Global investment returns

In 2016 we looked at time-varying expected returns (Dimson, Marsh and Staunton (2016a)). We used over a century of daily returns for the US and 85 years of UK data to examine the immediate effect of interest rate increases (and decreases) on stock and bond markets. The announcement effects were in the predicted direction but quite small, indicating that markets are efficient in anticipating rate changes and their impact.

We also conducted a coarser analysis based on annual data, extended to 21 countries over the period from 1900 onward. Real equity and bond returns both tended to be higher in the year following rate falls than in the year after rate rises. This relationship also held for subsequent periods longer than a year.

This raised an obvious question: how do different asset classes perform over hiking and easing cycles? To research this, we evaluated a trading strategy that, in principle, could have been implemented in real time (Dimson, Marsh and Staunton (2016)).

First, we looked at the performance of equities, bonds, bills and currencies and the corresponding equity and maturity premia. Our updated analysis is reported in Chapter 3 of this current Yearbook. We then examined performance within the equity market, analyzing factor returns, including industries, as well as the returns from size, value, and momentum. Finally, we examined the returns on real assets (including precious metals such as gold and silver), collectibles (including art, stamps, and wine), and real estate (including housing and farmland). We found consistent differences between returns during hiking and easing cycles. Returns and premiums were generally appreciably higher during easing cycles.

It is often argued that the risk of equities declines when the investment horizon is long, because equity returns are said to revert to the mean. Such mean reversion would not only reduce risk but could provide timing signals that allow investors to boost returns. In 2013 we wrote about mean reversion (Dimson, Marsh and Staunton (2013a)).

We concluded that the popular evidence for mean reversion is an “optical illusion” that employs hindsight. We used the Yearbook’s global dataset to analyze the evidence on return predictability in the absence of any look-ahead bias. We examined the profitability of buying shares when the cyclically adjusted price/earnings (CAPE) or cyclically adjusted price/dividend ratio (CAPD) looks cheap based solely on preceding data. We found that the evidence on mean reversion is weak. Market-timing strategies based on mean reversion typically gave lower, not higher, returns.

As we have found in other areas of our research, if investors are willing to accept some increase in risk, there are signals that can be used to identify when markets offer a larger or smaller reward. However, there is insufficient predictability to make equity investing “safe” over any horizon.

Starting with Triumph of the Optimists and an early Yearbook chapter called ‘The Quest for Yield’ (Dimson, Marsh and Staunton (2011)), we started quantifying the contributions to long-term returns of income, growth, and price/dividend multiple expansion. We documented the overwhelming and worldwide importance of dividends to the long term real returns on common stocks. We examined long-term returns in common currency and reported that, over the entire period starting in 1900, there was a striking outperformance of higher yielding stock markets relative to their low yielding counterparts. We showed that this cross-country pattern persisted in successive periods of a quarter-century.

In 2012 we studied currencies over the long haul (Dimson, Marsh and Staunton (2012)). We found that equities perform best after periods of currency weakness, which suggests that more unhedged cross-border stock exposure can be desirable at those times. In contrast to equities, cross-border bond investment can add to portfolio risk primarily through currency exposure. Short-term currency hedging is therefore found to be particularly meaningful in bond portfolios.

However, hedging benefits are found to fall off with longer investment horizons. We examined whether past currency movements are related to subsequent asset returns and found that equities performed best after currency weakness. The same was true for bonds over several decades. Our analysis provided some comfort to “buy-on-weakness” investors and offered no support for “stick-to-strong-currency” strategies.

The Yearbook provides compelling evidence that, over the long haul, exchange rates reflect relative inflation rates. For longer-term investors, the risk reduction benefits of hedging rapidly decline. This is because currencies tend to converge towards reflecting relative inflation rates. It is also because hedging introduces a new form of risk, namely, a bet on real interest rates at home versus abroad. We also looked at whether currencies are predictable. While, over the long run, currencies do tend to converge to PPP, this is of limited usefulness for short-term predictions.

The 2017 Yearbook focused on factor investing (Dimson, Marsh and Staunton (2017)). We discussed five approaches that were widely cited as contributors to long-term returns: size, value, yield, momentum and volatility. We assembled evidence on long-term factor premiums over a period (then) of up to 117 years for the UK, and shorter periods for 22 other markets. We presented out-of-sample evidence on the performance of smart-beta strategies and reported on the attenuation of performance in the post-publication era. A subsequent paper based on the Yearbook chapter was published in the Journal of Portfolio

Management and named one of three outstanding articles in the 19th Annual Bernstein Fabozzi/ Jacobs Levy Awards.

Our study of long-term factor returns has since been extended in coverage. We now (in Chapter 8 of this current Yearbook) report the size premium in 34 markets, the value premium in 35 markets, and the momentum premium in 35 markets.

In 2012 we wrote about real estate (Dimson, Marsh and Staunton (2012)). We examined the investment performance of commercial real estate using Investment Property Databank indices. Real property returns appear to be hurt less than stocks, bonds or bills by contemporaneous inflation. However, real estate prices can lag traded assets, and a rise in consumer prices was associated with a delayed decline in real property values that exceeded other assets. On balance, and given its relative illiquidity, an appropriate role for commercial property was as a diversifier and as a source of returns, forming part of an investor’s core long-term holdings.

