.1 One attracts increasing numbers of visitors merely because the first visitor made an essentially random choice to try this restaurant instead of the other. Subsequent visitors all choose the same restaurant, assuming that if earlier visitors chose this restaurant that must mean it's good. The other restaurant, meanwhile, stays empty. It may be the better one—who will ever know?

One way in which information cascades may influence investor behavior is analogous to the restaurant story: investors observe other investors' behavior and respond by adopting the same behavior. In stock markets, investors observe other investors' behavior through stock prices. This can create a feedback loop: price increases, via investor enthusiasm, feed back into further price increases.2

Investors may also observe other investors' behavior more directly, from the trading book and the stock exchange reporting system, and because securities laws require major share accumulations to be disclosed. When it becomes known that Warren Buffet has acquired a stake, the stock usually soars.3 The reason is investors presume that he is wefi-informed and that his move must mean the stock is undervalued.4 If Buffet is right, investors' copycat behavior will accelerate the process whereby the share price moves toward its fundamental value.5 But what if he were wrong? In that case we are worse off because by copying his investment behavior, some investors will have ignored their private information and consequently the market system will have aggregated less information than available among investors.6 As to those shareholders who didn't have any private information to begin with, their failure to make an effort to independently interpret public information will result in less diverse predictions and hence a less accurate group prediction.7

The ability to observe other investors' trades can influence investment behavior because trading is a sequential process. By contrast, one might argue, voting takes place simultaneously, since shareholders vote at the shareholders' meeting. On closer inspection, however, it becomes clear that shareholders can observe other shareholders' votes prior to making their own voting decisions. Perhaps the clearest example is a proxy contest. The challenger will often hold a significant number of shares, otherwise he wouldn't seem credible. In addition, it is not unusual for other large shareholders to publicly take sides prior to the actual vote, perhaps with the very purpose of influencing other shareholders. Again, if the other shareholders rely on the large shareholder 's judgment and ignore their private information or fail to independently interpret public information, epistemic quality might suffer.

Even if shareholders do not observe each other's voting behavior, they may stil make the same voting decisions because they base their decisions on the same information transmitted through a cascade. Again, it is easiest to see how this can happen when we look at investment decisions. After all, when markets soar, shoeshiners notoriously offer stock tips, not voting recommendations. But the vehicles that can rapidly spread information that influences investment decisions—human conversation, the media, the Internet—can equally rapidly spread information that influences voting decisions. In fact, the information is often the same. A popular story about a CEO 's golden touch can either cause an investor to buy stock or, if he already owns the stock, to vote for a risky acquisition proposed by that CEO. Either way, the investor's move hardly reflects an independent judgment.

Notice that this kind of voting behavior, while perhaps not independent, may be perfectly rational. As long as an investor does not expect the cost of actively gathering information to outweigh the benefit of casting an informed vote, he might as well base his voting decision on information presented for free. Thus, as the ownership of a firrn becomes increasingly dispersed and individual shareholders' expected influence on the outcome of the vote decreases, increasing numbers of shareholders might rationally engage in what is referred to as "epistemic free riding."8 That is, if they deelde to vote at all, for when shareholders do not expect to influence the outcome of the vote, it is rational for them to save themselves from not only the cost of gathering information but also the hassle of returning the proxy form.9 This, indeed, is why most shareholders are rationally apathetic.