What Drives ETF Flows?

• Date: 01 August 2014
• DOI: http://doi.org/10.1111/fire.12049
• Published date: 01 August 2014

The Financial Review 49 (2014) 619–642

What Drives ETF Flows?

Christopher P. Clifford

University of Kentucky

Jon A. Fulkerson∗

Loyola University Maryland

Bradford D. Jordan

University of Kentucky

Abstract

The value of exchange traded fund (ETF) assets has increased from $66 billion in 2000

to almost a trillion dollars in 2010. We use this massiveexpansion in ETF assets to study what

drives ETF flows. Using a data set of over 500 ETFs from 2001 to 2010, we show that ETF

investorschase returns in the same way as mutual fund investors. While there is an active debate

about whether return chasing by mutual fund investors represents the pursuit of superior talent,

the existence of return chasing in this passively managed environment should not represent a

search for skilled managers. We also show that ETF flows increase following high volume,

small spreads, and high price/net asset value ratios. Finally, we findlittle evidence of superior

market timing in ETF flows. Our results suggest that return chasing in both mutual funds and

ETFs is more likely the result of na¨

ıve extrapolation bias on the part of investors that has

contributed to the growth of the ETF industry.

Keywords: ETF, mutual fund, cash flow

JEL Classifications: G11, G12

∗Corresponding author: Sellinger School of Business and Management, Sellinger Hall, Loyola University

Maryland, 4501 N. Charles Street, Baltimore, MD 21210; Phone: (410) 617-2818; Fax: (410) 617-5035;

E-mail: jafulkerson@loyola.edu.

We thank Xin Hong and Monika Rabarison for valuableresearch assistance.

C2014 The Eastern Finance Association 619

620 C. P.Clifford et al./The Financial Review 49 (2014) 619–642

1. Introduction

Exchange traded funds (ETFs) are one of the fastest growing sectors of the

mutual fund industry, growing from $66 billion in assets in 2000 to almost a trillion

dollars by the end of 2010. By way of comparison, traditional equity mutual funds

increased in assets from $4 trillion to $5.7 trillion over the same period. The number

of equity ETFs grew from one in 1993 to 923 in 2010.1In early 2012, the SPDR

S&P 500 ETF, with assets of just under $100 billion, was bigger than most U.S.

equity mutual funds.

ETFs have several distinguishing features relative to mutual funds. Primarily,

shares of ETFs trade on exchanges while open-end mutual funds do not. Because

of this difference, ETF shares can (and do) sell for more or less than net asset

value (NAV). Further, because ETFs are exchange traded, we are able to observe,

and explore the effect of market-related variables such as trading volume, bid-ask

spreads, and historical returns.

Weuse a large sample of equity ETFs as a test environment to explore the growth

of ETFs over the past decade and to address return chasing as discussed in the mutual

fund literature. Like Elton, Gruber and Busse (2004) use a subset of S&P 500 index

mutual funds to examine investor choice, ETFs provide a unique laboratory for ad-

dressing cash flow behaviorbecause of the generally passive investment goals. Passive

investment means that the typical ETF tries to match—not beat—a particular market

index and precludes any purposeful intent to create alpha. By eliminating any search

for alpha, we can investigate how passive investors react to prior performance (and

other characteristics). If we are to find behaviorsimilar to mutual funds, it would imply

that investors have either consistently ignored or misevaluated active management.

We model ETF growthsimilar to cash flow studies in the mutual fund literature.

Our data include implied flows for more than 500 ETFs over the period 2001–

2010 and actual inflows and outflows for over 300 ETFs. Differences between ETF

structure and mutual fund structure mean these flows can haveadditional determinants

not present in mutual funds. However, as we discuss more fully later, ETF flow

has an essential relationship with individual investor behavior, despite part of this

relationship resulting from microstructure factors and the existence of “authorized

participants” (APs) as the sole intermediaries.

Using this large data set, we find that like mutual fund investors, ETF flow de-

creases with increases in fund size, expenses, and turnover. Our results are consistent

for both ETF gross inflows and outflows and when examining different styles of

ETFs (e.g., sector or foreign). Most of our results are driven by gross inflows and

are relatively consistent across the different ETF styles. Generally, the characteristics

shared by mutual funds and ETFs have the same qualitative influence on ETF flows

as the literature has previously demonstrated for mutual fund flows.

1All data come from the ICI Institute 2011 Factbook at http://www.ici.org/pdf/2011_factbook.pdf.