Pioneers Analytic Investors and Robeco have special websites with lots of good research on low volatility investing. See Analytic's here, and Robeco's here. It seems that low volatility investing is the new style, an addition to the hoary value/size/industry/region styles that predominate equity markets, so soon every major fund will offer its version on this theme.
Two things are key to a fund's low volatility success. First, that a fund merely targets a long low volatility subset of equities. Adding insurance over a standard portfolio misses the real value to low volatility investing, which is based on the fact that lower volatility subsets of equities are a free-lunch in Sharpe space: lower volatility and (slightly) higher returns. If you pay for insurance on a standard portfolio you are lowering your return, because insurance is costly: on average, low vol assets are underpriced (according to standard theory), volatility itself is not. Lastly, while high volatility subsets have low returns, shorting is nontrivial to implement and these stocks, by definition, have a lot of volatility, making them best to simply avoid.
Secondly, sellers need to understand there is a trade-off, and this is benchmark risk. As Harin de Silva of Analytic Investors confided to blogger/S-guru Pat Burns at a conference last years:
He says that marketing low volatility portfolios is hard because they have a high tracking error to the (market cap) index.
In practice, low volatility portfolios are risky because they dramatically underperform in bull markets. Relative returns dominate absolute returns in terms of allocating money as evidenced by the fact that any one time the inflows to funds goes to the latest 12-24 month top-performers regardless of markets. In bull markets people like Mary Meeker, Henry Blodget, Jack Grubman are anointed trenchant investing geniuses when assets are bubbling, in bear markets its people like Nouriel Roubini and Nassim Taleb, and despite their different reasoning they are more alike than different, right for wrong reasons, but most importantly right on the prior 24 months.
You have to sell investors on the idea that this risk is worth it, an eminently feasible objective, but it's hard because institutional managers are run by committees, and committees are dominated by tradition that implicitly acknowledges benchmark risk. The funny thing is that benchmark risk not an explicit consideration, more an unstated one, but often the assumptions and protocols have the greatest effect.
What about portfolio with substantial bonds (like 50%)? Is the low volatility advantage partly negated by the higher correlation to bonds?
You can overcome that by choosing a bond ETF, for example, like TLT which has a volatility much closer to SP500. Correlation would be substantially negative still.
I really like your work about low volatility portfolios. I have studied these portfolios too, but instead of analyzing them from a financial perspective, I study their statistical properties. That is, I focus on the estimation risk associated with the covariance matrix.
Anyway, the performance that I found for these portfolios is very similar to that in your posts.
In this blog:
I recommend a weekly portfolio based on minimum-variance strategies, analyzing several performance measures (after transaction costs) through extensive back-tested results.
Hope you like it too.
Javier: interesting. What is the advantage of trading every month vs every 6 months? Do you find the reduction in volatility is significant with this higher turnover?
Good question. The reason for rebalancing every 4 weeks is that the realized return is better than rebalancing at other frequency, once transaction costs are discounted. The volatility is practically the same.
Post a Comment