The Historical Record for Active Management Introduction What makes for successful active management? We have argued that the process of successful active management consists of efficiently utilizing superior information. It has two key elements: finding superior information, and efficiently building portfolios based on that information. Some sections of this book have described the meaning of superior information: better than the consensus, high information ratio, positive information coefficient, and, most likely, high breadth. We have also devoted many chapters to efficiently utilizing superior information. We showed how to process raw signals into alphas, how to build alphas into portfolios, how to trade off alphas against risk, and the cost of transacting. We now want to step away from our prescriptions and view the historical record for active management. Ultimately we will want to see the historical evidence for our view of successful active management. Finance academics have a fairly long history of studying active manager performance. Their motivation has typically been in the context of efficient markets theory. According to the strong version of the theory, active management is impossible. There is no source of superior information. Hence, efficient utilization is irrelevant. Of course some active managers will outperform and others will underperform, but no more so than at the roulette wheel, where some gamblers win through sheer luck.
Page 560 Studies of Performance As we discussed in Chap. 17, studies of active managers' performance began soon after the development of the CAPM, which provided a framework for performance analysis. Pioneers in this field included Jensen (1968), Sharpe (1966), and Treynor (1965). Studies of managers' performance have focused on three separate questions: Has the average active manager outperformed, are the top performers skillful or lucky, and does performance persist? Note that a negative answer to the first of these questions would not prove that successful active management is impossible. Let's start with the first question, about average manager performance. The early studies found that on average, mutual funds underperformed the index on a risk-adjusted basis, and that there existed a direct relationship between the level of fund expenses and the amount of underperformance. Subsequent work, summarized by Ippolito (1993), found that the performance of the average fund, net of expenses and on a risk-adjusted basis, is statistically indistinguishable from that of the index. At best, the performance of the average fund only matches the index. More recent academic work on this question has extended the previous work in three directions: overcoming survivorship bias, controlling for style, and looking at portfolio holdings. Brown, Goetzmann, Ibbotson, and Ross (1992) demonstrate that survivorship bias in manager databases can significantly affect the results of performance studies. Several subsequent studies have very carefully built databases free of survivorship bias by including all funds that are no longer in business. Malkiel (1995), for example, shows that from 1982 through 1991, the average equity mutual fund still existing in 1991 underperformed the S&P 500 by 43 basis points per year. But when he includes all funds that existed over that period, even those no longer in business in 1991, the average underperformance drops to 1.83 percent per year. Survivorship bias is important, and the average U.S. equity manager significantly underperforms the S&P 500. Several more recent academic studies analyzing manager performance have controlled for style and for publicly available information. Chapter 17 has described these methods. For example, Ferson and Schadt (1996) and Ferson and Warther (1996) control for
Page 561 publicly available information—namely, interest rates and market dividend yields—in analyzing 67 U.S. equity mutual funds1 from 1968 through 1990. Their approach improves average manager performance from below the market to matching the market. While their database suffers from survivorship bias, they claim that this should not affect their result concerning the improvement in average manager performance. From a slightly different perspective, Jones (1998) has analyzed median institutional manager performance. He finds that he can explain median performance relative to the S&P 500 almost entirely with three variables: market return, small-capitalization stocks versus large-capitalization stocks, and value stocks versus growth stocks. The average manager owns some cash, has a bias toward small stocks, and has a bias toward growth stocks. Hence, the average manager tends to underperform when the market is up, large-capitalization stocks outperform, and/or value stocks outperform growth stocks. Daniel, Grinblatt, Titman, and Wermers (1997) control for size, book-to-price, and one-year momentum effects, and use quarterly portfolio holdings for more than 2500 U.S. equity mutual funds from 1975 through 1994 in their analysis. Their approach begins by estimating quarterly asset-level returns beyond the effects of size, book-to-price, and momentum. They do this by assigning all assets to 1 of 125 groupings based on quintile rankings for these three characteristics, and then looking at each asset's active return relative to its (capitalization-weighted) group return. To analyze mutual fund returns, they use the quarterly holdings and these active asset returns. They find statistically significant evidence for positive average performance of 1 to 1.5 percent, but only for growth and aggressive growth funds over the entire period. Most of this arises from performance from 1975 through 1979. But we must view even this small evidence for average outperformance sceptically, since the study ignored both fees and transactions costs. The study used not the actual mutual fund returns, but returns to quarterly buy-and-hold portfolios with no quarterly rebalance charges. There is little reason to believe that the average 1Ferson and Warther examine 63 funds.
