Information Cascades

How people rationally follow others' choices and ignore their own information, why this can make crowds wrong, and how cascades can suddenly reverse.

You are choosing between two restaurants. Your guidebook recommends Restaurant A, but Restaurant B is full and A is empty. You go to B. The next diner does the same. Soon everyone crowds into B, even if A is actually better. This is an information cascade.

The theory

In 1992, economists Sushil Bikhchandani, David Hirshleifer and Ivo Welch, and separately Abhijit Banerjee, showed how this can happen rationally:

Each person has some private information, such as the guidebook, but it is imperfect.
They also see what others did.
After enough people choose one option, the public evidence outweighs any individual’s private information.
From then on, people ignore their own information and follow the crowd.

Once a cascade starts, later choices add no new information, since everyone is simply copying.

Why cascades matter

Crowds can be wrong: a cascade can lock in on a poor choice based on the first few decisions.
Fragility: because cascades rest on little information, they can suddenly reverse when new public information appears.

Examples

Fashion and fads: trends spread quickly and fade suddenly.
Financial markets: investors buy because others are buying, contributing to bubbles and crashes.
Technology adoption: early choices can decide which product wins.
Bank runs: seeing others withdraw money encourages more withdrawals.
Online ratings: early reviews can strongly influence later ones.

Breaking cascades

Sharing private information publicly, not just choices.
Independent reviews and expert evaluations.
Encouraging dissent in organisations, so people voice their own views.

The empty stall

Two food stalls open side by side. By chance, a few friends queue at the first one. Passers-by see the queue and join it, assuming it must be better. The second stall, possibly just as good, stays empty and closes within a month.

Thinking following the crowd is always irrational

Following others can be rational when they may know more. The problem is that cascades can lock in on mistakes.

Key takeaways

An information cascade occurs when people follow others and ignore private information.
It can happen rationally when public evidence outweighs private signals.
Cascades can lock in mistakes and reverse suddenly.
Sharing private information and encouraging dissent can break cascades.