Behavioral FinanceMarket Efficiency

Institutional Herding Is Weak, Pension Fund Data Shows

Summary by Robert Gorak · Published September 4, 2026 · Last reviewed September 4, 2026

Josef Lakonishok and Andrei Shleifer and Robert W. Vishny·1991·NBER Working Papers 3846
Sample: 769 all-equity pension funds managed by 341 money managersData: SEI quarterly portfolio holdings data set of tax-exempt equity fundsPeriod: 1985:1–1989:4

Institutional herding is the tendency of professional money managers to buy or sell the same stocks at the same time as other managers. Lakonishok, Shleifer and Vishny (1991) examine this in Do Institutional Investors Destabilize Stock Prices? Evidence on Herding and Feedback Trading. Using SEI's quarterly holdings for 769 pension funds and 341 managers over 1985:1–1989:4, they found a mean herding measure of just .027. Herding reached 6.1 percent in the smallest size quintile, but stayed weak at 1.6 percent in the largest quintile where most institutional trading occurs.

What the Study Found

The mean herding measure across all 1985:1–1989:4 stock-quarters is .027, with a median of just .001. Herding rises as stock size falls: the herding measure is 6.1 percent in the smallest market-capitalization quintile versus 1.6 percent in the largest. A cross-sectional regression of a manager's buy or sell decision on the fraction of other managers buying produces a slope of .05 (t=13.8). That regression's R-squared is only .7 percent. Positive-feedback trading (Dratio) among the smallest-quintile stocks swings from -.180 for the worst past-quarter performers to .036 for the best performers. Stocks institutions bought on net earned a size-adjusted excess return of 1.8 percent that quarter, versus -.3 percent for stocks sold on net.

"We conclude that there is no solid evidence in our data that institutional investors destabilize prices of individual stocks."

Lakonishok, Shleifer and Vishny (1991), Do Institutional Investors Destabilize Stock Prices? Evidence on Herding and Feedback Trading, p. 23.

Methodology

The study uses quarterly portfolio holdings from SEI, a pension fund performance evaluation service, covering 769 all-equity pension funds managed by 341 money managers. The sample period runs from 1985:1 to 1989:4, with holdings adjusted for stock splits and dividends at each quarter-end. Stocks are sorted into market-capitalization quintiles determined from the NYSE and AMEX universe and updated quarterly, to separate large- and small-stock trading patterns. Herding and feedback-trading measures are also computed by past-quarter and past-year performance quintiles, industry groupings, and money-manager asset-size quintiles.

Key Statistics

Metric Finding Context
Mean herding measure (H) .027 All stock-quarters, 1985:1–1989:4
Median herding measure (H) .001 All stock-quarters, 1985:1–1989:4
Herding measure formula H = |B/(B+S) − p| − AF B = net buyers, S = net sellers, p = expected buy proportion, AF = adjustment factor
Dratio formula Dratio(i) = ($buys(i) − $sells(i)) / ($buys(i) + $sells(i)) Dollar-based excess demand per stock-quarter
Excess return, net-bought stocks 1.8% Size-adjusted quarterly return, all firms
Excess return, net-sold stocks -.3% Size-adjusted quarterly return, all firms

Smallest-Quintile vs Largest-Quintile Stocks

Measure Smallest Quintile Largest Quintile
Herding measure 6.1% 1.6%
Share of dollar value traded .07% 84%
Share of number of holding changes 1.3% 35.7%

Why This Matters

The results undercut the popular narrative that institutional herding and trend-chasing routinely push stock prices away from fundamentals. Institutional trading activity is concentrated among the largest, most liquid stocks. That segment is also where herding and feedback trading are weakest, so it matters most for aggregate pricing. For portfolio managers, peer benchmarking pressure does not appear to translate into detectable correlated trading at the individual-stock level. The paper leaves open the possibility of herding at higher frequencies or across the whole market. Institutional influence on prices is therefore not entirely ruled out.

Frequently Asked Questions

Pension fund herding in individual stocks is weak, with a mean measure of just .027 across 769 funds from 1985:1 to 1989:4. The median herding measure is only .001 across the same sample. Herding is stronger among small stocks, reaching 6.1 percent in the smallest quintile versus 1.6 percent in the largest quintile.

Positive-feedback trading among institutions was concentrated in small stocks, where the Dratio measure swung from -.180 for worst past-quarter performers to .036 for best performers. In the largest-quintile stocks, where most institutional trading occurs, this pattern disappeared entirely. Aggregating across all size groups, the paper finds no evidence of positive-feedback trading on average, because trading is concentrated in large stocks.

Institutional herding occurs when money managers buy or sell the same stocks at the same time, measured at a mean of .027 across stock-quarters. The paper found only weak evidence of this behavior overall, with a median herding measure of .001. Herding was somewhat stronger among small stocks, reaching 6.1 percent in the smallest size quintile.

Reference

Josef Lakonishok and Andrei Shleifer and Robert W. Vishny (1991). Do Institutional Investors Destabilize Stock Prices? Evidence on Herding and Feedback Trading. NBER Working Papers 3846.

Read the full paper

Cite this summary

Gorak, R. (2026). Institutional Herding Is Weak, Pension Fund Data Shows. Tradicted. https://www.tradicted.com/research/lakonishok-institutional-1991/