The market penalty for mutual fund scandals

Boston University Law Review

Volumn 87, Issue 5, 2007, Pages 1021-1057

  Choi, Stephen ^a   Kahan, Marcel ^a  

^a NEW YORK UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 38849087230     PISSN: 00068047     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Review

References (118)

1. INVESTMENT COMPANY INSTITUTE, 2005 INVESTMENT COMPANY FACTBOOK 59 tbl.1 (45th ed. 2005). For the remainder of the article, we will use the term "mutual fund" to refer to open-end funds.
2. Id.
3. BD. OF GOVERNORS OF THE FED. RESERVE SYS., FLOW OF FUNDS ACCOUNTS OF THE UNITED STATES 1965-1974, at 82 tbl.L.213 (2006), available at http://www.federalreserve.gov/releases/z1/20061207/annuals/a1965-1974. pdf.
4. BD. OF GOVERNORS OF THE FED. RESERVE SYS., FLOW OF FUNDS ACCOUNTS OF THE UNITED STATES 1995-2005, at 82 tbl.L.213 (2006), available at http://www.federalreserve.gov/releases/z1/20061207/annuals/al995-2005. pdf.
5. Id.
6. See, e.g., JANUS EQUITY FUNDS, PROSPECTUS 78-79 (2006) (describing the management structure of the fund).
7. Janus.com, Fund Comparison, https://ww4.janus.com/Janus/Retail/ FundHolding? fundID=1 (click on "Compare Funds" tab) (last visited Sept. 20, 2007).
8. JANUS EQUITY FUNDS, supra note 6, at 8.
9. Id. at 83.
10. See Google Finance, Janus Capital Group Inc., http://finance. google.com/finance? q=JNS (last visited Sept. 9, 2007). In our Article, we will use the term "mutual fund" to refer to open-end fund entities corresponding to the Janus Fund, "fund management company" to refer to entities corresponding to Janus Capital Management and Janus Capital Group, and "fund manager" to refer to persons like David Corkins.
11. See Michael C. Jensen & William H. Meckling, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure, 3 J. FIN. ECON. 305, 312-13 (1976) (discussing the divergence between the interests of a firm's shareholders and the firm's managers).
12. See generally Lucian Arye Bebchuk & Marcel Kahan, A Framework for Analyzing Legal Policy Towards Proxy Contests, 78 CAL. L. REV. 1071 (1990).
13. See generally Ronald J. Gilson, A Structural Approach to Corporations: The Case Against Defensive Tactics in Tender Offers, 33 STAN. L. REV. 819, 843-48 (1981).
14. See, e.g., Lucian A. Bebchuk, The Myth of the Shareholder Franchise, 93 VA. L. REV. 675, 682-94 (2007).
15. See Investment Company Act of 1940 § 5(a)(1), 15 U.S.C. § 80a-5(a)(1) (2000) (defining an "open-end company" as a management company with redeemable securities).
16. See Marcel Kahan & Edward B. Rock, Hedge Funds in Corporate Governance and Corporate Control, 155 U. PA. L. REV. 1021, 1051-52 (2007) (arguing that regulatory barriers cause most mutual funds to charge fees based on assets under management).
17. 17 C.F.R. § 270.22c-1(a) (2007) (requiring redemption price to be based on the current net asset value). Losses that have already materialized, however, will have reduced the fund's net asset value and cannot be recouped by redeeming shares. By contrast, shareholders in regular corporations cannot insulate themselves from expected future losses by selling their shares. As long as the market anticipates future losses, the market price will reflect these losses, and shareholders who sell will receive a lower price both as a result of past losses and as a result of expected future losses.
18. Redeeming shares from one mutual fund, deciding where to invest the proceeds, and investing these proceeds all involve transaction costs that may induce investors not to redeem such shares.
19. For example, capital gains from the redemption of mutual funds held in a qualified pension plan or an individual retirement account are not taxable. See I.R.C. §§ 401(a), 408, 408A (2000) (dealing with qualified pension plans, IRA's, and Roth IRA's).
