I became a self-directed investor seven years ago and have followed what I consider to be a disciplined approach to dividend growth investing. This approach has evolved over time as I have continued to learn from the excellent articles published on Seeking Alpha and other online sources. In return, I consider it a privilege to contribute whatever knowledge and experience I may have gained back to the investment community through articles such as this.
I think the ultimate challenge facing equity investors can be summarized as follows: Over extended periods of time, equities have demonstrated consistently strong total returns ever since the establishment of organized trading exchanges 200+ years ago. The challenge, of course, arises over the shorter term. The market can produce returns for five, ten, fifteen, and even twenty years that significantly overperform or underperform longer term averages.
In his book Stocks for the Long Run (McGraw Hill, 2014), professor Jeremy Siegel summarizes long-term equity performance as follows (p. 81):
The long-term stability of stock returns has persisted despite the dramatic changes that have taken place in our society during the past two centuries. The United States has evolved from an agricultural to an industrial economy and then to the postindustrial, service- and technology-oriented economy it is today. The world shifted from a gold-based standard to a paper money standard. And information, which once took weeks to cross the country, can now be instantaneously transmitted and simultaneously broadcast around the world. Yet despite mammoth changes in the basic factors generating wealth for shareholders, equity returns have shown an astounding stability.”
For some time now, I have thought that it would be interesting to actually measure stock market performance and volatility from the post-civil war period to the present day. Of course, numerous writers and scholars have poured over this data countless times, but I wanted to take a fresh look and concentrate on:
- Price performance broken out by decade intervals
- Volatility of decade price performance
I began by gathering S&P 500 price data by decade, starting with January 2019 and moving back sequentially ten years for each data point until January 1879. (This data was obtained on the website multpl.com, which is an excellent site for stock market data).
That gave me 15 total values, which I listed on the left side of a spreadsheet. I then constructed 14 columns, each column representing a certain decade multiple (10 years, 20 years, 30 years, and so forth). The columns ranged from 140 years (only one value in this column: 2019-1879) all the way down to 10 years (14 values in this column). I then calculated the compound annual growth rate of price increase (or, rarely, decrease) for each of the 15 price data points within each of the 14 decade columns. This produced a total of 105 growth rate calculations which appear in a triangular formation in the chart below:
|STOCK MARKET PERFORMANCE 1879 – 2019|
|PRICE ONLY: EXCLUDES DIVIDENDS|
|————–COMPOUND ANNUAL GROWTH RATE———————|
|GRAPH AVERAGE GROWTH RATE: 4.75%|
|GRAPH AVERAGE VOLATILITY: 0.076|
(Due to space restrictions, the columns for 140 years, 130 years, and 120 years are not shown)
Finally, at the base of the chart, I calculated the average price growth rate for each decade column and also the volatility of price growth rates for each decade column. (Volatility is calculated by the Coefficient of Variation = Standard Deviation divided by the Average).
Prior to making these two final calculations, I had not expected to see any particular pattern or correlation between the 14 average values of the decade columns. It seemed to me that over this extended period of 140 years that these column averages might be fairly random. Second, I had expected to see a gradual increase in volatility as the decade multiples declined from 140 years to 10 years. This just made intuitive sense in terms of how numbers behave – or how the market behaves – over shorter and shorter time periods.
The bottom line is that I was completely wrong in terms of the column averages and completely right in terms of volatility.
I was frankly astounded to see that the 14 column averages all fell in a very tight range around 4.7 percent. The lowest value was 4.60% (130 years) and the highest value was 5.03% (10 years). The average of these 14 averages was 4.75% and the volatility was a low value of .076. This is far less volatile than all but one decade column and would validate what our eyes can see – this is a very tight cluster of data points.
I want to make clear that this data reflects stock price performance only. Dividends are not included. So, today you would need to add approximately 2.0% in dividend yield to estimate S&P total return. (However, dividend yield has declined in recent years and was higher during previous decades).
