Your clients probably know how hard it is to pick outperforming stocks. What they may not fully grasp is how hard it is to choose a winning mutual fund or separate account manager. Even the pros can get it wrong. For example, when the venerable Value Investment Survey launched a mutual fund ranking service in the mid-1990s, its editor told Financial World that the newsletter would provide a “first ever” chance “to make meaningful assessments of manager skill.”
Ignore the editor's boast (first ever? Please. Institutional consultants had been handicapping asset managers for decades) and examine the reality of the Value Line Mutual Fund Survey's early prediction. Topping its list of the “Ten Best Fund Managers,” as of Dec. 31, 1994, was Robert Sanborn, manager of the Oakmark Fund, a large-capitalization value portfolio, which had racked up impressive numbers. But fast-forward to 1999: Oakmark finishes dead last in the large-cap value category, with a loss of 10.5 percent. That's a five-year period in which the Dow tripled and closed at an all-time high.
So much for the strategy of naive forecasting (picking last year's best and hoping that the trend will continue). Do not regard this as an argument against active management and passive investing. Indeed, citing the Oakmark and Value Line Mutual Fund Survey example isn't meant to cast aspersions on either's talents; it's just that the Oakmark Fund's experience dramatically illustrates the fundamental weaknesses of manager selection methods based on performance history. Turns out that the boilerplate disclaimer, “Past performance does not guarantee future results,” is very true — at times, even an understatement. The lesson is that advisors must remember that raw performance rankings don't mean much on their own; the total return of a fund only tells a fraction of the story.
The problem: Risk measurement, regardless of the analytical sophistication brought to the task, is an inexact science. In Oakmark's case, what happened to lay it low would have been very hard to predict with the then-available data. Oakmark had posted record-shattering returns in the early 1990s when value stocks, especially financial issues, were outperforming the S&P 500. But in the later years of the decade, when greater-fool psychology drove tech stocks to ludicrous heights and “old economy” stalwarts looked passe, Sanborn decided (intelligently, one could argue) to wait the craziness out.
Investors showed no mercy to those who didn't buy into the New Paradigm, though. On March 21, 2000, Smart Money, under the headline, “Oakmark's Robert Sanborn Gets the Heave-Ho,” said, “As technology stocks stole the market lead, Sanborn stubbornly refused to participate.” If you factor in a couple of bad stock picks, well, it was a disaster. The magazine continued, “Top picks like Philip Morris and Mattel gave shareholders a beating. Redemptions have pounded the fund, which is now half the size it was two years ago.”
Yet, all but the most devout “efficient market” theorists would agree that some managers beat the market with greater consistency than others. So how do you evaluate a manager's “value added” after allowing for risk? How do you increase your odds of picking a manager who might continue to outperform? The short answer is to understand the language of risk and to use tools, such as the Sharpe ratio, when choosing asset managers.
Volatility, also called variability, is the accepted surrogate for risk. Volatility is measured using a yardstick called standard deviation, a statistical concept that focuses on a period of time, typically three years. Standard deviation computes an arithmetic average (mean) of the returns over that period, and then determines the range in which returns have varied from that mean. If you walk your dog three blocks down the middle of the sidewalk, and the dog runs one way to chase squirrels and the other to chase cars, the average range between the two distances is the standard deviation. So, if the mean return is 10 percent and the distribution of returns has been from a plus-25 percent to a minus-5 percent, the standard deviation is 15. (Of course if the dog got hit and died, standard deviation wouldn't know what to do with itself.) As one portfolio manager says, “All standard deviation measures is historical price behavior; it doesn't measure risk, which is the potential for loss.”
That is often described by beta, which, as everyone knows, is the standard deviation of a given portfolio compared to a portfolio or index that it is trying to beat. By definition, the market (or benchmark) has a beta of 1.0. The portfolio being measured has a beta score higher or lower than 1.0 depending on whether it moves up or down more than the market. Based on its beta, a portfolio has an expected return. Any difference between the expected return and the actual return is called alpha. Alpha, which can be positive or negative, represents the investment return from taking nonmarket risk. It is the best we can do to quantify the manager's contribution to total return (sorting out skill from luck is a whole other analytical process).
The Sharpe Ratio
In this low-return environment, getting any excess return is something to brag about. But you want to be sure that the manager earned his return without betting the farm. And that's where the Sharpe ratio comes in. The Sharpe ratio, named after its inventor William Sharpe, is a common risk-adjusted return measure used in a manager due-diligence process. It tells you how much reward a portfolio is getting per unit of risk the manager is taking. The ratio is calculated by subtracting Treasury bills, which are considered to be risk free, from a manager's return. You take the average difference between the manager's return and the risk-free return over a given period of time and then divide by how much the portfolio swayed from that average. It measures the “unexplained” excess returns captured by the manager — returns not explained by the market's movements. In general, the higher the number, the better (ratios over 1 mean the risk-reward tradeoff is attractive). The Sharpe ratio doesn't differentiate manager skill from luck, though. Yet it is the gold standard for comparing the relative efficiency of different funds, whether they hold stocks or bonds or both.
