FACTOR CROWDING AND LIQUIDITY EXHAUSTION

• Date: 01 March 2019
• Author: Joseph M. Marks,Chenguang Shang
• Published date: 01 March 2019
• DOI: http://doi.org/10.1111/jfir.12165

FACTOR CROWDING AND LIQUIDITY EXHAUSTION

Joseph M. Marks

Northeastern University

Chenguang Shang

Texas State University

Abstract

Well-known anomalies and stable patterns in equity returns are widely employed to

guide stock selection. The use of overlapping multifactor models built on these patterns

induces correlated trade across investors. A stock with a strong signal from a

parsimonious multifactor stock selection model exhibits changes in trade activity, net

order imbalances, lower volatility, lower liquidity level, and changes in liquidity

comovement consistent with correlated trade. These results illustrate that correlated

trading among investors can affect the liquidity and risk of the securities they trade, and

imply that measures of portfolio liquidity risk that ignore these changes can understate

risk.

JEL Classification: G01, G11, G12, G23

I. Introduction

Stock liquidity has long been a topic of interest and research. Over the past 30 years or so,

this body of literature has evolved significantly as the focus of empirical research

develops and shifts. Early articles concerning stock liquidity have focused on cross-

sectional variation and its determinants. Amihud and Mendelson (1986), Glosten and

Harris (1988), and Stoll (1989), among others, decompose the bid–ask spread to measure

the components related to adverse selection and dealer inventory costs. Researchers have

also used transaction-level data to characterize liquidity in terms of the price response to

order flow, among them Brennan and Subrahmanyam (1996), Hasbrouck (1991), and

Easley and O’Hara (1987). Amihud (2002) proposes a widely used measure of liquidity

that captures much of the power of computationally intensive variables that require

transaction-level data, yet is much simpler to calculate because it employs daily data.

Beginning with Chordia, Roll, and Subrahmanyam (2000) and Hasbrouck and Seppi

(2001), the focus has shifted from the level of individual stock liquidity to documenting a

systematic component to changes in liquidity. The finding that a systematic liquidity

factor explains a significant portion of the time variation in liquidity motivates the studies

of Pastor and Stambaugh (2003) and Acharya and Pedersen (2005), both of which find a

stock’s sensitivity to a systematic liquidity factor helps explain cross-sectional variation

in returns and support the existence of a significant illiquidity premium in the sense of

classical asset pricing models. The abundant research into stock liquidity has much

The Journal of Financial Research Vol. XLII, No. 1 Pages 147–180 Spring 2019

DOI: 10.1111/jfir.12165

147

© 2019 The Southern Finance Association and the Southwestern Finance Association

improved our understanding of the critical role it plays in determining the returns and risk

of an investment.

There is also a better understanding nowadays that the trading activities of

groups of investors following similar investment strategies can affect the securities that

they trade. One active area in this literature is focused on market efficiency and

mispricing, and recent studies highlight the positive role played by hedge funds as “smart

money”investors that follow similar strategies to exploit mispricing. For example, Sias,

Turtle, and Zykaj (2016) study hedge fund holdings and document that overlap among

portfolios has increased substantially over time because of the entry of new funds, and

that consistent with superior information or skill, aggregate demand for a stock by hedge

funds is positively correlated with future returns. Cao et al. (2017) also study changes in

the quarterly holdings of hedge funds and find that various measures of the informational

efficiency of stocks they purchase increase but that this effect is much weaker for changes

in mutual fund holdings. Both Akbas et al. (2015) and Kokkonen and Suominen (2015)

further document that hedge fund trading uniquely improves equity pricing efficiency.

Using the profitability of trading strategies based on common stock market anomalies as

a signal of inefficiency, Akbas et al. find that positive flows to hedge funds improve

efficiency, but positive flows to mutual funds actually hurt efficiency. They conclude that

mutual funds at least reinforce if not create some anomalies through their tendency to

purchase more of an existing holding when faced with inflows. Similarly, Kokkonen and

Suominen find that trading by hedge funds reduces a measure of misvaluation derived

from the residual income model and that trading by mutual funds does not have this

effect. The potential for crowded trades to affect pricing is also studied theoretically and

empirically by Jylha and Suominen (2011) in the context of foreign exchange markets.

