March 7, 2006
IN SEARCH OF STRATEGIC FRIENDSHIP
If there's any consensus in the business of managing money, diversification probably comes closest to the ideal of strategic agreement and accord. Diversification, after all, is your only friend in the long run. Timing the market and choosing securities is far more exciting, but who among us thinks such high arts can be sustained?
The alternative, of course, is the time-tested antidote of diversification, which is the only effective defense against the great unknown, otherwise known as tomorrow. Since no one really knows what's coming, prudence dictates the embrace of diversification, at least for us mere mortals who have no chance at becoming the next George Soros or Warren Buffett.
To be sure, not all diversification is enlightened. Owning three tech-stock funds, for example, doesn't come close to informed diversification.
Yes, it's true that there's no shortage of debate about what constitutes a prudent level of diversification. But there are some ground rules, starting with the fact that investors should have exposure to the various asset classes. We're defining asset classes as those groups that exhibit relatively unique characteristics that distinguish them from other groups. Price correlation is one way to measure such characteristics.
Bonds and stocks, to cite the obvious example, post a sufficiently low correlation with each other over time so as to warrant embracing them as separate and distinct asset classes. What that means is that there's a good chance that when one's losing money, the other will be holding its own, if not posting gains. In fact, history shows just that. Owning bonds in 2000-2002, for instance, offered valuable ballast when the stock market was suffering.
But how do you protect a portfolio when bonds are under attack? Traditionally, stocks can help, although the record shows that stocks and bonds have been known to tank together at times--this limited form of diversification fades just when you need it most. Meanwhile, cash is an asset class in its own right, and always holds its value, at least in nominal terms, which makes it an essential candidate for diversification. But holding too much cash opens one to the inflation threat over time.
The true solution is embracing a broader array of asset classes. That's not been easy for individual investors in the past, but as mutual funds and exchange traded funds bring formerly exotic asset classes to the masses, quite often in cost-efficient index-fund packages, the opportunity for greater diversification is here.
The question then becomes: is more diversification better? In search of an answer, or at least some perspective, we crunched the numbers on 10 asset classes (courtesy of Morningstar and Dow Jones) for the decade through the end of 2005 and compared three diversification strategies:
1) an equal mix of U.S. stocks, U.S. bonds and cash
2) an equal mix of 10 asset classes
3) a traditional pension-fund-inspired 60% U.S. stocks/40% U.S. bonds mix
The asset classes for the three strategies are based on returns from the following indices, most of which are available via index funds or an actively managed fund that serves as a proxy:
1. U.S. Stocks (Russell 3000)
2. U.S. bonds--Treasuries, corporates, mortgaged backed (Lehman Brothers Aggregate)
3. Cash (3 Month T-bill)
4. Emerging markets debt (Citigroup ESBI-Cap Brady, in dollars)
5. Foreign government bonds, developed countries (Citigroup Non-$ World Govt, in dollars)
6. U.S. high yield debt (CSFB High Yield)
7. Foreign developed markets stocks (MSCI EAFE, in dollars)
8. Foreign emerging markets stocks (MSCI EM, in dollars)
9. Real estate investment trusts (Wilshire REIT)
10. Commodities (DJ-AIG Index)
Starting on December 31, 1995, we rebalanced once a year, every December 31, back to the relevant weight as per each strategy, running the test for 10 years through December 31, 2005. The results were encouraging, and perhaps a bit surprising. Overall, an equal mix of the 10 asset classes delivered a 9.1% average annual total return over the 10 years, well above the stocks/bonds/cash return of 6.9%, and slightly higher than the 8.9% of 60% stocks/40% bonds.
Much of the superior performance in the 10-asset-class strategy's stellar returns in 2003-2005, as the chart below illustrates. To be sure, sometimes owning everything disappoints. But over time, our sample suggests that it pays to take an expanded approach to diversification, if the short-term results won't win you any applause at the next cocktail party.
