Learning about Mutual Fund Managers

• Published date: 01 December 2016
• Author: DARWIN CHOI,BIGE KAHRAMAN,ABHIROOP MUKHERJEE
• DOI: http://doi.org/10.1111/jofi.12405
• Date: 01 December 2016

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 6 •DECEMBER 2016

Learning about Mutual Fund Managers

DARWIN CHOI, BIGE KAHRAMAN, and ABHIROOP MUKHERJEE∗

ABSTRACT

We study capital allocations to managers with two mutual funds, and show that

investors learn about managers from their performance records. Flows into a fund

are predicted by the manager’s performance in his other fund, especially when he

outperforms and when signals from the other fund are more useful. In equilibrium,

capital should be allocated such that there is no cross-fund predictability.However, we

find positive predictability, particularly among underperforming funds. Our results

are consistent with incomplete learning: while investors move capital in the right

direction, they do not withdraw enough capital when the manager underperforms in

his other fund.

MUTUAL FUNDS ARE IMPORTANT investment vehicles for many households. While

previous studies show that investors infer funds’ ability to generate excess

future returns from past returns and allocate their capital accordingly (Sirri

and Tufano (1998), Huang, Wei, and Yan (2007,2012), Franzoni and Schmalz

(2016)), some attribute the performance-chasing behavior to behavioral biases

(Frazzini and Lamont (2008), Bailey,Kumar, and Ng (2010)).1In this paper, we

∗Darwin Choi is at CUHK Business School, Chinese University of Hong Kong, Shatin, Hong

Kong, Abhiroop Mukherjee is at Hong Kong University of Science and Technology, and Bige Kahra-

man is at Sa¨

ıd Business School, University of Oxford. We thank Kenneth Singleton (the Editor),

an Associate Editor, and two anonymous referees for many helpful suggestions. Weare also grate-

ful for comments received from Tim Adam, Vikas Agarwal, Nicholas Barberis, Jonathan Berk,

Utpal Bhattacharya, Lauren Cohen, Magnus Dahlquist, Francesco Franzoni, Mariassunta Gian-

netti, William Goetzmann, Luis Goncalves-Pinto, Jennifer Huang, Marcin Kacperczyk, Raymond

Kan, Dong Lou, Kasper Nielsen, Lubos Pastor, Jonathan Reuter, Mark Seasholes, Paolo Sodini,

Laura Starks, Per Stromberg, Mandy Tham, Heather Tookes, Michaela Verardo, Baolian Wang,

Mitch Warachka, Russ Wermers, Youchang Wu, Tong Yao, Hong Zhang, and Lu Zheng, as well

as from seminar participants at American Finance Association Annual Meeting 2014, Rothschild

Caesarea Center 11th Annual Conference 2014, China International Conference in Finance 2013,

Recent Advances in Mutual Fund Research 2013, Seventh Singapore International Conference on

Finance 2013, Auckland Finance Meeting 2012, HKUST Finance Symposium 2012, Seventh An-

nual Early Career Women in Finance Conference 2012, Curtin University,HKUST, London School

of Economics, Shanghai Advanced Institute of Finance, SIFR/Stockholm School of Economics, and

University of Western Australia. We acknowledge the General Research Fund of the Research

Grants Council of Hong Kong (Project Number: 640610) for financial support. We have read the

Journal of Finance’s disclosure policy and have no conflicts of interest to disclose. All errors are

our own.

1Elton, Gruber,and Busse (2004) and Choi, Laibson, and Madrian (2010) find that some mutual

fund investors are unable to make the right choice in the simplest possible context, and hence

question whether investors have the required level of sophistication.

DOI: 10.1111/jofi.12405

2809

2810 The Journal of Finance R

provide evidence that investors learn about mutual fund managers in a sophis-

ticated manner. Learning about managers is particularly value-relevant as re-

cent empirical research documents large differences among managers in terms

of skill.2Our paper studies managers who manage two mutual funds and exam-

ines whether investors learn about managerial ability from past performance

in the other fund managed by the same person. Moreover, we ask whether such

learning behavior is, as typically assumed in theoretical models, complete.Our

analysis contributes to the debate on the rationality of investors’ behavior.