20. See I.R.C. § 1001 (2000) (discussing taxation of capital gains).
21. See Financial Industry Regulatory Authority, Understanding Mutual Fund Classes, http://www.nasd.com/InvestorInformation/InvestorAlerts/ MutualFunds/UnderstandingMutualFundClasses/index.htm (last visited Sept. 25, 2007) ("Often Class C shares impose a small charge if you sell your shares within a short time of purchase, usually one year.").
22. See supra text accompanying notes 12-13 (discussing the election system of directors as a device to restrain agency costs).
23. See Lucian Ayre Bebchuk, The Case for Increasing Shareholder Power, 118 HARV. L. REV. 833, 901-03 (2005) (discussing giving shareholders the power to make scaling-down decisions, such as forcing dividend payments);
24. Zohar Goshen, Shareholder Dividend Options, 104 YALE L.J. 881, 884 (1995) (proposing to give shareholders the option to receive distribution of corporate profits in cash).
25. See Investment Company Governance, 69 Fed. Reg. 46,378, 46,381 (Aug. 2, 2004) (adopting a rule to mandate that investment funds must have a board composed of at least 75% independent directors and an independent chairman), invalidated by Chamber of Commerce v. SEC, 443 F.3d 890, 909 (D.C Cir, 2006);
26. Kara Scannell & Tom Lauricella, Cox's 'Independent' Day, WALL ST. J., Feb. 9, 2007 at Cl (discussing rationale and history of rule and prospects for the adoption of a revised rule on mutual fund governance).
27. This is the first article to analyze systematically the effect of scandals on mutual fund flows. Individual instances of funds experiencing outflows have been reported in the financial press in the context of market timing scandals. See, e.g., Brett Arends, On State Street, BOSTON HERALD, Dec. 28, 2004, at 23 (reporting significant outflows from Putnam, Janus, and other funds in wake of market timing scandals). Additionally, Professors Todd Houge and Jay Wellman have reported that funds in the same family as late trading and market timing funds suffer outflows following the announcement of an investigation.
28. Todd Houge & Jay Wellman, Fallout from the Mutual Fund Trading Scandals, 62 J. BUS. ETHICS 129, 133-34 (2005). Their study, however, does not control for other factors affecting fund flows; examine the effect of other types of scandals; or analyze the effect of particular scandal features - such as indicators of scandal severity, prominence, the likelihood of future harm and how the scandal was discovered - on fund flows as ours does.
29. See id. at 130.
30. See infra Part ILA.
31. See infra Part III.
32. See infra Part ILB.
33. See infra Part III.
34. See infra Part II.C.
35. See infra Part II.D.
36. See Robert McGough & Jeffrey Taylor, SEC Boosts Its Scrutiny of Magellan Fund, WALL ST. J., Dec. 11, 1995, at C1. The article does not specify whether the SEC had commenced a formal investigation, or whether its review was of a less formal type, such as an examination.
37. See id. Subsequent Wall Street Journal articles have, however, referred to the SEC conducting an examination.
38. See Jeffrey Taylor & Charles Gasparino, Why SEC Lambasted News Report on Fidelity, WALL ST. J., Apr. 22, 1996, at C1.
39. See McGough & Taylor, supra note 32.
40. Robert McGough, Fidelity Fund Managers Spoke Highly of 2 Other Stocks That Were Unloaded, WALL ST. J., Dec. 15, 1995, at C1.
41. Id.
42. Jeffrey Taylor, SEC Action is Unlikely on Vinik, WALL ST. J., May 9, 1996, at C1.
43. Karen Damato, Dreyfus Settles Charges Related to Disclosures, WALL ST. J., May 11, 2000, at C1.
44. Id. Dreyfus also agreed to pay $400,000 to the State of New York for the State Attorney General's investigation expenses and to make a $1.6 million contribution for an investor education program at the State University of New York. Id.
