Q. So this has become a very popular theme in portfolio construction – allocation according to risk, rather than by asset class.
A. Yep, especially since 2000, this has been a super hot topic. Most basically, it was driven by the observation that, in a 60-40 portfolio, the vast majority of the volatility was driven by stocks, and when they crashed (in 2000 and 2008), the losses totally overwhelmed the bond allocation, and the overall portfolio was cratered.

Q. Well, that sounds pretty logical…
A. It sounds great, but as usual the theory is not the same as the practice. What all these sorts of approaches do is allocate today in response to historic performance– they essentially assume that historic volatility is predictive of future volatility. Makes you feel good, because you have some math to point at, but you’re still basically looking forward by looking backward.

Q. And the basic idea that bonds are not volatile has certainly been challenged by the last six weeks of action in that market. Even the PIMCO total return fund was down a full one percent yesterday.
A. Right, and 10 years are down around 5% since mid May. The problem for these theories is that, over the previous 30 years, bonds were totally stable, and in fact slowly accreted value over the entire period. Meantime, stocks were terribly volatile. But bonds while bonds may well rebound from these lows over the next few weeks or whatever, on a bigger timeline– with the 10yr at 2%– they have almost no where to go but down. That rather obvious fact is the sort of thing these models don’t account for.

Q. So, bottom line is that the idea of risk parity is great, but maybe a lot harder to achieve than the models would lead you to believe.
A. Yes. And that’s precisely why many leading institutions lean so heavily on alternatives to achieve what you might call super-diversification. Their basic approach is, hey, we cannot predict the future, so we need to be in really uncorrelated assets and strategies to reduce risk. For example, Princeton’s been completely out of bonds for a while– looking pretty smart right now.