For individual investors, the most prevalent exposure to real estate is their own home. We investigated the behavior of house prices in all but one of the (then) 19 Yearbook countries. We also assembled a six-country database of house prices since 1900. House price indexes did appear to keep pace with inflation over the long term.

In 2018, we returned to housing (Dimson, Marsh and Staunton (2018)). Using a database of house prices for 11 countries spanning 1900–2017, we published a global comparison of the long-term investment performance of residential homes. On a population-weighted basis, and extrapolating index coverage to rural as well as city locations, real house prices had appreciated by some 0.4% per year before costs and quality adjustments (–2.1% per year on a quality-adjusted basis). We counselled that investment in private residences should be justified by the consumption benefits this provides and warned against exaggerated expectations of a large risk premium.

Treasure assets are sometimes referred to as investments of passion. They are beautiful and collectible items, though they do not generate income and any financial reward would need to come from capital gains. Collectors point to cultural and artistic investment not only as a pleasurable activity but also as a contribution to financial diversification. However, within the category of passion investments, investors almost invariably hold focused portfolios. The average of their holdings should not be regarded as a desirable allocation for a financial investor.

Long-term data for these emotional assets is hard to assemble not only because reliable historical records are elusive, but also because of heterogeneity in the items that have changed hands. In 2018, in addition to real estate, we focused on assets with long price histories, and with records that are as good as one can access in this asset class. We examined long-run investment performance since 1900 for art, stamps, wine and violins. We also studied assets with shorter (albeit still lengthy) price histories, namely rare books, historic cars and jewellery.

We made return comparisons not only on a pre-tax basis, but also after accounting for the tax payable by wealthy investors over the previous century. Our work supported the view that a moderate allocation to tangible alternative assets is likely appropriate for high-net-worth investors.

In 2010 and 2014 we wrote about the growth puzzle (Dimson, Marsh and Staunton (2010) and (2014)). We showed that, over the long run, investors had underperformed by investing in markets with high past GDP growth, compared to lower growth markets. We warned against excess enthusiasm for investment in high past growth markets – which were often developing countries.

Relatedly, we also wrote about emerging markets (EMs) (Dimson, Marsh and Staunton (2010a) and (2014a)). In 2019, our special feature highlighted the growing presence of China in benchmarks, the very divergent performances reported by competing equity indices for China, and the underwhelming performance of Chinese equities, despite China’s astonishing record of economic growth (Dimson, Marsh and Staunton (2019)).

We reviewed emerging market performance since 1900 and showed that emerging markets had underperformed developed markets (DMs). However, this was mostly due to very poor performance during the 1940s. We noted that even though global markets are now more connected, the gains from spreading assets across DMs and EMs are still substantial. The typical DM investor can reduce risk by holding EMs, and the typical EM investor can also benefit from investing abroad.

In 2021, we returned to the emerging markets theme, presenting a substantial extension of our DMS dataset (Dimson, Marsh and Staunton (2021)). Nine emerging markets were added, seven from Asia and two from Latin America, each providing at least 50 years of investment performance. We also introduced historical data for a further 58 countries with shorter histories. We documented factor investing and rotation strategies in the emerging world. Notably, the value effect has been strong both within EMs and also as a basis for rotation between markets.

Responsible investing was a special theme in the 2015 Yearbook (Dimson, Marsh and Staunton (2015a)). Asset managers were coming under pressure to demonstrate responsible investment behavior. This could take the form of “exit” via ethical screening, or “voice” through engagement and intervention. We demonstrated that “sin” can pay, not least because those choosing to exit from stocks they view as offensive can cause them to offer higher returns to those less troubled by ethical considerations. We reported superior longrun returns from tobacco and alcohol stocks in both the US and the UK.

When we revisited the environmental, social and governance (ESG) topic in 2020 (Dimson, Marsh and Staunton (2020)), we focused on the question of whether responsible investing can enhance returns, or whether ethical commitments involve making a sacrifice. We concluded that, despite claims to the contrary, there was no reliable evidence that ESG screening enhances returns or reduces risk. This remained true whether we looked at the performance of companies based on their ratings or at ESG funds or indexes.

For ESG investment strategies based on exclusions, we concluded that theory and evidence suggested that a small return and diversification sacrifice was involved, but the magnitude of this was unlikely to be material. In other words, the price for ethical principles appeared small. We did, however, provide evidence that corporate engagement can pay, whether the focus is on environmental and social issues or on corporate governance. We presented research that finds engagement is more likely to pay off when action is coordinated with likeminded activists.

Our 2020 article reported that there was a remarkable divergence between the ESG scores given to a particular stock by different rating agencies. We were the first to publish evidence on this, and a paper based on our Yearbook research was published in 2020 in the Journal of Portfolio Management and awarded first prize in the prestigious 22nd Annual Bernstein Fabozzi/Jacobs Levy Awards. Our observations have now been confirmed and replicated in subsequent studies conducted by leading empirical researchers.

