Contrarian Investment Strategies

by David Dreman

  In the first place, it has been known for decades that there is no correlation between risk, as the academics define it, and return. Higher volatility does not give better results, nor lower volatility worse results.

  J. Michael Murphy, in an important but little-read article in The Journal of Portfolio Management in the fall of 1977, reviewed the research on risk.2 Some of the conclusions were startling, at least for EMH and CAPM believers. Murphy cited four studies that indicated that “realized returns often tend to be higher than expected for low-risk securities, and lower than expected for high-risk securities, . . . or that the [risk-reward] relationship was far weaker than expected.”3 He continued, “Other important studies have concluded that there is not necessarily any stable long-term relationship between risk and return;4 that there often may be virtually no relationship between return achieved and risk taken;5 and that high volatility unit trusts were not compensated by greater returns”6 (italics in original).7

  In 1975, a paper by Robert Haugen and James Heins analyzing risk concluded with the statement “The results of our empirical effort do not support the conventional hypothesis that risk (volatility)—systematic or otherwise—generates a special reward.”8 Remember, this research was done in the middle to late 1970s, just as CAPM and the concept of risk-adjusted returns were starting the investment revolution and more than a decade before Nobel Prizes were awarded to its advocates.

  The lack of correlation between risk and return was not the only problem troubling academic researchers. More basic was the failure of volatility measures to remain constant over time—the assumption of constancy being central to both CAPM and MPT. Recall that Nobel laureate Robert Merton of Long-Term Capital Management almost believed he could set his watch by it—until its instability played an important role in the firm’s implosion. The instability of volatility also played a major role in the collapse of subprime mortgage bonds in 2007–2008, as well as in the 1987–1988 crash, when S&P futures and index volatility shot up enormously, and in the 2000–2002 crash, when dot-com and high-tech stock volatility increased sharply.

  The impact of volatility goes far beyond the market itself. CAPM had long been used by corporate managers to determine the attractiveness of new ventures. Because the accepted wisdom holds that companies with higher volatility must pay commensurately higher returns, CEOs of higher-volatility companies might be ultracautious in investing in a new plant unless they are certain they can receive the extra return the investment must yield.

  On a broader scale, volatility theory, it seems, resulted in bad business decisions in corporate America for a long time, because “good companies” were told that the markets would always have capital available for their growth. This encouraged them to reduce their liquidity. By the 1980s, economists, notably the EMH pioneer Michael Jensen at Harvard, argued that since EMH always got prices right in the market, the best thing corporate CEOs could do, not just for their companies but for the sake of the economy, was to maximize their stock prices.9 This would give them easier and cheaper access to the capital markets. The message came out loud and clear: do what’s best to get your stock price higher, even if it means compromising on your company’s long-term viability and profitability. The theory played a role in the significant decrease in liquidity reserves held by companies during the financial crisis of 2007–2008, which led to a major magnification of the economic downturn. Again we see the type of damage that a flawed theory can wreak.

  Beta Rides to Town

  Although beta is the most widely used of all volatility measures, a beta that can accurately predict future volatility has eluded researchers since the beginning. The original betas constructed by William Sharpe, John Lintner, and Jan Mossin were shown to have no predictive power; that is, the volatility in one period had little or no correlation with that in the next. A stock could pass from violent fluctuation to lamblike docility.

  Since the cornerstone of MPT and an implicit assumption of EMH is that all investors are risk-averse, in the same manner, the absence of a demonstrable beta was a serious problem for the researchers from the beginning. If investors are risk-averse, beta or other risk-volatility measures must have predictive power. That they have not, that there is no correlation between past and future betas, was a major anomaly, a “black hole” in the theory. Without a tenable theory of risk, the efficient-market hypothesis was an endangered species.

  Barr Rosenberg, a well-respected researcher, developed a widely used multifactor beta, which included a large number of other inputs besides volatility to measure the risk of specific securities. These multifactor betas were often called “Barr’s bionic betas.” Unfortunately, they were as hapless as their predecessors. Other betas were experimented with, all with the same result. Future betas of both individual stocks and portfolios were not predictable from their past volatility.

  The evidence, for the most part, was kept on the back burner until Eugene Fama put out his own paper on risk and return in 1992. Fama and his coauthor, James MacBeth, had published a paper in 1973 indicating that higher beta led to higher returns.10 It was one of the instrumental pieces of CAPM. Later, collaborating with Kenneth French, also at the University of Chicago, Fama examined 9,500 stocks from 1963 to 1990.11 Their conclusion was that a stock’s risk, measured by beta, was not a reliable predictor of performance.

  Fama and French found that stocks with low betas performed roughly as well as stocks with high betas. Fama stated, “Beta as the sole variable in explaining returns on stocks is dead.”12 Write this on the tombstone: “What we are saying is that over the last 50 years, knowing the volatility of an equity doesn’t tell you much about the stock’s return.”13 Yes, make it a large stone, maybe even a mausoleum.

