The Predictive Power of Price-Earnings and Price-Sales Ratios

    by Richard Goedde

    The price-earnings and price-to-sales ratios are popular methods of valuing stocks. But are they really useful to an individual investor?

    In his book, “What Works on Wall Street” (revised edition, McGraw-Hill 1998), James P. O’Shaughnessy did extensive analysis on how these ratios predict future stock price performance. He found a significant correlation between each ratio and price performance, with lower ratios leading to better performance during the following year.

    This article summarizes part of his work, and provides further information on the specific price-earnings and price-to-sales ratios that lead to superior results. The purpose of this article is to:

    • Provide specific price-earnings and price-to-sales ratios for the current market that correspond to O’Shaughnessy’s predictive deciles;

    • Show the effect of the recent bear market on both ratios; and

    • Show that price-to-sales ratios are slightly superior to price-earnings ratios as a predictive tool. In addition, an individual investor could, with a good stock database, relatively easily update the information following the approach used here.

    Creating Deciles

    The price-earnings ratio is a stock’s current price divided by earnings per share for the latest four quarters.

    The price-to-sales ratio is a stock’s current price divided by sales per share for the latest four quarters. O’Shaughnessy used deciles to demonstrate the correlation between each ratio and future price performance. In this article, price-earnings and price-to-sales ratios were ranked from lowest to highest, and then divided into deciles—10 groupings with an equal number of stocks.

    Table 1 shows how the number of stocks per decile was determined. This table was created using the Stock Investor Pro database developed by AAII. The database includes almost all stocks (approximately 8,500) on the New York Stock Exchange, American Stock Exchange, Nasdaq National Market, and Nasdaq Small Cap Market.

    TABLE 1. Defining Deciles for Price-Earnings and Price-To-Sales Ratios
      6/2/2000 8/2/2002    
    S&P 500 Index (Close) 1477 864    
    Total Stocks (No.) 9,323 8,732    
    Stocks Above $150 million (No.) 4,071 3,359    
      Price-Earnings Ratios Price-to-Sales Ratios
    6/2/2000 8/2/2002 6/2/2000 8/2/2002
    Stocks Above $150 Mil With        
    Ratios Above 0 (No.) 2,613 2,395 3,595 3,150
    Stocks Per Decile (No.) 261 240 360 315

    Because price-earnings and price-to-sales ratios change with the market, it is useful to provide deciles at both market peaks and troughs—for this article, those are provided when the S&P 500 index was close to its peak on June 2, 2000, and 26 months later on August 2, 2002 when the index was 42% lower.

    O’Shaughnessy chose to include only stocks with market capitalizations over $150 million (adjusted for inflation). He chose the $150 million minimum to focus on those stocks that a professional money manager could buy without running into liquidity problems. As you can see from Table 1, this reduces the number of stocks considerably.

    In addition, price-earnings and price-to-sales ratios equal to or less than zero are not considered to be meaningful and are eliminated. The remaining number of stocks is divided by 10 to calculate the number of stocks per decile.

    On August 2, 2002, only 2,395 stocks out of 3,359 stocks (71%) with market capitalizations over $150 million had price-earnings ratios greater than zero. The remaining stocks (29%) with price-earnings ratios of zero or less are primarily due to negative earnings per share. This is compared to 3,150 out of 3,359 stocks (94%) that had a positive price-to-sales ratio on August 2, 2002.

    All price-to-sales ratios are positive, but the Stock Investor Pro software rounded some to zero. Consequently, a meaningful (positive) price-to-sales ratio can be calculated for more companies than the price-earnings ratio. Table 2 shows the price-earnings and price-sales ratio ranges for the deciles. Using price-earnings ratios on August 2, 2002 as an example, the 2,395 stocks with market capitalizations over $150 million and price-earnings ratios greater than zero were ranked from lowest to highest. They were then divided into 10 groups with about 240 stocks in each group. The lowest price-earnings ratio in Decile 1 is 0.39, while the highest for Decile 1 is 10.14—the same as the lowest in Decile 2.

    Table 2 provides useful “rule-of-thumb” information on what a reasonable price-earnings or price-to-sales ratio is. For example, on August 2, 2002, the median price-earnings ratio was 18.83—the bottom of Decile 6. The median price-to-sales ratio was 1.40.

