Obfuscation in mutual funds

https://doi.org/10.1016/j.jacceco.2021.101429Get rights and content

Highlights

  • We find fund managers create unnecessarily complex disclosures and fee structures to keep investors uninformed.

  • We isolate manipulated complexity by examining S&P 500 index funds, which holds constant regulations, risks, and gross returns.

  • Using bespoke measures of complexity, we find evidence consistent with funds obfuscating high fees.

Abstract

Mutual funds hold 32% of the U.S. equity market and comprise 58% of retirement savings, yet retail investors consistently make poor choices when selecting funds. Theory suggests poor choices are partially due to fund managers creating unnecessarily complex disclosures and fee structures to keep investors uninformed and obfuscate poor performance. An empirical challenge in investigating this “strategic obfuscation” theory is isolating manipulated complexity from complexity arising from inherent differences across funds. We examine obfuscation among S&P 500 index funds, which have largely the same regulations, risks, and gross returns but charge widely different fees. Using bespoke measures of complexity designed for mutual funds, we find evidence consistent with funds attempting to obfuscate high fees. This study improves our understanding of why investors make poor mutual fund choices and how price dispersion persists among homogeneous index funds. We also discuss insights for mutual fund regulation and academic literature on corporate disclosures.

Introduction

Over 9,000 mutual funds, holding $21.3 trillion in assets, were traded on U.S. exchanges during 2019. Mutual funds hold 32% of the total U.S. equity market value and comprise 58% of retirement savings (Investment Company Institute 2020). Despite the popularity of mutual funds, many studies find that they underperform and that retail investors consistently make poor choices when selecting funds.1 Investor advocates argue that poor mutual fund choices are due in part to complex disclosures and fee structures that make it difficult to understand and compare funds, and that unnecessary complexity persists despite decades of regulatory efforts (e.g., SEC 1998; 2009; 2014; 2018; 2020). Theory suggests that complexity persists because it is part of a strategy to obfuscate unfavorable information and extract rents from retail investors (Carlin 2009). Given the size of the mutual fund market, rent extraction could have significant implications for investor wealth. We empirically investigate whether mutual funds create unnecessarily complex disclosures and fee structures to obfuscate weak net performance.

An econometric challenge in investigating strategic obfuscation in mutual funds is controlling for variation in non-discretionary complexity caused by differences across funds. We mitigate this issue by examining S&P 500 index funds, which have largely homogeneous gross investment returns and risks but charge different fees so have heterogeneous net returns. For example, Schwab's S&P 500 fund charged 2 basis points (bps) in 2019 while Deutsche's charged up to 508 bps, despite earning nearly identical pre-expense returns (31.46% and 31.47%). Thus, S&P 500 index funds provide a setting to examine how disclosures and fee structures vary across funds with weaker versus stronger net performance (i.e., due to differences in fees), while holding constant many drivers of non-discretionary complexity.

Theory demonstrates why high-fee index funds are motivated to create unnecessarily complex disclosures and fee structures. Carlin (2009) models a competitive equilibrium in which complexity prevents investors from understanding and comparing fees across otherwise identical funds, thus enabling funds to charge excessive fees. Recognizing the benefits of having more uninformed investors, high-fee funds create unnecessary complexity to increase investors’ learning costs. An increase in learning costs endogenously increases the fraction of uninformed investors in the market, and allows price dispersion and rent extraction to persist.

The core intuition of Carlin (2009) is that high-fee funds want to shroud the market in complexity, such that investors find it difficult to learn from disclosures and make informed decisions. They instead invest randomly or, in practice, likely turn to advisors for assistance. Thus, a practical benefit of complexity is to complement funds’ marketing efforts, such as providing commissions to advisors for selling funds. Extensive prior evidence finds that advisors steer investors toward high-fee funds and charge excessive incremental fees for doing so (Wall Street Journal 2019a, 2019b; Elton et al., 2004; Bergstresser et al., 2009; Edelen et al., 2012).

We investigate two potential methods that high-fee funds could use to increase complexity and keep investors uninformed (Carlin 2009). The first is to increase “narrative complexity” by using unnecessarily bad writing to make disclosures less readable. The SEC has repeatedly expressed concerns about the narrative complexity of mutual fund disclosures, but to date it has received little academic attention. The second method is to increase “structural complexity” by creating complex intra-fund structures and fee schedules that make it hard for investors to compare funds and identify the fees they must pay. While structural complexity has been the focus of prior research (see Section 2), we also investigate structural complexity for completeness because it affects some proxies for narrative complexity.