  An article in Fortune on June 1, 1992, concluded, “Beta, say the boys from Chicago, is bogus.”14 The Chicago Tribune summed it up well: “Some of its best-known adherents have now become detractors.”15

  If not beta, then what? If risk cannot be measured by volatility, how can it be determined? According to Professor French, “What investors really get paid for is holding dogs.”16 Their study, as we will see, indicated that stocks with the lowest price-to-book-value ratios and lowest P/Es provide the highest returns over time, as do smaller-capitalization companies. Stock returns are more positively related to these measurements than to beta or other similar risk criteria.17

  Fama added, “One risk factor isn’t going to do it.” Investors must look beyond beta to a multifactor calculation of risk, which includes some value measurements and other criteria.18

  Fama, along with Kenneth French, in a paper in 1996, refuted another written by academics who sought to defend beta, stating, in part, “‘It’ [beta] cannot save the Capital Asset Pricing Model, given the evidence that Beta alone cannot explain expected return.”19

  Buried with this canon of modern finance is modern portfolio theory, as well as a good part of EMH. Fama’s new findings rejected much of the academic work of the past, including his own. He said at beta’s graveside, “We always knew the world was more complicated.”20 He may have known it, but he did not state the fact for more than two decades. His statement that “beta is dead” was the shot at volatility heard round the financial world.

  Although the beta model and thus CAPM were shattered, Fama was not in mourning for long. In 1993, he and French substituted a new theory of risk to take its place, the three-factor model, scarcely a year after beta was buried. A suspicious type might wonder if the new romance had not already been blossoming before the funeral. The formula they introduced included small-capitalization (small-cap) and low-market-to-book-value (value) stocks in addition to beta.*37 Fama did say that “the three-factor model is not a perfect story.”21 Very true. It was an anomaly the professor found that seemed to give EMH more accurate volatility measurements, although no explanation of why it should be used was given.

  As we’ll see, EMH urgently needed new measurements that would show some correlation between highe
r return and higher volatility to save it from extinction. This methodology and reasoning, as we’ll soon see, seems to have twisted scientific method into a pretzel to keep the key pillar of EMH from collapsing. As one financial professor, George Frankfurter, put it in discussing the Fama and French findings:

  Modern finance today resembles a Meso-American religion, one in which the high priest not only sacrifices the followers—but even the church itself. The field has been so indoctrinated and dogmatized, that only those who promoted the leading model from the start are allowed to destroy it.22

  This is not just ivory-tower stuff, as we’ve seen. Beta and other forms of risk measurement determine how trillions of dollars are invested by pension funds, other institutional investors, and the public. High betas are no-nos, while the money manager who delivers satisfactory returns with a low-beta portfolio is lionized.23

  Take, for example, Morningstar, the largest service monitoring mutual funds. Although it is an excellent and easily readable source that I refer to often, its concept of risk is problematic. Morningstar’s five stars, its top ranking, widely followed and much sought after, uses Fama’s three-factor model, which is dubious at best, as part of its risk measurement.

  The Failure of CAPM and Ensuing Volatility Theory

  We saw that volatility theory has never worked and has cost most people who relied on it dearly over four decades. However, the researchers are putting together even more problematic risk/volatility hypotheses that almost send chills down my spine. My advice is to avoid it entirely. The next few pages will tell you why.

  As we’ve seen, Professor Fama and many others have abandoned CAPM, the original volatility theory, acknowledging that it has failed, while Nobel laureates such as William Sharpe still dispute this contention.24 Much as the stream of new findings of celestial motion based on telescopes’ improved accuracy destroyed the Ptolemaic system, so, as new, more powerful statistical information poured in that contemporary volatility measures do not work, CAPM and EMH were similarly threatened.

  The Flaws in EMH-CAPM Volatility Assumptions

  EMH proponents recognize the danger their theory is now facing. For investors to be omnisciently rational, there must be a systematic correlation between risk and return. Without it, EMH goes the way of the brontosaurus. If some investors get more return with the same or lower volatility and others get lower returns over time with the same or higher volatility, as we saw in the previous chapter, this indicates that investors are not omnisciently rational, and a dagger is pointed at the heart of EMH and CAPM.

  To defend efficient markets, Fama, as we saw, after abandoning CAPM, went to the three-factor model of risk in his 1992 paper, while others have gone on to a plethora of theories, including four- and five-factor risk evaluation models, to show that there is still a correlation between risk and return.

  However, this leads to a number of serious errors in the formulation of the new models. These models are built to replace CAPM; but they are built in an identical fashion, that is, they all attempt to show that higher volatility provides higher return and that lower volatility provides lower return. If CAPM could not do this, why should any new model built in an identical manner do so? CAPM was dumped because it didn’t work, but its central tenet, that there is a direct correlation between risk and return, is being kept as the core of EMH risk analysis.

  So researchers must search for new sets of risk and return variables that will give them the correlation between higher risk and higher return. In effect, they must create a new CAPM (naturally with a new name) that behaves almost exactly as the old one was supposed to behave but didn’t. In short, it appears they are attempting to create something along the lines of a financial Stepford wife. But this leads them deeper into the theoretical jungle.

