Contrarian Investment Strategies

by David Dreman

  Dropping companies with small earnings per share to avoid large percentage errors does not eliminate the problem; the error rate is still over 20 percent. Worse yet, analysts often err. Figure 8-2 showed that only 30 percent of consensus analysts’ estimates fell within the 5 percent range of reported earnings. Remember, too, that many analysts think this range is too wide. Missing the 5 percent range would spell big trouble for stock pickers relying on precision estimates.

  Unfortunately, the problems do not end here. Forecasting by industry was just as bad; the current studies to 2010 (and an earlier study between 1973 and 1996; see footnote, page 195) showed error rates almost indistinguishable between industries with supposedly excellent visibility, for which investors pay top dollar, and those considered to have dull prospects. If earnings estimates are not precise enough to weed out the also-rans from the real growth stocks, the question naturally arises why anyone would pay enormous premiums for “high-visibility” companies.

  Finally, we have seen two additional problems with analysts’ forecasts. First, the error rates are not due to the business cycle. Analysts’ forecast errors are high in all stages of the cycle. Second, and more important, analysts have a strong optimistic bias in their forecasts. Not only are their errors high, but there is a consistent tendency to overestimate earnings. This is deadly when you pay a premium price for a stock. The towering forecast errors combined with analysts’ optimism result in a high probability of disaster. As we saw, even a slight “miss” for stocks with supposedly excellent visibility has unleashed waves of selling, taking the prices down five or even ten times the percentage miss of the forecasting error itself.

  The size and frequency of the forecasting errors call into question many important methods of choosing stocks that rely on finely tuned estimates running years into the future. Yet accurate earnings estimates are essential to most of the stock valuation methods we looked at in chapter 2. The intrinsic value theory, formulated by John Burr Williams, is based on forecasting earnings, cash flow, or dividends, often two decades or more ahead. The growth and momentum schools of investing also require finely calibrated, precise estimates many years into the future to justify the prices they pay for stocks. The higher the multiple, the greater the visibility of earnings demanded.

  If the average forecast error is 40 percent annually, the chance of hitting a bull’s-eye on an estimate ten years out seems extremely slim. Two important questions might be asked at this point. The first is whether efficient-market theorists and androidlike analysis keep prices where they should be by flawlessly processing key information accurately. The charts just presented indicate that the information input by analysts to make their estimates is anything but correctly processed. Given the results we’ve viewed, if serious errors are repeatedly made with the vital analytical inputs, what keeps markets efficient? Second, analysts do not learn from their errors, as we’ve seen for periods exceeding thirty years. Rational investors should adjust almost immediately to keep markets efficient. Why is this not done? Unfortunatley, a gaggle of black swans has landed in this chapter, detailing forecast errors that once again must be labeled “anomalies” to protect EMH. Which brings us to another Psychological Guideline.

  PSYCHOLOGICAL GUIDELINE 13: Most current security analysis requires a precision in analysts’ estimates that is impossible to provide. Avoid methods that demand this level of accuracy.

  Forecasting Follies 7: Hey, I’m Special

  What can we make of these results? If the evidence is so strong, why aren’t more investors, particularly the pros, aware of it, and why don’t they incorporate it into their methods, rather than quick-marching into an ambush? Why do Wall Streeters blithely overlook these findings as mere curiosities—simple statistics that affect others but not them? Many pros believe that their own analysis is different—that they themselves will hit the mark time and again with pinpoint accuracy. If they happen to miss, why, it was a simple slip, or else the company misled them. More thorough research would have prevented the error. It won’t happen again.

  Let’s examine why this mentality is prevalent in the face of overwhelming evidence to the contrary. It should be an interesting drama in several acts. The show has a terrific cast of experts from many fields, the action is compelling, and there’s a heartwarming (perhaps make that “portfolio-warming”) lesson for the audience.

  Forecasting Follies 8: Some Causes of Forecasting Errors

  As we just saw, investors either ignore or are not impressed by the statistical destruction of forecasting, even though the devastation has been thorough and spans decades. There are a number of reasons, some economic, some psychological, why investors who depend on finely calibrated forecasts are likely to end up with egg on their face. Two academics, John Cragg and Burton Malkiel (the latter of whom we met briefly in chapter 6), did an early analysis of long-term estimates, published in the Journal of Finance.14 They examined the earnings projections of groups of security analysts at four highly respected investment organizations, including two New York City banks’ trust departments, a mutual fund, and an investment advisory firm. These organizations made one- to five-year estimates for 185 companies. The researchers found that most analysts’ estimates were simply linear extrapolations of current trends with low correlations between actual and predicted earnings.

  Despite the vast amount of additional information now available to analysts, say Cragg and Malkiel, and their frequent company visits, estimates are still projected as a continuation of past trends. “The remarkable conclusion of the present study is that the careful estimates of security analysts . . . performed little better than those of past company growth rates.”

