Comments on information cascade theory
The information cascade theory has three major premises. First, the event being predicted is unknown to all observers; second, the information comes out sequentially; third, the later observers are able to base their predictions on their own signals and previous information. Without these premises, we can hardly describe an information cascade phenomenon sufficiently. In an article “Information Cascades and Rational Conformity”, the authors note that “cascade behaviour tends to increase the accuracy of predictions”. If a simple model based on an experiment such as “guessing blue or red balls in the urns” is based on perfect rationality, this is true according to Bayes’ rule. It is better to follow implications for the most of time, since the incorrect types are usually related to situations that are outside of our control, for instances, a person mistakenly makes a prediction that is inconsistently with the implications of prior predictions, a deviation that break a cascade, or an “unlucky” incorrect decisions observed by early observers.
It can be easily noticed that the first few observers’ signals and decisions are of crucial importance. This fact is termed “path-dependence” of the information cascade. False information conveyed by the first few observers is another source of incorrect cascade. How to promote correct cascades? In an article entitled “Following the Herd”, Lemieux provides a few suggestions. First, it is better to have “higher-precision individuals” who have more precise information than others to attract more followers. Second, it is suggested to have more private signals publicly known, so as to make the cascade less path-dependent, and lower the probability of errors. As a real world implication, Lemieux confirms Hayek’s arguments that “diversified individual actions will create more information than centralized decisions”, and “free markets and decentralized decisions maximize the use of knowledge in society”. Further, going back to the previous article “Information Cascades and Rational Conformity”, it is also necessary to add a pay-off mechanism to give incentives to decision making. In this light, if a reward is given to the majority of the people who make the right decision in the groups, individuals will care more about other individuals’ decisions and will try to coordinate by signalling their information to others, and thus other people will make better decisions.
Cascade theory can also be used to explain why some ideas suddenly become popular and die out just as quickly. According to Lemieux, information cascade has a fragile nature, they are “triggered by a small amount of information and can also be reversed or stopped by little new information”. However, cascade can sometimes be resilient. In the public domain, cascade can reach a tipping point in public opinions and thus leads to social collective actions such as revolutions or certain inertias. Sometimes special interest groups (e.g. trade unions) can manipulate public opinions that promote preference falsification to control public choices. Such manipulation ability of certain groups can amplify the collective action biases.
Links:
https://object.cato.org/sites/cato.org/files/serials/files/regulation/2003/12/v26n4-2.pdf