How does herding behavior affect market stability, and what mechanisms trigger informational cascades?

Herding occurs when investors abandon their private information and follow the actions of others. While often portrayed as irrational, herding can be perfectly rational when private information is weak, when career concerns dominate, or when informational cascades make individual information irrelevant. The consequence is amplified market movements that overshoot fundamentals.

Three Types of Herding:

Loading diagram...

1. Informational Cascades:

Bikhchandani, Hirshleifer, and Welch (1992) showed how cascades form:

Investor 1 receives a positive signal and buys. Investor 2 receives a negative signal but observes Investor 1 buying. The positive evidence from Investor 1's action outweighs Investor 2's private negative signal. Investor 2 also buys. Investor 3, regardless of their signal, sees two buys. Their private information is overwhelmed. They buy.

The cascade is now established. Every subsequent investor buys regardless of their private signal. No new information enters prices — the market has become informationally inefficient.

Crucially, cascades are fragile. A single credible piece of public information can shatter the cascade, causing sudden reversal.

2. Reputational Herding:

Portfolio manager Dahlgren receives negative information about Meridian Corp, a widely held stock. If she sells and is right, she earns modest praise. If she sells and is wrong while peers held, she looks incompetent and may lose her job.

Expected payoff of selling: P(right) x small_gain - P(wrong) x large_career_loss Expected payoff of holding (herding): safe regardless of outcome

For most career-concerned managers, holding with the crowd dominates. Keynes captured this: it is better for reputation to fail conventionally than to succeed unconventionally.

Worked Example — Amplification Mechanism:

Stonebridge REIT has fundamental value of $45 and currently trades at$50 (5 units of noise-driven overvaluation).

Round 1: Herding momentum pushes price to $55. New investors see the price trend and buy.\nRound 2: Price reaches$62. Fundamental analysts are skeptical but career risk prevents selling. Round 3: Price hits $70. Index funds automatically increase allocation (mechanical herding).\nRound 4: A negative earnings report breaks the cascade.\nRound 5: Selling cascade. Price crashes to$38 (overshoots fundamental value on the downside).

Total range: $38 to$70 around a $45 fundamental. Herding amplified volatility by approximately 3x versus what fundamentals would justify.\n\nMeasuring Herding:\n\nThe LSV measure (Lakonishok, Shleifer, Vishny, 1992) quantifies institutional herding:\n\nH(i) = |p(i) - E[p(i)]| - AF\n\nwhere p(i) is the proportion of institutional trades that are buys in stock i, E[p(i)] is the expected proportion, and AF is an adjustment factor for random variation.\n\nStudies find H values of 2-5% for large-cap stocks and 5-12% for small-caps, confirming stronger herding in less-liquid, harder-to-value segments.\n\nMarket Stability Implications:\n- Herding creates excess volatility, excess comovement, and fat-tailed return distributions\n- Flash crashes and bubbles are partially herding phenomena\n- Circuit breakers and position limits are regulatory attempts to break incipient cascades\n- Contrarian strategies profit from herding-induced overshoots, but with significant timing risk\n\nMaster herding and market dynamics in our CFA Level III course.