We first extend Berk and Green’s (2004) model to a setting with two funds

per manager, and derive empirical tests for the flow-performance relationship

under fully rational and frictionless conditions. We find that flows do indeed

respond to the other fund’s past performance in the data, and in ways that are

consistent with our model predictions on learning. Instead of simply suggest-

ing that investors are learning rationally in a frictionless market, we take an

additional step and examine whether the response in flows is “sufficient.” Su-

perior past performance in one fund signals positive managerial ability.If flows

drive down fund performance due to decreasing returns to scale, sophisticated

investors should allocate more capital into the manager’s other fund, up to the

point that it earns zero expected returns in the future.3A similar argument

applies if one of the manager’s funds performs poorly. This null hypothesis of

no cross-fund predictability, which mirrors Berk and Green’s (2004) equilib-

rium, indicates sufficient allocation. However, if investors do not move enough

capital into and out of a fund given the other fund’s performance, there would

be positive predictability, while negative predictability could arise if investors

move too much capital in response to signals.

Our main results are summarized as follows. We find that flows into a fund

are predicted by past performance in both of the manager’s funds. In a linear

flow-performance regression, which includes past four-factor alphas of both

funds as independent variables, the sensitivity to the other fund’s performance

is about 17% of the sensitivity to the fund’s own performance. This result is not

explained by the fact that the two funds come from the same family. We control

for family fixed effects and the presence of a star fund, which can create spillover

flows to other funds in the family (Nanda, Wang, and Zheng (2004)). Moreover,

consistent with our model’s predictions, flows respond more to the other fund

and less to the fund itself when the two funds are more similar in style or

when the fund has more volatile returns. Flows also respond less when the

manager has been managing the funds longer. Taken together, these findings

2Skill seems to be related to managerial characteristics such as education (Chevalier and Ellison

(1999a)), past experience (Pool, Stoffman, and Yonker (2012), Kempf, Manconi, and Spalt (2014)),

and social networks (Cohen, Frazzini, and Malloy (2008), Pool, Stoffman, and Yonker(2015)). Using

novel measures of ability,several papers find that some fund managers are better than others (e.g.,

Kacperczyk and Seru (2007), Kacperczyk, Sialm, and Zheng (2008), Cremers and Petajisto (2009),

Baker et al. (2010)).

3As argued by Berk and Green (2004), Chen et al. (2004), and Yan (2008), fund size may erode

the performance because managers of larger funds spread their information-gathering activities

too thin, and large trades have higher price impact and execution costs.

Learning about Mutual Fund Managers 2811

suggest that investors draw inferences about managerial ability from the past

returns of both funds. Through a piecewise linear regression framework, we

further show that the effect of the other fund is more pronounced when its

performance has been exceptionally good.

From the performance predictability tests, we find positive cross-fund pre-

dictability, which indicates that investors do not respond enough to the

manager’s performance in his other fund. We sort all two-fund managers into

portfolios based on past performance in one of their funds. Managers’ future

performance in their other funds is examined, with various holding periods.

Our tests show that a manager’s past performance in one fund predicts future

performance in his other fund, a result that is also confirmed by using a double

sort or running a regression, both of which control for past performance in the

fund itself. Such predictability is unlikely to be due to price pressure, as it

does not reverse in the long run (in contrast to the own-fund case; Lou (2012))

and it is not driven by cases in which the two funds’ portfolios have a high

degree of overlap. Rather, it comes mostly from underperforming multifund

managers, which suggests that investors do not withdraw enough money from

a fund when the other fund underperforms. This finding is consistent with our

previous result that investor flows respond more to a manager’s performance

in his other fund when it is better.

Although other studies have examined the flow-performance relationship

and return predictability in mutual funds, using two funds from the same

manager offers some unique advantages. First, we provide new evidence on the

performance-chasing behavior, which may be rational or related to behavioral

biases. As predicted by our model, we find that flows respond more to past

performance in the other fund when the signal is more precise and relevant,

indicating that investors learn in a sophisticated manner. Second, we are able

to identify investor learning at the manager level. Previous studies conducting

fund-level analyses typically cannot distinguish between information about the

fund and its management. Our paper contains two sets of “placebo” samples

to single out the effect of managers. In one sample, we use the same two

funds in a period when they are managed by different managers, while in the

other sample, we replace one of the manager’s funds with another fund in the

same fund family or with a fund that has similar characteristics but is not

managed by the same manager. Our main results do not obtain using these

placebos. Third, our research design allows us to control for the impact of flow-

driven price pressure on the performance persistence and explore the role of

investor learning. There is some return persistence in mutual funds (Carhart

(1997)), but own-fund return persistence is partly attributed to price pressure

arising from fund flows.4Funds facing outflows liquidate their positions that

4Carhart (1997) documents some persistence in the performance, especially among underper-

forming funds, but the driving forces are not well understood. Toexplain the continued investment

in poor performers, some authors focus on the role of biases in investor information sets or search

frictions (e.g., Gruber (1996), Goetzmann and Peles (1997)), while an alternative view relates per-

sistence in poor performance to flow-driven price pressure (e.g., Lou (2012)). Prior research does

not provide conclusive evidence.