In an international comparison, we estimated the concentration of industries by country, and the concentration of countries by industry (Dimson, Marsh and Staunton (2015)). In 35 out of 40 worldwide industries, two countries accounted for a majority of the industry’s global capitalisation. In 33 out of 47 countries, the weighting of three large industries accounted for a majority of the country’s market capitalisation. We also examined the relative importance of industries versus countries in determining equity performance. We found that since 2003, industries and countries have been roughly equally important.

We also examined the rise and fall of industries over time in the US and UK, compiling long-term stock market indices from 1900 onward. Successive waves of new industries, technologies and companies have transformed the world. We therefore addressed the question of whether it was better to invest in new or old industries. We found that, historically, there had been some tendency to overvalue new industries and technologies and undervalue the old. We showed that an industry value rotation strategy that leaned against this historically would have generated superior returns. However, an industry momentum approach was an even more effective strategy.

We noted that new industries are typically born on a wave of IPO activity and provided evidence on the poor post-IPO performance of stocks around the world. Clearly, investors need to be especially cautious about the valuations of IPOs. We also examined the performance of stocks based on their degree of seasoning – the length of time they had been listed since their IPO. We found that over our 35-year research period, terminal wealth was almost three times higher from investing in the most, rather than the least seasoned stocks. At the stock level, “old” clearly beat “new”.

In our 2022 Yearbook (Dimson, Marsh and Staunton (2022)), we focused on diversification. Investors can easily be misled by claims that only 10 to 20 stocks are needed for a diversified portfolio. We showed that far more are needed. Moving to international diversification, we found that over the last 50 years, investors in most countries obtained higher Sharpe ratios from investing globally rather than just domestically. However, in the US, the world’s largest market, investors would have been better off investing at home. Prospectively, our advice to investors from all countries, including the US, is that they should invest globally. This is likely (but not guaranteed) to reduce risk and increase Sharpe ratios.

The long-run behavior of stock-bond correlations show there has been good scope for diversification across stocks and bonds. However, the mostly negative correlations in the first two decades of the 21st century were the exception, not the rule. The stock-bond correlation tends to be negative during crises. This makes government bonds valuable diversifiers that can enhance the power of portfolio diversification when it is needed most.

In chapter 6 of the current Yearbook, we have revisited and updated these findings. We have also added new material on market concentration.

In 2011 we studied the impact of inflation on stock market returns (Dimson, Marsh and Staunton (2011a)). If countries are classified by their inflation over the coming year, then rotating wealth in favor of low-inflation markets provides higher real returns. However, at New Year we do not know the coming year’s inflation rate. Instead, we can classify markets by last year’s inflation. We found that a strategy of investing in countries with high past inflation performed best. Investing in troubled markets was riskier but over the long term it was more rewarding.

In the 2023 Yearbook we examined the role commodities can play in investors’ asset allocations from the perspective of diversification and inflation hedging (Dimson, Marsh and Staunton (2023)). The backdrop of a more inflationary environment made this topical. As rising commodity prices, including energy prices, were important contributors to the then recent bout of inflation, we explored whether investing in commodities offers an effective hedge.

Individual commodities have, on average, provided low long-run returns. However, portfolios of futures have generated attractive risk-adjusted long-run returns, albeit with some large, lengthy drawdowns. We concluded, based on historical returns, that a balanced portfolio of collateralized commodity futures is likely to provide an annualized long-run future risk premium of around 3%. We cautioned, however, that there was a limit to exploiting this otherwise attractive asset class in that the investable market size was quite small.

Historically, commodities have had a low correlation with equities and a negative correlation with bonds, making them effective diversifiers. They have also provided a hedge against inflation. Commodities are unique in this respect, compared with other asset classes. However, their inflation-hedging properties also mean that in extended periods of disinflation they tend to underperform.

In 2024 we focused on corporate bonds and the credit premium (Dimson, Marsh and Staunton (2024)). Corporate bonds are a major asset class with an outstanding value of some USD 44 trillion, almost half that of the value of global equities and two-thirds that of sovereign bonds. They trade on higher yields than equivalent sovereign bonds because of credit risk. We examined the yield spread, default and recovery rates in the US since the 1860s and reviewed their determinants.

Because of default losses, the yield spread is not a measure of the extra return investors can expect. To determine the credit premium, we examined the performance of corporate versus equivalent sovereign bonds from 1900 to 2024 for the US and from 1860 to 2024 for the UK. Investment grade corporate bonds have provided a credit risk premium over government bonds of around one percentage point. The premium from high-yield (or junk) bonds is some two percentage points higher. Long-run evidence since 1815 showed that risky sovereign bonds are also subject to credit risk and have enjoyed an even greater long-run credit premium than that on high yield corporate bonds.

We also showed that corporate bonds are a distinct asset class. We examined whether factors can boost corporate bond returns. Finally, we showed that it may be possible to beat corporate bond benchmarks just by buying and holding the constituents.