By Chris Turner
Investors suffer difficult decisions -- where to put your money to get not only a return of capital, but a return on capital. Professor Jeremy Siegel generated a study back in 1994 that highlighted over the last 100+ years, stocks performed better than bonds (5th edition of the book due in July). Many have argued the fallacies and problems with the book, notably the data set used. Rather than focus on his data, perhaps more understandable is that most of us do not live 140 years, nor do we invest that long. This led to an epiphany of sorts to generate a study, which resulted in asking myself the following questions:
- What about simply measuring performance of stocks and bonds over a normal worker's life?
- How would different asset allocations compare?
- What if someone placed 10% of their income into only stocks or only bonds over the course of their working life?
- How would someone that started investing in the early 1900s compare to those of us entering the workforce in the latter part of the 1900s?
- How many "life periods" would find that stocks actually beat bonds?
The study seems easy enough, but piecing the data together required extensive research. First, how does one actually calculate 10% of income? Data existed for mean and median incomes back to 1913 from splicing government data out of the IRS and Census. Inserting the CPI calculation allows both an inflation-adjusted median income and a nominal income for comparison purposes. Next, what about a data set for stocks and bonds? That data proved much easier -- using Professor Robert Shiller's monthly averages of daily closes produces both nominal and CPI-adjusted sets at the same time going back to 1871. Now, the calculations begin.
The following assumptions were made for this study:
- Investor began saving at the age of 20 and stopped at age 65 (45 years total)
- Investor saved 10% of median income per month
- Investments do not include fees
- Dividends reinvested proportionally monthly
The calculations consist of the following over a 45-year period:
- Mattress: Saving 10% per month cumulative using median income
- Stocks: Saving 10% per month, adjusting monthly for dividends and performance of S&P composite
- Bonds: Saving 10% per month, adjusting each month for performance of 10-year yield
- Stocks and Bonds: Using 100 less age to allocate that much toward stocks (adjusted monthly until 65)
By tabulating the final result over each 45-year period beginning in 1958 (representing a worker's investments from 1913 to 1958) until 2012 (1967 to 2012) – a total of 55 "life" experiences occurred. The results favor Siegel, on a nominal basis -- stocks beat bonds 33 times to 22.
The chart below shows the tabulated nominal series results:
The chart displays the total amount saved for four investment types during each 45-year period on a line graph beginning in 1913 (age 20) and retiring in 1958 (age 65). Whichever color displays at the top for each year shows which method produced the best results. For example, workers that retired in 1966 or 2000 would have experienced a better return from all stocks (green line on top), while someone in 1986 and 2010 would have fared better with all bonds (blue line on top).
From the chart above, another thought occurred to me. Out of the 33 times that stocks beat bonds, how many dollars did they beat on average? The following chart provides the answer.
The chart above calculates the final retirement difference between stocks and bonds and displays a mean of how much stocks either beat or lost to bonds. When stocks beat bonds (calculated by just a mean including all positive numbers), stocks beat by $51,374. When bonds beat stocks (calculated by just a mean of negative numbers), bonds beat stocks by $22,665. Readers may glean just how much of a skew (both positive and negative) occurred post-1996.
So what would an individual have experienced over their lifetime of investment, which was the original intent on this study? To answer, I created a separate chart for each time period or life experience. The following chart depicts exactly what someone that began work in 1913 would have experienced through the 45-year investment strategy.
The chart displays that depending on when the investor either retired or stopped investing, a different result would have occurred. For example, this investor would have experienced a better return with bonds nearly all the years except around the peak in late '20s and at the last five years. The chart also shows that 30 years into the investment, placing money into a mattress would have produced a better result than stocks.
To view a time-lapsed nominal series showing each time frame (includes all workers up to 1990), please visit this YouTube video:
For a downloadable PowerPoint presentation, click here.
Moving into the CPI-adjusted world, the following displays the exact same information as the nominal series, except adjusted for consumer price index.
Based upon CPI adjustments, stocks beat bonds 36 times to 19. Note that between 1975 and 1996, stocks or bonds didn't really matter, they produced similar results.
How much did stocks beat bonds over all the life experiences? The following CPI-adjusted chart shows the differences.
On a CPI-adjusted basis, stocks handily outperformed by a wide margin. The mean of occurrences when stocks beat bonds meant an increase over a bond portfolio of $224,099. The mean of occurrences when bonds beat stocks meant an increase above a stock portfolio of $164,166. Kudos to those that retired in 2000, as 100% stock investors would have accumulated nearly $1,000,000 more (theoretically, of course).
Back to our individual that began working in 1913, what did their CPI-adjusted investment life experience?
This chart adjusts the median income for inflation and shows a dramatic impact from stock growth around 1929 and the last five years of investment.
To view a time lapsed CPI-adjusted series showing each life experience (original 55, plus all workers up to 1990 for a total of 78 charts), please visit this YouTube video:
For a downloadable PowerPoint presentation, click here.
Based on this study, clearly those who dismiss Siegel may want to reconsider. Perhaps in the future, we can look at different time series (25- or 30-year investments) and determine just what someone could have done with the money in retirement.