What's more, the equal mix of 10 asset classes is mindless, the byproduct of automatic rebalancing at the end of each year. But owning 10 asset classes opens the door for enhancing results a bit, at least for those with the ambition and skills to customize a portfolio. Consider junk bonds. When the spread of high-yield debt is relatively small relative to Treasury yields, there's a strong case for pulling back on the junk bond weight, and vice versa.
Whether you're tweaking or mindlessly rebalancing, there's a case to be made that investors should own some of each of these asset classes. Why? For the simple reason that you don't know what's coming, and so diversification truly is your only friend.
Posted by jp at March 7, 2006 8:55 AM
You say tomato, I say "to-mah-toe."
Yes, I recognize the distinction, at least as it's recognized in the trade. But ultimately asset allocation is a means to an end, i.e., diversification. AA is the tool, diversification is the objective. For a given asset class, diversification is first and foremost about the number of individual securities. For a broad portfolio of multiple asset classes, diversification is arguably about the number of asset classes. Thus my preference for the financial faux pas. In any case, a rose by any other name, etc.
Posted by: James Picerno at March 8, 2006 8:34 AM
Nicely done although what you describe is usually referred to as asset allocation. Diversification is done within each asset class.
Posted by: thc at March 7, 2006 10:45 PM
A couple of points. First, the commodities index used here is the Dow Jones-AIG benchmark, which caps energy at about 33%. The main competitor, Goldman Sachs Commodities Index, is production weighted and recently allowed energy to represent 74% of the index. I specificially used DJ-AIG because it minimizes oil's influence, relative to GSCI.
As for the questions about commodities over the long run, there's empirical support for Keynes' theory on backwardation, which posits positive expected returns for futures over the long haul. See the paper "Facts and Fantasies about Commodities Futures" at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=560042
As for TIPS, I didn't use them in the study here because the history is shorter, inflation-indexed Treasuries being a late-1990s invention.
Posted by: James Picerno at March 7, 2006 4:30 PM
Also, where are TIPS (Treasury Inflation-Protected Securities) as an asset class? They show low correlation to most other financial assets. Since the "commodities index" is a financial markets claim on the physical asset, it may not capture inflation over long periods (see Simons link above).
Finally, where is real estate in the list of 10 asset classes? This is arguably a higher weighting in individual investors' portfolios than anything else.
Posted by: Bond investor at March 7, 2006 3:12 PM
The time misspecification questions have already been covered in the comments above.
The relative size, liquidity and sophistication of these markets may also bias the findings. Three contentions:
1. In the past, the emerging markets debt market was too small for adequate investment on an institutional level. Much of the outperformance occurred simply because the liquidity/marketability premium faded over time, only to come back in 1997-8 with the Asian currency crisis and Russian debt default, then fade (again) over the last 7 years.
2. The Commodities index used is basically an oil proxy. Except for oil, Au, Pt and natural gas, all commodities have shown a significant downward expected return. See Howard Simons.
3. While the use of the Lehman Aggregate (bonds of maturities from 1-30 years, with wtd average index duration of 4.56 years as of 2/28/06) may work for institutions like pensions, individual investors have specific liabilities that we may want to target (college tuition, retirement), so a bond index's return does not capture all of the utility of bonds. A municipal zero maturing in 2021 is great for my 3-year-old's freshman year tuition.
Posted by: Bond investor at March 7, 2006 2:43 PM
The previous commentator raises a good issue: Do 10 asset classes work over longer periods beyond just the 10 years analyzed here? My gut reaction is "yes," but without the hard data I can't say for sure. In fact, I'm in the process of building a longer-term database, having wondered about the longer-term implications as well. Rest assured that once we have the numbers and crunch the data, we'll report it here.
Posted by: James Picerno at March 7, 2006 1:37 PM
It seems to me that the advantage of the 10-asset classes strategy over the other two approaches is due to its stronger performance since 2001. Before 2001, this approach earns “mixed reviews” (I presume that in 1998 the 10 assets earned exactly nothing). So, if one calculates from 2001 back 10 years, a different winner may emerge.
Posted by: Anonymous at March 7, 2006 1:26 PM