Our empirical predictions are that high-fee funds have greater narrative complexity and greater structural complexity. As both fees and complexity are choice variables, these predictions are not causal. Instead, as depicted in Fig. 1 and modeled in Carlin (2009), the association between fees and complexity is a joint outcome of funds’ strategic choice. Sources of tension in our predictions are that S&P 500 funds are simple and that mutual fund disclosures are heavily regulated, so strategic obfuscation is plausibly unrealistic.

After imposing data requirements, our sample spans 1994–2017 and includes 38 S&P 500 index mutual funds. We exclude fund classes that are only available to or through institutions (e.g., as part of a retirement plan), such that all classes in our sample are available to self-directed retail investors. Our sample comprises $463B in assets under management in 2017, which is 47% of the total S&P 500 index mutual fund market in 2017 (Investment Company Institute 2018). Our sample funds have an average tracking error of only 3.4 bps, consistent with findings that S&P 500 funds very closely mirror the underlying index (Elton et al., 2019).2

We measure funds’ total fees including annual fees and amortized one-time charges (Fees). The within-year standard deviation of Fees is 51 bps. This variation in fees is economically meaningful; e.g., our data indicate that retail investors paid an extra $358M in 2017 alone by holding high-fee versions of S&P 500 index funds.

We examine narrative complexity within funds’ prospectuses, and especially within summary prospectuses, which research finds are a common source of information for retail investors (see Section 2.3). Further, we show that brokerage websites copy text directly from summary prospectuses, so investors can be affected by prospectus readability even if they do not read the document itself. We develop two custom measures of readability based on guidance from practitioners and the SEC. FundsinFiling is the number of unique funds that managers include in a single prospectus (e.g., S&P 500 fund, Russell 3000 fund, etc.). Repetition measures the degree to which the prospectus summary section exactly repeats language from the details section. We also use two standard narrative complexity proxies, Length and WordsPerSentence, measured for both the entire prospectus and just the summary expense disclosure.

We measure structural complexity based on the fund's number of share classes and types and tiers of fees, combined into a principal component Structural_Complexity.

Our analyses are based on OLS and robust regressions with year fixed effects to eliminate common temporal variation in index returns, risks, regulations, and other non-discretionary fund characteristics. Consistent with strategic obfuscation, we find strong and positive associations between fees and multiple measures of both narrative complexity and structural complexity. We also investigate a number of alternative explanations for our findings.

First, it is possible that funds obfuscate using only structural complexity, and the associations between fees and narrative complexity are simply a byproduct of fund structure choices. Two of our narrative complexity proxies, FundsinFiling and WordsPerSentence, are designed to be free of this confound. To further investigate this alternative explanation, we identify specific disclosures in the summary prospectus that are unaffected by differences in structural complexity: fund objective and equity risk descriptions. We find positive associations between Fees and narrative complexity just within objective and equity risk descriptions, indicating that structural complexity is not the sole driver of narrative complexity.

Second, perhaps high fees are justified because their issuers offer a wider variety of funds and services, and narrative and structural complexity are unavoidable outcomes of issuer offerings. Inconsistent with this explanation, we find that high-fee fund issuers offer fewer other funds and a similar variety of services as do low-fee fund issuers (e.g., trading platforms, mortgages, and insurance). Relatedly, we find that high-fee funds are more expensive regardless of an investor's holding period, which is inconsistent with the explanation that multi-class funds are used to cater to different horizon preferences.

Third, it is possible that narrative complexity is a byproduct of high-fee funds’ efforts to reduce litigation risk. Section 6.5 explains that disclosure regulations were purposefully designed to allow funds to mitigate incomplete disclosure litigation risk without conflicting with plain English rules. Specifically, the multi-part structure of disclosures allows summary sections to be kept concise and readable, while the prospectus details and Statements of Additional Information contain extensive information. We also find associations between fees and narrative complexity in specific disclosures that should be unaffected by litigation risk, again indicating that litigation risk is not the sole driver of narrative complexity.

Fourth, to further investigate whether narrative complexity is a non-discretionary byproduct of an omitted variable or irrelevant to investors, we examine the effects of two 2009 SEC regulations that were designed to reduce narrative complexity. We find that funds with the most narratively complex disclosures pre-2009 reduce their narrative complexity more than other funds after the regulations became effective, which suggests that prospectus narrative complexity is at least partially discretionary. We also find that funds with the most complex disclosures pre-2009 are more likely to lose market share or close post-2009, which supports the underlying assumption that complexity affects investment decisions.