45. Id. While unequal IPO distributions are not illegal, the SEC argued that the practice should have been disclosed. Id.
46. Id.
47. Gregory Zuckerman & Ken Brown, Merrill Ousts 3 Fund Brokers, WALL ST. J., Oct. 6, 2003, at C1.
48. See 17 C.F.R. § 270.22c-1(a) (2007).
49. See Zuckerman & Brown, supra note 41.
50. Gregory Zuckerman, Alger Executive Becomes Focus of Fund Probe, WALL ST. J., Oct. 16, 2003, at C1.
51. John Shipman, Morningstar Sounds Alarm on Alger, WALL ST. J., Oct. 21, 2003, at D13.
52. Gregory Zuckerman, Executives on Trial: Former Officer at Fred Alger Sentenced to 1 to 3 years for Obstructing Spitzer, WALL ST. J., Dec. 18, 2003, at C14.
53. Id.
54. John Hechinger, MFS Allowed Timing Trades. SEC Believes, WALL ST. J., Dec. 9, 2003, at C1.
55. See, e.g., David Ward, Note, Protecting Mutual Funds from Market-Timing Profiteers: Forward Pricing International Fund Shares, 56 HASTINGS L.J. 585, 586 (2005). Market timing can also work for thinly-traded stock. Id.
56. See Hechinger, supra note 48.
57. See, e.g., John Hechinger & Tom Lauricella, MFS Inquiry Says Holders Lost Millions, WALL ST. J., Jan. 23, 2004, at C1;
58. John Hechinger & Tom Lauricella, MFS's Wunderkind CEO Ballen May Face SEC Temporary Ban, WALL ST. J., Jan. 29, 2004, at C1;
59. John Hechinger & Tom Lauricella, Sun Life Unit Reaches Pact in Fund Probe, WALL ST. J., Jan. 27, 2004, at C1;
60. Tom Lauricella et al., Developments in Mutual-Fund Cases, WALL ST. J., Jan. 15, 2004, at D7.
61. Frederick P. Gabriel Jr. & Bruce Kelly, Reverse Spin: MFS Not Too Proud to Settle with SEC, INVESTMENT NEWS, Feb. 9, 2004, at 4.
62. Id.
63. The term "load" is used to refer to the fee charged to purchase mutual fund shares.
64. The Strategic Insight's Fund Objective Code is a three-digit code which identifies the fund's investment strategy. For instance, "AGG" denotes an "aggressive growth" strategy fund.
65. See, e.g., Daniel C. Indro, Does Mutual Fund Flow Reflect Investor Sentiment?, 5 J. BEHAV. FIN. 105, 107 (2004) (stating that reinvested fund distributions should be excluded from funds flows because they represent shareholder liability to the IRS and not incoming assets);
66. Erik R. Sirri & Peter Tufano, Costly Search and Mutual Fund Flows, 53 J. FIN. 1589, 1594 (1998) (defining net flow as growth in excess of growth based on reinvested dividends).
67. A review of some of the outliers suggests that they are due to data recording errors. For example, in some instances, fund assets were recorded as dropping in a given month by a large percentage and then increasing to the former level in the following month.
68. See, e.g., Jayendu Patel et al., Investment Flows and Performance: Evidence from Mutual Funds, Cross-Border Investments, and New Issues, in JAPAN, EUROPE, AND INTERNATIONAL FINANCIAL MARKETS 51, 61-62 (Ryuzo Satl et al. eds., 1994) ("A return 1% above the cross-sectional mean return in the previous period implies a $200,000 increased flow in this period.");
69. Richard A. Ippolito, Consumer Reaction to Measures of Poor Quality: Evidence from the Mutual Fund Industry, 35 J.L. & ECON. 45, 61 (1992) ("For funds with a positive performance residual equal to 100 basis points over the past five years ... the current growth rate increases by .90 percent.");
70. Sirri & Tufano, supra note 56, at 1598 ("The results . ., confirm that equity mutual fund inflows are sensitive to historical performance, ...");
71. A. Edward Spitz, Mutual Fund Performance and Cash Inflows, 2 APPLIED ECON. 141, 144 (1970) ("[I]nflows in a given period were correlated with performance in the previous period ....").
