Finally, we compare the two sets of growth and profitability indicators in columns 2 and 3 of Table 11-2, to see whedier the changes in these indicators are the chief cause of "best" and "worst" stock price movements or whether it is investor overreaction.

Are the higher retums for the low price-to-book value stocks the result of their fundamentals improving 8 1 , while those of high price-to-book value drop through the floor? Not at all.

As Table 11-2, column 3 shows: three of the five growth and profitability indicators for the low price-to-book value group do not improve but continue to decline in the five years that their stock prices surge. Two of these measures, profit margins and sales growtii, actually deteriorate significantly. Sales growth, for example, dropped from 10.4% a year to 6.8%, as the lowest price-to-book value stocks shot out die lights. Only eamings-per-share growth improves: 6.4% to 11.6% annually over this five-year period.

The growth and profitability indicators of high price-to-book value stocks also deteriorate some (see Table 11-2, columns 2 and 3), but are still well above the market average in every case, and much higher than these indicators for the lowest price-to-book value group, except for eamings growth. Yet the high price-to-book value stock performance is abysmal. As columns two and three of the table show, the collapse in fundamentals of best stocks and a major improvement in those of worst stocks just doesnt happen. The cause of the dramatically better retums of this contrarian group lies elsewhere. The only conclusion left is that investors do indeed overreact predictably and continually to the prospects of "best" and "worst" stocks.

As we have seen, even mild disappointments in "best" stocks will drive them lower, while positive news has almost no effect. For the lowest price-to-book value quintile the resuhs are precisely the opposite. Investors have overreacted to their poorer prospects so strikingly that even when their fundamentals continue to get worse over this five-year period, the stocks, like a corked bottle, still pop out of die water. Nothing else to this time shows investor overreaction so clearly."

Other research also confirms that investor psychology is the cause of the overreactions. In the study discussed in chapter 6, showing the effects of surprise, we found that even five years after we measured an eamings su rise, "best" stocks still had much higher P/E, P/BV, and P/CF ratios than "worst" stocks. For example, though high P/E stocks experiencing negative su rises decline more than 44%, relative to the market, their average P/E at the end of five years is 17, well above the 13 for the market. Conversely, though the lowest P/E group with positive su rises provide a 35% above-market retum after the five years of

our study, the average P/E multiple for the lowest group is 10, still well below the markets at the end of this time.

What this shows is that the prior overvaluation of the "best" and undervaluation ofthe "worst" groups was so large that even after these major price re-evaluations, they still trade at premiums and discounts respectively to the market. That investors still price "best" stocks at a premium to the averages after five years indicates they are believed to continue to have above-market prospects, albeit not as good as they were thought to be earlier. The price dechne can only be attributed to the ove ricing of "best" stocks originally. In the case of "worst" stocks, the situation is reversed. Investors price them at a discount to the market, indicating their outlooks continue to be thought below average.

That the P/E muUiples of "worst" stocks are still below-market after outperforming for five years (and "best" stocks are above-market after the unde erformance) indicates major ove ricing of "best" stocks and unde ricing of "worst" occurs prior to the su rise. "Best" companies are still believed to have better prospects than the market, and so trade at higher P/E multiples five years later, but the multiples have come down significantly, indicating the prospects are not nearly as exciting as was believed originally.

Although the outlooks for "worst" companies are evaluated more favorably, they are still considered to be poorer tiian those of the average stock. The measurements for low price/eamings, price-to-book value, price-to-cash flow, and high yield are similar."

An independent verification of overreaction comes from a study done by Fuller, Huberts, and Levinson. These researchers showed that the lowest P/E group experiences the smallest eamings growth for periods in excess of five years. Similarly, "best" companies show tiie highest rate of eamings growth for five years. Analysts apparently do predict which companies will have above- and below-normal growth over time. This finding is important because it demonstrates tiiat, although the fundamentals dont change dramatically in eitiier the case of "best" or "worst" stocks, investors continually ove rice favorites and discount unpopular issues.

What we see then is that it is not changes in the underlying outlooks that appear to be the primary cause of the major price re-evaluations of "best" and "worst" stocks; rather it is the overreaction to the exciting or lackluster prospects of companies tiiat often results in best stocks being priced far too dearly, and worst stocks much too cheaply.

The Dark Side ofthe Moon

Some readers and certainly many academics will ask when the overreaction begins. The academic literature seems uncertain on this point, but appears to say overreaction occurs during the time die contrarian stocks bounce back. Nothing is further from the truth. Contrarian stocks recover from their previously depressed prices in that period, while high-priced issues decline. The overreaction, as I use the term, takes place earlier, during the period when the favorite stocks are becoming overpriced, as they move up to the highest price-to-book value (or highest price/eamings, price-to-cash flow or price-to-dividend quintiles) and when unfavored stocks are being knocked down to the bargain basement group.

What happens to the prices of favored stocks before they have reached their highest price-to-book value, or price/eamings, price-to-cash flow, or price-to-dividend ratios, and to out-of-favor issues before they hit the bottoms of these groupings? If overreaction is the powerful force I believe it is, we should expect two things to occur. Favorite stocks should outperform the market as they move up to die priciest quintile, and the unloved and unwanted stocks should unde erform, as they drop towards the lowest quintile. If the investor overreaction hypothesis is correct, then, we should have two distinct and opposite over-reactions taking place concurrendy, and continuing for a number of years. Favored stocks should become increasingly ove riced, and out-of-favor issues should get increasingly less expensive. Does it really happen this way?

To find out we took the same period measured in Table 11-2, 1973-1996, and the same stocks from the Compustat 1500. However, instead of measuring the retums of best and worst price-to-book value groups for the next five years (Period A, column 2, Figure II-I), we now asked the computer to do a backplay of the same portfolios, this time calculating the retums for the stocks in the highest and lowest price-to-book value quintiles for the prior five years (Period B, years -5 to 0). The performance is shown in Figure I I-l, column IWe call this study the "Dark Side of the Moon," because no one had examined how stocks behave as they are moving towards their highest or lowest price-to-book levels.

The only way we canconstmct the portfolios in Figure II-I, column 1, is by initially sorting the Compustat 1500 into highest, lowest, and middle groups by price-to-book value (or other contrarian indicators) at point 0. If we didnt know the lowest and highest price-to-book value