Star Ratings and the Incentives of Mutual Funds
| DOI | http://doi.org/10.1111/jofi.12888 |
| Date | 01 June 2020 |
| Author | CHONG HUANG,XI WENG,FEI LI |
| Published date | 01 June 2020 |
THE JOURNAL OF FINANCE •VOL. LXXV, NO. 3 •JUNE 2020
Star Ratings and the Incentives of Mutual Funds
CHONG HUANG, FEI LI, and XI WENG∗
ABSTRACT
We propose a theory of reputation to explain how investors rationally respond to
mutual fund star ratings. A fund’s performance is determined by its information
advantage, which can be acquired but decays stochastically. Investors form beliefs
about whether the fund is informed based on its past performance. We refer to such
beliefs as fund reputation, which determines fund flows. As performance changes
continuously, equilibrium fund reputation may take discrete values only and thus
can be labeled with stars. Star upgrades thus imply reputation jumps, leading to
discrete increases in flows and expected performance, although stars do not provide
new information.
MUTUAL FUND FLOWS ARE CLOSELY associated with fund star ratings. Analyzing
Morningstar ratings, Del Guercio and Tkac (2008) show that when a fund
is upgraded from the four-star group to the five-star group, on average, it
observes positive abnormal flows of 53% to 61%.1Reuter and Zitzewitz (2015)
further find that bottom five-star funds tend to receive significantly more
∗Chong Huang is at the Paul Merage School of Business, University of California, Irvine. Fei
Li is at the University of North Carolina, Chapel Hill. Xi Weng is at the Guanghua School of
Management, Peking University. We would like to thank Philip Bond (Editor), an anonymous
Associate Editor, and three anonymous referees for insightful comments and suggestions. We
also thank Jonathan Berk, Aislinn Bohren, David Brown, Shaun Davies, Vincent Glode, Richard
Green, Diane Del Guercio, Jie He, David Hirshleifer,George Mailath, Moritz Meyer-ter-Vehn, Uday
Rajan, Christopher Schwarz, Neal Stoughton, Zheng Sun, Yilin Wang,Lucy White, Youchang Wu,
Fernando Zapatero, Lu Zheng, and audiences at the UCI Finance Seminar,UNC Finance Seminar,
PKU Finance Seminar, JHU Micro Theory Seminar, CUHK Economic Theory Workshop, 2014
USC/UCLA/UCI Finance Day Conference, 2014 SITE Summer Seminar, First Summer School on
Financial Intermediation and Contracting, 2016 CICF,2017 AMES, 2019 Midwest Macroeconomics
Meetings, and 2019 Shanghai Microeconomics Workshop for comments and suggestions. Wethank
Tan Gan for his excellent research assistance. Chong Huang thanks the CORCLR Awards for
financial support. Xi Weng thanks the National Natural Science Foundation of China (Grant No.
71973002) and the Key Laboratory of Mathematical Economics and Quantitative Finance (Peking
University), Ministry of Education of China for financial support. The authors contributed equally
to this paper. The authors have read The Journal of Finance disclosure policy and have no conflict
of interest to disclose.
Correspondence: Chong Huang, Paul Merage School of Business, University of California, SB2
344, 4293, Pereira Dr, Irvine, CA 92697, U.S.A.; e-mail: chong.h@uci.edu
1The abnormal flow can be interpreted as a fund’s actual flow relative to its expected flow if it
had maintained its prechange star rating. Newer evidence on the effect of Morningstar ratings on
fund flows based on data from 2003 to 2014 is presented in Morningstar research. The observed
pattern is very similar.
DOI: 10.1111/jofi.12888
C2020 the American Finance Association
1715
1716 The Journal of Finance R
flows than top four-star funds, even if their previous investment outcomes are
similar. Two recent empirical studies, Evans and Sun (2018) and Ben-David
et al. (2019), also document the importance of Morningstar ratings for investor
decisions. We refer to the estimated effects of star ratings on mutual fund
flows star rating effects.
However, mutual fund star ratings do not produce new information. Star
ratings are coarse summaries of publicly available information. For example,
the Morningstar rating is a purely quantitative, backward-looking measure of
funds’ past performance, it employs only five stars to summarize a fund’s past
investment outcomes, and since its rating mechanism is publicly available, in-
vestors can calculate star ratings based on their own knowledge of funds’ previ-
ous investment outcomes. How, then, can star rating effects arise in a rational
economy? Put differently,why are rational investors’ investment decisions s en-
sitive to changes in star ratings when more precise information is available?
