Wednesday, September 2, 2015

The Fascinating (To Me, at Least) History of Sustainable Withdrawal Rates

I have always found the history of “sustainable withdrawal rates” (SWR) to be quite interesting (a fact that secures my qualifications as a geek, as if there might otherwise have been some doubt), so I enjoyed retirement researcher, Wade Pfau’s recent post, Safe Withdrawal Rates for Retirement and the Trinity Study.

As Wade's post explains, William Bengen published a seminal paper on the topic in 1994 (download PDF). The Trinity study, subsequently published in 1998 (download PDF), took a slightly different direction with important repercussions. The latter shifted focus from “the highest withdrawal rate possible in the worst-case scenario from history” to the proportion of thirty-year periods in which a portfolio would have (past perfect tense) survived.

As he also notes, both studies were based on historical market returns available in the mid to late nineties and basing a retirement plan on the assumption that future portfolio returns will equal or exceed those of the past is faulty logic. Given current capital markets, they are unlikely to, according to Pfau, Michael Finke and David Blanchett in The 4% Rule is Not Safe in a Low-Yield World.

There is another interesting piece of history regarding safe withdrawal rates that Pfau doesn’t include in his post. In the September 1995 issue of Worth magazine, Fidelity Magellan's fund manager, Peter Lynch, published an article entitled “Fear of Crashing” in which he suggested that a retiree should be able to invest her savings 100% in growth funds and safely spend 7% of the portfolio annually (damn, some days I miss the ’90s).

I apologize for not being able to provide a link to the Worth article. Apparently, 1995 was before the time that the Internet knew everything. Prehistoric, in today's world.

Scott Burns, then a financial columnist for the Dallas Morning News, published a challenge the following month, October 1995, entitled “Dangerous Advice from Peter Lynch.” Burns provided data showing that a retiree implementing Lynch’s advice would quickly have gone broke in many historical time periods.

(Timing is everything. Lynch wrote his column a year following Bengen’s paper. Burns apparently had read Bengen’s work; Lynch apparently had not. )

Lynch not only withdrew the “Fear of Crashing” statement, he hired Burns as a contributing editor for Worth, yesteryear's equivalent of hiring the guy who hacked you as your new security consultant, I suppose.

The story became even more interesting when, in 2010, Burns and economist Laurence Kotlikoff published “Spend 'Til the End”, in which the authors refer to the “rules of thumb”, like 80% replacement rates and 4% safe withdrawal rates that Burns had written about for so long in his columns, were better referred to as “rules of dumb.” (I recommend this book.)

SWR studies provide important insight into the portfolio survival process, but as Pfau points out, translating the results directly into a retirement plan is a risky proposition. I have concerns beyond Wade’s research showing that historical data is a poor basis for future expectations.

As I have previously pointed out on this blog, the model is based on the absurd policy that the retiree will continue spending the same amount annually even when it is clear that he is headed for ruin. As Pfau has noted, constant-dollar spending is a research strategy, not a retirement planning strategy. If your retirement plan is to spend a constant-dollar amount annually, you need a new plan.

Nonetheless, SWR studies can be enlightening and I find their history far more interesting than that of most retirement research – admittedly a low hurdle.


  1. Great summary regarding SWR history Dirk. I hope you have in mind continuing where SWR transitioned from static (what you allude to as constant dollar strategies) to the history of dynamic approaches to retirement planning of which I see three basic kinds: 1) the more common dynamic approach using fixed periods (our early work in JFP as part of that); and 2) the approach that incorporates longevity statistics, or an actuarial approach, to determine the length of fixed periods (Ken Stein's work as an example); and 3) (my favorite - our later work) the approach that incorporates a dynamic, annually adjusting, rolling application of the actuarial tables. This last approach looks deeper into cash flow and probable portfolio values rather than simulation terminal value dispersions.

  2. Thanks for the walk through the history of SWR. Strangely, I recalled Peter Lynch's FEAR OF CRASHING article, but was completely unaware of the aftermath and reversal by Lynch. Interesting stuff.

    Pfau's work in this area is also interesting, including some of the other papers he has published showing that most of the portfolios with 100% survial rates will actually triple in value, leaving behind a substantial legacy.

    1. Paul, I'm not sure which Pfau studies you have seen or how you formed that understanding, but I can tell you that I have spoken with Wade directly about this issue in the past and what you suggest is completely inconsistent with his beliefs (and mine).

