Chapter 24. The Paradox of Diversification
Jon Stewart: “My mother is 75 and she bought into the idea that long-term investing is the way to go. And guess what?” Jim Cramer: “It didn’t work.”
Interview on Comedy Central’s The Daily Show Jon Stewart, March 11, 2009
October 2008 saw virtually every kind of asset collapse together, in truly unprecedented fashion. This inflicted losses on most investors around the world and showed that they need a new way to mitigate risk.
By Monday, October 6, 2008, all reason for optimism had been exhausted—there was a run on banks, governments evidently could not deal with it, and markets were wildly overvalued after decades of moral hazard and herding behavior by investors. The result was the markets’ most disastrous week ever. It set all the investment orthodoxies on their heads and proved that markets across the world had been caught in a synchronized bubble.
As markets opened that Monday, the S&P 500 stood at 1099.23. By Friday morning, it was at 839.8, down 23.5 percent. At that point, it was the worst week for the U.S. stock market on record, including the bear market following the 1929 crash.
More remarkable was that it was indiscriminate. MSCI’s EAFE Index, covering developed markets outside the United States, fell 22.4 percent for the week; its World index fell 22.5 percent; and its emerging markets index fell 21.3 percent. Its BRIC Index, where investors could allegedly “decouple” from these problems, fell 21.6 percent. With total synchronicity, virtually all world markets lost one-fifth of their value during the week.
Even within markets, the sell-off did not discriminate—just look at the equal-weighted version of the S&P 500, in which each stock accounts for 0.2 percent of the index (unlike the standard index that is concentrated in the biggest stocks). The equal-weighted index should be better protected in a sell-off after a bubble in certain sectors or stocks. In the years after the Internet bust, it gained while the main S&P fell.
But in 2008, the average stock fell even more than the main index. The equal-weighted index beat the main S&P throughout the rally of 2003–2007, but when disaster struck it fell by more, ending 48.5 percent down from its peak. The phenomenon was global. The MSCI World Index of developed world stocks fell 49 percent during the sell-off, the same as the capitalization-weighted version of the index. By the calculations of GMO, the Boston-based fund managers, the main index should suffer a fall this severe about once every 100 years; it should happen to the equal-weighted index only once every 34,000 years.1
When a bubble bursts, it is because it is overvalued. But overvaluation implies that the assets in the bubble are overvalued with respect to something else. In this case, every asset that bore any element of risk came down simultaneously. Rather than going somewhere else, the value simply evaporated because it should never have existed in the first place. Markets had grasped that there was more risk in the planet than prices had reflected.
All the asset classes now tied to equities came down with equal ferocity. Over the week, the Australian dollar fell 21.4 percent against the Japanese yen. In other words, the carry trade had collapsed, and the losses traders took on the carry trade were virtually equal to the losses they took on stocks. As U.S. investors made forced sales and repatriated their money, they bought dollars, pushing all other currencies down.
As for commodities, oil fell 16.8 percent that week, continuing the epic slide that had started on Bastille Day. Industrial metals shed 15.6 percent (part of a 64.9 percent fall since their peak in the spring). The cost of hiring a container ship, as measured by the Baltic Dry Index and a proxy for world trade, dropped 26 percent.
Even the assets that should be safe havens during market carnage suffered falls. Gold, traditionally the ultimate refuge from volatility elsewhere, rose more than 11 percent early in the week, but lost it all in a sell-off that Friday. Bonds roughly held their value for the week, but there was no great rush to take money out of currency carry trades, commodities, or stocks and put it in bonds. That would have shown up in rising bond prices. Instead, one-fifth of the wealth that had on paper existed in these assets evaporated.
Research from Research Affiliates, the investment company run by the former academic Rob Arnott, tracked 16 different asset classes, including government and corporate debt, equities, loans, and commodities from around the world. In September 2008, all fell except for long-dated U.S. government bonds (which gained 0.4 percent). In October, all of them fell together, and a portfolio equally weighted among all 16 asset classes would have lost 14.4 percent that month—an almost inconceivable event. Even in August 1998, the month of the LTCM meltdown and the closest previous approach to such a correlated meltdown, six asset classes had shown a profit.2
This collapse was the ultimate proof that different asset classes, many of which had only opened to investors after financial innovations of the last decade, had come to reinforce each other. The value of each was not contingent on real-world conditions as much as valuations in other markets. This was the ultimate consequence of the herd-like behavior that the investment industry had encouraged for decades; the herd finally stomped over all of the world’s markets.
This blew up investors’ most basic notions about diversification and long-term returns. Under the academic orthodoxy that had developed over the previous half-century, diversification is about balancing different asset classes. The correlations between these different assets are assumed to be relatively stable over time; that was why investors hunted for new uncorrelated asset classes in the first place.
