Through extensive investigation and analysis that builds on existing research, the authors evaluate whether mutual funds herd and the degree to which such herding affects industry valuations.
What’s Inside?
Mutual funds herd in industries. Neither individual stock herding nor fund flows drive this process. Although herding affects industry returns, it does not skew or disrupt their valuations.
How Is This Research Useful to Practitioners?
Several metrics that the authors use lead them to the conclusion that mutual funds do herd, or engage in similar trading practices, in industries.
Two detailed measures of herding are widespread in the literature. The first is the Lakonishok, Shleifer, and Vishny (Journal of Financial Economics 1992), or LSV, approach, which involves looking for quarterly imbalances in the number of buyers and sellers of individual stocks. The second is the Sias (Review of Financial Studies 2004) approach, which expands on the LSV approach by quantifying the degree to which such imbalances persist across quarters. There is mixed evidence of herding when either of the two methods is used. The authors use both on a large dataset over the period of 1980–2013. Mutual funds’ tendency to exhibit behavioral traits similar to herding and the fact that mutual fund holdings are reported at the fund, rather than at the family, level are why mutual funds, not institutions, are the focus of the research.
Individual and institutional investors alike would gain from the authors’ insights to improve their understanding of how herding and groupthink among mutual fund portfolio managers could affect performance.
How Did the Authors Conduct This Research?
The authors’ approach considers the relevance of the existing literature on the subject and expands on it. Their investigation covers the period from 1980 to 2013. The dataset consists of all actively managed funds in the Thomson Reuters Mutual Fund Holdings database, excluding international, sector, index, and nonequity funds.
Using two well-known industry measures, the authors determine whether mutual funds herd into and out of industries. The LSV approach measures the imbalance of mutual funds that are buyers and sellers but does not distinguish whether the imbalance occurs on the buy or sell side. The Sias approach quantifies the extent to which investors’ trades follow those of other investors.
Using these quantitative metrics, along with mock portfolio construction and analysis of industry fund performance through factor analysis, the authors investigate these hypotheses: Mutual funds herd in industries, fund flows and individual stock herding do not drive mutual fund industry herding, industry-level herding does not steer industry values away from their fundamentals, and herding positively correlates with momentum at the industry level.
Previous researchers have provided strong evidence of mutual fund herding in industries. The authors substantiate those findings through the use of the aforementioned herding metrics because these metrics demonstrate statistical significance in the results. Going further, the authors test for herding among funds with different investment objectives and arrive at the same conclusion.
Fund flows do not drive mutual fund industry herding—nor does individual stock herding underpin mutual fund industry herding. Using the LSV and Sias approaches, the authors demonstrate that funds following other funds remains a meaningful piece of total cross-sectional correlation. This result holds for individual stocks as well. Although funds follow others into the same stock, funds pursuing other funds into different stocks in the same industry refutes the notion that stock herding is behind fund industry herding. Parsing the data further by style investing, time period subsets, the effects of investor sentiment, and industry conditions, the authors arrive at the same conclusion, with statistical evidence of these additional tests failing to produce significant results in terms of their effect on mutual fund industry herding.
The authors’ findings also confirm that industry-level herding does not move industry valuations away from their fundamentals. Herding and industry returns correlate positively, but no evidence of return reversals exists in industries where this correlation is high. Finally, industry-level herding is behind industry momentum. Exercises in grouping industries by returns and creating mock portfolios of winner and loser industries—along with exercises that use the CAPM and three-factor alpha calculations—demonstrate that winners (losers) experiencing the largest subsequent institutional ownership increase (decrease) display the strongest momentum.
Abstractor’s Viewpoint
Evidence of mutual fund herding in industries has existed for some time. The authors add significantly to the debate through their robust implementation of two recognized herding metrics across a large dataset spanning over three decades. The outcome of their investigation substantiates and expands on existing findings. Herding does exist but does not disrupt industry valuations. Their endeavor is a micro-level approach to a behavioral finance phenomenon that has important implications for individual and institutional investors. Understanding how herding affects industry returns is an important input in the portfolio management and asset allocation decision. A modified effort applicable to the fund industry in other developed economies would be a welcome addition to this area of research.