Diversification Can Be Everything

Diversification isn't everything, but it's a lot. And sometimes, it's everything.

It doesn't take much analysis to recognize that asset allocation's value has risen sharply this year. More precisely, the right asset allocation has generally made the difference between losing a lot of money and either losing a little or even turning a profit this year. Even a passive asset allocation across the major asset classes has generated potent benefits. Indeed, the year-to-date performance numbers for the major asset classes midway in 2008 are wide ranging, as we noted last week.

That's in sharp contrast to the horse race as it looked mid-year in 2007, when almost everything was running higher. No wonder that halfway through 2007 there was chatter that asset allocation was washed up as a strategic tool. Who needs to own everything when robust returns are falling out of trees?

So it goes in the cyclical mindset of investing analysis for the crowd, which too often succumbs to belief that the recent past informs the future. Yes, wisdom on Wall Street tends to wax and wane over time, but on these digital pages diversification's value endures, as our update on correlations remind.

But let's be clear: diversification is anything but static. It prevails over time, but in the short run its offerings vary, as our chart below reminds. (For easy reading, click to launch a larger view of the chart.)

Let's start by noting that as equity markets have corrected this year, correlations between the U.S. stock market (Russell 3000) and foreign markets have trended higher. (Correlations range from 1.0, or perfect positive correlation, to -1.0, perfect negative correlation. Generally, mixing assets with less than perfect correlation improves expected risk-adjusted performance.) The same can be said of U.S. stocks and REITs. That's par for the course in bear markets, where like-minded products tend to dive together. Of course, there are always a few surprises along the way. The fact that the correlations between REITs and equities is so high of late can be chalked up as bad luck born of the fact that bull markets in real estate and stocks seemed to have peaked together this time around.

In the meantime, look at the brown, purple and blue lines in the bottom right-hand corner of the chart. Note that all have been falling for the past year or so relative to the Russell 3000. (For simplicity, all correlations in the chart are calculated relative to the Russell 3000.) Unsurprisingly, all three of these lines (representing commodities, U.S. bonds and foreign-government bonds denominated in foreign currencies) have historically provided valuable diversification benefits relative to stocks when the diversification is needed most. By the standards of the first half of this year, those benefits are once again alive and kicking.

Nonetheless, correlations fluctuate over time, sometimes dramatically. The fluctuations occasionally offer clues about what's coming. Such was the case in 2006 and early 2007, when correlations generally were rising. But just when it seemed that diversification's value was fading, the worth of broad ownership was only beginning to shine.

For those who stayed committed to holding a diversified portfolio across the major asset classes, the payoff has been sweet. But we have no doubt that when the bear market is over, and bull markets in equities bloom anew, diversification and asset allocation will once again fall victim to criticism and neglect. Perhaps that's why diversification remains so valuable over time, i.e., many if not most investors ignore its benefits at critical junctures, effectively leaving large arbitrage opportunities across asset classes on the table for strategic-minded investors.

If everyone woke up to strategic diversification's value, the benefits would undoubtedly diminish. Even so, that would inspire a fresh wave of skepticism and create new asset allocation opportunities as investors bailed out of the idea. Hope, in short, never fades for strategic-minded investors.

Comments

[Afam]

It takes for the tide to go out to see who's swimming naked. smartinvestorafrica.co...

[Aalan]

Thanks for the chart. I'd seen (and saved) an earlier article about the THEORY of correlations changing over time, but this is the first place I've seen it quantified. Very helpful!

[buyitcheap]

This is an excellent teaching graph - thanks!

[mmbc]

Excellent article highlighting non-static correlations between distinct asset classes.

Being Lazy may be one way to ride out the uncertainity:
www.maxmoneyblog.com/b.../

[Kyrriana]

Agreed. And I think that the effect you mention is even more pronounced recently. I track a shorter term correlation matrix for the major asset classes at www.assetallocation.co... and at the moment there haven't been many places to hide as everything seems to be moving together. I agree, though, that as sanity gradually returns it will be important to be adequately diversified again.

[Kyrriana]

Whoops. The link in that previous comment should have been www.assetcorrelation.c...

[Igneous]

What about the big commodity party everyones been having?

[Smart ETF]

Good stuff Jim. I don’t believe folks have issue with asset allocation; but rather its reliance on mean variance; note that your chart uses a rolling 36 month moving average. Had you used traditional linear correlation and regression you would have not identified the dramatic swings in correlation over time. In fact, if you had used shorter intervals your swings would have been much larger; especially in times of extreme events (as correlations move towards +1); as the saying goes: the only thing that goes up in a down market is correlation.

Linear Correlation falls under the family of Dependency models. A more sophisticated dependency model that better represents a dynamic marketplace is a method called Copula Dependency; think of it as a dynamic correlation model that continually test the relationship between two securities.

The advantage of using a copula dependency model is that it would identify the increasing volatility in the marketplace (in conjunction with a GARCH model) and recognize that correlations would be advancing during large market moves and invest accordingly. In other words, it would recognize that assets that are traditionally non-correlated may become highly correlated during extreme events and therefore opt to invest in short-term treasuries as an alternative.

The mean-variance disciples use the laws of large numbers to forecast performance. Over the past 80 years the domestic equity market has returned 10% annually. Note that the market is down over the past 10 years (and 3 years, and 1 year, and YTD); just as it was from 1800 – 1815 (15), 1835 –1843 (8), 1852 – 1861 (9), 1880-1896 (16), 1907 – 1921 (14), 1930 – 1949 (19), 1968 – 1981 (13), 2000 – today (8). The down markets caused by deflation over the past 200 years lasted, in sequential order: 8, 16, 19, and so far 8 years. So mean-variance is like a clock that is right twice a day, even if broken. If an investor enters the market at the top of one of these long-term cycles it could easily take up to 30 to 40 years to break-even. Let mean-variance R.I.P. and take a look at Extreme Value Theory.

[Tao Man]

Very good stuff Smart ETF - right up to where you exaggerate the time it takes to break even from market tops. Taking dividends and inflation into account is important. Take someone who had invested at the top of the market in 1929. The S&P Composite Price index didn’t return to its old 1929 high until 1954, but the S&P 500 Return Index rebounded to its old 1929 high in 1945, and on an inflation-adjusted total return basis, the S&P composite returned to the 1929 high at the 1937 bull market top.