  The researchers found that analysts could have done better with their five-year estimates by simply assuming that earnings would continue to expand near the long-term rate of 4 percent annually.15

  Yet another important research finding indicates the fallibility of relying on earnings forecasts. Oxford Professor Ian Little, in a 1962 paper appropriately titled “Higgledy Piggledy Growth,” revealed that the future of a large number of British companies could not be predicted from recent earnings trends.16 Little’s work proved uncomfortable to both EMH theoreticians and EMH practitioners, who promptly criticized its methodology. Little good-naturedly accepted the criticism and carefully redid the work, but the outcome was the same: earnings appeared to follow a random walk of their own, with past and future rates showing virtually no correlation, and recent trends (so important to security analysis in projecting earnings) provided no indication of the future course.17

  A number of studies reach the same conclusion:18 changes in the earnings of U.S. companies fluctuate randomly over time.

  Richard Brealey, for example, examined the percentage changes of earnings of 711 U.S. industrial companies between 1945 and 1964. He, too, found that trends were not sustained but actually demonstrated a slight tendency toward reversal. The only exception was companies with the steadiest rates of earnings growth, and even their correlations were only mildly positive.19

  Juxtaposing the second set of studies with the first provides part of the explanation of why analysts’ forecasting errors are so high. If analysts extrapolate past earnings trends into the future, as Cragg and Malkiel have shown, and earnings do follow a random walk, as Little and Brealey demonstrated, one would expect sizable errors. And large forecast errors are what we have found consistently.

  Thus, once again and from quite another tack, we see the precariousness of attempting to place major emphasis on earnings forecasts. Which calls for another Psychological Guideline:

  PSYCHOLOGICAL GUIDELINE 14: It is impossible, in a dynamic economy with continually changing political, economic, industrial, and competitive conditions, to use the past to estimate the future.

  There are several other economic reasons that can cause earnings forecasts to be off base. One is what the Harvard economist Richard Zeckhauser calls “the big bath theory.” In a paper that he wrote wi
th Jay Patel of Boston University and François Degeorge of the HEC School of Management in Paris, the researchers provided evidence that many companies try to manage earnings by attempting to show consistent, gradual improvements.20 Analysts have an appetite for steady growth, and that is what management tries to serve up. When the managers can’t do it, they take a “big bath,” writing off everything they can, perhaps even more than is necessary (accounting again), in order to produce a steady progression of earnings after the bath. The big bath could be another unpredictable effect that throws analysts’ forecasts off.

  Reviewing the evidence makes it appear that forecasting is far more art than science and, like the creative fields, has few masters. Apart from highly talented exceptions, people simply cannot predict the future with any reliability, as the figures starkly tell us.

  BEHAVIORAL FINANCE: CAREER PRESSURES ON ANALYSTS’ RECOMMENDATIONS

  There are some substantive factors that affect analysts directly, the most important being career pressures (behavioral finance labels this as a part of agency theory*48). These can result in forecasts that stray markedly. After surveying the major brokerage houses some years back, John Dorfman, then editor of the market section of The Wall Street Journal, whom we met earlier, provided a list of what determines an analyst’s bonus, normally a substantial part of his or her salary.21 In Dorfman’s words, “Investors might be surprised by what doesn’t go into calculating analysts’ bonuses. Accuracy of profit estimates? That’s almost never a direct factor. . . . Performance of stocks that the analyst likes or hates? . . . It is rarely given major weight.” The ranking of seven factors determining an analyst’s compensation places “accuracy of forecasts” dead last.

  What is most important is how the analyst is rated by the brokerage firm’s sales force. That was precisely the same pressure on analysts seen earlier during the Internet bubble, which led many into a serious conflict of interest with part of their clientele.

  Many firms conduct a formal poll of the sales force, which ranks the analysts primarily on how much commission business they can drum up. At Raymond James, the sales force’s ratings accounted for 50 percent of the analysts’ bonuses. Top executives also review a printout of how much business is done in the analysts’ stocks. The report is called “stock done” for short. PaineWebber*49 kept careful records of what percentage of trades it handled in every stock, and its market share in stocks it provides research for compared with market share of competitors. Michael Culp, then the director of Prudential Securities,*50 instituted a rule requiring his analysts to make 110 contacts a month—but, he added quickly, most of his analysts were not affected, because they were already making 135. Another firm ranks analysts’ recommendations when calculating their bonuses. A buy recommendation is worth 130 points, a sell recommendation only 60, because sell recommendations don’t generate nearly as much business as buy recommendations; no points are added for accuracy.22 We also have seen where this road led in the earlier section on the marketing of dot-com stocks.

  Making the Institutional Investor All-Star Team, according to The Wall Street Journal list, ranks second. Having an analyst on “the Team,” as we’ve seen, results in big commissions for the firm. Even as he denied the importance of the Institutional Investor poll, one research director said, “Most of the guys know they’ll be visiting for I-I in the spring.” That is, the analyst will be making the annual pilgrimage to visit institutional clients, implicitly lobbying for their vote to fame and fortune. “I’m a lonely guy in March and April, shortly before the balloting,” he continued.

