Effect of high‐frequency trading on mutual fund performance

• Published date: 01 May 2023
• Author: Nan Qin,Vijay Singal
• Date: 01 May 2023
• DOI: http://doi.org/10.1111/fire.12331

DOI: 10.1111/fire.12331

ORIGINAL ARTICLE

Effect of high-frequency trading on mutual fund

performance

Nan Qin1Vijay Singal2

1Department of Finance, Northern Illinois

University, DeKalb, IL, USA

2Department of Finance, PamplinCollege of

Business, Virginia Tech,Blacksburg, VA, USA

Correspondence

Nan Qin, Department of Finance, Northern

Illinois University,236G Barsema Hall, 740

Garden Road, DeKalb, IL 60115, USA.

Email: nanqin@niu.edu

Wethank Frank Hatheway from NASDAQ for

providing the high-frequency tradingdata.

NASDAQmakes the data freely available to

academics providing a project description and

signing a nondisclosure agreement. We thank

Jonathan Brogaard, Ryan Riordan, Gjergji Cici,

and LeiZhou for comments and suggestions, as

well as conference and seminar participants at

the 2016 FMA annual meetings, 2019 SFA

annual meetings, Carleton University,

McMaster University,University of Windsor,

and Northern Illinois University.All errors are

our own. An appendix containing additional

tables, figures, and details of methodology is

availablein the supporting materials section

online.

Abstract

We find that high-frequency trading (HFT) in stocks held

by mutual funds negatively affects fund performance: when

sorted by HFT intensity of holdings, funds in the top quin-

tile underperform funds in the bottom quintile by 2.64%

per year. The negative relation can be at least partially

explained by the illiquidity premium induced by high-

frequency traders’ preference for more liquid stocks. This

reason for underperformance of mutual funds has not been

previously explored or documented. In addition, we do not

find evidence to support the concern that HFT raises trading

costs of mutual funds.

KEYWORDS

high-frequency trading, illiquidity premium, mutual fund perfor-

mance, trading costs

JEL CLASSIFICATION

G12, G14, G23

1INTRODUCTION

High-frequency trading (HFT) has been of interest to practitioners, policymakers,and academic researchers over the

past two decades. However,there is a lack of consensus about its impact on traditional institutional investors. Some

academic researchers find that HFT reduces trading costs for most market participants (Brogaard & Garriott, 2019;

Baldauf & Mollner, 2020; Boehmer,Fong, & Wu, 2021; Conrad, Wahal, & Xiang, 2015; Goldstein, Kumar, & Graves,

2014;Hendershott, Jones, & Menkveld, 2011; Hendershott & Riordan, 2013). On the other hand, many studies suggest

that HFT may increase the trading costs of traditional institutional investors by anticipating their trades(Biais, Fou-

cault, & Moinas, 2015; Bongaerts & Van Achter,2021;Hirschey,2021;Korajczyk & Murphy, 2019; Shkilko & Sokolov,

2020; Van Kervel& Menkveld, 2019; Yang & Zhu, 2020). A similar lack of agreement exists among practitioners, and

Financial Review. 2023;58:369–394. wileyonlinelibrary.com/journal/fire ©2022 The Eastern Finance Association. 369

370 QIN ANDSINGAL

some of them havetaken actions to protect themselves from HFT.1Giventhe importance of institutional investors and

financial markets, it is important to gain better insight into this topic.

Inthis study, we examine the impact of HFT on the performance of U.S. equity mutual funds that collectively manage

trillions of investment dollars. We approach the role of HFT from a different and previously unexplored channel by

investigating whether HFT affects cross-sectional stock returns through the illiquidity premium and whether mutual

funds holding HFT-intensive stocks are negatively affected as a consequence.