The volatility of the decade columns performed as I had expected, although I must admit that I had not anticipated such a perfect lock-step pattern. You can see that as you move to the right and decrease the decade multiple, each volatility value is equal to or higher than the preceding column value (with only one minor exception – 90 years). Even with only one, two, or three numbers in a column (140 years, 130 years, 120 years) where you might expect some considerable variance, this increasing pattern still holds.
After reviewing and considering these two very clear patterns – virtually equal price averages and lock-step volatility increases, it occurred to me that these results might be unique to or dependent upon the starting point – 2019. By starting with that particular year to determine decade intervals, we just happen to wind up picking some very high points (1929, 1999, 2019) and some very low points (1939, 2009). We also include the entirety of the 1970s decade, a miserable period for stocks.
Therefore, I decided to perform this entire exercise again – using a different starting date. I moved the start date back five years to study decade periods that would be five years out of sync with the decades previously studied. I, therefore, considered the 140-year period between 2014 and 1874, now looking at decade intervals that ended with the number “4” instead of “9”.
This new “4” decade sequence looks very different than the “9” sequence. All the major high and low points have been removed. What is most interesting about this new “4” sequence is that S&P prices increase steadily between every decade in the last century – without exception. For example, the 1929 spike/market crash disappears and we see an actual increase in prices between 1924 and 1934. The 1999 high (before the dot.com crash) disappears as does the 2009 low, so prices climb smoothly between 1994, 2004, and 2014. With this new starting point, the last 120 years appear to be an extraordinary period of controlled and steady growth!
Given such differences between the “9” and “4” sequences, I expected that column average and volatility figures shown below might look very different.
|1874 – 2014|
PRICE ONLY: EXCLUDES
|GRAPH AVERAGE GROWTH RATE: 4.48%|
|GRAPH AVERAGE VOLATILITY: 0.016|
To my surprise, they did not!
Once again, the column averages show a very tight cluster. The average of these “4” column averages is 4.48%, slightly lower than the “9” columns average of 4.75%. The volatility is exceptionally low at .016, far below the “9” value of .076 (itself a very low value). While still surprised by the overall results, the actual differences between the “9” chart and the “4” chart seem to make intuitive sense. The average rate of 4.75% for the “9” chart is higher than the 4.48% for the “4” chart because the former includes a high value of 2602.6 compared to a high value of 1822.4 for the latter. So growth rates had to be somewhat higher to reach that higher final value.
Similarly, the “4” chart eliminates the dramatic high and low values of the “9” chart and as mentioned above gives the appearance of a relatively smooth appreciation in value for over 120 years. So it would be expected that its volatility would be well below the “9” chart, which is, in fact, the case.
It would certainly be possible to expand this study by selecting all 10 possible start dates, which would involve a total of 1050 growth rate data points. I might be tempted to try this but I think my trusty old BA II Texas Instruments calculator would explode.
But as we have looked at two different charts where the decade multiples are exactly 5 years out of sync, my guess is that additional start-point selections would further confirm the two fundamental findings of this study, which are:
- Regardless of the length of the decade multiple investment periods, stock prices have achieved a very consistent average annual growth rate of between 4.50% to 4.75% for the 14-decade multiple periods between the post-civil war years to the present day.
- The volatility of average returns increases in a gradual but steady geometric fashion from longer decade multiples to shorter decade multiples. The increase in volatility is quite pronounced between the 30, 20, and 10-year decade columns.
So another way to say this would be that you can expect an average return of 4.5-4.75% no matter how many decades you may invest. The catch, of course, is that the volatility – the chance of not achieving that average – increases dramatically the shorter the holding period.
Again, it is important to remember that dividends are not included in the returns calculated above. The S&P 500 dividend yield is now approximately 2.0%, which should be added to the price returns to compute total returns.
I welcome your comments – especially from any statisticians that may be out there – and will do my best to respond.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.