Measuring Innovation and Product Differentiation: Evidence from Mutual Funds

• Author: LEONARD KOSTOVETSKY,JEROLD B. WARNER
• DOI: http://doi.org/10.1111/jofi.12853
• Date: 01 April 2020
• Published date: 01 April 2020

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 2 •APRIL 2020

Measuring Innovation and Product

Differentiation: Evidence from Mutual Funds

LEONARD KOSTOVETSKY and JEROLD B. WARNER∗

ABSTRACT

Westudy innovation and product differentiation using a uniqueness measure based on

textual analysis of prospectuses. We find that small and start-up families have higher

start rates than larger families, and their products are more unique. Unique strategies

attract more inflows in the first three years, and investors respond more to text-

based uniqueness than other measures such as holdings or returns uniqueness. For

established funds, word uniqueness has weak negative power for explaining returns,

so investors in competitive equilibrium do not sacrifice much performance to get

specialized products. Uniqueness attenuates the flow-performance relation, reducing

the risk of investor outflows.

IN THIS PAPER,WE STUDY ENTRY, innovation, and product differentiation by U.S.

actively managed equity mutual funds. From 1975 to 2016, the number of

mutual funds, excluding money market funds, rose from 530 with $45 billion

in assets to 7,736 with $13.4 trillion in assets. The number of fund families

(i.e., fund sponsors) increased from 297 to 824 during the same period, and

the number of Morningstar style classifications rose from fewer than 20 to

over 100, nearly tripling in the past 20 years. The proliferation of funds and

fund styles raises important questions about how existing fund families and

potential sponsors of new fund families make product introduction decisions,

and the consequences of such decisions for investor flows. While these questions

have received attention in prior literature (see, for example, Khorana and

Servaes (1999,2012), Massa (2000,2003), Caffrey (2006), Wahal and Wang

(2011), Hoberg, Kumar, and Prabhala (2018)), many issues remain unexplored.

We first focus on how to measure innovation and product differentiation.

Next, we examine the causes and consequences of differentiation, both for new

funds and for established funds. The power of our approach comes in part from

the use of text-based measures. The motivation for using such measures is

strong. For any given fund, the correlation of its securities’ measurable char-

acteristics (e.g., size, book to market, returns) with other funds, particularly

∗Leonard Kostovetsky is at the Carroll School of Management, Boston College. Jerold Warner is

with the Simon Business School, University of Rochester.We thank Dan Burnside, Ronald Goettler,

Ron Kaniel, Fabio Moneta, Jonathan Reuter,participants at the Boston College seminar and Telfer

Conference, as well as Editor Stefan Nagel, an anonymous associate editor, and two referees for

their helpful comments. The authors do not have any potential conflicts of interest to disclose, as

identified in The Journal of Finance Disclosure Policy.

DOI: 10.1111/jofi.12853

C2019 the American Finance Association

779

780 The Journal of Finance R

those of the same style, is likely to be high, in which case differences in stan-

dard characteristics will not fully capture the relevant degree of “uniqueness.”

If fund sponsors want to create a fund that is viewed as “new” and “differ-

ent,” they could use the prospectus, in particular, the written description of the

fund, to differentiate their product and reduce search costs for clienteles with

different tastes for fund attributes.

To illustrate, consider the Patriot Fund, which opened in 2012. The Pa-

triot Fund invests primarily in common stocks included in the S&P 500 index

that pass the fund’s “patriotic” investment screen, which attempts to weed out

companies that do business with terror-sponsoring nations as defined by the

U.S. Department of State. Based on standard financial measures such as hold-

ings’ characteristics, this fund looks like (and is classified by Morningstar as)

a Large Cap Blend fund. However, the text-based uniqueness of the Patriot

Fund is more than two standard deviations higher than that of a typical Large

Cap Blend fund. This high level of uniqueness makes sense, and suggests that

an investor would be unable to replicate the fund’s investment strategy with

any other existing mutual fund.

The paper proceeds in three steps. First, we develop a text-based measure

of a fund’s uniqueness and compare it to holdings and return-based measures.

