If you look up Professor Jeremy Siegel’s work you’ll find one thing: He firmly believes that stocks are the best long-term asset class.
In 1994, Seigel published a classic book that helped shape his investing view and asset allocation decisions for millions investors around the world. “Stocks for the Long Run: The Definitive Guide to Financial Market Returns and Long-Term Investment Strategies” continues to be a top-selling investment book. His conclusions are based on stock market data going back about 200 years.
New research has compiled stock market data going back to 1793, adjusts for some erroneous methodologies and reaches a different conclusion – one that contradicts Siegel’s long-standing belief that stocks are the preferred asset class over the long run.
Edward F. McQuarrie, Professor Emeritus, Santa Clara University, presents these findings in a recent paper called “New Lessons from Market History: Sometimes Bonds Win“.
McQuarrie summarizes his findings as follows:
When Jeremy Siegel published his Stocks for the Long Run thesis, little information was available on stocks before 1871 or bonds before 1926. But today, digital archives have made it possible to compute real totalSource: New Lessons from Market History: Sometimes Bonds Win, Edward McQuarrie
return on stock and bond indexes back to 1793. This paper presents that new market history and compares it to Siegel’s narrative. The new historical record shows that over multi-decade periods, sometimes stocks outperformed bonds, sometimes bonds outperformed stocks, and sometimes they performed about the same. More generally, the pattern of asset returns in the modern era, as seen in the Ibbotson SBBI and other datasets that begin in 1926, emerges as distinctly different from what came before. Contrary to Siegel, the pattern of asset returns seen in the 20th century does not generalize to the 19th century. A regime perspective is introduced to make sense of the augmented historical record. It argues that both common stocks and long bonds are risk assets, capable of outperforming or underperforming over any human time horizon.
For most investors, it’s not news that bonds sometimes outperform stocks. Indeed, this is why we hold bonds within a portfolio – to diversify sources of risk and return. However, for most that’s as deep as the asset allocation decision goes, and bonds – particularly Treasury bonds – are simply viewed as a safety mechanism for short periods of risk asset downside. It remains that, over the long-run, the common view is one that aligns with Siegel’s thesis that equities is the better asset class over long periods of time.
McQuarrie disagrees: Simply, he finds that there are long periods of time during which bonds outperform stocks. The particularly troubling finding is that bonds can outperform for the duration of a human lifespan. It is therefore financially dangerous to adhere to the ideology that stocks are always the better choice.
History is an important guide, but it cannot predict the future. The purpose of this research is not to identify what will outperform over the next decade. Rather, it is to shift portfolio construction from a monolithic to multi-faceted choice.
The crux of Siegel’s case for stocks is exhibited by the following chart. Looking at this alone, it is understandable that someone might conclude that stocks are the better investment. McQuarrie, however, argues that Siegel’s methodology was misleading and the reality is that stocks and bonds performed in parallel until the post-war period (second chart below). Moreover, much of the final stocks-bonds cumulative performance variance is explained by a single time period from post WWII to around 1982, with bonds subsequently resuming parallel performance. (I.e. if you re-started the chart from 1983, the stocks and bonds lines would follow a similar path.) The difference flips the investing paradigm on its head.
It is critical to understand why the results are so different.
McQuarrie’s explanation for the superiority of his bond data is as follows:
Explanation: I observe municipal bond prices from 1857 to 1897 as compiled by the Boston stockbroker, Joseph G. Martin. Siegel relied on a tertiary source, the summary table in Sidney Homer’s History of Interest Rates. Homer does not state there that the municipal bond index he had tabled represents a theoretical construction erected by Macaulay upon Martin’s compilation rather than an observed portfolio. I also adjust for the greenback price of interest paid in gold coin—a distinguishing feature of Federal and some municipal bonds between 1862 and Finally, I use an aggregate bond portfolio up to 1897, and corporate bonds after that point. Corporate bonds returned more than the government bonds used by Siegel, especially when government bond yields were depressed by tax and other privileges.
Essentially, McQuarrie used actual bond portfolio data, whereas Siegel used computed theoretical data.
