A truly global and long-term (116 years) data set for both successful and failed financial markets shows that equity has delivered positive long-term performance in each and every country that did not expropriate capital owners, even those that were ravaged by wars. Also, equity significantly outperformed government bonds in every country, with a world average annual return of 5% versus 1.8%. The long-term Sharpe ratio on world equity has been 0.24 versus 0.09 for bonds. Valuation-based strategies for market timing have historically struggled to improve equity portfolio performance. Active management strategies that rely on both valuation and momentum would have been more useful.

Dimson, Elroy, Paul Marsh, and Mike Staunton (2016), “Long-Term Asset Returns” and
Illmanen, Antti (2016), “A Historical Perspective on Time-Varying Expected Returns” in
“Financial Market History – Reflections on the Past for Investors Today”, edited by David Chambers and Elroy Dimson, The CFA Institute Research Foundation.

The post ties in mainly with the subject of implicit subsidies (view lecture here), particular subsidies for risk sharing related to companies’ earnings prospects and the discount factors applied to their long-term dividend payments.

The below are excerpts from the articles, with the first two sections being based on Dimson, Marsh and Staunton, while the last section is based on Illmanen. Emphasis and cursive text have been added.

A truly global and long-term data set

“Our database of annual returns has expanded to cover 23 countries from the beginning of 1900 to the beginning of 2016. It comprises annual returns for stocks, bonds, and bills, plus inflation and exchange rates… In total, the database covers 98.3% of global equity market capitalization at the start of 1900…As of the start of 2016, these countries make up 92% of the investable universe for a global investor.”

Every one of the 23 countries experienced market closures at some point, typically during wartime. However, in all but two cases, it is possible to bridge these interruptions and construct an investment returns history that spans the closure period. For 21 countries, therefore, we have a complete 116-year history of investment returns. For Russia and China, market closure was followed by expropriation of investors’ assets, so we have market returns only for the pre- and post-communist eras…Incorporating China and Russia into our database…addresses…survivorship bias…China was a small market in 1900…but Russia accounted for some 6% of world market capitalization…Similarly, Austria–Hungary had a 5% weighting in the 1900 World Index, and although it was not a total catastrophe, it was the worst-performing equity market and the second-worst bond market among the 21 countries with continuous investment histories. Incorporating Austria, China, and Russia drastically reduces the potential for bias in world market returns from ignoring non-surviving and deeply unsuccessful markets.”

“An issue more serious than survivorship bias is success bias. The United States is the world’s best-documented capital market…The evidence cited on long-run asset returns [is] predominantly US-based…Extrapolating from an unusually successful market—ignoring the fact that the economic and financial performance of that nation was exceptional—introduces success bias.”

“At the beginning of 1900…virtually no one had driven a car, made a phone call, used an electric light, heard recorded music…Transformations have given rise to entire new industries… Of the US firms listed in 1900, more than 80% of their value was in industries that are today small or extinct; the UK figure is 65%…Markets at the start of the 20th century were dominated by railroads, which accounted for 63% of US stock market value and almost 50% in the United Kingdom.”

Market performance since 1900

“Returns include reinvested income, are measured in local currency, and are adjusted for inflation.”

“In each country, equities performed best, long-term government bonds less well, and Treasury bills the worst. In the United States, an initial investment of $1 grew in real value to $1,271 if invested in equities, $10 in bonds, and $2.7 in bills. In the United Kingdom, an initial investment of £1 grew in real value to £445 if invested in equities, £7 in bonds, and £3.3 in bills.”

“[The figure below] shows annualized real equity, bond, and bill returns over the period 1900–2015 for the 21 countries with continuous index histories, plus the World Index (Wld), the World ex-USA (WxU), and Europe (Eur)…Markets are ranked in ascending order of real (inflation-adjusted) equity market returns, which were positive in every location, typically at a level of 3% to 6% per year… Although there have been setbacks, over the 116 years, equities beat bonds and bills in all 21 countries for which we have a continuous stock market history. For the world as a whole, equities outperformed bills by 4.2% per year and bonds by 3.2% per year.”