  Do you find these remarks disturbing? The analysts’ behavior is certainly better than it was during the dot-com bubble, but the temptations not to put out sell reports and the rewards for putting out as many buy recommendations as possible are high. Most investors directly or indirectly put their savings into the hands of research analysts whose forecasts are the cornerstones of their recommendations. But the forecasts are a trivial factor or even a nonevent in determining their compensation. Unfortunately, that’s always been the name of the game. The key question for the investor is whether an analyst is giving you his best recommendations or, given his compensation, which is based upon maximizing his commissions generated, recommending what is likely to sell the most. Obviously, this is impossible to pinpoint. But there is a good Psychological Guideline that will help:

  PSYCHOLOGICAL GUIDELINE 15: If you’re interested in an analyst’s recommendation, get all of his reports for at least the last three to five years and see how they’ve done. If they haven’t done well or he can’t provide them, move on.

  There are yet other direct pressures on analysts. An important one, well known on the Street, is the fear of issuing sell recommendations. Sell recommendations are only a small fraction of the buys. A company that the analyst issues a sell recommendation on will often ban him from further contact. If he issues a sell recommendation on the entire industry, he may receive an industry blackball, which virtually excludes him from talking to any important executives. If the analyst is an expert in the industry and it represents the main part of his intellectual property, he is facing major career damage by pressing the sell button.

  Recommending a sell, even when the analyst proves to be dead-on, can be costly. In the late 1980s, an analyst at Janney Montgomery Scott issued a sell recommendation on one of the Atlantic City casinos owned by Donald Trump. Trump went bananas and insisted that the analyst be fired for his lack of knowledge. Shortly thereafter, he was fired—but naturally, said the brokerage firm, “for other reasons.” The analyst proved right, and the casino went into Chapter 11 bankruptcy. Out of a job, he won the equivalent of a few years’ salary from an arbitration panel. But he was never rewarded for his excellent call and, in fact, suffered because of it.

  Another analyst was banned from an analysts’ meeting of the then high-flying Boston Chicken (later named Boston Market). His offense: he had issued a sell recommendation on the company. “We don’t want you here,” Boston Chicken’s CFO told him. “We don’t want you to confuse yourself with the facts.”23 The facts were that Boston Chicken filed for Chapter 11 bankruptcy not long afterward. A number of studies indicate that analysts issue five or six times as many buy as sell recommendations.24 Obviously, career pressures have an impact on the buy-sell-hold rating.

  Many companies retaliate when analysts write negative reports on them. The retribution can take many forms. One analyst at Prudential Securities wrote a number of negative reports about Citicorp in 1992. Frustrated that Prudential could not become the lead underwriter in some asset-backed bond deals, a Prudential investment banker went to Citicorp and was told that the reason was the analyst. The same analyst, a year later, criticized Banc One and its complex derivative holdings, which eventually cost it hundreds of millions of dollars in write-offs. Banc One stopped its bond trading with Prudential. By coincidence, the analyst left the firm shortly thereafter. At Kidder Peabody an analyst repeatedly recommended the sale of NationsBank (now Bank of America). The bank stopped all stock and bond trading for its trust accounts with Kidder.25

  For analysts at brokerage firms that are also large underwriters, the pressure is even greater. Negative reports are a major no-no. Bell South officials were unhappy about the comments of a Salomon Brothers analyst who stated that its management was inefficient and ranked it sixth out of the seven regional Bells. Salomon, the bond powerhouse, was excluded from the lucrative lead-manager role in a large Bell South issue. In late 1994, Conseco fired Merrill Lynch as its lead underwriter in a big bond offering shortly after its analyst downgraded Conseco’s stock. Smith Barney, according to sources, believes it lost a chance to be part of the underwriting group of Owens-Corning Fiberglas after one of its analysts wrote a negative report on the company.26

  Just how heavy career pressures can be for analysts working for major underwriting firms if they recommend a sale is shown in an academic study. The work examined 250 analyst reports from investment banki
ng houses, matching them up with 250 from brokerage firms that did not conduct investment banking. The conclusion: investment banking house brokers issued 25 percent more buy recommendations and a remarkable 46 percent fewer sell recommendations.27

  What is apparent from the above circumstances and those we saw earlier is that an analyst’s most important responsibility is to be a good marketer of the brokerage firm. The analyst must tell a good story, not one that is necessarily right. The bottom line is commissions. A good marketer and a good forecaster are different animals. We have already seen one example of the “All-Stars” significantly underperforming the market with their picks. Another example is that major money managers, to whom the All-Stars devote the bulk of their attention, consistently do worse than the averages.

  Analysts of necessity use disingenuous gradations that actually mean sell, such as underweight, lighten up, fully valued, overvalued, source of funds, swap and hold, or even strong hold. Peter Siris, a former analyst at UBS Securities, summed it up well several decades ago: “There’s a game out there. Most people aren’t fooled by what analysts have to say . . . because they know in a lot of cases they’re shills. But those poor [small] investors—somebody ought to tell them.”28 You’ve been warned!

  Even if brokerage firms don’t focus on accurate estimates, however, analysts are not punished for producing them. And although firms may be pressured by their underwriting clients not to make sell recommendations on their stocks, to my knowledge an analyst has never been reprimanded for writing a highly optimistic report on them.