Validation tests show that the text-based measure contains information that

predicts some aspects of fund strategy better than other measures, which gen-

eralizes the intuition provided by the Patriot Fund example. We thus use text-

based and other differentiation measures to address a broad set of questions.

Second, we study new funds. Previous papers typically define innovation as

a fund start (e.g., Sirri and Tufano (1993)), and use the terms “innovation” and

“product differentiation” interchangeably. While there is a positive correlation

between the number of new funds introduced and family size, we document

that the rate at which large families offer new funds (e.g., the percentage of

their funds that are new funds) is lower than for other families, including new

families. The start rate for families in the lowest (first) size quintile is 168%

higher than that for families in the highest (fifth) quintile. On average, the

family size of a new fund offering is 41.3% ($47.5 billion) smaller than that of an

existing fund in the same month. Furthermore, since there are more small than

large families, this size distribution reinforces the view that innovation comes

from smaller (and new) families, which runs counter to previous literature (e.g.,

Khorana and Servaes (1999)).

Our text-based measure uses ex ante reported strategies available for in-

vestors prior to the fund’s inception. We find that smaller families and new

families not only offer more new products, but offer more unique products than

larger existing families. These findings make sense because we do not expect

small families to be able to compete based on economies of scale or brand

recognition. Small and new fund families charge more (i.e., have higher ex-

pense ratios) on their more innovative new funds. For a fund offered by a new

family, a one-standard-deviation shock to text-based uniqueness is associated

with an approximately 12 basis point (bp) increase in annual expense ratio.

Large families charge less (have lower expense ratios) on unique funds than

Measuring Innovation and Product Differentiation 781

small families. However, firms (large and small) earn higher total dollar fees

on unique funds after the inception date.

More innovative offerings attract more assets under management in the first

few years of operation. Funds with above-median uniqueness attract about 50%

(approximately $75 million) more in asset flows than less innovative (below-

median uniqueness) new funds in the first 36 months of operation. Text-based

uniqueness is a significant independent predictor of fund flows for recently

opened funds, even after using other uniqueness measures and controlling

for expenses, which include marketing expenses. These results indicate that

investors pay attention to the uniqueness of the underlying product revealed

in the text of the prospectus.

The third part of our study focuses on established funds. It seems unlikely

that a unique fund can attract new money forever in a competitive market.

For example, competing products may come along, or there might be a limited

market for highly innovative strategies. We examine product uniqueness in

an industry equilibrium setting, reflecting in part the intuition of Berk and

Green (2004). In their model, flow follows alpha and net alpha is driven to

zero because of increasing returns to scale. In our setting, investors care about

net alpha as well as uniqueness. However, with clientele effects and a limited

market for a unique strategy, existing unique funds can survive even if they

do not generate alpha. Furthermore, some investors may be willing to sacrifice

net performance to get characteristics they value.

Our empirical tests study the relation between flows, uniqueness, and alpha

for established funds. Fund returns show only a weak negative relation with

text-based uniqueness. Overall, offering a fund that uses a unique strategy

with popular appeal allows a firm to initially collect rents from its first-mover

advantage, but competition eliminates such rents fairly quickly.

We also find that unique products have significantly lower flow-performance

sensitivity than other funds. One possible explanation for this finding is that

due to less product competition for more unique funds, it is more difficult for in-

vestors to react to underperformance by liquidating their holdings and putting

their money in a comparable product offered by another firm. If there is a lim-

ited market for highly innovative strategies, outperformance will not attract

as many new investors. An alternative explanation is that more unique prod-

ucts are harder to benchmark. Ferson and Lin (2014) note that client-specific

alphas can be very different from the alphas used in academic research due

to incomplete markets and investor disagreement. In any case, fund sponsors

gain an additional benefit from offering unique products because they are less

susceptible to the costs associated with flow shocks.

Our analysis of innovation and differentiation fills an important gap. We

provide new evidence from the asset management industry that small organi-

zations drive financial innovation. Frame and White (2004) find surprisingly

few studies on financial innovation, and only two that look at the conditions

that lead to financial innovation. Lerner (2006) tries to fill this gap by collecting

data on financial innovations from Wall Street Journal articles. Consistent with

our results, he finds that smaller and less profitable firms drive innovation to a