In this paper, we study these questions by providing a natural framework for
analyzing a fund’s dynamic incentives to acquire private information, which has
been shown to be crucial for an active mutual fund to outperform its peers.2Un-
like a fund manager’s ability, which is usually considered innate and unchang-
ing, information can be endogenously acquired and is potentially useful for
multiple periods, although it decays stochastically. Accordingly, our framework
features a fund that can acquire information at a cost if it is uninformed but that
may become uninformed in the next period if it is currently informed. Rational
investors cannot observe whether the fund is informed or whether it has ac-
quired information. They therefore form beliefs about whether the fund is cur-
rently informed based on its past investment outcomes and its perceived likeli-
hood of information acquisition. We refer to such beliefs as the fund’sreputation.
More specifically, we develop an infinite-horizon repeated game between a
monopoly mutual fund and a continuum of investors. In our model, the fund’s
current flow is determined by the fund’s current reputation and is independent
of the fund’s information status. Hence, an uninformed fund cannot obtain any
extra flows in the current period by acquiring information. Therefore, as Berk
(2005) points out, the fund’s incentives to acquire information arise only from a
potentially higher future reputation, which can generate larger future flows. On
the one hand, when informed, the fund is more likely to realize a higher invest-
ment outcome, in which case the fund has a better reputation in the next period.
On the other hand, by acquiring information in the current period, the fund may
remain informed in the next period. Both effects will lead to larger future flows.
Therefore, when uninformed, the fund will acquire information if and only if
the incremental discounted future management fees, which we refer to as the
information premium, can compensate for the information acquisition cost.
Our main contribution is to highlight an equilibrium phenomenon whereby
the fund’s reputation only takes values from a discrete set, despite the fact that
2For example, Phillips, Pukthuanthong, and Rau (2014) document evidence demonstrating that
the source of superior fund managers’ skill lies in the private information that these managers
obtain.
Star Ratings and the Incentives of Mutual Funds 1717
past investment outcomes are continuously distributed. Hence, if we label these
discrete values with stars, a change in the star assigned to the fund represents
a discrete change in its reputation, which leads to a jump in its flows. We refer
to the equilibrium property of discrete reputations as the star rating property.
To illustrate the star rating property, we first analytically characterize a
set of equilibria in which (i) the information premium is always the same
as the information acquisition cost, and hence the fund always randomizes
between acquiring information and remaining uninformed; and (ii) the fund’s
equilibrium reputation can take one of only two possible values.
In such an equilibrium, even though fund past performance is continuously
distributed, funds are endogenously partitioned into just two types: “good”
and “bad.” Good funds are those that investors believe are likely to have
information, while bad funds are those that investors believe are less likely
to have information. Funds benefit from being perceived as good because they
can then attract more investors. Furthermore, there is heterogeneity among
funds in each category: uninformed funds with better past performance are
less likely to acquire information. This endogenous information acquisition
choice leads to the same reputation for funds in the same category. Thus, a
rating company that wants to accurately predict the fund’s performance can
group funds with the same reputation together and hence use only two stars
to capture the fund’s reputations.
The star rating property provides rational explanations for several empirical
observations. First, the promotion of a fund to a higher rated group will
naturally attract more investors. This is the star rating effect. Moreover,
because in our model, funds with the same rating have the same reputation,
improvements in a fund’s performance and hence in its within-group rank
cannot result in significant abnormal cash flows if its star rating does not
change. This result is consistent with the findings documented by Del Guercio
and Tkac (2008). Specifically,they show that abnormal flows are insignificantly
different from zero if a fund’s Morningstar percentile ranking increases but
its star rating does not change.
Second, the dynamic structure of our model helps analyze star ratings’
power to predict a fund’s future performance. In equilibrium, higher rated
funds have a higher reputation and are more likely to be informed, and so
on average perform better. In other words, our theory suggests that the flow
jump caused by a star rating upgrade reflects investors rationally raising their
expectation of the fund’s future performance. This result is consistent with
empirical findings. For example, Blake and Morey (2000) show that low-rated
funds have poor future performance, and Reuter and Zitzewitz (2015)show
that on average the bottom five-star funds perform at least as well as the
top four-star funds, and strictly outperform them in some fund classes. In
addition, star ratings are imperfectly persistent: in equilibrium, high-rated
funds are more likely to receive a high rating in the next period, but once they
are downgraded, it is hard for them to regain a high rating.
Having illustrated the star rating property in the equilibrium with two
reputation values, we argue that the star rating property is a general property
To continue reading
Request your trial