Diversification comes to those who wait
By Howard L. SimonsAnyone who lived through the debacle of 2008, or for that matter anyone who owned any financial asset anywhere during the lift-off rallies of 2003 and 2009, can be excused for thinking diversification lives in financial textbooks and nowhere else. Not only have asset returns converged during financial crises since reliable records have been kept, the mechanics of high-frequency trading, algorithmic trading, and exchange-traded funds all have combined to push instruments up and down in rhythmic lockstep.
Or have they? The answer, like so many other aspects of market analysis requires a time horizon. (As an aside, I always have considered the never-ending debate between technical and fundamental analysis to be pointless without a time horizon: The longer the time frame involved, the more a market is likely to be driven by its fundamentals; the shorter the time frame involved, the more a trader has to rely on the regular patterns and formations of technical analysis.)
Just as there is no one as bearish as a sold-out bull or as convinced a move higher is a bubble than someone not invested therein, there is no one who complains about a lack of diversification as much as someone who has underperformed their benchmark. Incredibly, the combined tyrannies of index management and performance analysis drive some portfolio managers to trade variance swaps not for reasons of return but rather to maintain their volatility of return within a stipulated band relative to their benchmark’s volatility.
A man from Mars might consider this last action prima facie evidence of insanity.
Diversification of returns over time
One way to look at diversification is through the lens of mean-variance analysis. If the variance of returns increases over time independently of any changes in the trend in the mean of those returns, we can trust short-term convergence will devolve into long-term divergence of returns.
Those who need to demonstrate issue selection matters as much as index timing need to compare the distribution of returns for that index’s members over a range of holding periods. Ideally, these returns would be traced forward across successively longer holding periods, out to 10 years or more. While this is possible for individual issues, the very nature of equities works against us: Each year hundreds of issues in the broad Russell 3000 index disappear for reasons such as merger or bankruptcy. Commodity traders do not have to deal with corn going bankrupt, or heating oil and natural gas merging into one giant energy commodity.
To address this issue simply, the period total returns for all members of the Russell 3000 index just for periods of one and five days, and one, three, and six months are examined in Figure 1. As a statistical aside, the distributions of returns across five-day and one-month time periods are different from that of the three-month distribution of returns at near 100-percent confidence.
The progression of return mean and variance from five days to six months for the Russell 3000 does, in fact, show an expected expansion of variance unrelated to mean. Even if intra-index variance is low for the first week a portfolio is held, it should expand and provide diversification over time.
For the complete article, see the March 2012 issue of Active Trader magazine. Click here to subscribe.
|
|