Using data on 11 countries for a 30-year period, these authors find that because many

hedge funds employ a currency carry strategy, inflows to hedge funds affect interest rates

and exchange rates in a manner consistent with the predictions of their general

equilibrium model. Pojarliev and Levich (2011) also study currency managers and

propose a way to measure the degree of crowding in different strategies. Their results

indicate that high returns in the carry strategy attract new funds and increase crowding,

and that future returns on the carry strategy are inversely related to the current degree of

crowding. These articles provide ample evidence that the tendency for groups of

investors to follow similar investment strategies can affect the risk and the returns of the

securities they trade.

A novel feature of our study is a focus on the liquidity dimension associated with

potentially crowded strategies. Previous studies have concentrated on returns, and the

primary question of interest has been whether active investors have driven expected

profits to zero as increasingly aggressive arbitrage corrects mispricing. Examples of

studies focused specifically on the profitability of potentially crowded equity trades

include Gustafson and Halper (2010) and Cahan and Luo (2013), and the evidence they

provide suggests that the factors they study are not significantly crowded in this sense.

Further evidence comes from Verbeek and Wang (2013), who demonstrate that using

quarterly disclosures mandated by the U.S. Securities and Exchange Commission (SEC)

to mimic the holdings of active mutual funds provides the same performance as the target

funds, and hence the strategies appear to have excess capacity. Rather than focusing on

148 The Journal of Financial Research

profitability, we examine how the liquidities of individual stocks may be affected by

investors with overlapping models and positions. The importance of this issue is growing

both because there is a heightened focus on liquidity risk management and because the

tendency to use overlapping models has greatly increased in recent years. There has been

a proliferation of “factor”-based investment products during the past 10 years, and the

growth in assets under management connected to such products has been high and is

accelerating.

1

These products exhibit substantial similarity in the stock characteristics

they employ, and the potential for them to increase risk grows as they become more

popular. In this article, we attempt to document this risk.

To capture correlated trading, we use a multifactor model based on conventional

empirical asset pricing factors to generate the types of signals that many active portfolio

managers and factor-based investment strategies employ when selecting and weighting

stocks.

2

These factor strategies link cross-sectional variation in returns to stock

characteristics, and the positive average returns typically documented for them are

consistent with conclusions in Daniel and Titman (1997) and Kelly, Pruitt, and Su

(2018). Throughout this article, we use the term “multifactor model”to refer to an

empirically motivated cross-sectional model of returns, rather than a theoretical

equilibrium asset pricing model based on expected comovement of returns with

systematic factors (e.g., the arbitrage pricing theory of Ross, 1976, or time-series models

widely used for risk adjustment such as the four-factor model of Carhart, 1997). We find

that a strong signal to buy or sell a stock from a representative multifactor model is

associated with a change in trade activity, a persistent order imbalance, lower return

volatility, a lower level of stock-specific liquidity, and greater comovement of stock

liquidity within the set of stocks favored by the model. Our empirical results complement

those of Kamara, Lou, and Sadka (2008), who tie correlated trading more broadly to firm

size based on the notion that large institutional investors cannot feasibly trade small,

illiquid stocks. The multifactor model used in our article provides more specific signals

of the individual stocks that many investors may trade and is particularly relevant as

factor-based investment strategies grow in popularity.

Our interpretation of the empirical results we provide is intuitive and draws on

the literatures studying investor disagreement and models of market making. The use of

overlapping multifactor models essentially reduces disagreement among investors

because these models tend to process the same information in the same way at the same

time. Prior empirical and theoretical work focused on investor disagreement has shown

that greater disagreement leads to greater return volatility (e.g., Ajinkya, Atiase, and Gift

1991; Shalen 1993; Banerjee 2011; Carlin, Longstaff, and Matoba 2014), and therefore a

reduced level of disagreement following a signal could be expected to generate lower

volatility as documented here. Models of investor disagreement help explain the peculiar

1

In May 2012, BlackRock (a major provider of factor exchange-traded funds [ETFs]) projected that assets

invested in factor-based strategies would reach $1 trillion by 2020 and $2.4 trillion by 2025. Wessel (2015) is one of

many recent articles highlighting the growth of factor investing and smart beta strategies at both the institutional

and retail levels.

2

The use of a multifactor model to generate signals of misvaluation on which to trade is similar to the

approaches of Stambaugh, Yu, and Yuan (2012) and Akbas et al. (2015).

Factor Crowding 149