It has its shortcomings, as noted. But a major one is that the ratio is computed on past performance — the average historical return and the historical variability of those returns. Future performance, as the Oakmark investors learned, can vary greatly from past returns. Compounding matters is the fact that relatively small changes in either variable, whether caused by luck or skill, can result in a misleading figure. Sharpe himself was even cautious about his ratio's use. He recently told the The Wall Street Journal that the ratio is frequently misused by hedge funds: “Anybody can game this,” through outright manipulation. “I could think of a way to have an infinite Sharpe Ratio,” he said.
And anyway, any individual “analytical” number can mislead. In the late 1990s, both risk and return in technology stocks were affected by abnormal market events, in that they posted an almost unbroken upward momentum. They didn't reveal much downward volatility until the craze ended several years later. But in the interim, tech-heavy managers sometimes produced extraordinary returns with low variability. Not surprisingly, portfolios heavily weighted in tech stocks had unrealistically high Sharpe ratios (and “genius” managers).
The Vanguard Group's former Chairman John Bogle, in his book Common Sense on Mutual Funds, complains that Sharpe's equal weighting of risk and return is arbitrary and unrealistic. “Standard deviation is only a rough proxy for a concept as elusive as risk,” he says, adding that “an extra percentage point of standard deviation is meaningless, but an extra percentage point of return is priceless.” He calls the Sharpe ratio “a blunt instrument to measure risk-adjusted returns,” but admits it “can be remarkably useful if its limitations are recognized.”
That said the Sharpe ratio is especially useful when tested for consistency, that is, when it is paired with the “information” ratio, also pioneered by Sharpe. A high Sharpe ratio signifying high return to risk accompanied by a high information ratio — signifying low variability of alpha — provides the best available measurement of historic “value added” by a manager.
There are many other variations to the Sharpe ratio, such as the Sortino and Treynor ratios. Perhaps the most notable risk-adjusted measure of funds is the Morningstar star system (Morningstar also offers other performance measurement statistics in its research reports). For years, critics scoffed that the stars had no predictive value; Morningstar management more or less agreed. In June 2002, Morningstar changed the system, putting more emphasis on risk and less weight on raw performance. So far the change seems to be an improvement. During the system's first two complete years, five star funds lived up to their billing, outperforming everything else in nearly every category. The results were most notable in the international equity field. Funds that began the period with five stars were still thriving after two years, ranking in the 32nd percentile for performance. Four-star funds were in the 47th percentile, while lowly one-star holders stood at the 56th percentile. “The findings suggest that you may want to avoid the one- and two-star funds,” says Russel Kinnel, mutual fund research director of Morningstar.
These funds receive high marks by Morningstar. Franklin Templeton's Mutual Shares A (Sharpe 1.43, alpha 5) is considered a low-risk fund with above-average returns; Dodge & Cox Stock (Sharpe 1.19, alpha 5.6) is considered an average risk fund with high returns; and Third Avenue Value (Sharpe 1.69, alpha 12.3) a below-average risk fund with high returns. (For more fund picks, see page 89.)
Value Line Mutual Fund Survey
Value Line has a different system, where by it assigns managers a numerical rating representing the percentage difference between a fund's actual return and its expected return. The expected return is determined by comparing a fund's standard deviation to the average of a peer group, then multiplying by the peer group's average return. Critics of the Value Line system are those who believe the management factor is given more importance than it deserves, because many funds are managed by committee and because expense ratios and interest-rate trends are an equal or greater factor in returns.
The shortcomings of the above-mentioned ratings underscore the importance of performance attribution, a field of study related to performance measurement that reveals the parts of portfolios responsible for overall returns, and of asset-allocation decisions, which may or may not be part of a manager's responsibility but, some argue, have historically trumped individual security selection as an explanation of results.
Since predicting managers is so tricky and performance measurement an imperfect science, many advisors simply devise their own informal systems. One top-producing investment advisor at one of the world's leading financial institutions was asked how he deals with risk measurement questions.
He doesn't. “Getting the best return for the least risk is every advisors goal,” he says, but he focuses on asset classes and customer satisfaction. He says the investor-centered Lipper rankings (lipperweb.com) are his best first screen. When portfolios with comparable variability are grouped, raw total returns are, in effect, risk-adjusted. He does consult Morningstar to satisfy himself on questions like continuity of management and for more detailed information on expenses and fees.
Performance analytics are now on the desktop of retail investors, who can access data for very little money. (Besides Morningstar and Lipper, there are analytics such as Zephyr Associates, Statpro, Barra, Thompson Vestek, Wilshire Associates and more.) Advisors would be wise to keep up on the latest tools, since the retail set is far more sophisticated than ever.
But one must be careful when using formulations like the Sharpe ratio. Institutional consultants use them as tools in much more elaborate modeling; the danger is that retail advisors may substitute the raw analytics for real due diligence on things such as manager tenure, investing philosophy and buy/sell disciplines. Of course, no amount of study and due diligence can get around the problem: Past performance does have limited predictive value.