The paradox of diversification

Sometimes, writing this blog, I sense a woeful dearth of new ideas. I despair that I’ll come across as an oldster repeating the same story over and over, believing that I’m telling it for the first time. It’s true that I don’t remember every post I’ve written, but that’s not the problem. Rather, this blog occasionally becomes repetitive because there’s just as much herding in the world of financial literature as there is in the markets themselves.

In the wake of the financial collapse one of the “hot topics” is the failure of diversification to protect portfolios. I wrote about this briefly in my recent post on The Endowment Model of Investing. John Authers in The Fearful Rise of Markets: Global Bubbles, Synchronized Meltdowns, and How to Prevent Them in the Future (FT Press, 2010) joins in the conversation. He points to the new “paradox of diversification”—that “the more investors bought in to assets on the assumption that they were not correlated, the more they tended to become correlated.” (p. 166)

In fact, he writes, everyone was exposed to the same risks. Liquidity risk was the most serious; the second was that the run-up in commodity prices would end.

Echoing Mohamed El-Erian, he claims that asset allocation should be done according to type of risk. Instead of balancing asset classes, he suggests that it might be more sensible to balance, for instance, the risks of inflation and deflation. Admittedly, his prose is much clearer than El-Erian’s, but that may be because his suggestion is simpler. (I wrote about El-Erian’s idea early in the life of this blog and offered a few thoughts about how traders could flesh it out.) For those who don’t recall El-Erian’s words in When Markets Collide, here they are:

“The ideal situation is to come up with a small set (three to five) of distinct (and ideally orthogonal) risk factors that command a risk premium. The next step is to assess the stability of the factors and how they can be best captured through the use of tradable instruments. This provides for a portfolio optimization process whereby the factors are combined in a manner that speaks directly to the investors’ return objective and risk tolerance. The end product is a more robust and time-consistent combination of asset classes that map clearly to the underlying factors.” (p. 233)

In an unleveraged or modestly leveraged world this style of portfolio building makes eminent sense and is theoretically elegant to boot. But as leverage increases and, in a crisis, funds sell whatever they can to meet margin calls, it doesn’t matter how carefully constructed a long-only portfolio is and on what principles it is diversified; it will get whacked. (I specifically use the example of a long-only portfolio because it was for this kind of portfolio that the traditional asset allocation model was devised. A hedged portfolio is a different kettle of fish altogether.)

Personally I prefer the notion of a portfolio with many moving parts rather than one that is fixed for a certain period of time—let’s say rebalanced once a year. Indeed, why not make a very difficult task a herculean one?