The long run outperformance of equity markets versus all other major asset classes has become orthodoxy in the investment world. Widely read studies from the likes of Roger Ibbotson, Jeremy Siegel and Credit Suisse’s Global Investment Returns Yearbook have cemented equities as an unequalled asset class for delivering the best long term returns for investors. Implications based on the above include the belief that investors with time horizons of 20 to 30 years or more can have a 100% allocation to equities, and the data presented in these studies seemingly provides the proof that this has been a particularly sweet pudding for investors to eat over the course of a century or more.
While there appears to be sufficient evidence to support this recommendation, work over the past few years has placed this commonly held belief into question. At the risk of over-egging the pudding analogies, the equity souffle has hit some cooler air and its rise has been thrown into doubt.
Edward McQuarrie, a retired university professor, has sought to both improve the quality, and extend the period of data used when analysing the returns of equities versus bonds. Key improvements include the expansion of coverage for both equities and bonds, thereby reducing survivorship bias, and calculating capitalisation weighted returns rather than price or equal weighted.
The first and perhaps most striking result of McQuarrie’s analysis1 is that equity and bond performance between 1792 to 1941 was roughly equal, a very different result compared to the vast outperformance of equities over bonds from 1942 onwards. Concentrating on a long-term time horizon should be encouraged, but the full periods measured in both eras are longer than most people would consider when saving for future expenditure*.
Siegel’s case of ‘Stocks for the Long Run’, partly rests on shorter term performance, which demonstrates over 30-year rolling periods, equities have outperformed bonds 90%2 of the time, suggesting an aggressive allocation to equities is warranted for most long-term investors. However, McQuarrie’s 227 year analysis indicates that while the odds of equity outperformance do increase as the period becomes longer, the extended periods of equity underperformance versus bonds across the 19th century mean the rolling 30-year odds only reach 67%. It is still strong case for preferring equities over bonds, but by no means a one-way bet over time periods that are meaningful for the average investor.
The sample data taken from a time when the world that looks very different to today’s can easily be consigned to the annals of history . However, data outside of the US, also provides an alternative picture to the US biased analysis used by Siegel et al. McQuarrie’s work finds multiple instances of 20 and 30 year negative real returns for non-US equities, along with underperformance versus bonds. There is no pattern within the data, with negative equity premiums observed across many individual countries, during the 19th, 20th and 21st centuries.
McQuarrie’s analysis also shows the changing correlation between equities and bonds over time. Correlation over rolling 20-year periods reached their lowest ever (i.e. negative) levels in recent decades, very different to the 0.61 correlation recorded for the 134-year period from the beginning of the data set.
McQuarrie’s contribution to the literature provides important new information for investors. By expanding historical and international datasets, the relationship between equities and bonds, in terms of the former’s assumed outperformance, and the correlation of returns between the two asset classes, is perhaps not as straightforward as other studies have implied. McQuarrie suggests a lack of stationarity exists, meaning there are no fixed statistical properties over time when it comes to asset class returns. One explanation cited for this is the Regime Thesis. This states that unlike the improvements made when increasing the sample size to measure the mean and standard deviation of say, height in a population, increasing the dataset through the addition of further historical data does not help to gain further confidence when analysing long-term asset class returns. Instead, temporary patterns, or regimes, of asset returns occur. These can often be for extended periods, and can contain temporary stationarity, but this breaks down across multiple regimes. One broad lesson to take from McQuarrie’s study is that it is always worth questioning investment convention.
While the odds remain in favour of equities outperforming bonds over long time periods, it cannot be assured, and the possibility of being in an unlucky generation of equity investors should not be discounted. The good news however is that multi-asset portfolios can offer some protection, offering a balance between core asset classes such as equities and bonds, along with other avenues of returns from physical gold, and uncorrelated strategies such as trend following or long volatility.
*Undrawn savings passed from one generation to the next is a valid counterpoint, but I believe the average saver tends to think of reaching pension age as a milestone in their investment journey.
Sources: 1Edward F. McQuarrie (2023) Stocks for the Long Run? Sometimes Yes, Sometimes No, Financial Analysts Journal. 2Siegel, Jeremy J. Stocks for the Long Run: The Definitive Guide to Financial Market Returns and Long-Term Investment Strategies.