Page 562 growth or aggressive growth fund delivered any of that outperformance to investors. Overall, there is no evidence for average active management's producing exceptional returns. Fortunately, this says nothing about the possibility of successful active management. Why would we expect the average manager to outperform? To focus more on the possibility of successful active management, Marcus (1990) looked at whether top-performing mutual funds exhibited statistically significant positive performance, given the large number of funds in existence. The Marcus study is a rigorous version of the classic (and anecdotal) defense of active management: ''Look at Peter Lynch, look at Warren Buffet." Using statistics for the maximum of a sample, he shows that the very top funds do outperform. Peter Lynch and Warren Buffet don't appear to be just the two out of tens of thousands of investors who are lucky enough to flip heads 10 or 15 times in a row. So this is one study that demonstrates that successful active management is possible. But beyond whether the average fund outperforms or whether the top fund outperforms, the question concerning the possibility of successful active management that has received the most attention is whether performance persists. We now turn to this question. Persistence of Performance Do the winners in one period remain winners in a subsequent period? After 30 years of intensive research, the results fall into two camps: those that do not find persistence, and those that do. Several studies have shown, based on different asset classes and different time periods, that performance does not persist. Jensen (1968) looked at the performance of 115 mutual funds over the 1945–1964 period and found no evidence for persistence. Kritzman (1983) reached the same conclusion after examining the 32 fixed-income managers retained by AT&T for at least 10 years. Dunn and Theisen (1983) found no evidence of persistence in 201 institutional portfolios from 1973 to 1982. And Elton, Gruber, and Rentzler (1990) showed that performance did not persist for 51 publicly offered commodity funds from 1980 to 1988.
Page 563 Several other diverse studies, however, have found that performance does persist. Grinblatt and Titman (1988) found evidence of persistence in 157 mutual funds over the period 1975 to 1984. Lehmann and Modest (1987) report similar results from looking at 130 mutual funds from 1968 to 1982. In the United Kingdom, Brown and Draper (1992) demonstrated evidence for persistence using data on 550 pension managers from 1981 to 1990. Hendricks, Patel, and Zeckhauser (1993) documented persistence of performance in 165 equity mutual funds from 1974 to 1988. Goetzmann and Ibbotson (1994) showed evidence for persistence using 728 mutual funds over the period 1976 to 1988. Bauman and Miller (1994) showed evidence for persistence using as many as 608 institutional portfolios from December 1972 through September 1991, but only when using periods corresponding to complete market cycles. Kahn and Rudd (1995), after accounting for style effects, fees and expenses, and database errors, found no evidence for persistence of performance for 300 equity funds from October 1988 through September 1994. They did, however, find evidence for persistence of performance for 195 bond funds from October 1991 through September 1994. Unfortunately, this persistence was insufficient to support an outperforming investment strategy: It could not overcome the average underperformance of bond mutual funds (especially after fees and costs). Kahn and Rudd (1997a, b, c) find similar results when they extend the analysis to additional time periods and to institutional portfolios as well as mutual funds, and when they focus on managers rather than just on mutual funds. Now remember that Brown, Goetzmann, Ibbotson, and Ross (1992) showed that survivorship bias could significantly affect performance studies. In particular, they demonstrated that survivorship bias would generate the appearance of significant persistence. This calls into question several of the studies that found evidence for persistence. The recent work on persistence has carefully utilized databases free of survivorship bias. Looking at all U.S. equity mutual funds from 1971 through 1991, Malkiel (1995) found evidence for persistence of performance in the 1970s, but it disappeared in the 1980s. However, Gruber (1996), also looking at 270 U.S. equity mutual funds from 1985 to 1994, found persistence so strong, that, he argued, it explained the growth in active mutual funds. Malkiel measured performance