As suggested by prior studies, illiquid assets should provide a return premium—that is, illiquidity premium—over

liquid assets because investors require higher returns to compensate for higher trading costs associated with illiquid

assets, and the magnitude of the illiquidity return premium depends not only on the cost per trade but also on trading

frequency of the marginal investor (Amihud & Mendelson, 1986; Amihud et al., 2015; Chalmers & Kadlec, 1998;Chen

et al., 2020). Compared to traders with lower trading frequencies, traders with higher trading frequencies are more

sensitive to trading costs as they will incur higher tradingcosts from the same asset over the same period. The impli-

cation in the context of HFT is that high-frequency traders (HFTs)will prefer to trade liquid assets,2and the presence

of intensive HFT will amplify the magnitude of the illiquidity premium by lowering the risk-adjusted returns of liquid

assets and increasing those of illiquid assets.3

Based on this intuition, we hypothesize that a majority of HFT-intensive stocks are relatively liquid and will deliver

lower risk-adjusted returns, and therefore mutual funds holding such stocks will underperform other funds that do

not hold these stocks or hold them in a smaller proportion. For HFTs, the lower return of liquid assets is a reason-

able compromise for substantial savings from low trading costs. However,for most mutual funds that trade much less

frequently, the gains from the low trading costs of liquid assets can hardly offset the loss in returns. Therefore, HFT

intensity in liquid stockholdings of mutual funds will exerta downward pressure on returns and consequently lead to

underperformance of these funds.

Our empirical analyses are based on a sample of 3536 U.S. actively managed equity mutual funds and a sample

of 8150 U.S. common stocks during 2003–2018. Following Hendershott, Jones, and Menkveld (2011) and Conrad,

Wahal, and Xiang (2015), we use the number of quote messages, defined as the number of changes in the best bid

or offer quote or size across all quote reporting venues in the United States, as our HFT proxy.This proxy captures

one of the essential features of HFT—a very large number of order submissions and cancellations compared to that

of non-high-frequency traders (nHFTs).Moreover, it is based on publicly available Tradeand Quote (TAQ)data so that

it covers most U.S. common stocks since 1993. A comparison between quote messages and the “NASDAQ data set”

(120 stocks randomly selected by the NASDAQin 2008 and 2009) reported in Table A.1 in the Internet Appendixsup-

pinfo1confirms its effectiveness—quote messages indeed have a high correlation with actual HFT and significantly

lower correlations with actual non-high-frequency trading (nHFT). Tocontrol for the positive relation that arises nat-

urally between firm size and quote messages, we extractthe component of quote messages that is orthogonal to size

and use it as the measure of HFT in this paper.

We begin our empirical analysis by investigating the relation between risk-adjusted stock returns and HFT inten-

sity. Sorting stocks into quintile portfolios by HFT intensity, we find that the top quintile (highest HFT intensity)

significantly underperforms the bottom quintile by a risk-adjusted return of 6.12% per year. We then examine the

relation between fund performance and HFT intensity in fund holdings or trade portfolios and find it to be, again,

1Forexample, in recent years many traditional investors turned to dark pools such as IntelligentCross or exchanges such as Investors Exchange to combat

predatoryhigh-frequency traders (“Steven Cohen targets high-frequency trading with ‘dark pool’ venture,” Wall StreetJournal, April 17, 2018).

2More precisely,liquidity-taking HFTs should prefer to trade liquid assets to save trading costs, while liquidity-providing HFTsmay prefer to trade illiquid

assets where more revenue can be collected. However,as reported in Baron et al. (2019), liquidity-taking HFTs contribute much more HFT volume than

liquidity-providing HFTs.For example, in August 2012, 41% of all HFT volume on the E-mini S&P 500 futures market is generated by liquidity-taking high-

frequency traders, whereas 15% is generated by liquidity-providing high-frequency traders. The rest of the HFT volume is generatedby high-frequency

tradersusing both strategies.

3It is important to note that the change in the illiquidity premium is not necessarily related toa nychange in liquidity or trading costs. That is, even without

anychange in the level of liquidity, magnitude of the illiquidity premium could still be increased by higher trading frequency.