Almost an entire year has passed since I began a three-part series of articles that were intended to demonstrate the impact of volatility on portfolio drawdown rates during retirement.
Part 1 (published on March 13, 2011) suggested that there was at least some evidence that the post-2009 PIMCO "New Normal" of muted returns and higher volatility might indeed be occurring (at least up to the point in time that Part 1 was written).
Part 2 (published on March 27, 2011) discussed a freely-available Excel spreadsheet model that could be used to estimate the impact of such "New Normal" conditions on the annual rate at which a retiree could withdraw money from his or her retirement portfolio.
Part 3's topic was to have been a discussion about the potential benefits of an options strategy known as "collaring" as a means of mitigating portfolio volatility.
A variant of this strategy was the subject of a backtest study conducted by Edward Szado and Thomas Schneeweis of the Isenberg School of Management at the University of Massachusetts.
Over a 138-month study period from March 1999, through September 2010, the authors found that a long protective collar strategy on the PowerShares QQQ exchange-traded fund (using six month put option purchases and consecutive one month covered call writes) earned far superior returns compared to a simple buy-and-hold strategy on the QQQ alone. Moreover, the collar strategy reduced portfolio risk (volatility) by over 60%.
Specifically, the collar strategy returned over 185% (9.6% annually), while the long-only QQQ position declined by 3% over the same period.
These results are illustrated in Figure 1 below (see www.OptionsEducation.org for the original Figure 1 and further description of the study):
Of particular interest to those readers who had obtained the Excel spreadsheet in Part 2 are the annualized return and standard deviation (volatility) values reported by the authors for the collar strategy (shown circled in red below):
Using the values above as inputs into the Milevsky spreadsheet described in Part 2, we obtain estimated spending rates that are nearly twice that of the best rates obtained using the assumptions in Part 2, as illustrated in the screenshot of the spreadsheet (below):
That is, if a collar strategy could achieve this type of return and standard deviation performance going forward throughout a retiree's entire lifespan, the impact on retirement lifestyle would be significant.
Or, stated differently, the effort invested in strategies (options-based or otherwise) that attempt to achieve lower portfolio volatility can be as effective, or even more effective, for a retiree than strategies that attempt to achieve higher return.
Now, this is all well and good, provided that the type of performance achieved in this back-tested study can be sustained going forward.
...Which leads me to the reason why Part 3 never quite got submitted to SA for publication.
It turns out that the University of Massachusetts passive collar strategy is rather similar to the approach that my firm uses in managing client portfolios.
Around the time I began writing this series of articles, however, I was beginning to notice that the collar-type approach might be underperforming relative to the long-only ETF strategy.
So as 2012 progressed, I began tracking the performance of the UMASS strategy on my own. Also, for purposes of continuity, I decided to start my tracking at the point that the UMASS study ended -- September 2010.
Figure 1 above illustrates the growth of $100 for each strategy over the period March 1999 through September 2010. To see what each strategy would have returned from September 2010 onwards, I reset both strategies to $100 on that date and continued forward using the same 2% OTM 6-month long put/2% OTM 1-month short call collar strategy described in the original study.
Figure 2 shows the results I obtained from September 2010 onwards:
Not only has the underlying ETF (QQQ) outperformed the collar strategy since September 2010, but the divergence between the long-only and collar strategies (particularly since the latter part of 2011) has increased as well.
These "paper trade" results for QQQ, in fact, were consistent with what I was beginning to observe in my own portfolio in early 2012.
Now, investors are generally not excited by a methodology that returns 0-3% over the same time period that the underlying returns in excess of 40%. Clearly, some research was needed to both a) explain the underperformance, and b) determine if some variant of the collar strategy might continue to be viable going forward.
In reviewing the UMASS collar methodology in conjunction with the historical options pricing data for the ETFs I hold, the conclusion I reached was that the underperformance was largely due to the effects of a persistent volatility skew that has gradually settled into the options market for ETFs over the last 18-24 months.
The skew is the result of OTM put options having significantly higher implied volatilities than OTM call options... More simply, this means that the cost of the "put side" of the collar is much more expensive than the cumulative revenue that can be generated by selling covered calls.
Understanding why such collar-type strategies are underperforming is one thing. Implementing appropriate investment process changes to correct for the situation is another matter entirely.
In future articles, I plan to address various strategies that I now employ to counter the effects of this volatility skew.
For now, the takeaway I would like to leave you with is this:
- There are, of course, no mechanical or prescriptive investment methodologies that produce superior results under all market conditions; as they say, "past performance is no guarantee of future results."
- That having been said, the thoughtful and selective use of option strategies has been shown to mitigate portfolio volatility to a significant degree under appropriate market conditions.
- Moreover, such volatility mitigation can be an effective means by which an investor can optimize his or her portfolio withdrawal rate during the distribution (retirement) phase.
By the way, if you are interested in the original study by Szado and Schneeweis, you can obtain a copy via this link.