Sunday, October 26, 2008

Stock Market Earnings Growth History and a Worst Case Scenario

In 100 Years of Stock Market History, I observed that rarely does the stock market, as measured by the DJIA (Dow Jones Industrial Average) or S&P 500, fall significantly below its 25-year moving average, even during bear markets (see graph below). That the October 10 close was in that general area was comforting, suggesting that we might be at, or near, the bottom. On the other hand, the graph makes it clear that we might not be; during the Great Depression following the 1929 stock market crash, the market fell significantly below the moving average. (Note: the graph below is as of October 2008; the graph in the 100 Years post is updated approximately monthly.)

Graph of DJIA (Dow Jones Index) since 1900 with 25-year moving average
Dow 25-Year Moving Average
Since the future is, by definition, unknowable, strategic planners like to develop planning scenarios. Typically, one develops 3-5 scenarios covering a wide range of possible futures. This post begins to develop one such scenario
for the stock market, a “worst case” scenario. (Note: The vertical axis is log scale; for a short discussion of log graphs, see About Stock Market Log Graphs.)

What’s the Dow's Long-Term Earnings Growth Rate?

As I mentioned in the previous post, one interpretation of the graph above is that the stock market goes through cycles. It periodically gets ahead of itself by increasing faster than the businesses that it represents. It then has to wait for the earnings of those businesses to “catch up” with the valuation before beginning a new growth spurt. (For now, let’s ignore the fact that the composition of the Dow index changes over time.) Clearly, it would be helpful to look at about 100 years of Dow Jones earnings history in order to estimate the long-term earnings growth rate. This would give us an estimate of what the stock market's average return performance has been, and will be, over the long term -- excluding dividends.
Long-Term Growth Rates
Start1941110.965.63%$10.22 5.72%10.9
Start19821046.54$100.17 10.4

To estimate the stock market's long-term earnings growth rate, we need to use data points at comparable points in the stock market cycle – e.g., trough to trough, or cycle-start (the beginning of a bull market) to cycle-start. I compared “normalized” earnings at the beginning of the last two bull markets. In the table above, I used 1941 and 1982 as the start years. I normalized the earnings by averaging the earnings from five years prior through five years afterwards. The earnings growth rate over that period was 5.72%. (Note: To do your own analysis, e.g., to calculate stock market earnings growth rates between any two years, see the link to the spreadsheet/model at the end of this post.) This ballpark growth rate based on earnings is consistent with the change in the Dow’s price (the average stock market return excluding dividends) during that period, 5.63%.

If you look at price growth from trough to trough, it tells a similar story. The two most severe bear markets on the graph bottomed in 1932 and 1974. The associated trough-to-trough growth rate is 5.71%. Therefore, the most recently completed stock market cycles indicate that the long-term average stock market return has been about 5.5-6% per year, excluding dividends. This estimate would seem to be further corroborated by estimates of long-term GDP growth, which I have seen generated by combining approximately 1% population growth with 1.5-2.5% in productivity improvements, and 2-3% long-term inflation.

IF that long-term growth rate is still applicable (a critical assumption), it is disconcerting since the Dow’s price growth for the 17 years from ’82 to ’99 was more like 15% per year.
(Note: For another approach to estimating long-term earnings growth, and arriving at the same basic conclusion, see Earnings Growth Contribution to 50-Year Returns.)

Long-Term Stock Market Performance

I keep saying "excluding dividends." Since dividends are a major portion of total long-term stock market returns, you might be interested to know what returns have been including dividends. Reported numbers vary somewhat depending, e.g., on what index is used to represent the stock market (Dow, S&P 500, etc.), and what time period is analyzed. However, they are typically in the area of 10% over long periods of time. For example in Bogle on Mutual Funds (p.31, 1st edition), John Bogle reports that the S&P 500 returned 10.3% between 1926 and 1992. Note that about 1% of this return was due to multiple expansion -- prices getting more expensive relative to earnings and dividends; depending on your purposes, you might not want to include that 1%. In a future post, I will explore the implications of multiple expansion in more detail.

(Update: See a later post for average long-term total stock market returns, including dividends, for specific periods -- e.g., from 1900 or 1929).

A Worst-Case Scenario

One way to arrive at a worst-case estimate is to extrapolate some of our previous “worst-evers” to the present. In the table below, I have taken the year-end lows from 1932 and 1974 and projected them to year-end 2008 using an assumed long-term growth rate of 5.72%.

In both cases, the result is a Dow of approximately 4100. You’ll arrive in the same neighborhood if you draw a line from the ’32 low through the ’74 low and extend it out to 2008. In 2008, you will see that the line is just a touch higher than the 1993 and 94 data points, which were 3754 and 3834 respectively. Dow 4100 would represent a 70% drop from the all-time (2007) year-end high of 13265; it is roughly 50% below the October 24 close of 8379. As a reference point, the drop from year-end 1928 to year-end 1932 was about 80%. (Again, to do your own analysis see the link at the end of this post.)

