Home » Articles » 6. What to invest in? » 6.1 Historical returns and risks

6.1 Historical returns and risks

072

In Article 4.3 I introduced the relationship between returns and risk.  In a nutshell, the prospect of higher returns comes with a higher risk of your investment declining in value.  At a broad level, history tells us the relative returns and risks for the three main investment types are:

  • Highest for stocks
  • Intermediate for bonds
  • Lowest for cash

For cash, the nominal annualized return since 1928 has been about 3.3% as measured by historical rates from 3-month Treasury bills.  When this article was last updated in January 2022, cash returns were in the 0.4% to 0.5% range for high-yield saving accounts, money market accounts, and short-term Certificates of Deposit (CDs).  The risk of a decline for cash holdings is near zero if we ignore the effects of inflation.

The annualized return of 3.3% for cash is in “nominal” terms, which means it’s not adjusted for inflation.  Inflation decreases the spending power of money over time, and inflation-adjusted returns are often called “real returns”.  Inflation is an important problem that I will come back to in Article 8.6, but for now, just remember that the average annualized return for cash, as well as the stock and bond returns presented below, would be lower if they were adjusted for inflation.

If you want to calculate the annualized returns for cash between any two years going back to 1928, you can use the cash return calculator provided here.

Now let’s take a more detailed look at historical returns and then risks for stocks and bonds.

Historical returns

You’ll find various statistics about the historical returns of stocks and bonds, and they can be frustratingly different from one source to another depending on the data used, the period examined, and myriad other details.  Nonetheless, a close examination of various data sets paints a pretty consistent picture.  Two of the most often cited data sets for historical stock and bond returns are from Yale Nobel Laureate Robert Shiller and Aswath Damodaran of the Stern School of Business at New York University.

Data Source Stocks Nominal Average Annualized Return 10-Year Bond Nominal Average Annualized Return
Shiller 1871 through 2021 9.37% Not Available
Damodaran 1928 through 2020 9.98% 4.84%

Of course, in some historical periods, stock and bond returns varied substantially from the average annualized return, as this table of annual return statistics shows.

Statistic Stocks Shiller Stocks Damodaran 10-Year Bond Damodaran
Mean 11.00% 11.82% 5.11%
5th Percentile -18.71% -23.07% -4.98%
25th Percentile -0.77% -1.17% 0.86%
Median 11.16% 14.52% 3.28%
75th Percentile 23.12% 25.72% 8.71%
95th Percentile 38.91% 39.48% 18.53%

You can see each year’s annual returns by downloading the full data sets at the Shiller and Damodaran websites.  Note that the “mean” value shown in this summary statistics table is an arithmetic average of annual returns, whereas the “average annualized returns” in the earlier table are calculated using a geometric average or Compound Average Growth Rate (CAGR).  The average annualized return is generally a little lower and a more accurate measure of the returns achieved through investing consistently over many years.

You may be interested in determining annualized returns for specific historical periods.  These two calculators provide annualized stock and bond returns (nominal and inflation-adjusted) between any two periods based on the Shiller and Damodaran datasets, respectively.



You may also be interested in calculating returns for more specific flavors of stocks and bonds.  While these more specific return histories tend to be much shorter, I’ve created additional Mindfully Investing return calculators at these links for the following asset types:

Historical risks

Stock returns have historically outpaced bond returns by 4 to 5%.  So, this seems like a compelling case for investing in stocks.  But we need to also look at the risk side of the equation.  Here are some simple statistics on the ups and downs (“volatility” as measured by standard deviation) of stocks and bonds from a Vanguard study using data starting in 1926, with annual standard deviations added from the Shiller and Damodaran datasets, respectively:

Statistic Stocks Government Bonds
Return in Best Year 54.2% 45.5%
Return in Worst Year -43.1% -8.1%
Years with a loss 25 19
Standard deviation 18.6% 7.7%

As you can see, stocks have a wider range of ups and downs (volatility), and in a single very bad year, you could lose about half your investment value.  In comparison, bonds have much less potential for large annual losses and had fewer years where a loss occurred.  So, we can start to see why return and risk are normally assumed to be linked.

