Susan Woodward Sand Hill Econometrics Palo Alto, CA 94301 February 29 , 2004 This is a Comment in response to S7-06-04 “Confirmation Requirements and Point-of-Sale Disclosure Requirements” Jonathan G. Katz, Secretary U.S. Securities and Exchange Commission 450 Fifth Street, N.W. Washington, D.C. 20549-0609 Dear Mr. Katz: Because the SEC is moving ahead to improve mutual fund disclosure, I want to share with you some things I learned that you might find useful in this effort. First, testing, further testing and testing some more is essential for design of a good disclosure. All sorts of things that look like good design, seem to just make sense, and incorporate the suggestions from investors and design people often turn out to just not work very well in terms of helping people make better choices. You cannot know whether information presented in a given format teaches until you show it to people and ask them questions to find out what they learn from it. They may learn a lot, or not learn much, or worse, they may learn things that are not true (such as to expect that past performance is likely to be repeated, and thus to conclude that it is worthwhile to chase performance when it is not). You won’t know until you test it. There is much to be learned from others in the government who have worked on disclosures and done research on comprehension and format. Much of what they have learned is informed by the work of social psychologists who can improve the first draft of a disclosure design a great deal. For example, a sans serif typeface slows down the rate at which people read and increases the likelihood that each word sinks in. All caps slows them down even more, but also impedes comprehension. Numbers are more challenging. People generally do a much worse job of comprehending and assessing numbers when all of the significant digits are on the right hand side of the decimal. That is, they do much better with a number presented as 6% than with 0.06. (This is why bankers invented interest rates and investment bankers invented basis points. Do they talk about .0006? No, they talk about six basis points.) Empty space around any entry on a given piece of paper causes people to look there. I learned these things when we did mall intercept studies at the SEC in 1995, and then discovered that they had been learned before by psychologists in all sorts of contexts. How information is presented can have a big impact on what investors learn. This is why companies spend vast amounts on marketing research. It is also why a standardized format for a disclosure could be so valuable to investors. A good format improves people’s comprehension. And it gets better with practice. As people become familiar with the format, their understanding continues to improve. The government can do investors a very valuable service by introducing a clear, simple, standardized format. The people at the FDA have much to share about their experience in designing the nutrition label (the label we find now on all packaged food, disclosing the calories, protein, fat fiber, etc. in that little chart that has the same format on all foods). Alan Levy, a social psychologist at the FDA designed many of the FDA’s studies and is particularly knowledgeable. They will tell you that when shown different candidate labels, one with more numbers, one with fewer, people often believe the one with more numbers is the more informative, yet the tests of comprehension reveal that it contributes less to their understanding -- they make fewer correct inferences from it. Giving people every conceivable figure does not necessarily inform them better. Emphasizing the key information can do a better job in helping consumers make correct inferences. We can learn the right balance only by testing. Another very knowledgeable person is Burkey Belser. He is with the research/design firm Greenfield/Belser that executed much of the research for the FDA. He is in Washington at (202) 775-0333. (Phone number may be stale.) The Federal Trade Commission has designed many disclosures, among them the energy efficiency sticker for major appliances. Economists there who are knowledgeable about these issues are Jan Pappalardo and Pauline Ippolito. Another person with current experience is John Weicher at HUD (Assistant Secretary for Housing, aka FHA Commissioner), who is at present involved in designing an improved disclosure for mortgage loans as part of RESPA reform. The brief feedback I have gotten on the HUD effort is that the format of the disclosure mattered much more than anyone expected. You should also read a study of the S&P500 index fund market by two economists, Ali Hortecsu and Chad Syverson of U. of Chicago, “Product Differentiation, Search Costs, and Competition in the Mutual Fund Industry: A Case Study of S&P500 Index Funds”. (There is a lot of economist-ese in this paper, so maybe your economist Larry Harris can do a translation for you.) They chronicle the evolution of expenses in S&P500 mutual funds, noting that between 1995 and 2000, the average fees rose, the market share of the most economical funds fell, and the fees on the most expensive funds rose even more than the average fees did. Today the lowest cost S&P500 index fund has expenses of 9.5 basis points and the highest cost has fees of a whopping 268 basis points. The impact of this difference on an investor’s retirement is more than 100% -- the person investing in the lower expense fund will enjoy a retirement income more than double that of the more expensive fund. In a classic competitive market (with well informed consumers) we would expect to see prices that are fairly uniform for identical or nearly identical products. We would also expect that when new firms enter with nearly identical products, they enter with lower prices. The mutual fund market has the essential characteristics of classic competitive markets -- low entry costs and low fixed costs. So why do the fees vary so much? And why have the new entrants come in with higher prices (expenses) than the incumbents? Yes, there are some differences in the services offered, but these are not plausible as explanations for the fee dispersion. So what does explain it? H&S conclude that the main reason is that many investors do not understand the importance of expenses or that much lower expense funds are available. In other words, their choices are poorly informed, and as a result, some choose badly. And the reason we see expenses rising is that the fraction of investors with poorer understanding is rising. This paper is available at http://home.uchicago.edu/~syverson/mutualfunds5.pdf Also consider the choices made by the federal government itself in designing the Federal Thrift Plan. The choices are very limited and very economical. The equity option offered is a very well-diversified index fund run for less than 10 basis points a year. Does anyone think that Federal employees are ill-served by this limited