McQuarrie’s data also adjusts for the striking survivorship bias found within Siegel’s stock market data:
Here most of the variation comes before the Civil War, and this deviation is readily explained: Siegel’s sources omitted the largest single stock that traded before the Panic of 1837, the 2nd Bank of the United States. At the peak before the Panic hit, the 2nd BUS accounted for almost 30% of total market capitalization. It failed spectacularly as the Panic proceeded, with shares dropping in price from $120 to $1.50, and never recovered. To duplicate this omission in the contemporary stock market, it would be necessary to drop Microsoft, Apple, Amazon, Alphabet/Google, and Facebook from the S&P 500; and even these five would not account for as high a percentage of S&P 500 capitalization as did the 2nd BUS at its peak. Omission of the BUS is the single most glaring error I found in Siegel’s stock market sources. More generally, I found and corrected survivorship bias. Banks failed during panics, turnpikes and canals succumbed to railroads, and struggling railroads went bust in the 1840s and 1850s to an extent not previously understood. In short, Siegel’s sources had left out the bad parts, producing an overly rosy picture of antebellum stock returns.
Few probably realize that Siegel’s data contains such biases – especially when it’s standard practice to adjust accordingly. For example, modern fund performance evaluation, like Morningstar’s SPIVA reports, always account for survivorship bias. To read that Siegel excluded failed companies in his analysis is material. To be fair, however, this survivorship bias exists in many historical analyses as it is an embedded feature within index data.
Let’s not make hasty conclusions and throw out all of Siegel’s conclusions. Stocks do outperform bonds during a meaningful range of time periods and scenarios. Stocks are still a core component to an investing portfolio. McQuarrie’s analysis simply illustrates that the probabilities have shifted away from stocks’ favour. This is an important consideration for professional asset allocators and Joe Smith saving for retirement.
Using McQuarrie’s new data, the table below shows the odds of stocks outperforming bonds across a range of rolling 1, 5, 10, 20, 30 and 50 year periods. The odds remain in stocks’ favour across all rolling periods between 1793-2019, and those odds increase with the length of period. However, it is critical to observe that these results are time-period sensitive. The pre-war experience is markedly different, especially prior to the US Civil War.
Modern-day assurances that stocks always outperform over the long run are based on post-war data (1943-2019). In contrast, the full history shows there’s a 32% chance that someone investing in stocks over 30 years – a lifetime of investing – would underperform bonds.
Why did stocks outperform during the post-war years?
During that period, the US dominated the world with military and economic hegemony acting as the global arbiter and enforcer of transnational trade and globalization, creating a ‘peace dividend’ for risk assets around the world. As that power fades, it’s quite possible the structural forces that created that environment will someday no longer exist.
The following chart shows something similar to the table above, focusing on rolling 10yr performance differentials between stocks and bonds. Again, it is apparent that stocks can underperform for life-altering periods of time.
How bad can stock market returns get?
If one asset goes up by 20% and the other by 10%, the second asset underperformed but still provided a decent return. Unfortunately, history shows that the stock market can provide dismal returns for very long periods of time.
The table below shows the worst 20, 30 and 50 year real total returns for stocks. The results show that stocks can perform poorly in both relative and absolute terms over investing lifetimes.
The standardization of time periods in the previous chart, however, doesn’t show the true extent of weak stock market performance potential over long periods of time. One only has to go back to the 13 year period ending 2013 to have witnessed one of the worst periods ever for US stocks. The table below shows the worst six periods for US stocks.
The implications for asset allocation decisions are immense, yet intuitive: Don’t rely on history alone to guide asset allocation decisions.
Investors must look at the prevailing environment to determine how to construct portfolios, understanding that long-term performance characteristics can change. Perhaps the most challenging part of this effort is understanding the relative variability of stocks and bonds, as illustrated by long-run correlations. Correlations change markedly over time, so the idea that one simply dumps a portion of a wealth into bonds with the purpose of creating a portfolio ballast is wrong. Stocks don’t always outperform and bonds don’t always move opposite to stocks. This is strikingly apparent (but shockingly rarely discussed) during the recent Covid-19 market crash, when US Treasuries (using TLT as a proxy) declined by almost 10% from March 1st to March 8th, while US stocks simultaneously declined by 8.8%. US Treasuries soon recovered and rallied as stocks continued to plunge, but this relative performance aberration – when the correlation between stocks and risk-free assets moved to 1 – is an example of how historical relationships can and do change. It is a critically important example of how expectations based on long histories can be false.
Will bonds outperform stocks? Are bonds still a good tool for diversifying portfolios?
Without spending considerable amounts of time and effort deciphering the current economic regime and its implications on forward asset prices, a more pragmatic approach to asset allocation might be to ensure broad representation across assets that have fundamentally different risk exposures. Stocks are probably still good to hold for the long run. Probably so are bonds (both corporate and a variety of sovereign). But it’s entirely possible that neither will perform as you expect.