by David Dreman
We have seen three distinctly different reactions to earnings surprises by the high, low, and middle stock groups using three of the most important value measurements. However, as with the weather, not all days can be sunny and not all news can be good. Negative surprises, which normally send chills down investors’ spines, are the other side of the coin we need to examine.
The Effects of Negative Surprises
Figures 9-5 and 9-6 show the effect of negative surprise on the “best,” “worst,” and middle groups by price to cash flow and price to earnings. Out-of-favor stocks again win in a breeze. Let’s start by looking at price to cash flow (Figure 9-5). Negative surprises in analysts’ forecasts have a minimal impact on the lowest 20 percent of stocks in the surprise quarter, and as a result this group falls below the market by only 3/10 of 1 percent. Moreover, the market shrugs off the surprise by the end of the year, with the lowest-price-to-cash-flow group outperforming the market by 1.3 percent. (The results for price to book value are similar but are not shown here.)
Negative surprises are like water off a duck’s back for this out-of-favor group. Investors have low expectations for what they consider lackluster or bad stocks, and when these stocks do disappoint, few eyebrows are raised. The bottom line is that a negative surprise is not much of an event in the surprise quarter and is a nonevent in the nine months following.
Consider the “best” companies, however. Investors expect only glowing results for these stocks. After all, they confidently—overconfidently—believe that they can divine the future of a “good” stock with precision. These stocks are not supposed to disappoint; people pay top dollar for them for exactly this reason. So when a negative surprise arrives, the results are devastating.
Figure 9-5 shows how the “best” stocks, by price-to-cash-flow ratio, reacted to negative earnings surprises. In the quarter in which investors received the news, the stocks underperformed the market by a startling 3.6 percent on average. They did 12 times as poorly as the lowest-price-to-cash-flow group when receiving “bad” surprises. Worse yet, whereas the most out-of-favor stocks outperformed the market slightly in the next nine months, the favorites continued to drop. At the end of the year, they were 9.4 percent under the average. Favorite stocks with negative surprises underperformed the market’s return of 14.8 percent by a shocking 64 percent annually, on average, over the thirty-eight years of the study. As Figure 9-5 also shows, the lowest-price-to-cash-flow group outperformed the highest by a remarkable 10.7 percent in years when both groups suffered negative surprises.
Figure 9-6, which measures negative surprises by price-earnings ratio, shows similar results. Negative surprises on the highest 20 percent of P/Es cause the stocks to drop sharply in the surprise quarter; this drop is followed by a larger decline in the next nine months.
What do we make of numbers like these? It’s apparent that investors are shaken when companies they expect to excel disappoint. The disappointment does not have to be large. You may remember that we purposely used a very small analyst forecast error—one cent and over—to see how precise estimates have to be. From the major declines of high-visibility stocks on even nominal forecast errors, it’s obvious that the pinpoint accuracy demanded of earnings forecasts should not play a major role in valuing a stock. Yet, as we have seen, it’s at the heart of security analysis practiced today. A pilot would not use a GPS in his plane that could only get him accurately to within some hundreds of miles of his destination, yet he is quite comfortable using a financial GPS that does just that.
We saw in the previous chapter (Table 8-3) that the probability of avoiding a negative surprise of more than 5 percent was 1 in 62 for ten quarters and 1 in 3,800 for twenty quarters. The current study indicates that investors’ tolerance for negative surprises on popular stocks is much lower than this. Considering the devastating effects of negative surprises on favorite stocks, these are odds no rational person should want to face.
Too, we saw in the preceding chapter that analysts are, on the whole, overoptimistic in their forecasts. The combination of analysts’ noted overoptimism and their large forecast errors—in one landmark study the error was 9 percent annually—is lethal to buyers of favorite stocks.6
Finally, Figures 9-5 and 9-6 show the expected: negative surprises have more effect on the 60 percent of stocks in the middle group than on the lowest-price-to-cash-flow stocks but far less than on the highest-price-to-cash-flow group for both the surprise quarter and the year. But this is mostly offset by their outperformance with positive surprises. Figures 9-1, 9-2, and 9-3 show very similar results. All the findings behind the charts in this chapter are statistically significant.7
The Effects of Event Triggers
Regardless of which valuation method was used, when earnings of out-of-favor stocks came in above analysts’ forecasts, they shot out the lights. Just as apparent was the sharp underperformance of the winners, the top 20 percent of stocks, as measured by the price-to-earnings or price-to-cash-flow ratio when analysts’ estimates were too optimistic.
What the study shows, then, is that the overvaluation of “best” stocks and the undervaluation of “worst” stocks are often driven to extremes. That brings us quite naturally to the next Psychological Guideline.
PSYCHOLOGICAL GUIDELINE 19: Positive and negative surprises affect “best” and “worst” stocks in a diametrically opposite manner.