Comparing historical returns and risks

One way to simply compare stocks and bonds is to put the returns and risks on a cross plot like the one introduced in Article 4.3, but using the real-world data from above.

This comparison suggests that bonds are a better balance of risk and return as compared to stocks.  For example, with bonds you get about half the return of stocks, but for less than half of the risk from stocks.  Put another way, for stocks, with every percent in return, you also get about two percent of standard deviation.  In contrast, for bonds, with every percent of return, you get about half a percent of standard deviation.

Returns over time – So now it sounds like we should invest in bonds.  However, we should also ask: what do the different historical stock and bond returns mean for the growth of investments over time?  Here’s a graph comparing the growth in stocks versus bonds from an initial $1 investment starting in 1927 (Damodaran dataset).

The long-term effect of the different average annual returns is apparent, with the stock value ending around $3,800, while the bond value ends at around $73.  The lower volatility of bonds is also apparent in the chart because the line for bonds is much less choppy than the line for stocks.  Obviously, none of us will be investing for 90 years, but the graph also shows how quickly the returns of stocks and bonds can diverge.  For example, looking at the period starting around World War II (about 1944), we can see that stocks subsequently outpaced bonds by a wide margin in just 10 years (by about 1954), and after that bonds never caught up again.  So, while the volatility risk with stocks is clearly higher, the nearly double average annual return in stocks versus bonds has provided a huge relative benefit over the long term.

It’s worth briefly highlighting the beginning of this chart.  It shows that for about a 13-year period starting in 1927, bonds kept pace with stocks and at some points (like around 1932 and 1941), bonds briefly performed even better than stocks.  The chief argument for bonds is that during volatile periods like the 20s and 30s, bonds provide relative stability of return.  But let’s put that 13-year volatile period in perspective.  While most of us won’t be investing for 90 years, many of us will invest for 40 or even 50 years.  From a lifetime investing perspective, it seems quite reasonable to wait 13 years for stocks to recover from something momentous like the Great Depression.  I discuss the frequency and duration of historical stock market crashes in more detail in Article 8.

We can see that the long-term returns for stocks mount up quickly relative to bond returns most of the time.  And even in very unusual times, you don’t have to wait too long before the benefits of those compounding stock returns start to substantially outpace bond returns.

Risks over time – What is the “risk” being measured by the standard deviation in the above analyses?  It’s essentially how much the value of the stock or bond investments goes up and down over time.  While this is one definition of risk, is it really the type of risk that we should be concerned about?

To answer that question, let’s say that I buy an investment and sign a contract that says:

  • Upon penalty of death, I will not sell the investment for 10 years and at the end of that period, I must sell the entire investment.

Let’s also say that a few days after buying that investment, it plunges 50% in value.  That’s not a great start, but given the alternative under my contract is death, I do not sell my investment prematurely.  Let’s further say that my investment recovers and even increases in value by 34% by the end of the 10-year contract.  This equates to a 3 percent average annualized increase over 10 years.  So, the one-time 50% plunge caused a low total rate of return for 10 years, but the final return is still positive.  Under this scenario, there was no realization of the perceived risk implied by volatility as measured by the standard deviation.  No actual “permanent loss” occurred from the 50% plunge, as frightening as that may have been.  No matter how wildly the investment varied over ten years, the primary risk that matters in this scenario is the potential for a permanent loss when you end the investment and spend the money on something.

Standard deviation or other measures of routine volatility are actually a very poor measure of the risk that matters most to real-life investors.  A better risk definition focuses on the potential for a permanent loss, which means that the money is not available to you when you need to spend it.  To estimate this better-defined risk, we need to overlay the expected investment volatility with the timing of when you will most likely need to use the money (investment time horizons), which is a more nuanced risk analysis.  The role of time as it relates to investing risk is the subject of Article 8.