    TABLE 2. Price-Earnings and Price-To-Sales Ratios for Each Decile
      Price-Earnings Ratios Price-to-Sales Ratios
    Bottom Ratio of Decile 1 1.08 0.39 0.03 0.01
    Bottom Ratio of Decile 2 8.00 10.14 0.37 0.34
    Bottom Ratio of Decile 3 10.12 12.98 0.65 0.56
    Bottom Ratio of Decile 4 12.07 14.77 1.01 0.78
    Bottom Ratio of Decile 5 14.01 16.56 1.47 1.03
    Bottom Ratio of Decile 6 16.33 18.83 2.11 1.40
    Bottom Ratio of Decile 7 20.54 21.59 2.98 1.89
    Bottom Ratio of Decile 8 28.26 25.27 4.43 2.59
    Bottom Ratio of Decile 9 41.36 32.24 8.41 3.45
    Bottom Ratio of Decile 10 75.59 50.88 21.11 5.28
    Top Ratio of Decile 10 4,025.00 2,230.00 5,140.22 712.09

    Predictive Powers

    Table 2 takes on greater importance when combined with the information in Table 3, which summarizes part of O’Shaughnessy’s work. O’Shaughnessy started at the end of 1951 and included all stocks with market capitalizations over $150 million (adjusted for inflation). He then divided the stocks into deciles by price-earnings and price-to-sales ratios—Decile 1 had the lowest ratios while Decile 10 had the highest. He then measured the average performance of all stocks in each decile during the following year—1952. This process was repeated for each year through 1996. Table 3 reports the average annual compound return for each decile over that 45-year period. Note that different stocks will be in each decile each year as price-earnings and price-to-sales ratios change for individual stocks over time. Note, also, that each price-earnings ratio decile will have different stocks than the price-to-sales ratio decile.

    Table 3 shows the superiority of price-to-sales ratios over price-earnings ratios in predicting performance over the next year and over the long term (a 45-year period). The pattern is perfect for the price-to-sales ratios in that the average annual compound return of each succeeding decile has a lower return than the one before it.

    The pattern for price-earnings ratios is also declining, but it is not as precise. More importantly, the price-to-sales ratio deciles provide better predictive power between deciles as evidenced by the larger difference between the highest return—17.63% for Decile 1 price-sales ratios—and the lowest return—5.12% for Decile 10 price-sales ratios. The difference in the highest and lowest returns for the price-earnings ratio deciles is not nearly as great—16.70% for Decile 2 versus 10.34% for Decile 8.

    TABLE 3. The Predictive Power of Price-Earnings and Price-To-Sales Ratios
    (Historical Period: 1/1/52 to 12/31/96)
    Decile Average Annual Return for Following Year
    Based on Prior Year Ratios (%)
    Predicted by
    P/E Ratios
    Predicted by
    P/S Ratios
    1 (Lowest Ratio) 14.89 17.63
    2 16.70 16.52
    3 15.40 16.50
    4 13.60 15.64
    5 12.92 13.91
    6 11.79 13.18
    7 10.84 11.62
    8 10.34 9.93
    9 10.93 7.78
    10 (Highest Ratio) 10.98 5.12
    All Stocks 13.23 13.23

    This table includes only stocks with market capitalization greater than $150 million, adjusted for inflation over the 45-year period. The price-earnings ratios use earnings per share from continuing operations.

    Stock Selection Guidelines

    Having looked at the evidence in Table 3, let’s look back again at Table 2, which is now interesting in another light.

    With the S&P 500 index falling 42% between June 2, 2000, and August 2, 2002, one may expect price-earnings ratios to drop as stock prices fell. However, that is not always the case. While price-earnings ratios did decline for Deciles 8 to 10 (and the bottom of Decile 1), they increased for Deciles 2 to 7. This is not too surprising, given that earnings per share also dropped during this period. Of more interest is that prices did better for the stocks in Deciles 2 to 7 relative to those in Deciles 8 to 10 (implied by rising price-earnings ratios in Deciles 2 to 7), just as O’Shaughnessy’s analysis would have predicted. The high price-earnings stocks were hit the hardest during the latest bear market.

    For the price-to-sales ratio deciles, all deciles are lower in 2002 than in 2000. Prices fell, but unlike earnings per share, sales are more stable in a bear market. However, the high price-to-sales stocks were also hardest hit, as evidenced by the larger drop in price-to-sales ratios in the higher deciles.

    It should be noted that price-to-sales ratios vary by industry because they are affected by profit margins, which vary by industry. Supermarkets have low profit margins, which means that they need a relatively high level of sales to generate a reasonable profit, leading to low price-to-sales ratios. Drug companies and jewelry stores are examples of high profit margin businesses with relatively high price-to-sales ratios. Consequently, the change in price-to-sales ratios between 2000 and 2002 in Table 2 is also affected by the fact that some industries, such as technology, were hit harder than others by the market decline.

    Combining the information in Table 2 and Table 3 provides guidance in selecting potentially high-performing stocks. It suggests that, as of August 2, 2002, the best stocks have:

    • Price-earnings ratios between 10 and 13 (Decile 2), and
    • Price-to-sales ratios between 0 and 0.34 (Decile 1).
    Stocks with price-earnings ratios of less than 10 (Decile 1) do not do quite as well as Decile 2 because there are many poor-performing companies with very low price-earnings ratios in Decile 1—as well as the desirable high-value companies.


    Both price-earnings and price-to-sales ratios are significant factors in predicting future stock price performance over the next year. It is useful to have specific information on the distribution of these two ratios so that their predictive power can be realized.

    Richard Goedde is an associate professor of economics at St. Olaf College in Northfield, Minnesota. He can be reached via E-mail at

→ Richard Goedde