People are far too confident of their ability to predict complex outcomes in the future. This has been shown to be true in many fields from medicine and law to building new plant facilities. The stock market, with its thousands of continually shifting company, industry, economic, and political events, certainly ranks among the most formidable areas in which to make forecasts.
Good or bad news, which occurs frequently in markets, results in diametrically opposite movements of “best” and “worst” stocks. When we recall that money managers are considered “stars” if they can outperform markets by 2 percent or 3 percent annually over a five-year period, the 3.4 percent annual outperformance of the “worst” stocks after all surprises, as measured by price to cash flow (shown in Figure 9-2, with similar results in Figures 9-1 and 9-3), coupled with the 3.6 percent underperformance of the “best,” or 7.0 percent total outperformance of low-P/E over high-P/E stocks, is enormous. The disparity, of course, is firmly anchored in investors’ overconfidence in pinpointing future events. We thus see that earnings surprises have an enormous, predictable, and systematic influence on stock prices.
Looking carefully at the charts, we can also see that earnings surprises cause two distinct categories of price reactions in both “best” and “worst” stocks. I’ll call the first an “event trigger” and the second, which will be discussed shortly, a “reinforcing event.”
I define an event trigger as unexpected negative news about a stock believed to have excellent prospects, or unexpectedly positive news about a stock believed to have a mediocre outlook. As a result of the event trigger, people look at the two categories of stocks very differently. They take off their dark or rose-colored glasses. They now evaluate the companies more realistically, and the reappraisal results in a major price change to correct the market’s previous overreaction.
THE FIRST EVENT TRIGGER
There are two types of event triggers. The first is a negative surprise for a favored company, which will drive its stock price down. The second is a positive surprise for an unfavored stock, which pushes its price much higher. The event trigger initiates the process of perceptual change among investors, which can continue for a long time. As has been shown, the process goes on beyond the quarter in which the surprise is reported and through the year following the surprise. In the next section, we’ll see that it actually continues for much longer periods.
Event triggers can result from surprises other than earnings. A non-earnings surprise might be the FDA’s approval of an important new drug or its denial of further testing. Winning or losing a landmark tobacco case woul
d be another. New technology that suddenly makes a semiconductor obsolete would be a third. There could be hundreds of such surprises, any one of which could have a sharp and lasting impact on a stock’s price. Although the idea has not yet been tested, observation suggests that the impact of such surprises on stock prices would be similar to the impact earnings surprises have on the best, worst, and middle groups.
Event triggers are, however, most frequently earnings surprises. The first type of event trigger is a negative surprise on a highly regarded company. An example is the free fall in the price of Amgen, the world’s largest independent biotech medical company, which has a strong product line for the management of cancers and other serious diseases. Its major products were Aranesp and Epogen to treat anemia in cancer patients. The company added another promising drug for cancer, Vectibix, in 2006. Amgen showed outstanding earnings growth from 2002 through 2005, and the stock price moved from $57 in early 2005 to $86 later that year owing to its earnings, strong product line, and promising pipeline of potential new drugs. Analysts continued to increase their earnings forecasts for the stock.
Then the roof fell in. In late 2006, the company, in attempting to further expand its market for Aranesp, found that the mortality rate, always a concern with this drug, was slightly higher in a new study than in previous ones. Significant concern was expressed by leading oncologists, and recommendations were made to decrease the allowable dosage or possibly ban the drug entirely, which shocked both analysts and investors. Earnings estimates were cut well below the prior exuberant forecasts for a number of quarters, and the stock was downgraded by many on the Street. The price of Amgen plummeted 54 percent by March 2008.
The doomsday scenario analysts painted did not turn out to be nearly as bad as thought. The FDA mandated additional warnings on Arenesp’s labeling and moderate downward adjustments in dosages were made, but the drug continued to be a major profit center. Earnings growth began to accelerate in 2008 and was up again in 2009. But the luster was gone. By the fall of 2010, Amgen was no longer considered a “favorite stock.” Trading at a P/E of 10х, it was now relegated to the “worst-stocks” category.
The event trigger resulted in a major reassessment of the company by investors. As we saw, investors systematically overrate the future of favored companies. When a negative earnings surprise occurs to a favored stock, accompanied by serious downbeat news, people are shaken by the realization that this could happen to a “best” stock. Their reaction is to sell—fast—sending the prices down, often dramatically. Even when the bad news proved to be not nearly as severe as originally anticipated and the company, in fact, came near to meeting its original earnings targets, the memory of the unpleasant experience lingered. Though many “best” companies bounce back, their stock underperformance continues for some time.
THE SECOND EVENT TRIGGER
Investors do not expect positive surprises from companies they consider to have poor outlooks. When such surprises happen, people begin a process of perceptual change. The stocks are reevaluated more positively, and they outperform the market significantly, largely because of the original undervaluations.
The second type of event trigger is a positive surprise—or a series of positive surprises—for an out-of-favor stock. Take the example of Reynolds American. The company is the second largest cigarette tobacco producer in the United States and was as out of favor as a stock can be.