Finally, we investigate whether one motivation for obfuscation is that uninformed investors are more susceptible to high-fee funds' marketing and advisors. As expected, we find positive associations between marketing efforts and both narrative and structural complexity. We also continue to find an association between complexity and high fees in disclosures that are unrelated to marketing, and after excluding marketing charges and service fees. Thus, complexity appears to be a strategy that complements high-fee funds’ marketing efforts, but high fees do not solely compensate for fund advisors and services.

Across numerous other robustness tests, we find positive associations between fees and complexity that are inconsistent with plausible alternative explanations. We conclude that, consistent with theory and concerns from practitioners and the SEC, our findings are most consistent with the inference that high-fee index funds use narrative and structural complexity as part of an obfuscation strategy. We find similar results among the portfolios of funds offered by our sample parent companies, indicating that these findings likely generalize to the broader mutual fund market. That said, Section 6 discusses limitations for readers' consideration. We also note that obfuscation is unlikely to be funds' only strategy for extracting rents, and encourage future research to devise tests to estimate the impact of obfuscation on investors' choices in isolation from complementary strategies. Finally, our findings do not necessarily mean that managers consciously create complexity to obfuscate high fees. Alchian (1950) argues that strategic behaviors can evolve unintentionally through experimentation and mimicry. Still, even if managers do not understand the effects of complexity, our findings indicate that high-fee funds have not embraced the SEC's efforts to reduce complexity in fund disclosures.

Our primary contribution is to investigate whether mutual funds manipulate narrative complexity to obfuscate weak performance.3 Although fund issuers argue that variation in prospectus readability is driven by innate factors, our results indicate that unreadable disclosures are part of a discretionary strategy to extract rents from retail investors. These findings add to a broad literature investigating why retail investors make poor mutual fund choices. Our findings should also be of interest to the SEC as it develops and enforces regulations to improve fund disclosures (SEC 2020). Section 7.1 further discusses potential regulatory implications.

Our second contribution is to build on research examining funds' strategic use of structural complexity (e.g., Gabaix and Laibson 2006; Edelen et al., 2012; Adams et al., 2012; Badoer et al., 2020). Most of the existing literature, discussed in Section 2, examines heterogeneous active funds, and examines specific aspects of structural complexity in isolation. Our approach investigates structural complexity within homogeneous index funds and examines multiple aspects of complexity together. Thus, our results bolster prior findings and provide a fuller view of funds’ obfuscation strategies.

Finally, our paper has implications for the corporate disclosure literature, which we discuss in Section 7.2.

Section snippets

Mutual fund classes and fee structures

Fig. 2 illustrates how mutual funds are structured. Funds are created and sold by institutions such as Schwab. A single fund can be subdivided into share classes, which can have different combinations of fees and may be available only to certain types of investors (e.g., institutions). All classes share the same asset pool so have the same gross returns and risks. Subdividing a fund into classes is discretionary, and many funds have only one class.

Fund classes can have a variety of fees. Most

Sample and variable construction

We briefly discuss our sample assembly here and provide details in our Online Appendix (OA 2). Appendix A has further details on variable definitions.

Summary statistics and descriptive information

Table 1 Panel A provides summary statistics. The rightmost column tabulates the residual standard deviation after each variable is orthogonalized to year fixed effects (deHaan 2020). We highlight a few details. First, there is considerable variation in Fees, as the interquartile range is 20–115 bps.12

Primary analyses

As depicted in Fig. 1, Carlin (2009) predicts that fund fees and complexity are simultaneous outcomes of the manager's choice of fund strategy. Thus, complexity does not cause high fees or vice versa. As we cannot observe managers' strategic choices, we cannot perform typical regressions in which outcome variables Y (in our case, high fees and complexity) are regressed on independent variables X (the manager's strategic choice). Instead, our empirical strategy is to test whether the two outcome

Additional analyses and discussion

This section investigates whether our results likely generalize to the broader mutual fund market, and summarizes tests of additional alternative explanations for our main results.