If markets are not driven by the same flows and prices are not set in reference to other markets, then diversifying your investments by dividing the world into asset classes or geographies should indeed reduce risks. But now that money travels so easily among them, this is plainly not the case. Traders thought they had made several uncorrelated trades, but in fact they had made the same bet—that the United States was in trouble but “decoupled” emerging markets would be fine—many times over.
All were exposed to the same risks—that money would be withdrawn quickly (“liquidity risk” in the jargon) and that the big rise in commodity prices would end. The crash demonstrated what might be called a new “paradox of diversification”; the more investors bought in to assets on the assumption that they were not correlated, the more they tended to become correlated.
In the future, it would make more sense to divide the world by risk. If an investment is not prone to the same risks as the others you already hold, then buying it will reduce your overall risk. If it is subject to exactly the same risk, then buying it is pointless, even if it is in a different asset class or country. Rather than balance between stocks and bonds, for example, it might be better to balance the risks of inflation and deflation, which both affect stocks and bonds. Diversification itself is as good an idea as ever. You should not put all your eggs in one basket. But in the globalized world, you can put your egg into a different country and still find that it is in the same basket.
Further, the point of a risk is that you do not know in advance if you will be paid for taking it. It cannot be gauged with precision—so the very concept of risk argues for leaving greater margin for error—and it does not lie in correlations between asset classes. Rather, risk management should be about guarding against conditions that could cause those correlations to change.
Another core notion must also be rethought. Ever since the 1950s, the cult of the equity had held sway, backed by both theory and experience. In theory, investors get their profits in return for taking a risk. That risk means that in the short term, returns on equities are volatile. But in the long term, they will get a premium for taking that risk so that equities will outperform safer investments, such as bonds. When teaching valuation, business schools typically assume that this “premium” will be about 7 percent each year.
Exhaustive historical studies show that equities easily beat bonds over the very long run. Compounded over decades, the gap is astronomical. Over the twentieth century, according to research by London Business School, U.S. equities rose at 10.1 percent per year, compared to a 4.8 percent rise for bonds, and a 3.2 percent rise for inflation. Compounded over the century, that meant that $1 in stocks in 1900 became $16,797 in 2000, compared to $119 for bonds. Thanks to inflation, you needed $24 to maintain the buying power of one 1900 dollar. A pound invested in UK stocks did almost as well, turning into £16,160, but UK inflation was roughly double that of the United States.3
Thus the premium equities paid over bonds was robust and high. The London Business School study covered 16 countries from 1900 to 2005 and suggested the extra return on equities each year was 6.1 percent—almost as high as the business school guesstimate.4 So pension funds, and individuals egged on by advertisers, allocated ever more to equities, safe in the knowledge that in the long-run they would do best.
Such faith in long-term equity outperformance made stocks more expensive. Since 1954, the dividend yield on equities had been less than the yield on bonds. That aligned with the orthodoxy that bonds needed to pay a higher yield because equities would rise more in the long-term. But in December 2008, as equities crashed, that assumption came to an end, as dividend yields rose above bond yields.
The comforting long-term averages had concealed another lesson of history: that stocks can do worse than inflation for very long periods. During the twentieth century, all but three countries outside the United States had at least one period of at least 20 years in which equities failed to beat inflation. Developed countries like Japan, Italy, and Germany all went for periods of about half a century in which equities lost money.5 The assumption that the “long-term” would always bail out equities was always a lazy one.
In February 2009, after the correlated crash, this was proved conclusively. Anyone who started investing in February 1969—when baby boomers were joining the workforce and starting to save—would find 40 years later that bonds had done better than stocks.6 This ran against all the advice boomers had ever received from the investment industry.
Some argue that the 2008 correlated crash was an aberrant event and provided a great opportunity to buy cheap. When stocks recovered in 2009, their yields dropped back below bond yields. But that does not mean these events can be dismissed. The point of investing in a diverse group of assets, rather than just picking a favorite, is to guard against sharp downturns—which may happen just as savers need access to their money. Protesting that this was an extreme event and that correlations can change in extreme conditions misses the point. Diversification is to protect against just such extreme events.
Relying on stocks to win in the long run is also dangerous. Investors persuaded themselves to pile into equities, to the exclusion of bonds, because the long term would always be there to bail them out. History makes it clear that over time stocks have done better than bonds and so it would be foolish to ignore them. But the historical era for which we have the most complete data involved a long period of peace and prosperity as the world recovered from two world wars, abandoned communism, and enjoyed many technological advances. There is no reason to assume that this can be repeated. Relying on stocks to outperform in the long run only makes sense if we know we will live to be 100 years old. This might be a sensible plan for endowments and charities—but not for any individual.
Nothing falls forever. After the disaster, markets picked up. But that recovery started far from Wall Street or the City of London.
In Summary
• Diversification in its traditional form—balancing different asset classes—has failed. After the 2008 collapse, a new form of diversification is needed.
• The way to achieve this is to allocate according to risks, not asset classes.