72. See, e.g., Martin J. Gruber, Another Puzzle: The Growth in Actively Managed Mutual Funds, 51 J, FIN. 783, 799 tbl.5 (1996) (showing a stronger investor reaction to above-market returns then to below-market returns);
73. Ippolito, supra note 58, at 61-62 ("[F]unds that do better than the market experience a stronger response than those that do worse.");
74. Sirri & Tufano, supra note 56, at 1598 ("For top performers ... performance is associated with economically and statistically significant inflows.... In the lowest quintile (the poorest performers), there is virtually no relationship between historical performance and flows.").
75. See, e.g., Patel et al., supra note 58, at 60 n.16 (defining a risk measure as the ratio of the mean to the standard deviation of excess returns); Sirri & Tufano, supra note 56, at 1597 ("[R]isk . . . [is] measured by the standard deviation of monthly fund returns over the prior 12 months.").
76. See, e.g., Sirri & Tufano, supra note 56, at 1597 (including the log of total net assets in the preceding period as a control variable).
77. See Brad M. Barber et al., Out of Sight, Out of Mind: The Effects of Expenses on Mutual Fund Flows, 78 J. BUS. 2095, 2114 (2005) (finding that salience of expenses affects fund flows).
78. The coefficient varied from -0.013 to -0.018.
79. SCAN_MON12, for example, is equal to 1 if the flow-month is from the twelfth month after the initial Wall Street Journal announcement.
80. As an additional robustness check, we added a dummy for the month that a regulatory settlement or fine (if any) is first announced. The coefficient is equal to -0.007 (or -0.7%) and is significant at the 10% level.
81. To avoid endogeneity problems, we excluded from our count of articles all Wall Street Journal articles that made explicit reference to fund flows.
82. As a robustness check, we redefined BIG SCANDAL to include scandals that resulted in a settlement or fine that was greater than 1% of the total net assets of the fund. Using this alternate form of BIO SCANDAL we found that the coefficient on SCANDAL was equal to -0.016 (significant at the <1% level) and the coefficient on BIG SCANDAL was equal to -0.005 (significant at the 5% level), consistent with the hypothesis that more severe scandals result in a greater outflow.
83. It appears that the settlement amount in the market timing scandal involving Putnam Investments was based in part on the amount of fund outflows. See U.S. Securities and Exchange Commission, Commission Announces Completion of Independent Assessment Consultant's Report on Losses Attributable to Market Timing and Excessive Short-term Trading by Putnam Employees (Mar. 3, 2005), http://www.sec.gov/news/press/2005-26.htm (stating that Putnam had agreed to compensate shareholders for $48.5 million in losses from the abnormal redemptions).
84. This relationship may result in an endogeneity problem in our regression test using BIG SCANDAL, AS another robustness check, we eliminated TIMING scandals involving Putnam Investments from the BIG SCANDAL regression. Once again, the coefficients on SCANDAL and SCANDAL X BIG SCANDAL were negative (-0.013 and -0.010 respectively) and significant at the <1% level.
85. In unreported regressions, we re-estimated each of the models in Panel B with 12 separate dummy variables for each month in the scandal period for the SCANDAL variable (and 12 separate monthly interaction terms for each of our severity proxies). In each specification, the Zero Hypothesis (i.e. that the sum of the coefficients on the 12-month variables is equal to zero) was rejected at the <1% confidence level for both the 12-month SCANDAL variables and for the 12-month interaction terms between the SCANDAL variable and the severity proxies.
86. See Eric Zitzewitz, How Widespread Was Late Trading in Mutual Funds?, 96 AM. ECON. REV. 284, 287 tbl.2 (2006).
87. In unreported regressions, we re-estimated each of the models in Panel B with 12 separate dummy variables for each month in the scandal period for each of the scandal type interaction term variables. Results are similar to those reported in Panel B.
