Stockmarket valuations


Great expectations


Jan 31st 2002

From The Economist print edition



Around the world, equities have beaten bonds for the past century. Where's the risk in that?


AS AMERICAN equity investors and their advisers will tell you, it is “stocks for the long run”: over time, equities perform better than bonds. Even had you put all your money in shares just before the stockmarket crashes of 1929 or 1973, you would in the end be far richer than the wallflowers who played it safe with bonds. That was the thrust of a best-selling book by a business-school professor at Wharton, Jeremy Siegel, first published in 1994. By the end of the decade, it was common wisdom that every dip in share prices was merely an opportunity to buy.


Others have taken the notion to wacky extremes. In 1999, another popular book, “Dow 36,000”, predicted that the Dow Jones industrial average would nearly quadruple in three or four years (it is now around 9,800, and thus about where it was three years ago). Though the book's author has since, ahem, revised his target date to, maybe, “the end of this decade”, he remains unrepentant about his core thesis: whether shares are cheap or dear at current prices, they are the safest way to guarantee long-term wealth. Books by others bid the target up still higher: “Dow 40,000”, for one, and “Dow 100,000”, a $25 volume now discounted at $7.50.


Some of America's affection for shares—more American households have their savings in equities than in any other country—is understandable. In stockmarkets, the victors write the history, and most academic analysis of equity performance has been based on stockmarkets in America, which happened to be the most successful large economy of the past century. Moreover, nearly all analysis has focused on the years after 1925, for which the best data is available. Far less attention has been lavished on other parts of the world, or on earlier periods in America.


Sponsored by ABN Amro, a new study by two professors and a research director at the London Business School, Elroy Dimson, Paul Marsh and Michael Staunton, takes a more cosmopolitan view. Their book escapes the tyranny of American data; it looks at indices of total returns for 16 rich countries, using newly gathered data going back to 1900, a full quarter-century earlier than most other studies.


Have they put paid to the notion that shares beat bonds? No: actually, they confirm it, hence the book's title: “Triumph of the Optimists”. In every country in their study, the authors show that real (that is, inflation-adjusted) returns from equities beat bonds. Still, the devil, like the risk, is in the details.


The equity premium (also known as the equity-risk premium) is a measure of the average annual return over and above riskless debt, such as government bonds, that shareholders expect to receive as compensation for holding risky shares. This risk is no illusion: shareholders are always the last to be paid, after other creditors get their cut. The authors conclude that all the world's stockmarkets offer strong evidence that riskier assets, on average, return more to investors.


Mind the gap


Some accounts of the equity premium often go astray from here. Since equities return more, on average, over the 75 years of data previously studied, shares must not be so risky after all. In fact, so the argument goes, shares are no riskier than bonds. Clever investors who recognise this will push up share prices until the equity premium goes to zero. The Dow will be bid up to 36,000, 100,000 or whatever level ensures equity returns equal those of riskless debt.


These ideas are reinforced by a widely held belief: that in every single 20-year period of American stockmarket history, equities have outperformed bonds. One problem is that this is a bit of a statistical mirage. In the 75 years since 1926, it is possible to measure only three distinct 20-year periods and the rule is found to hold. Yet earlier, there were periods when shares lost out to bonds. Modern promoters of the 20-year rule are unburdened by knowledge of the past. Nor does the 20-year rule hold for other countries. Messrs Dimson, Marsh and Staunton found four stockmarkets—the Netherlands, Germany, Sweden, and Switzerland—where at times 40 years of market exposure were needed to ensure that shares outperformed bonds.



In addition, previous studies may have overestimated absolute equity returns, partly because they relied upon inadequate data. For instance, during the first world war, for which data is hard to find, returns were generally lower. Also, earlier studies look at the historical share-price performance of companies that survive today, rather than examine the performance of now-extinct shares that would have been in a past investor's portfolio.


Correcting for these factors, as well as using a full century of data, gives a wide range of equity-risk premiums, with Denmark at the bottom of the league and Germany at the top. Messrs Dimson, Marsh and Staunton estimate a global historical average equity premium, over bonds, of 4.6 percentage points (see chart). That is nearly half the widely received forward-looking estimate of 8.8 percentage points from Ibbotson Associates, a consulting firm, and the 8.5 percentage points taught in finance courses everywhere (chart 2).



Even this estimate may be too generous. Some stockmarkets, such as those of China, Russia and Poland, are not included in the study, since they were closed down under communist rule. That led to returns best described as “steeply negative”. If these markets were taken into account, the historical equity premium would be even lower.


Another way of measuring the equity premium—which is, after all, the extra return on equity that investors require—is to ask them. Surveys suggest that investors in America expect an annual extra return of 7.1 percentage points. Still, they appear capricious, raising their target after a period of strong stockmarket returns and lowering it during the lean years.


But all of these estimates are too high, the authors say. To get a truer picture, they take into account the decreased volatility of today's stockmarkets, as well as the effects of unanticipated profits in recent years. After adjusting for these factors, the authors argue that the best estimate of the equity risk premium worldwide in future is 4-5 percentage points.


The true level of the equity premium is not merely an academic debate. Pension funds, for instance, have to consider it. At present, their fixed liabilities—the money they owe to pensioners—are mostly offset by risky shares. Too much optimism about expected equity returns could have painful consequences. Recognising this, one British retailer, Boots, recently switched its entire portfolio into bonds.


Participants in pension schemes with defined contributions rather than defined benefits, such as America's 401(k) plans, face shock and disappointment if lower returns than expected leave them stranded at retirement. What is more, rich countries can expect a surge in the numbers of retired folk, starting in a decade's time, as the world's baby boomers start withdrawing assets for consumption. If the retired have all bet on shares, they will find that their long run is, it turns out, a crowded race.


Perhaps the biggest damage that comes when returns from equities are overestimated is that investment, and hence growth, in the world economy gets crimped. Every business-school student learns to size up a new project by comparing its expected returns to expected stockmarket returns. If the equity premium is overstated, managers may be overlooking profitable investments, from new factories to drug research to investments in poor countries.


Over the past century, newspapers have given repeated warnings about shares' lofty prices. Most of the warnings, with hindsight, were misplaced. Yet financial markets have a way of killing off the easy bets. Each time a financial pattern emerges—that small companies' shares outperform those of large ones, that shares outperform in January, and so forth—investors exploit the pattern, thus undermining it. Might the same thing happen with the relatively recent discovery of the supposed 20-year rule, leaving investors to wait for even longer before their shares beat bonds?