Page 564 using simple CAPM regressions, while Gruber also controlled for size, book-to-price, and bond market effects. Finally, Carhart (1997) looked at 1892 diversified equity mutual funds from 1962 through 1993, controlling for size, book-to-price, and 1-year momentum effects. The only significant persistence he found was for the strong underperformance of the very worst mutual funds. In summary, the past 30 years of research on persistence of performance has generated at best a mixed record. These data have been tortured mercilessly, analyzed in as many different ways as there have been studies, and with widely varying results even among recent studies produced by highly respected academics. In spite of mutual fund advertising (though consistent with proxy statements), the connection between historical performance and future performance is weak. The probability that a winner will repeat is not 90 percent but perhaps 55 percent, with academics strongly arguing about the statistical significance of such a result.2 What does the limited connection between historical performance and future performance tell us about the possibility of successful active management? A Simple Model of the Manager Population The mixed results on persistence of performance clearly show that a simple strategy of picking last year's winners (above-median performers) will not have much (if any) more than a 50 percent chance of being a winner (outperforming the median) this year. These results do not definitively say that active management is impossible. At the same time, they are hardly a dramatic vindication of active management. These results do say that it isn't easy to find successful active managers from their track records alone. We can better understand this with a simple model. Instead of investments, we will consider flipping coins. Imagine that the 2Goetzmann and Ibbotson (1994), in one of the more optimistic studies, find, for example, that the probability of equity mutual fund winners' (defined by above-median alphas) repeating, on average, year by year from 1978 through 1987, is 62 percent. The more pessimistic studies find numbers like 50 percent. That roughly defines the range of the academic argument.
Page 565 population consists of two different groups. There are the "persistent winners," who consistently call heads or tails correctly. They exhibit the ultimate in skill: They are correct every time. Then there are the "coin tossers," who call heads or tails at random. Since we want to rig this to be a zero-sum game, we will set up the coin tossers to be correct slightly less than half the time. Then, when we average over the entire population, with persistent winners always correct and coin tossers correct slightly less than half the time, the calls are correct exactly half the time. We will assume that most of the population are coin tossers. We will analyze our overall population using a 2 X 2 contingency table, Fig. 20.1. Every member of the population will call a coin toss. Half will be correct (winners), and half will be incorrect (losers). The experiment will then be repeated. Each member of the population will be either a winner or a loser on the first toss and either a winner or a loser on the second toss. A contingency table records how many members of the population are winners on both tosses, losers on both tosses, winners then losers, and losers then winners. The persistent winners will show persistence. They will all appear in the two-time winner category. The coin tossers will show up in all four categories. Some, through sheer luck, will be consistent winners. The rest will be scattered through the other categories. If there are no persistent winners—if coin tossers constitute the entire population—then we expect one-quarter of the population in each category. The probability of a winner repeating is exactly 50 Figure 20.1 Contingency tables.