88. Because severe scandals often involve market timing (see Table 6 of Panel B), our results on scandal severity presented in Table 4 may simply reflect a negative market reaction to market timing scandals. As a robustness test, we defined BIG_TIMING as equal to 1 if the TIMING scandal had greater than the median settlement amount for the group of all TIMING scandals ($175 million), We defined TIMING_FEATURED as equal to 1 if the TIMING scandal involved greater than the median number of Wall Street Journal articles for the group of all TIMING scandals (10 articles). We defined TIMING_ENFORCED as equal to 1 if the SEC or some other governmental entity filed formal charges relating to a TIMING scandal. We then added to Model 1 of Panel B of Table 5 (the model for the different scandal types) interaction terms for TIMING X BIG_TIMING, TIMING X TIMING_FEATURED, and TIMING X TIMING_ENFORCED. In unreported regressions, we found that the coefficients for the three interaction terms were negative and significant at the 10%, <1%, and <1% levels respectively, indicating that more severe scandals among the group of TIMING scandals resulted in greater outflows.
89. An investor may nonetheless withdraw funds for other reasons, such as to exact retribution for past wrong-doing or to enforce a social norm not to engage in wrong-doing, especially if withdrawing funds is not costly to the investor. Cf. Richard H. McAdams, The Origin, Development, and Regulation of Norms, 96 MICH. L. REV. 338, 355 (stating that second-order collective action problems can be solved if punishing violators is relatively costless).
90. Indeed, even though facing a scandal is a rare event for the group of all fund families, five of the fund families in our data set were involved in repeat scandals during the time period encompassed by our study. For example, INVESCO was involved in scandals in 1994 and 2003. See Laurie P. Cohen & Sara Calian, Kaweske Probed By SEC Over Personal Trades, WALL ST. J., Jan. 7, 1994, at A3;
91. Tom Lauricella & Susan Pulliam, Invesco Charged in Scandal as Strong Quits, WALL ST. J., Dec. 3,2003, at C1. Kemper was involved in scandals in 1995 and 2000.
92. See Judith Burns, SEC Stays Tough on Mutual-Fund Ads, WALL ST. J., May 22, 2000, at C35;
93. Jeffrey Taylor & Robert McGough, Kemper Unit Target of SEC Complaint, Charging Fraudulent Stock Diversion, WALL ST. J., Mar, 3, 1995, at B16. MFS Investments was involved in scandals in 2002 and 2003.
94. See John Conner & Gregory Zuckerman, SEC Probes MFS Over Bond Issue, WALL ST. J., April 23, 2002, at C17;
95. Tom Lauricella & John Hechinger, Alliance Settles Charges, WALL ST. J., Dec. 19, 2003, at C1. Strong was involved in scandals in 1994 and 2003.
96. See Sara Calian, SEC Studies Securities Moves in Strong Funds, WALL ST. J., Mar. 11, 1994, at C1;
97. Tom Lauricella, Probe Hits Strong's Chairman, WALL ST. J., Oct. 30, 2003, at C1. Van Kampen was involved in scandals in 1995 and 1999.
98. See Sara Calian, Ex-Fund Manager at American Capital Is Barred from Industry Over Pricings, WALL ST. J., Mar. 3, 1995, at B6;
99. Pui-Wing Tam, IPO's Gave Big Boost To Van Kampen Fund, WALL ST. J., Sept. 9, 1999, at C1.
100. See Jason T. Greene & Conrad S. Ciccotello, Mutual Fund Dilution from Market Timing Trades, 4 J. INVESTMENT MGMT. 31, 39-41 (2006) (developing a model of the impact of market timing trades on returns);
101. Zitzewitz, supra note 69, at 287 tbl.2 (estimating that, between 1998 and 2003, late trading reduced fund holder returns by 3.77 basis points per year in international equity funds and by 0.88 basis points per year in domestic equity funds).