Through a number of acquisitions, including Conwood Smokeless Tobacco, Reynolds increased its revenues significantly in an industry in which consumption is dropping each year. Reynolds stated publicly that it could increase its profits substantially by consolidating its operations in North Carolina, thereby eliminating well over a billion dollars of expenses and excess plants. The company did exactly that. Beginning with its March 2004 quarter, it produced a string of large earnings surprises, which pushed the stock up 154 percent, including its high 6 percent dividend (a return we can only dream of today) in mid-2007. Investors who bought this stock—and naturally more than a few wouldn’t—did very well.
An Earnings Surprise for a Stock Is Reinforced
by Additional Earnings Surprises
Investors’ perceptions about a company, an industry, or the market itself often do not change with a single positive or negative surprise. Jeffery Abarbanell and Victor Bernard of the University of Michigan, for example, have studied analysts’ estimates and found they are slow to adjust to earnings surprises. Whether the estimate was too high or too low, analysts do not revise it accurately immediately but take as long as three quarters after the surprise to do so.8 When a forecast is too high, it continues to be high for the next nine months, and when it is too low, it continues to be low for the next three quarters.
As Abarbanell and Bernard put it, analysts “underreact to recent earnings reports.” This underreaction generates new surprises, which reinforce investors’ changing opinion of a company. If, for example, investors are taken aback by a negative earnings surprise on a favorite stock and more negative surprises occur in the following quarters (as a result of analysts’ not revising their earnings estimates down enough), people’s increasingly poor reappraisal of the company pushes the stock even lower. The event trigger continues over a number of quarters, as we have seen in the various annual quarterly and annual surprise charts. The same is true for a series of positive surprises on an out-of-favor company.9
The Effects of Reinforcing Events
The second category of earnings surprises is what I call a “reinforcing event.” Rather than changing investors’ perceptions about a stock, these surprises reinforce the current beliefs about the company. Since they do, they should have much less impact on stock prices. Reinforcing events are defined as positive surprises on favored stocks or negative surprises on out-of-favor stocks. A positive surprise on a favored stock reinforces the previous perception that this is an excellent company. Good companies should do well; if they have positive surprises, it is only to be expected.
Microsoft is the world’s largest developer and manufacturer of software and has a major stake in computer equipment. The company is a classic example of a favorite stock experiencing a reinforcing event. In late 2003, “Mr. Softy,” as it’s sometimes called, racked up good growth in both the consumer and the small and medium-sized business markets. It beat estimates handily in late 2003 and early 2004. However, within only a few months the stock was down about 14 percent, and two years later in mid-2006 it was still down about 13 percent, well behind the market. Again a premier company at a premier P/E showed so-so returns even with very positive earnings surprises.
Boeing was an example of a reinforcing event on an out-of-favor stock several years back. The company hit a rough patch with a machinists’ strike and charges related to its 747 and 787 programs, as well as continued woes caused by the global economic downturn that put pressure on its vital commercial airline markets. After missing estimates on the downside four quarters in a row, starting in June 2008, it dropped into the most out-of-favor group, bottoming at $29 per share in March 2009. But it came back strongly, rising over 165 percent by April 2010. Not a bad return from a company whose earnings continued to disappoint!
Figure 9-7 shows how different the impact of earnings surprises is on event triggers and reinforcing events for the surprise quarter (on the left), as well as for the full year (on the right). The figure uses price-earnings ratios to measure the surprise effect, but using price-to-cash-flow or price-to-book-value ratios results in very similar comparisons. The two types of event triggers (negative surprises on favored stocks and positive surprises on out-of-favor stocks) have substantially more impact on stock prices than reinforcing events (positive surprises on favored and negative surprises on out-of-favor issues).
Look first at the event triggers in Figure 9-7, the first two adjoining columns on the left side of the chart, for both the quarter and the year. We see for the event trigger that, adding them together for the average quarter, the
total price impact is 5.7 percent (+2.6 percent and –3.1 percent, removing the signs) in the surprise quarter. By contrast, adding the reinforcing events together (the last two adjoining columns on the right side of the quarterly chart) results in a much smaller surprise impact: 1.3 percent (1.1 percent and –0.2 percent) for the same quarter. For the full year, we also see that the size of the event triggers more than doubles, resulting in a total impact of 14.1 percent. This is because, as we saw, positive surprises for out-of-favor stocks and negative surprises for favored stocks are much larger for the full year than for the surprise quarter alone. Reinforcing events, on the other hand, have a negligible 0.6 percent impact on prices after one year.
Figure 9-7 demonstrates not only two distinct classes of surprises, event triggers and reinforcing events, but the fact that their response to unanticipated good and bad news is remarkably different. Event triggers result in a perceptual change, which continues through the end of the year and has a major impact on stock prices.
The effect of reinforcing events on prices, on the other hand, is minor by the end of the twelve months following the surprise.
Neuroeconomics and the Market