About Information Cascade Theory
Information Cascade, proposed by Bikhchandani, Hirshleifer & Welch (1992), describes how rational individuals ignore their private information and follow predecessors' actions, creating a 'cascade' — even when the majority choice is wrong.
The BHW model assumes a binary world state (Good/Bad). Individuals decide sequentially, each observing all previous public choices and receiving a private signal. They use Bayesian updating to compute the posterior and choose the option with higher posterior probability. When accumulated public information overwhelms any private signal, a cascade forms.
Bayesian updating: Individual i computes P(G | previous choices, private signal). If posterior > 0.5, choose A; else B. When the lead in public counts reaches the threshold (default 2), no private signal can reverse the decision — the cascade begins.
Cascade fragility: Information cascades are fragile. A small external shock, release of public information, or authoritative opinion can break an established cascade. This explains why financial bubbles burst suddenly and fashion trends shift rapidly.
Information cascades appear everywhere: financial herding causes bubbles and crashes; consumers choose the busiest restaurant; social media trends self-reinforce; technology standards adoption (VHS vs Betamax); online review feedback loops.
Information cascade vs herd behavior: Cascades emphasize rational individuals ignoring private info (Bayesian rationality), while herding usually implies irrational conformity. In a cascade, each person makes the optimal decision — public information simply overwhelms private signals.