This is the first in a short series of posts on the recent update of the Dimson-Marsh-Staunton long-term return study sponsored by Credit Suisse.
The Dimson-Marsh-Staunton study, first published twelve years ago under the title Triumph of the Optimists, covers a number of important investment issues based on an abundance of statistical information on equities, government bonds, bills and inflation in nineteen markets going back to the year 1900. For any investor interested in the long-term results of financial assets, this study should be of great interest. Naturally, it shares the usual problems of other historical analyses. There is survival bias since all nineteen markets included in the study are still in existence today. Russia is one obvious omission. It was the seventh largest equity market in 1900 and had comfortably outperformed the US market in the last decades of the nineteenth century but obviously did not survive the revolution. It is also possible that the experience of the past 112 years may be completely irrelevant for the future. The period may have been atypical or asset returns may be inherently non-stationary so that the future will bear no resemblance to the past. Zhou Enlai allegedly once said that it was much too early to ascertain whether the French revolution had been truly a success. Similarly, we may consider 112 years of data insufficient to represent reliable evidence of long-term return behavior. But even though the future may indeed be quite different from the past, I am willing to assume that there is some valuable information in these past returns to help us assess the prospects for risk and return of financial assets in the future.
At first sight, the results of the study for the world portfolio – made up of the nineteen countries – are as expected: equities had the highest compound real annual return (5.4% in USD), followed by bonds (1.7%) and bills (0.9%). The order is preserved in the annual volatility measures: the standard deviation is highest for equities (17.7%), followed by bonds (10.4%) and bills (4.7%). Investors were rewarded for bearing risk, on average. But once we look beyond these basic numbers, some unexpected results emerge that challenge some of our standard assumptions.
Investors have a fixation on annual return and risk measures. We use them routinely in most of the standard asset-liability models. Thus, the main result of the modeling exercise – the “optimal” investment strategy – is optimal only if our investment horizon is in fact one year. Yet most investors pride themselves on being long-term oriented. We also focus on volatility as the principal risk measure. Standard deviation being a symmetrical measure, it handles asymmetries or optionality in the returns rather badly. This is unlikely to be a major problem for equities but bonds are, by their very nature, instruments with an asymmetric risk structure. If held to maturity, bonds offer at most whatever the promised return happens to be. But investors also face the possibility of losing part of all of their capital in case of a default. The likelihood of default may be small but it cannot be ruled out. In addition, the promised return of a traditional bond is a nominal return. Should there be unanticipated inflation during the life of the bond, the investor is completely unprotected. To put it somewhat provocatively: bonds offer no upside but a potentially important downside that is not completely captured by traditional risk measures.
Thus, for long term investors, volatility seems to be too narrow to be a useful measure of risk. Ideally, we would want a broader concept that encompasses all factors that increase the likelihood that we will not be able to meet our long term (real) objectives. And once we take this long-term viewpoint, bonds look rather less attractive. In contrast, some of the higher equity risk may actually wash out over time, even though whether equity risk truly decreases with longer investment horizons remains the subject of a heated academic debate.
The Dimson-Marsh-Staunton study provides some empirical support for this somewhat jaundiced view of the attractions of bond investment. When we look at the nineteen individual countries covered by the study, we find that bonds in six of them – Germany, Italy, France, Belgium, Finland and Japan – generated negative real rates of return for the overall period. Thus, in almost a third of the countries in the sample, investment in government bonds destroyed real wealth. In some cases, the destruction was massive: in Italy, 86% of the initial real value was destroyed, in Japan, 70%. Actually, for the first eighty years of the study, bonds had negative real returns in eight more countries and for the world as a whole – only when inflation started to be controlled in the eighties did bonds start to perform well. The only five countries with positive real bond returns up to 1980 were Denmark, New Zealand, South Africa, Sweden and Switzerland. Even Sweden, highlighted in the study as the only country that places equities, bonds and cash in the top four in terms of real returns, went through a difficult period for bond investors. For over fifty years, from the mid thirties to the mid eighties, the compound real return on Swedish bonds was negative – more than two generations experienced markets that led to the destruction of real wealth.
Bonds may be “safer” than equities in terms of volatility but their sensitivity to unanticipated inflation made them a disastrous investment in many countries for a substantial part or all of the last one hundred plus years. It appears that, for much of the last century, investors in many markets systematically overpaid government bonds because they underestimated inflation and/or the default risk. Naturally, governments had and continue to have obvious incentives to inflate away some of their debt burden unless that debt is inflation linked. Given the current economic situation and the current yields on bonds, is it really reasonable to assume that “this time, it will be different” and that there will be essentially no inflation in the foreseeable future, as suggested by the current yields of government bonds?
Despite appearances, this post is not an covert advertisement for equity investment. It is true that equities offered positive real returns in all nineteen markets and outperformed bonds and bills comfortably. But the graphs in the study also show that equities have often given investors a wild ride even though, in the past, they always recovered from what in some cases were disastrous drawdowns. It would be overly naive to recommend that investors should blindly increase the equity weight of their portfolio at the expense of bonds or even to invest their entire wealth in equities. Not every investor’s horizon exceeds one hundred years and there have been extended periods of horrific performance by equities in most countries with recovery periods that stretched, in some cases, into decades. Also, 2008 has shown that most investors, even those proudly proclaiming themselves to be very long-term oriented, become extremely myopic in periods of stress. Some because regulations on solvency force them to be (insurance companies, pension funds), others simply because the pain of watching their capital being destroyed becomes unbearable. As a result, many investors found that their strategy, formulated in happier days, was a fair weather strategy that could not survive the stress induced by the crisis with the all too common result that it was abandoned in favor of a more conservative strategy at the worst possible moment.
So what instrument is riskier in the long run, equities or bonds? If we take a broad view of risk, to this observer, bonds actually appear riskier. If, as an example, we look at the impact of the second world war on the nations that lost the war or on those whose territorial integrity was compromised, the negative impact on government bonds was dramatic and usually permanent. Equities suffered equally badly but bounced back. The lack of upside potential combined with the moral hazard created by the fact that the debtor controls, to some extent, the real return of the instrument, makes a irrecoverable loss on bonds more likely than for equities where bankruptcies of individual firms occur regularly but will have limited impact on a well diversified equity portfolio.
Thus, long term investors should exercise care when using an asset allocation that is derived from a model that is focused on the short term. Recall that bonds look too good in such a model, equities look too risky. Investors who have the courage and discipline to look away in periods of extreme stress may want to consider shifting to a strategy with a higher equity content. The ride would become a bit wilder but the risk of the bond position being eroded through unanticipated inflation or exploded through default would be reduced. I am personally convinced that many investors focus too much on short-term risk. In the long run, the comfort of their low risk allocation light on equities but heavy on fixed income instruments may prove to be an illusion.