102. For example, the alleged actions by Fidelity Magellan Fund's Jeffrey Vinik, see supra text accompanying notes 32-36 (discussing how Vinik may have manipulated the market by publicly touting a stock while his fund sold it), relate to a scandal where third parties, rather than fund shareholders, would have suffered harm.
103. See Jennifer Arlen, Evolution of Corporate Criminal Liability: Implications for Managers, in LEADERSHIP AND GOVERNANCE FROM THE INSIDE OUT 191, 192-96 (Robert Gandossy & Jeffrey Sonnenfeld eds., 2004) (describing various ways in which management is held responsible for corporate crime).
104. As a check on the validity of our PEN_ENTITY and HARM variables we examined a fund's market performance during the three-year period after the initial scandal announcement in the Wall Street Journal. Unreported, we found that the mean monthly excess return (defined as the difference in return of a fund with all other funds in the same Strategic Insight Fund Code for the same month) was negative and significant (at the <1% level) for HARM and PEN_ENTITY funds, and positive and not significant for NO_HARM and NO_PEN_ENTITY funds. The difference in excess returns for HARM versus NO_HARM funds and PEN_ENTITY versus NO_PEN_ENTITY funds was significant, respectively, at the 10% and the 5% levels.
105. This is represented by the interaction term SCANDAL X HARM X PEN_ENTITY, which includes 112 scandals.
106. This is represented by the interaction terms SCANDAL X NO_HARM and SCANDAL X NO_PEN_ENTITY. In total, these interaction terms involved 23 scandals in the data set.
107. In unreported regressions, we re-estimated each of the models in Table 7 with separate dummy variables for each month in the scandal period included in the scandal interaction terms. The results were similar to those reported in Table 7, except that the SCANDAL X NO_PEN_ENTITY variable was significant at the 10% level.
108. Re-estimations of the models in Table 8 with 12 separate dummy variables for each month in the scandal period for the various scandal interaction terms yielded results similar to those in Table 8.
109. To the extent that the Wall Street Journal coverage did not report all of the required information regarding who initially uncovered a scandal, we would conduct our own additional research to supplement it.
110. For one scandal, it was not possible to determine who discovered it.
111. For example, the interaction term SCANDAL X SECFIRST estimates the relationship between non-timing scandals discovered by the SEC and fund flows.
112. For example, the interaction term SCANDAL X TIMING X SECFIRST estimates the additional flows for timing scandals first discovered by the SEC.
113. For example, the sum of SCANDAL X SECFIRST and SCANDAL X TIMING X SECFIRST estimates the relationship between timing scandals discovered by the SEC and fund flows.
114. Re-estimations of the models in Table 9 with 12 separate dummy variables for each month in the scandal period for the various scandal interaction terms yielded similar results.
115. Re-estimations of the models in Table 11 with 12 separate dummy variables for each month in the scandal period for the SCANDAL, SCANDAL FAMILY, and related interaction terms yielded results similar to those reported here.
116. Although we found no significant increase in the spillover effect when severity was proxied by FEATURED or ENFORCED, in an F-test the sum of the coefficients for future harm and for severity was negative and significant at a <1% confidence level when severity was proxied by ENFORCED.
117. In unreported regressions, we re-estimated each of the models in Panel C with 12 separate dummy variables for each month in the scandal period for the various SCANDAL and SCANDAL FAMILY interaction terms and tested whether the sum of the coefficients on the 12 month variables was equal to zero. For each model, the results were similar to those in Panel C, with the exception that the Zero Hypothesis test (that the sum of the coefficients for the 12-month SCANDAL FAMILY X PROMINENT X HARM X PEN_ENTITY X FEATURED variables was equal to zero) was rejected at the 10% confidence level.
118. Note from Models 2, 3, and 4 that the coefficient on SCANDAL FAMILY X PROMINENT X HARM X PEN_ENTITY was significant at the <1% level, In Model 4, the coefficient was negative and of a similar magnitude to the coefficients in Models 1, 2, and 3, but it was not statistically significant.