Investors did not benefit from investing in more volatile stock markets as compared to more-stable markets… US equities had a standard deviation of returns of 20.1% (annualized standard deviation)… The most volatile markets were Portugal (34.4%), Germany (31.7%), Austria (30.0%)…which were the countries most seriously affected by the depredations of war, civil strife, and inflation.”

Government bonds were on average disappointing for investors over the 116 years from 1900 to 2015. Across the 21 countries, the average annualized real return was 1.0% (1.2% excluding Austria’s very low figure). Although this exceeds the return on cash by 1.3%, bonds had much higher risk. As already noted, real bond returns were negative in four countries, with German bonds doing worst once the 1922–23 hyperinflation is incorporated. In the United… Since 1900, the average standard deviation of real bond returns was 13.1%, versus 23.6% for equities and 7.7% for bills (these averages exclude Austria). US real bond returns had a standard deviation of 10.4%, versus 20.1% for equities and 4.6% for bills.”

Creating value through timing

“Time variation in expected returns can be explained by either rational or irrational theories.

  • Rational explanations include time-varying volatility, time-varying risk aversion, and time-varying risk of rare disasters… forward-looking required risk premiums should be higher after bad times; for example, the equity premium should be higher after recessions and financial crises…
  • Irrational explanations often rely on time-varying investor sentiment, cycles of greed and fear, as well as social interactions.

Survey-based evidence clearly sides with irrational explanations. The consensus in survey-based return expectations is often bullish when rational valuation measures suggest low required (and thus rationally expected) returns.”

“Bond investors’…estimates of prospective long-run returns almost always start from market yields. In contrast, equity investors rarely use starting (dividend) yields when assessing long-run expected returns. This makes perfect sense if they assume expected returns to be constant over time…When it comes to equities, investors and academics have traditionally assumed constant expected returns and have estimated prospective returns based on long-run historical realized returns…The standard…capital asset pricing model (CAPM), was a one-period model that automatically implied a constant equity premium.”

The tech boom and bust cycle around the turn of the millennium was important in shifting the conventional wisdom. As the 1990s bull market continued, estimates of the long-run equity premium based on historical average returns kept inching higher. Yet, any valuation measures based on lower starting yields should imply lower future returns. Periods of sustained declines in required returns boost contemporaneous returns (the so-called discount rate effect) and are especially dangerous for investors who believe in constant expected returns: A rearview-mirror perspective made the equity premium seem highest at the end of a long bull market, just when market valuation ratios were flashing red. After the bust in the early 2000s, it was evident that forward-looking valuation measures had given an empirically and logically better signal than had historical average returns.”

“Valuation-based indicators are the most widely used measures of market conditions…Value- or yield-based expected returns [include]…the earnings yield, using past-decade earnings…[and] a sum of dividend yield and the real trend growth of dividends-per-share… [The figure below] traces these two expected real return proxies all the way back to 1900 and compares them with the next-decade average realized real return for the US equity market… Both expected return series appear to predict next-decade realized returns of the S&P 500 Index. The forecasts were especially good during the first half of the sample. The biggest forecast failure occurred in the 1990s, when both predictors forecasted low returns but the bull market extended until the year 2000 before reversing.”

“Although contrarian indicators exhibit an apparently promising ability to predict market returns over long horizons, real-world applications using such tactical market-timing rules have a surprisingly poor empirical track record… exploiting this time variation is so difficult that most investors are better off resisting the temptation to try.”

“The second-most important indicator is recent price momentum, or trend. Across assets and over time, recent outperformance tends to persist (for time-series evidence. Good performance in recent months (up to a year but not over multiple years) tends to be followed by further good performance, and vice versa. Value and momentum signals tend to be negatively correlated because often an asset that is cheap today tends to have performed poorly in the past. When the two signals agree—for example, in the market-timing context when the market is cheap and has recently begun to improve—the double signal is especially strong… Over long horizons, value and yield indicators tend to have the best predictive ability; over short horizons, momentum and macro indicators are more helpful. No tactical indicators are particularly reliable for near-term market timing.”