Conclusions on historical returns/risk

Our interim mindful conclusions based on the history of stock and bond returns and risks are:

  • First, the seemingly small additional annual return of stocks can reap huge benefits over periods of 10 or more years.  The return benefit of stocks can be under-estimated by focusing solely on annualized average returns or similar metrics.
  • Second, risk defined by ordinary volatility is too simplistic and does not determine our actual risk of permanent losses.  Depending on our investing horizons and goals, the risks associated with stocks may be over-estimated using simple measures like standard deviations.

Both conclusions indicate we should tilt our portfolios more aggressively toward stocks and away from bonds, which differs significantly from some advice you will find in the media or investing books.  Deciding how much to tilt toward stocks is a much more complex question that is linked to the risks and returns of mixed portfolios, your specific situation, time horizons, and your investing goals as detailed more in Articles 7 and 8.  But first, let’s examine expected future returns/risks with stocks and bonds in Article 6.2.

5 comments

  1. Nathan Fryzek says:

    “For cash, the annualized return since 1928 has been about 3.4% as measured by historical rates from 3-month Treasury bills.”

    Is this nominal or real?

    • Karl Steiner says:

      My bad. The cash annualized return is on a nominal basis, which was explained for stocks and bonds later in the article. I updated the article to include the note about nominal returns earlier in the text. Thanks.

  2. Amanda Luc says:

    I am confused about something I say in the NYTimes recently:

    https://www.nytimes.com/2020/05/01/business/bonds-beat-stocks-over-20-years.html

    It talks about how the “bonds beat stocks over the last 20 years.” Their numbers seem different from the graph in your article, but I am guessing it is because they are talking about “annualized returns.” According to the annualized returns “bonds beat stocks” in the last 20 years… but are they just being misleading by talking about annualized returns (instead of showing what the value of, say, a $1 investment would have been for each of these assets?

    • Karl Steiner says:

      Yes, the devil’s in the details when it comes to comparing stock and bond performance. First off, we need to make sure we’re talking about the same types of bonds. The NY Times article states that “long-term Treasuries [20-30 year], long-term corporate bonds, and high-yield (or junk) bonds” outperformed stocks. But the calculator and graph in my article use historical data for 10-Year U.S. Treasury bonds. And I’d argue that the 10-year bond is a better surrogate for the way most investors include an array of bonds (along with stocks) in their portfolios. Second, my graph is showing the long-term performance of a one-time investment starting in 1927 in the S&P 500 (stocks) versus 10-year T-bonds (bonds). The NY Times article is looking at an investment in the S&P 500 vs. long-term T-bond, corporate bonds, and junk bonds for someone who started investing in 2000. Different bond comparisons, time periods, and start dates all yield different results.

      The problem illustrated here is that you can find longish periods in the historical record where almost any asset “X” outperformed any particular asset “Y” and vice versa. In fact, if you compare any two random assets over long periods, you will almost always see a see-saw performance where the two assets taking turns outperforming each other for multiple years. If you go back through my blog posts you will find numerous examples. This is what makes it impossible to pick the “best” portfolio (a combination of investment assets) for the future (see this post).

      I could accuse the NY Times article of cherry-picking their time frame to get results that fit a preconceived headline. But I actually agree that the recent outperformance of some bond types is notable, and for that reason, it was the subject of my most recent blog post. It turns out that long-term bonds have been outperforming stocks for nearly 40 years! However, it’s pretty much impossible for that trend to continue for reasons that I laid out at the end of my last post.

      One definite point of disagreement I have with the NY Time article is this part: “[Bond outperformance] is a sign of how unreliable many assumptions about financial markets actually are these days — of how risky the markets have become and of how difficult it is to invest sensibly for the future.” To me, the see-saw performance of assets over time is normal and should be expected, and just because asset classes don’t perform the way people think they should in any given period does not make investing inherently more “risky”. Expecting the unexpected is what investing is all about.

      Thanks for the thoughtful comment.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.