From Mutual Fund Herding to a Cascade

This thread follows decision-makers who observe one another, abandon private evidence, and sometimes make sensible or disastrous choices as a result.

increasingBearing stren­gth in each eventdecreasing

A Safe Career Choice

The Manager’s Double Task Investing and Appearing Sensible Mutual-fund managers are judged not only by whether they invest well, but by whether investors believe their decisions are reasonable. Short-term performance can be a poor measure of skill, which makes comparison with peers attractive. A manager who holds the same kinds of stocks as competitors can defend a bad outcome as a shared mistake. A contrarian who is wrong risks looking uniquely foolish. Incentives to Imitate David S. Scharfstein and Jeremy C. Stein’s model explains why managers may copy competitors even when imitation reduces their informational advantage. The conflict is professional as well as intellectual: managers seek accurate choices, but also protection from reputational punishment.

Information Loses Its Source

Copying Replaces Research Private Signals Disappear Once managers base choices on what others appear to know, their own research contributes less to the market. Later observers may mistake a repeated choice for many independent confirmations. In reality, the decisions may all trace back to one original signal or to fear of appearing different. A Collective Blind Spot The imitation shrinks the range of investments under consideration and can make the market less informative. Each participant seems to be responding to evidence, while much of the group is responding to other participants. The process illustrates how a collection of rational career choices can yield a collectively weak result.

The Plank Road Cascade

Adoption Becomes Evidence A Visible Success Plank roads initially improved travel, and their rapid spread made them seem increasingly proven. Towns considering transportation projects observed roads that appeared to solve real problems. As adoption grew, fewer people had reason to question the idea or investigate alternatives. The visible pattern became a substitute for information about long-term durability. A Rational Sequence, Bad Result In a model by Sushil Bikhchandani, David Hirshleifer, and Ivo Welch, early people receive imperfect signals; later people infer from their choices. A few unlucky early signals can send everyone toward the worse option. The explanation shifts blame from simple gullibility to a process that can make imitation reasonable at each step and still collectively wrong.

The Telecom Boom

A Number Becomes Conventional Wisdom Growth Expectations Early Internet traffic growth was reported at roughly 1,000 percent annually. When growth slowed, the figure remained lodged in the industry’s assumptions. Telecommunications companies invested tens and then hundreds of billions of dollars in capacity to meet a future that many participants treated as certain. Not investing appeared more dangerous than questioning the shared forecast. Collapse Reveals the Error After the bubble burst, many companies were bankrupt or close to it, and the growth assumption was finally reconsidered. The cascade had made each new investment appear to validate the next. The episode connects information cascades to large-scale economic consequences: a common forecast can become less informative precisely because everyone acts as though it is confirmed.

Making Decisions Together

Changing the Rules of Learning Sequential Guesses In Angela Hung and Charles Plott’s urn experiment, participants received private marble draws and observed earlier participants’ guesses. Cascades often began when people trusted the visible sequence over their own imperfect signals. That behavior could improve individual decisions while erasing the independent information needed for a strong group answer. The Group’s Incentive When payment depended on the group’s majority answer rather than each person’s individual accuracy, participants relied more on private information. The collective result became significantly more accurate. The experiment points toward a remedy: reduce the influence of earlier choices and design institutions that reward contributions to group accuracy, not merely personal conformity.

The chapter 3 discussion of fund managers

As managers imitate one another

Mid-nineteenth-century America

From about 1996 to the early 2000s

Chapter 3’s classroom experiments