Thursday, March 15, 2012

The Impact of Starting P/E Ratio on 5-Year Stock Market Returns

In this post, we'll look at the impact of P/E ratio at time of purchase on subsequent 5-year stock market performance. Previous posts in this series presented the results for 10 years and for 20 years.

Scatter Diagram: Initial P/E Ratio vs. 5-Year Stock Market Performance

Scatter diagram: initial price/earnings ratio, 5-year DJIA/stock market performance

The above scatter plot (click on image to expand) shows the historical relationship between the price/earnings ratio of the stock market at the time of purchase and the market's return over the next 5 years. In the chart, each dot represents
a hypothetical purchase at the end of a year between 1901 and 2005.

The dot's position on the horizontal axis shows the stock market's normalized P/E at the end of that year; the dot's position on the vertical axis shows the annualized total return of a hypothetical investor over the next 5 years. For example, the leftmost dot represents a year when the normalized P/E was 6.6; the annualized return over the next 5 years was 22.4%. (FYI, that year was 1914.)

As The Initial/Starting P/E Increases Future Returns Decrease

As you can see, as a general rule, the dots move from the upper left quadrant of the graph (low P/E, high return) to lower right quadrant (high P/E, low return). The green line is the statistically calculated trend line that best fits the data. It shows that, on average, the expected future returns decrease by about 1.1% per year for each unit increase in the price/earnings ratio. For example, the trend-line annual return for 5-year periods beginning with a P/E of 15 is 9.7%; for a p/e of 16, it's 8.6%.

Comparison to 10 and 20-Year Results

So far, the impact of the initial P/E ratio is increasing as the length of time decreases. As we go from 20 to 10 to 5 years, the impact of a unit increase in P/E has gone from -0.4%/year to -0.6%/year to -1.1%/year. Put another way, based upon 5-year results, the trend line slopes downward relatively sharply. However, as we look at longer and longer time periods, the impact of the initial P/E ratio is muted.

This makes intuitive sense to me since in the long run you expect the price/earnings ratio to regress back to the average price/earnings ratio. To get to the average in 5 years requires a steeper curve than would be required to get to the average in 10 or 20 years.

The next post in this series will cover the relationship between the initial price/earnings ratio and returns for the following year.


The above scatter plot shows the historical relationship between the p/e ratio of the stock market at the time of purchase and the market's return over the ensuing 5 years. By "stock market" I mean the DJIA (Dow Jones Industrial Average), though historical results for other broad-based indexes like the S&P 500 would be comparable. By P/E ratio, I mean the normalized price/earnings ratio as I calculate it (see About Normalized P/E Ratios). Finally, by typical investor's return I mean the total return of a hypothetical investor who bought the Dow Jones Index and held it for 5 years, reinvesting all dividends (with no expenses such as taxes or commissions). Also note that earnings and dividends prior to 1929 have been estimated based upon another stock market index.

The Objective Here is Analysis, Not Prediction!

Periodically I get a comment to the effect that the predictive powers of the graphs in this series "would be enhanced by using either the trailing one-year PE, the trailing 10-year PE" or some other variation. That may well be true.

However, the objective of this series was analysis, not prediction. For analysis purposes, I would argue that my normalized earnings are a better estimate of sustainable earnings for a given year than either that single year's actual earnings or the trailing 10 years earnings. And, since the earnings estimate is more accurate, I think it yields a better estimate of the price/earnings ratio.

Those same readers sometimes assume the approach is useless for predicting since, unlike their preferred approaches, "the normalized PE that you calculate won't become known for five years!" They're right to the extent that I can't calculate my normalized P/E for 2011 until 2016.

However, I can calculate the normalized earnings for 2006 (more accurately, I will be able to calculate 2006 normalized earnings in April after I get the official 2011 earnings). And, it's worth pointing out that my 2006 normalized earnings are based upon almost exactly the same earnings as the above-mentioned "trailing 10 years earnings;" I use the average of the earnings from 2001 to 2011. It's just that I would argue that the average of those earnings is a much better estimate of sustainable earnings 5 years ago than it is of sustainable earnings at the close of 2011. If you want to use these results for projections, start there.

Related Posts

The other posts in this series:

The Impact of Initial P/E Ratio vs. 20-Year Stock Market Returns
The Impact of Initial P/E Ratio vs.  1-Year Stock Market Returns
The Impact of Initial P/E Ratio vs. 10-Year Stock Market Returns
Other related posts:
Initial P/E Ratio vs Rolling 10-Year Returns: Same data, but looked at chronologically.
P/E Ratio Impact On Future Returns: Returns of purchases made when p/e is high compared to returns of low p/e purchases. Impact in $$$.
The 10 Year Stock Market Projection My approach to "predicting" market returns.
For lists of other popular posts and an index of stock market posts, by subject area, see the sidebar to the left or the blog header at the top of the page.
Copyright © 2012 Last modified: 3/25/2012

Share This Article

Delicious Bookmark this on Delicious To share via Facebook, Twitter, etc., see below.


  1. I've come across this website through a few various other websites. Thank you for sharing your analysis (and time and effort) in this graph format.

    I see others have have inquired about the "trailing P/E" or Schiller P/E. I would be curious to see the past returns following Schiller P/e like you have graphed here. Any interest in providing one that shows starting Schiller P/e and past returns?

    1. Thanks for your suggestion; I'll add it to my list of possible posts. Unfortunately, I can't promise when (or even if) it'll bubble up to the top.

      Thanks for stopping by.


No spam, please! Comment spam will not be published. See comment guidelines here.
Sorry, but I can no longer accept anonymous comments. They're 99% spam.