Summary and concluding discussion

We examine “strategic obfuscation” in mutual funds; i.e., whether fund managers attempt to obfuscate unfavorable information via unnecessarily complex disclosures and fee structures. Our tests examine homogeneous S&P 500 index funds and use a within-year research design to hold constant many non-discretionary drivers of complexity. Consistent with theory in Carlin (2009), we find that funds with higher fees have greater narrative complexity (i.e., less readable disclosures) and structural

References (75)

  • F. Li

    Annual report readability, current earnings, and earnings persistence

    J. Account. Econ.

    (2008)
  • K. Lo et al.

    Earnings management and annual report readability

    J. Account. Econ.

    (2017)
  • V. Nanda et al.

    The ABCs of mutual funds: on the introduction of multiple share classes

    J. Financ. Intermediation

    (2009)
  • D. Solomon et al.

    Winners in the spotlight: media coverage of fund holdings as a driver of flows

    J. Financ. Econ.

    (2014)
  • V. Agarwal et al.

    Mandatory portfolio disclosure, stock liquidity, and mutual fund performance

    J. Finance

    (2015)
  • A.A. Alchian

    Uncertainty, evolution, and economic theory

    J. Polit. Econ.

    (1950)
  • S. Anagol et al.

    The impact of shrouded fees: evidence from a natural experiment in the Indian mutual funds market

    Am. Econ. Rev.

    (2012)
  • M. Armstrong et al.

    Consumer protection and contingent charges

    J. Econ. Lit.

    (2012)
  • B. Barber et al.

    Out of sight, out of mind: the effects of expenses on mutual fund flows

    J. Bus.

    (2005)
  • S. Benartzi et al.

    Naive diversification strategies in defined contribution saving plans

    Am. Econ. Rev.

    (2001)
  • D. Bergstresser et al.

    Assessing the costs and benefits of brokers in the mutual fund industry

    Rev. Financ. Stud.

    (2009)
  • J. Beshears et al.

    “How does simplified disclosure affect individuals' mutual fund choices?”

  • B. Blankespoor et al.

    “Disclosure processing costs, investors’ information choice, and equity market outcomes: a review

    Journal of Accounting and Economics

    (2020)
  • S.B. Bonsall IV et al.

    A plain English measure of financial reporting readability

    J. Account. Econ.

    (2017)
  • L.K. Chan et al.

    On mutual fund investment styles

    Rev. Financ. Stud.

    (2002)
  • H. Chen et al.

    “Don't take their word for it: the misclassification of bond mutual funds

  • J. Choi et al.

    Why does the law of one price fail? An experiment on index mutual funds

    Rev. Financ. Stud.

    (2010)
  • K.M. Cremers et al.

    How active is your fund manager? A new measure that predicts performance

    Rev. Financ. Stud.

    (2009)
  • Q. Curtis

    The Past and Present of Mutual Fund Fee Litigation under Section 36(b).” Research Handbook on the Regulation of Mutual Funds

    (2018)
  • Q. Curtis et al.

    An empirical study of mutual fund excessive fee litigation: do the merits matter?

    J. Law Econ. Organ.

    (2014)
  • H. Cronqvist

    Advertising and portfolio choice

  • E. deHaan

    Practical guidance on using and interpreting fixed effects models

  • D. Del Guercio et al.

    Mutual fund performance and the incentive to generate alpha

    J. Finance

    (2014)
  • E.J. Elton et al.

    Are investors rational? Choices among index funds

    J. Finance

    (2004)
  • R.B. Evans et al.

    “Institutional investors and mutual fund governance: evidence from retail–institutional fund twins

    Rev. Financ. Stud.

    (2012)
  • E.F. Fama et al.

    Luck versus skill in the cross-section of mutual fund returns

    J. Finance

    (2010)
  • S. Ferris et al.

    The effect of 12b-1 plans on mutual fund expense ratios

    J. Finance

    (1987)
  • Cited by (0)

    Thanks to the JAE editors, Beth Blankespoor, Matthias Breuer, Alex Coble, Quinn Curtis, Alper Darendeli, Roger Edelen (discussant), Zachary Kaplan, Mark Lang (Editor), Brian Miller (referee), Olivia Mitchell, Yini Wang, Rachel Zhang, Frank Zhou, several industry professionals, and workshop participants at the 2020 JAE Conference, Georgia Tech, Texas A&M, UT Austin, Washington, Wharton, the 2019 Miami Winter Conference, and the 2020 FARS Midyear Meeting for helpful comments. We thank Dengsheng Chen, Adam Greene, Andri Hail, Douglas King, Shibao Liu, Claudia Peng, Jia Teo, and Jun Wu for data assistance.

    View full text