Hi Guys,
Many MFO members are justifiably interested in portfolio retirement drawdown rates that yield a high likelihood of portfolio survival. Ultimately, it’s a subject that gets everyone’s focused attention.
There are numerous ways to approach the problem, some more sophisticated than others, but all are somewhat fragile because of uncertain and unknowable future market returns. But these imperfect tools do generate respectable estimates of the odds, the probabilities of success if you are a glass half filled personality or a glass half empty if you are a pessimistic type.
As an aside, of course we all recognize that the glass is never just half filled or half empty. It is always completely filled with a mixture of liquid and gas.
I did my own portfolio survival estimates using Monte Carlo analytical techniques. Many such tools, which use statistical, random sampling techniques to choose annual returns, are freely accessible on the Internet these days.
I like them and recommend you give them a try. However, they do have shortcomings. They rely on statistical inputs like guesstimates of returns, the variability in those returns, and the correlation coefficients between the various asset holdings that are included in the portfolio. Most need additional inflation rate projections. Some do not permit withdrawal rate adjustments after the analysis is initiated. Almost all use Normal distribution assumptions. None handle Black Swan events very convincingly.
But that’s the state-of-the-art today. The final predictions seem plausible even given the list of shortcomings.
But some folks do not trust Monte Carlo-based methods and/or statistics. Some folks prefer real world data; they favor a historical analysis. These analysis were commonly completed before Monte Carlo tools were available.
Here is a Link to one such computation that was completed in late 2011:
http://www.marketwatch.com/story/how-to-invest-so-your-money-lasts-in-retirement-2011-11-01The analysis used real market returns in the sequence that they were recorded. It doesn’t get much more real than that. Portfolio drawdown rates of 4 %, 4.5 %, and 5 % were postulated.
Note the hole in the middle of the portfolio failure rate charts; that’s goodness; the most successful portfolios were constructed with a balanced mix of equities and bonds. Monte Carlo codes produce similar findings.
Larry Katz, the author of the piece, provides three conclusions: (1) Diversify among beneficial asset classes, (2) Take moderate distributions from your portfolio, and (3) Choose a reasonable stock/bond allocation.
The general finding from a drawdown perspective is that a 4 % withdrawal rate is relatively safe, whereas a 5 % annual withdrawal rate is far more risky.
That’s an excellent guideline if you are committed to an inflexible drawdown schedule. However, if you have the flexibility to adjust that rate after a bad portfolio year or two, you can bump that withdrawal rate plan upward by approximately 1 % per year without much compromising the survivability odds of the portfolio. Many independent portfolio studies have verified this more aggressive drawdown strategy.
I have done a host of studies using a number of Monte Carlo codes that reaffirm this result. Of course, you had better be prepared to execute the plan after a disappointing market return. Plans are simply plans, execution is the more demanding component.
Enjoy the referenced article.
Best Regards.
Comments
Larry Katz is the Director of Research at Merriman Inc., a Seattle Based Financial Advisor. These folks also ran FundAdvice.com which provided educational material although with the retirement of founder Paul Merriman last year, FundAdvice.com has lost some of its previous content that I used to refer to. (Update: The article you posted in available in perhaps a bit better format at here)
Two of my favorite articles are "Fine Tuning Your Asset Allocation" and "Ultimate Buy and Hold Strategy".
Fine Tuning Article contained a table that provided the performance, risk and max drawdown information for various equity and bond mixes. Armed with that information an investor could make a more intelligent choice what risk/return level he/she is comfortable with and the what sort of downside could be expected at the portfolio level. Here is the 2011 Update that I can find. They used to update this article with each new year data. If it is updated since 2011, it is not publicly exposed anymore.
For the other article they continue to provide year to year updates to numbers. There is now a 2013 Update to Ultimate Buy-and-Hold Strategy article. Direct link is here.
The article you have linked and these two provide a good set of educational material.
Has anyone seen this type of analysis approached from the other direction? Given a drawdown of a fixed percentage of the original portfolio, what kind of mix leads to most attractive distribution of possible returns? Intuitively, I am supposing that you still get the lopsided bathtub-shaped distribution of outcomes, but with more failures throughout as the drawdown rate gets higher. This sort of analysis would be of interest for scenarios in which there are non-optional drivers of the drawdown (e.g. unreimbursed costs of care).
The most sobering message I get from this material is that the high side of the bathtub (the higher failure numbers) are on the side of lowest equity allocations. Extreme "conservatism" is counterproductive, on average. Interesting message for a time when there are multiple reasons to be a bit thoughtful about high allocations to treasuries.
gfb
Thank you guys for taking time to reply to my post. Both your responses were thoughtful and thought provoking. They both added to the scope and utility of my original submittal.
Investor, the articles that you suggested greatly expand the usefulness of my posting by identifying two strategies that focus on the execution of retirement portfolio maintenance. Retirees and near-retirees will benefit from exposure to these two excellent summary articles.
Also, I was not aware that the author of the article I referenced was part of Paul Merriman’s FundAdvice staff. I did not recognize the potential conflict of interest. I like Merriman, but he has special incentives because of his financial relationships with DFA. There’s the possibilities of some biased opinions because of that relationship. Thanks for the heads-up. In this instance, I am not overly concerned of any major distortions or misrepresentations because I trust Merriman and I trust the DFA organization. Both do yeoman work.
Greg, your interpretations of the reported work, and your intuitions on possible outcomes for the scenarios that you postulated are spot-on-target.
Even today, for Monte Carlo codes that nominally adjust drawdown rates to reflect inflation rate changes, the Monte Carlo tools can be effectively used to examine the constant drawdown case you proposed. That’s simply done by setting inflation rate and it variability to zero in the inputs. The very first Monte Carlo code that I developed did not include an inflation adjustment. As you correctly perceived, the general contour, in terms of an optimum asset allocation distribution, are about the same both with and without inflation considerations. Portfolio failure rates are obviously impacted, but your excellent Bath-tube description as a function of equity percentages is still valid.
Indeed, being ultra-conservative by including too few equities in a portfolio is a risky proposition in a retirement portfolio, especially these days with extremely low fixed-income yields. Remember that fixed-income products also have a returns volatility aspect. That variability essentially means that any drawdown schedule from most fixed-income dominated portfolios must be lower than the expected overall average return from that portfolio because of that lowered, but still non-zero, volatility. In the end, that translates into either a very low permissible (and not acceptable) withdrawal rate or an unrealistically gigantic portfolio size.
Once again, thank you guys for your perceptive and useful inputs.
Best Wishes.