Welcome to this week's video blog made in conjunction with The Wealth Channel at the American College. Today I provide a discussion of Jonathan Guyton's recent Journal of Financial Planning column, "Sequence-of-Return Risk: Gorilla or Boogeyman?"
For email readers, the videos never show up in the email, but you can see the video by clicking here.
What now follows is not an exact transcript. It is the written version of what I meant to say above, though as I was not reading from a script, I sometimes veered away from the plan when speaking:
Jonathan Guyton Tames a Gorilla
In the October 2013 Journal of Financial Planning, Jonathan Guyton asks the question: is sequence of returns risk in retirement an 800-pound gorilla posing a great threat to even the best-laid retirement plans, or is it only a boogeyman which can effectively be avoided. His conclusion is that sequence of returns risk may be more of a boogeyman.
To provide a little background, sequence of returns risk is experienced by those seeking to fund a constant expenditure goal from a volatile portfolio. When the portfolio drops, a greater percentage of assets must be spent to achieve the desired spending amount, permanently removing these assets from the portfolio and making it extra difficult for the portfolio to fully recovery. The specific returns experienced in the first years of retirement explain a disproportionate amount of the final retirement accounts.
However, as I outlined in a recent blog post about lifetime sequence of returns risk, two effective strategies for dealing with sequence of returns risk are either to reduce spending when the portfolio declines or to use a less volatile portfolio.
In his article, Jonathan Guyton focuses on both of these approaches in the first study I am aware of which combines a valuation-based asset allocation study with a dynamic withdrawal strategy. On the asset allocation side, he uses a globally-diversified portfolio with a “normal” allocation of 65% stocks.
When markets are sufficiently overvalued or undervalued with respect to the Shiller cyclically-adjusted price-earnings ratio, he will modify the stock allocation to 50% (when overvalued) or to 80% (when undervalued). Historically, large stock market drops tend to happen from points of overvaluation, and so this strategy aims to eliminate some of the downside, and therefore reduce the impact of sequence risk.
As for the dynamic spending strategy, Guyton applies the spending decision rules that he developed and tested with William Klinger in the Journal of Financial Planning. These rules stop an inflation-adjustment for spending after years when the market declined in value, and also sets thresholds to either increase or decrease spending by 10% when the withdrawal amount as a percent of remaining financial assets deviates outside of a 20% range from the initial withdrawal level. When it comes to dynamic spending strategies,
I have tended think in terms of David Blanchett’s mortality-updating constant probability of failure approach, but at the recent FPA Experience conference, well-respected and influential planner Dave Yeske chided me for not paying enough attention to the Guyton and Klinger decision rules. These rules may be more practical and easy to understand, and therefore will be more usable in the real world, even if they are not fully optimal at a theoretical level. Any approach which reduces spending after market drops does also help to alleviate sequence risk.
[Note: since first writing these paragraphs, I wrote a column about variable spending strategies for Advisor Perspectives and came away from that with much greater respect for the Guyton and Klinger decision rules.]
Nonetheless, we now have a framework for using both dynamic asset allocation and dynamic spending rules.
Jonathan Guyton then applies this to the case of a hypothetical 2000 retiree over the first 14 years of their retirement. He finds that with this dynamic approach, the 2000 retiree could have started with a 5% withdrawal rate with a spending freeze in 2001 and cuts in 2002 and 2008, and still see their portfolio balance to have grown by the start of 2013. Compared to a static strategy with a 4% initial withdrawal, the dynamic approach supported both higher total cash flows for spending as well as a larger remaining portfolio balance. These are very impressive results which suggest that sequence of returns risk is a beast which can potentially be tamed.
So, as long as future market conditions aren't worse than those for a 2000 retiree, and we can live with an 80% stock exposure at times (not me!), this should work fine, right?
ReplyDeleteThere seems to be a classic argument ongoing with both sides in "violent agreement". One side yells, "If the market performs significantly worse in the future than it has in the past, you will lose your standard of living." The other side replies, "Yes, but that's not likely to happen."
Strangely, each side seems to agree with the other's position.
Yes, after spending time recently mulling over safety-first type approaches, this post swings completely around to the opposite end of the spectrum. It is, indeed, a purely probability-based approach.
DeleteThat is a great summary of the matter though: Each side seems to agree with the other's position.
Yet, both sides come away with the opposite conclusions about what to do.
It so happens that I had this blog scheduled for the 7th Wade. http://blog.betterfinancialeducation.com/sustainable-retirement/the-apparent-paradox-of-retiring-twins-is-it-true/
ReplyDeleteIt is from the Probability-Based school of thought (which by the way, is how assets should be managed, I would argue, for those assets NOT used to support the Safety-First school of thought income liability matching approach for prudent income over and above that fixed income approach).
The approach in the blog post works for any asset allocation using either historical or expected returns data in the users’ Monte Carlo generator. It also uses actual distribution periods derived from period life tables. The approach is a bit more empirical, in my humble opinion, than Guyton rules (which are okay - but take longer to provide a signal the retiree really needs earlier).
Remember the days in medicine thought heart surgery wasn't possible because it was too complicated. The "Blanchett" method you mention in your post is more involved on the surface than Guyton rules. However, as demonstrated in the blog post link above, once a person understands the procedure and how to interpret the results, it becomes another important tool. The research technique was complicated to demonstrate results against baseline data (which is often not done in some published methods). However, the conclusions are not that complicated and can be used with most Monte Carlo software today. The advantage is that the retiree has, AHEAD OF TIME, pre-determined portfolio values they can use to buffer their emotions from daily, even hourly, market news cycles.
Either method probably won't come into play until the portfolio value changes larger than about one standard deviation of the returns characteristics of the underlying portfolio. Dirk's comments above are true if either side uses fear to support their position. Going to a fear extreme seems extreme to me ... What are rational decision points? What are the decisions to be made at those decision points? And make these decisions beforehand in a "cold state" of emotion so that fear during the "hot state" of emotion doesn't lead to an action that is actually more harmful.
I submit that the main idea is to have some rational method based on some research with which to make some sort of adjustment when pre-established decision points have been reached, when the markets misbehave as well as when they do well. An annual update during annual reviews is also an important part of building prudent expectations.
I wish he would back test it against another real-life decade: the 70's ...and toss in that nasty inflation rate on expenses for grins.
ReplyDelete(Just want to say I appreciate the changes made by the production team. There's been some real improvements since a couple of us commented back in October! Thanks!)
ReplyDelete