Again, these results would not have surprised the value investors
Graham and Dodd, who, in their classic 1934 text Security Analysis,
stated the following:
Hence we may submit, as a corollary of no small practical importance, that
people who habitually purchase common stocks at more than about 16
times their average earnings are likely to lose considerable money in the
long run.17,18
In a manner analogous to the research on dividend yields among
S&P 500 stocks, I computed the P-E ratios for all 500 firms in the index
on December 31 of each year by dividing the last 12 months of earnings
by the year-end prices. I then ranked these firms by P-E ratios and di-
vided them into five quintiles, computing their subsequent return over
the next 12 months.^19
The results of this research are similar to that reported on the divi-
dend yield and are shown in Figure 9-4. Stocks with high P-Es (or low
earnings yields) are, on average, overvalued and have given lower re-
turns to investors. A portfolio of the highest-P-E stocks had a cumulative
return of $65,354, earning an annual return of 8.90 percent, while the
lowest-P-E stocks had a return of 14.30 percent and accumulated to al-
most $700,000.
In addition to a higher yield, the standard deviation of low-P-E
stocks was lower, and the beta was much lower than that of the S&P 500
Index stocks, as shown in Table 9-4. In fact, the return on the 100 lowest-
P-E stocks in the S&P 500 Index was about 5^1 ⁄ 2 percentage points per year
above what would have been predicted on the basis of the capital asset
pricing model.
PRICE-TO-BOOK RATIOS
Price-to-earnings ratios and dividend yields are not the only value-
based criteria. A number of academic papers, beginning with Dennis
Stattman’s in 1980 and later supported by Fama and French, suggested
150 PART 2 Valuation, Style Investing, and Global Markets
(^17) Graham and Dodd, Security Analysis, 1st ed., p. 453. Emphasis theirs.
(^18) Yet even Graham and Dodd must have felt a need to be flexible on the issue of what constituted
an “excessive” P-E ratio. In their second edition, published in 1940, the same sentence appears with
the number 20 substituted for 16 as the upper limit of a reasonable P-E ratio! (Graham and Dodd,
Security Analysis, 2d ed., p. 533.)
(^19) Firms with zero or negative earnings were put into the high-P-E-ratio quintile. Returns were cal-
culated from February 1 to February 1 so that investors could use actual instead of projected earn-
ings for the fourth quarter.