by excellent menu? I don’t think so. Your efforts to find out what investors want in the way of disclosure are a good thing. But you have to do more than ask investors what they want in the way of disclosure, because most of them are not well enough informed to know what matters, what they should want. For example, to know that part of the fees they are paying are going to sales people may clue them in that they could find the same product elsewhere cheaper, but what they really need to focus on is total expenses, period, regardless of who is getting the fees. Even when expenses are spent on research the funds on average do not earn them back for investors. A study my office did of 12b-1 fees while I was at the SEC shows that funds with 12b-1 fees did no worse than other funds without them but with the same level of total expenses. The research says over and over that it is total expenses that matter. And they matter even more than they appear because the under-performance is more than one-for-one on expenses. This should not be a surprise, because reported “expenses” understate what funds actually spend because expenses include no trading costs. The government’s research must be independent. It must not solicit the “assistance” of the funds. The large fund organizations have done research of this nature and used it to keep investors confused – to suggest to them that past performance is likely to be repeated when the scientific evidence is overwhelming that it will not be, to suggest that research can improve performance when the science informed by 60 years of data says that it has not, that clever trading strategies can improve performance when the scientific evidence is that they do not, and to avert their interest in expense ratios. When I was searching for contractors to execute the SEC telephone survey in 1994-95, I talked at length with a person at the University of Michigan Survey Research Center. He was the most knowledgeable person I had encountered regarding fund disclosure formats and investor comprehension, and I was eager for his contribution. After some months of phone conversations, he stopped returning my calls. By calling at an odd time, I caught him when he answered the phone himself. He sheepishly admitted that SRC had done a lot of work for Fidelity, and Fidelity did not want them to help us. Yes, it is my understanding also that the Survey Research Center is a non-profit, eleemosynary organization. Some of these ideas may collide with free-market philosophies. But consider this: I am not suggesting price controls, nor banning expensive funds, or banning any fund investment strategy no matter how crazy, I am only suggesting that the middle-class investor who has not been to college and studied business get the benefit of what the PhD economists who have studied fund performance exhaustively already know and are teaching to everyone who goes to business school. Why should we let the funds mislead people who have not had the opportunity to study statistics or the time to apply statistics in a systematic way to fund performance data? It would take more than a lifetime to learn these facts through casual observation. The facts are highly relevant to mutual fund choices. Mutual fund choices will have a tremendous impact on the comfort with which these people retire. Policy choices have been made that have caused a dramatic shift away from employer-managed retirement systems to employee-selected ones. We have given these people almost nothing in the way of tools or education to guide their choices. Many are making choices that are unquestionably poor by the standards of the most informed scientific observers. This is just way too important to ignore. To see that free market, even libertarian principles are not necessarily incompatible with guiding others to better choices, have a look at this highly readable paper (no translator needed) by Dick Thaler, “Libertarian Paternalism”, soon to appear in American Economic Review but available now at http://gsbwww.uchicago.edu/fac/richard.thaler/research/Libertarian%20Paternalism.pdf Dick’s main point is that framing (format, context, etc.) influences decisions. So in settings where people have to choose, set the choices to aim people toward what is good for them. Do you want the cafeteria to put the rich desserts first, in the most visible and accessible location, or the salads? Should the default in the savings plan be opt-in or out-out? The default will influence their choices. We can design a disclosure that will help people understand their options and the consequences of their choices, or we can stay with our vague disclosure standards and leave it to the funds to “inform” them. The trouble is, how informed they actually become from the “information” they get depends heavily on the way that information is presented. Only by imposing some standard format on that presentation can we be reasonably sure that the story gets through. It will take some serious research to develop an effective presentation. Consider the investors who invest in the S&P500 index fund identified by Hortacsu and Syverson as the most expensive at 284 basis points. On the one hand, they are better off there than in a precious metals fund with similar expenses -- at least the risk they are bearing is on average compensated with higher return. But would the investors have chosen such a fund if they were better informed about their options? Can they have any idea how much their poor choices are costing them? Surely no one believes that they would or that they do. Figure out how to help them. Just last week, the Federal Trade Commission released the report on its research on the proposed new HUD mortgage disclosure rules. The proposed rule would require disclosure of the compensation to mortgage brokers as well as the rate and up-front charges on the loan. The FTC study found that rather than improving consumer’s ability to choose the lowest cost mortgage, the additional information, though accurate, interfered with consumers’ ability to determine which loan was cheaper. This is consistent with the findings of the FDA that additional, but only tangentially relevant, information did not help consumers select the most healthy foods. The inference for selection of mutual funds is that dividing mutual fund expenses into numerous categories may distract and impair the ability of investors to select the funds best for their long-run investments. The research on fund performance indicates that only total expenses matter for fund performance, and that the categories of fund expenditures are irrelevant. This study is somewhat important for its substance (more numbers are often worse rather than better) but even more important for what it conveys about the importance of testing. The entire FTC study is available at http://www.ftc.gov/opa/2004/02/mortgagerpt.htm Thanks for your attention. Sincerely, (imagine signature here) Susan Woodward