  Where can you find these critical new risk variables that will work and give you better results than the old CAPM, which was proved not to work? You can’t very well advertise for them in the classifieds of The New York Times or The Wall Street Journal under “Wanted: new risk factors.” And certainly not under “Wanted: financial Stepford wives.”

  Unfortunately, this is where efficient market researchers are today. Have they subsequently found a way out of the volatility woods? Not exactly. The first problem is that they have presented a grab bag of simple correlations that attempt to show a link between risk and return. This is very dangerous scientific ground. As Milton Friedman warned, “If there is one . . . [correlation] . . . consistent with the available evidence there are always an infinite number that are.”25 Even if the academics found a correlation between volatility and return, which is doubtful from what we have seen, there could be hundreds of others that explain risk and return better.

  The first problem researchers have is that they cannot prove that the three-factor, the four-factor, or any other models they put forth actually work rather than simply being chance correlations. As Friedman also noted, one of the basic rules of scientific method is that correlation is far removed from causation. There can be innumerable chance correlations for any effect, but without reasonable evidence that one or another is true, they are pure coincidence and likely to go away with time. A humorous example of chance correlation was the Hemline Indicator, which was shown in a well-known Wall Street chart for decades. When fashion dictated that hemlines should be high, in periods such as the 1920s, the 1960s, the 1980s, and the 1990s, markets roared ahead. When hemlines dropped in the 1930s, in the 1940s, and again in the 1970s, markets went lower. According to this tongue-in-cheek hypothesis, the height of hemlines dictated where markets would go. Obviously, few took it seriously.

  However, many EMH researchers seem to accept chance correlations as proof, although there is no evidence that a myriad of other variables might also correlate as well as or better than the ones they’ve selected. No new theory of volatility and return has been discovered that shows a consistent correlation between the two. Still, the EMH researchers continue to struggle to find one. They have scanned thousands of possible financial variables in an attempt to do this, but none has worked consistently.

  Not only is this bad science; it is likely to blow up over time because of the lack of critical scientific underpinnings to these correlations. Importantly, Fama, in his 1998 survey of market efficiency, wrote that all of the volatility models tested to date “are incomplete descriptions of average returns.”26 In brief, they do not work consistently. Using these methods, the researchers seemingly entrapped themselves.

  Unfortunately, this logic has yet another hurdle to clear. Even if it were true, it is not enough to say that a correlation between risk and return has been discovered or shortly will be. If markets are efficient, the sophisticated investors that supposedly keep prices in line with value must have known about the new correlation generations back, even if the academics didn’t. If they did not, how have markets been efficient over the decades? Since there is no evidence that a new proof of volatility has been brought forth or will be, we have to conclude that EMH risk measurements have not been correct and may have been significantly wrong for decades. The logical jungle seems to get more impenetrable with each new finding.

  But without this volatility correlation EMH goes the way of the Ptolemaic system. It is also remarkably similar to the Ptolemaics working on epicycles and eccentric circles to attempt to save their system. It’s sad to see gifted researchers cross over from the bounds of scientific discovery to enter a world of ideologues, some of whom at times appear to be almost zealots.

  The Achilles’ Heel of EMH

  In chapter 4 we saw the supposedly overwhelming evidence that it is impossible to beat the market over time. Let’s now look more closely at the original work the researchers did, which “proved” that no investor could beat the markets. I’m sure the results will surprise some of you.

  There’s no clearer statement of the testament, according to Fama,27 than his own concise description of efficient markets: if the necessary conditions for market efficie
ncy are present—i.e., information is readily available to enough investors and transaction costs are reasonable—there is no evidence of consistently superior or inferior returns to market participants.

  The argument assumes that thousands of analysts, money managers, and other sophisticated investors search out and analyze all available information, constantly keeping prices in line with value.28 Since the academics claimed that it was difficult to assess how investors analyze information to determine undervalued stocks, tests of this premise focus on whether groups of investors have earned superior returns. The group whose members most frequently serve as guinea pigs is that of mutual fund managers, because information about their decisions and performance is readily available. The research shows that mutual funds do not outperform the major averages, whether risk-adjusted or not, although the risk-adjusted studies that support the efficient-market hypothesis are now certainly open to question.

  Flawed Statistics and Kangaroo Courts

  The statistics of the original mutual fund researchers in the 1960s and early 1970s failed to turn up above-average performance by investors, thereby contributing the essential evidence to make the EMH case.

  But on closer examination, the efficient-market victory vanishes. Studies have demonstrated that the standard risk adjustment tools the researchers used back then were too imprecise to detect even major fund outperformance by money managers of their benchmark average. The statistical tests used made it extremely difficult to show superior manager performance when it existed, because the hurdles that outperforming portfolios had to clear were set far too high. One, for example, showed that using the techniques of Michael Jensen, only one manager of the 115 measured demonstrated superior performance at a 95 percent confidence level, the lowest statistical level normally acceptable.29