Page 566 percent. In this simple model, any deviations from this random pattern result from the presence of the persistent winners. In fact, we can relate the probability p of winners repeating to the fraction δ of consistent winners in the population: If the probability of winners repeating is just 50 percent, then δ = 0. If the probability of winners repeating is 100 percent, then δ = 50 percent. The consistent winners represent half the entire population. Note that in a zero-sum game, the winners can't make up more than 50 percent of the population. Now, if the probability of winners repeating is 60 percent, which corresponds to about the highest claimed probability of investment winners' repeating in any academic study, then Eq. (20.1) implies that about 17 percent of the population is skilled. Note that this model implies a way to identify the persistent winners: Look over more periods. This effectively divides the population into more groups. If we look at the players who call coin tosses correctly three periods in a row, these will consist of all the persistent winners, plus the few coin tossers who were lucky three times in a row. Even fewer coin tossers will be lucky four times in a row. In this way, we can filter out the persistent winners. But this simple model does not perfectly capture the investment world. Even skilled managers do not outperform every day, or every month, or even every year. We can never escape some element of noise in historical performance numbers, and the shorter the horizon, the more that noise is a significant factor. Hence, Eq. (20.1) will underestimate the fraction of skillful managers. Put another way, we would be happy to be able to identify managers with information ratios of 0.5. But given the levels of noise in performance numbers, in any 1-year period, such a manager has only a 69 percent chance of ouperforming (assuming normal distributions). The probability of outperforming in two consecutive periods is 48 percent. If we look at 5-year periods, then these probabilities change to 87 percent and 75 percent, respectively. Even if we choose two consecutive 5-year periods and build a contingency table as described before, only 75 percent of our target pool of managers (those with information ratios above 0.5) will
Page 567 show up as repeat winners. And that is if their skill (and their career with one firm) has lasted 10 years. This simple model of the population of active managers leads to two conclusions. First, the limited evidence for persistence of performance does not rule out the possibility of successful active management. It simply puts a rough bound on the fraction of skillful managers in the population. Second, the levels of noise in performance numbers imply that historical performance will never precisely identify skillful managers. What Does Predict Performance? After 30 years, the academics have beaten to death this question of persistence of performance. It's time to frame the question more broadly: What, if anything, does predict performance? Here, we can rely on some results given in this book. The fundamental law of active management shows that high information ratios depend on skill and breadth. We might expect to see high information ratios for managers who hold more positions or exhibit more turnover, both of which are proxies for breadth. Of course, higher turnover also implies higher transactions costs, so the connection may not be very strong. Stewart (1998) examined 1527 U.S. institutional equity products from January 1978 through March 1996, and found just such a connection. Rather than looking at information ratios, he grouped all managers into quintiles based on their fraction of monthly positive active returns, a measure he labeled "consistency of performance." This fraction should be a monotonic function of the information ratio,3 and so this should correspond to information ratio quintiles as well. He found that the number of holdings, and the turnover, increases as we move to the more consistent (higher information ratio) quintiles. Furthermore, Stewart found that the more consistent performers in one period had higher active returns in the next period. 3 Assuming normally distributed active returns, this fraction is just
Page 568 Kahn and Rudd (1997c) have presented similar results for 367 institutional equity portfolios from October 1993 through December 1996. After controlling for style effects, they regressed information ratios in the second half of the period cross-sectionally against a set of possible explanatory variables from the first half of the period. While the information ratio in the first half of the period does not show up as statistically significant, the number of assets in the portfolio does have statistically significant forecasting power, consistent with Stewart's results. These results may be the clearest empirical evidence of the importance of the precepts in this book.4 Why Believe in Successful Active Management? We have reviewed the empirical results. The references include a long list of learned studies. We have compiled some evidence that successful active managers exist. We have encouraging evidence that the approaches outlined in this book help. But the evidence for successful active management isn't overwhelming, and besides, all this evidence looks backward. Why should we believe in active management, looking forward? First, we know that markets aren't perfectly efficient, because market participants are human. Humans are not perfectly rational. Perhaps more surprising, humans are irrational in consistent and specific ways. The field of behavioral finance has developed to understand these persistent human behaviors and their impact on financial markets.5 Behavioral finance has claimed that markets are not efficient because of human behavior. Until evolution cooperates, market inefficiencies will remain available for successful active management. 4 In a more flattering vein, Chevalier and Ellison (1997) have uncovered a positive connection between active returns and manager SAT scores. We naturally believe that the followers of the ideas in this book represent the high end of the active manager intelligence spectrum. 5 One seminal article is Tversky and Kahneman (1974). See also DeBondt and Thaler (1985) and Kahneman and Riepke (1998).