How do we interpret such a disturbing and seemingly ridiculous possibility? What does this mean? Well, for sure it’s not a prediction that the Dow will fall to 4100. And, it’s certainly not an estimate of what the market is “worth” since the market tends to be undervalued at the end of bear markets. The exercise above is simply my attempt to develop a worst-case scenario, based on the last century or so of data, of what might happen. That the data support this possibility does not mean that it will happen. But it could.

Note: I have used data from a spreadsheet that I created by hand in 1997. If anyone finds any errors – either computational or conceptual – please let me know.

Recommended Articles:

Three Scenarios for the Economy (and the Stock Market): for a fuller, qualitative, description of the worst case scenario.
Worst-Case Scenarios Based on 100 Years of Dow Price/Earnings History: My alternate approach. Estimates outcomes from best-ever to worst-ever.
The 1929 Stock Market Crash Chart: 1929-1932 revisited. The ultimate worst case?
Projected Market Returns for the Next Ten Years uses the above growth rate as a starting point.
See the sidebar to the left for links to additional posts.

How Low, How Bad, How Long is an excellent article by fund manager John Hussman. His methodology is similar to the methodology used in my two Worst-Case posts (Note: this article is 10 pages long and mildly technical.)
Buy American. I am. Warren Buffet’s New York Times op-ed piece reminding us of his dictum -- “Be fearful when others are greedy, and be greedy when others are fearful.”
Brighter Days Ahead: From the March 16, 2009 issue of Newsweek.

Note: I will add additional relevant articles as I discover them.

Dow Earnings and Year-End Closing Price Data

For those who would like to perform additional analysis, my Dow Jones / DJIA yearly closing price and earnings data, and the associated calculations, are in this spreadsheet/model. The spreadsheet is designed to automatically compute price and normalized earnings growth rates between any two years (e.g., 1941 to 1982). It can also project any year's closing price forward to any future year at whatever growth rate you would like to assume (e.g., project the 1932 close forward to 2008 at 5.72%). If you have trouble downloading the Excel spreadsheet, see this post.

Share This Article

To share via Facebook, Twitter, etc., see below.

Last modified 8/30/2010


  1. Al:

    I think your worst-case scenario is certainly plausible - unfortunately. Doesn't mean it will happen but it's well within the range of possibility.

    The only quibbles I have with the methodology don't invalidate the conclusion.

    Choice of 1941 as a start year is questionable because of the war.

    Markets are like pendulums so they tend to over-correct roughly as much as they were too optimistic or pessimistic. At the low the market will under value the companies behind the stock. But that's no comfort to anyone who has to sell at the low.

    Despite all that I do have some stock in my portfolio but mainly special situations like closed-end bond funds and MLPs which are likely (I hope) to be valued on their earnings before the market recovers as a whole.

    Good but scary post

  2. Tom,
    You mean I have to respond to comments too? Who knew! That wasn’t in my blogging plan – and you know how planners hate unplanned activities. The light came on last night when I was checking out one of my favorite blogs (yours) and noticed how you handle comments.
    Anyway, thanks for the comment. I will respond. However, I have to do a little more research, and make a few more calculations, first.

  3. Tom,
    Upon further review, I’m not sure I really understand your “quibbles” -- but I'll try to respond anyway.

    For these purposes, I think year-end 1941 is clearly the start of that bull market; the stock market closed higher virtually every year for the next 20 years. The fact that it was less than a month after Pearl Harbor does mean that the market was artificially low. However, the nature of the analysis dictates that we will be looking at times of maximum pessimism. Secular bull markets always start from an overly pessimistic base. (The next one started after years of high inflation and steadily increasing interest rates, with the 20-year Treasury yield topping out at over 13%.) So, from a market price growth-rate point of view, I think we’re ok.

    Our use of normalized earnings covering more than a decade muted the impact of start-year on earnings. If the war had an inordinate impact, it presumably REDUCED earnings to below what they would have been otherwise – and caused us to overstate the long-term growth rate. That would mean the real growth rate is actually lower than we assumed, and the Dow projection for 2008 should be lower.

    Given our very high-level perspective, even if you make minor changes in start-year the result is essentially the same -- horrendous. Lucky for us worst-case scenarios are, almost by definition, unlikely.

    Let me know if I misunderstood your point.

  4. Totally logical that gold will double four times as it did in the 1970's and the Dow will drop 70% as it has before to give a 1:1 Dow Gold ratio of 4,238 by 2018. This is consistent with the past as in 1980 when the Dow Gold ratio was 1:1. The Dow would also yield about 7% at this price which is where it's a good investment and people would resume buying stocks because they are a good value.


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.