Page 569 Second, and perhaps paradoxically, only active management can make markets efficient. Since the development of the CAPM and efficient markets theory in the 1960s, passive management has grown in popularity. According to Ambachtsheer (1994), in the 1970s, 100 percent of institutional assets were actively managed. That fraction had dropped significantly. As he points out, if only 10 percent or 1 percent of assets were actively managed, then elementary investment research, not even efficiently implemented, should pay off. As the fraction of active management increases, some inefficiencies will remain, but they will be exploitable only with an efficient implementation. Moreover, that exploitation will be exactly at the expense of managers with inefficient processes and/or inferior information. Ambachtsheer argues that successful active management must be possible, because efficient markets require it. Furthermore, successful active management will require both superior information and efficient implementation. Looking forward, there are compelling arguments for believing in successful active management. But, not surprisingly, successful active management will require cleverness and hard work: to uncover information superior to that of other managers, and to implement more efficiently than other managers. References Ambachtsheer, Keith P. "Active Management That Adds Value: Reality or Illusion." Journal of Portfolio Management, vol. 21, no. 1, 1994, pp. 89–92. Bauman, W. Scott, and Robert E. Miller. "Can Managed Portfolio Performance Be Predicted?" Journal of Portfolio Management, vol. 20, no. 4, 1994, pp. 31–40. Bello, Zakri, and Vahan Janjigian. "A Reexamination of the Market-Timing and Security-Selection Performance of Mutual Funds." Financial Analysts Journal, vol. 53, no. 5, 1997, pp. 24–30. Bogle, John C. "The Implications of Style Analysis for Mutual Fund Performance Evaluation." Journal of Portfolio Management, vol. 24, no. 4, 1998, pp. 34–42. Brown, G., and P. Draper. "Consistency of U.K. Pension Fund Investment Performance." University of Strath Clyde Department of Accounting and Finance, Working Paper, 1992. Brown, Stephen J., William N. Goetzmann, Roger G. Ibbotson, and Stephen A. Ross. "Survivorship Bias in Performance Studies." Review of Financial Studies, vol. 5, no. 4, 1992, pp. 553–580.
Page 570 Brown, Stephen J., William N. Goetzmann, and Alok Kumar. "The Dow Theory: William Peter Hamilton's Track Record Reconsidered." Journal of Finance, vol. 53, no. 4, 1998, pp. 1311–1333. Brown, Stephen J., William N. Goetzmann, and Stephen A. Ross. "Survival." Journal of Finance, vol. 50, no. 3, 1995, pp. 853–873. Carhart, Mark M. "On Persistence in Mutual Fund Performance." Journal of Finance, vol. 52, no. 1, 1997, pp. 57–82. Chevalier, Judith, and Glenn Ellison. "Do Mutual Fund Managers Matter? An Empirical Investigation of the Performance, Career Concerns, and Behavior of Fund Managers." National Bureau of Economic Research Preprint, 1997. Christopherson, Jon A., and Frank C. Sabin. "How Effective Is the Effective Mix?" Journal of Investment Consulting, vol. 1, no. 1, 1998, pp. 39–50. Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers. "Measuring Mutual Fund Performance with Characteristic-based Benchmarks." Journal of Finance, vol. 52, no. 3, 1997, pp. 1035–1058. DeBondt, W. F. M., and Richard Thaler. "Does the Stock Market Overreact?" Journal of Finance, vol. 40, no. 3, 1985, pp. 793–805. Dunn, Patricia C., and Rolf D. Theisen. "How Consistently Do Active Managers Win?" Journal of Portfolio Management, vol. 9, no. 4, 1983, pp. 47–50. Elton, E., Martin Gruber, and J. Rentzler. "The Performance of Publicly Offered Commodity Funds." Financial Analysts Journal, vol. 46, no. 4, 1990, pp. 23–30. Ferson, Wayne E., and Rudi W. Schadt. "Measuring Fund Strategy and Performance in Changing Economic Conditions." Journal of Finance, vol. 51, no. 2, 1996, pp. 425–461. Ferson, Wayne E., and Vincent A. Warther. "Evaluating Fund Performance in a Dynamic Market." Financial Analysts Journal, vol. 52, no. 6, 1996, pp. 20–28. Goetzmann, William N., and Roger Ibbotson. "Do Winners Repeat?" Journal of Portfolio Management, vol. 20, no. 2, 1994, pp. 9–18. Grinblatt, Mark, and Sheridan Titman. "The Evaluation of Mutual Fund Performance: An Analysis of Monthly Returns." Working Paper 13-86, John E. Anderson Graduate School of Management, University of California at Los Angeles, 1988. Gruber, Martin J. "Another Puzzle: The Growth in Actively Managed Mutual Funds." Journal of Finance, vol. 51, no. 3, 1996, pp. 783–810. Hendricks, Darryll, Jayendu Patel, and Richard Zeckhauser. "Hot Hands in Mutual Funds: ShortRun Persistence of Relative Performance, 1974–1988." Journal of Finance, vol. 48, no. 1, 1993, pp. 93–130. Ippolito, Richard A. "On Studies of Mutual Fund Performance 1962–1991." Financial Analysts Journal, vol. 49, no. 1, 1993, pp. 42–50. Jensen, Michael C. "The Performance of Mutual Funds in the Period 1945–1964." Journal of Finance, vol. 23, no. 2, 1968, pp. 389–416.
Jones, Robert C. "Why Most Active Managers Underperform (and What You Can Do About It)." In Enhanced Index Strategies for the Multi-Manager Portfolio, edited by Brian Bruce (New York: Institutional Investor, 1998). Kahn, Ronald N., and Andrew Rudd. "Does Historical Performance Predict Future Performance?" Financial Analysts Journal, vol. 51, no. 6, 1995, pp. 43–52.
Page 571 ———. "The Persistence of Equity Style Performance: Evidence from Mutual Fund Data." In The Handbook of Equity Style Management, 2d ed., edited by J. Daniel Coggin, Frank J. Fabozzi, and Robert D. Arnott (New Hope, Pa. Frank J. Fabozzi Associates, 1997a). ———. "The Persistence of Fixed Income Style Performance: Evidence from Mutual Fund Data." In Managing Fixed Income Portfolios, edited by Frank J. Fabozzi (New Hope, Pa: Frank J. Fabozzi Associates, 1997b), chap. 18. ———. "The Persistence of Institutional Portfolio Performance." BARRA International Research Seminar, Montreux, Switzerland, September 1997c. Kahneman, Daniel, and Mark W. Riepke. "Aspects of Investor Psychology." Journal of Portfolio Management, vol. 24, no. 4, 1998, pp. 52–65. Kritzman, M. "Can Bond Managers Perform Consistently?" Journal of Portfolio Management, vol. 9, no. 4, 1983, pp. 54–56. Lehmann, Bruce N., and David M. Modest. "Mutual Fund Performance Evaluation: A Comparison of Benchmarks and Benchmark Comparisons." Journal of Finance, vol. 42, no. 2, 1987, pp. 233– 265. Malkiel, Burton G. "Returns from Investing in Equity Mutual Funds 1971 to 1991." Journal of Finance, vol. 50, no. 2, 1995, pp. 549–572. Marcus, Alan J. "The Magellan Fund and Market Efficiency." Journal of Portfolio Management, vol 17, no. 1, 1990, pp 85–88. Sharpe, William F. "Mutual Fund Performance." Journal of Business, vol. 39, no. 1, Part II, 1966, pp. 119–138. Stewart, Scott D. "Is Consistency of Performance a Good Measure of Manager Skill?" Journal of Portfolio Management, vol. 24, no. 3, 1998, pp. 22–32. Treynor, Jack L. "How to Rate Management of Investment Funds." Harvard Business Review, vol. 43, January–February 1965, pp. 63–75. Tversky, Amos, and Daniel Kahneman. "Judgment under Uncertainty: Heuristics and Biases." Science, vol. 185, 1974, pp. 1124–1131.
Page 573