      Michael Kitces has argued that most portfolios in these studies end up near 100% of (not triple) their original value, but Wade responded that Kitces refers to inflated portfolio values. In real dollars, according to Wade, most portfolios end up with around half their initial value in these studies – again, not triple.

      If a retiree wants anything close to a 100% survival rate, he would need to invest conservatively and spend about 2% or less annually. I'm pretty sure Pfau would tell you that a lifetime 100% portfolio survival rate is generally an unrealistic goal. I believe that either of these researchers would tell you that tripling your initial portfolio balance, or leaving behind a substantial legacy (unless you started with one), is not very likely.

      Thanks for writing.

    2. Dear Dirk,

      I greatly enjoy your blog here in the United Kingdom.

      Paul D. Allen's observation about 'tripling in value' of portfolios rang a bell. I re-read Michael Kitces' blog post on a 'ratchehting safe withdrawal rate':

      In his blog Michael makes the following observation:

      'Of course, in some of these situations, final wealth is augmented by decades of compounding inflation. Nonetheless, even on a real (inflation-adjusted) basis, retirees finish with more than 100% of their inflation-adjusted principal 60% of the time, and double their real wealth almost 1/4th of the time, even after supporting a lifetime of inflation-adjusted spending at a 4% initial withdrawal rate!'

      Earlier in the same post Michael suggests ending portfolio values of upto '5X' starting retirement portfolio values based on inflated (i.e. nominal) values.

      Perhaps Paul Allen reads the same stuff as me. It was interesting to get both your views.

      Kind Regards.

    3. Thanks for writing, Chris!

      Michael and I actually don't disagree. All of those outcomes he describes are possible, they're just not all probable. Paul's comment was that "most of the portfolios with 100% survival rates will actually triple in value." Even if Michael is right about doubling wealth a quarter of the time, you won't triple it most of the time. (And, I strongly suggest that you ignore the inflated returns in that piece (or any) and concentrate on real dollars. No sense getting excited about dollars that won't buy anything.)

      Michael argues a 4% safe withdrawal rate, but Pfau, Blanchett and Finke recently found "a significant reduction in “safe” initial withdrawal rates, with a 4% initial real withdrawal rate having approximately a 50% probability of success over a 30-year period." Pfau thinks 2.8% is a better bet.

      Keep in mind, when you bet your future on the results, that these are incredibly over-simplified models we study that don't remotely mimic real life.

      I don't think anyone argues that SWR strategies can at least theoretically end up with large residual values, the studies have always shown that they can if you get 30 years of great market returns. The question is how much you're will to bet that they will.

      Thanks again for following my column, Paul!

  3. The most fascinating thing I have found about the history of Safe Withdrawal Rates is the apparent total lack of interest in the topic by the major financial firms that hold most of the retirement funds. The charge to understand them and develop better models appears to be coming form a handful of sole proprietor financial advisors (starting with Bengen) and university professors. The companies that have trillions of dollars of assets under management and huge research arms appear to spend no time thinking about this critical issue for their customers.

    1. Most of the major brokerages do refer to the studies and several implement tools on their websites, but I agree that they don't take a lead in the research. It is mostly done by academics. Then again, we shouldn't expect their research to be unbiased. I would be shocked if their results didn't favor buying stocks.

      Thanks for writing.

  4. This is a digression from your primary point, but you mention an 80% replacement rate. It that based on the gross or net income? Do you have any insights on how this number was derived? Taking this from the abstract to the concrete, my family lives on 50% of my income. It seems odd to expect my family to need an additional 30% in addition to our current demands on our lifestyle.

  5. OK, first of all, there are no digressions here at the Retirement Cafe. I am perfectly happy to chase off squirrel-like in any direction you suggest and, in fact, I prefer it that way.

    I do have some insight into the derivation of the 80% replacement rate and to answer your first question, it refers to 80% of your pre-retirement income. In your situation, as is often the case with these "rules of thumb", it makes no sense at all. It's a great number to use if you're sitting in a coffee shop musing about retirement, but it is not something upon which you should base a retirement plan.

    The general idea is that after retiring you will no longer need to save for retirement or pay FICA taxes. If you were spending all of the remainder or your income (you aren't) and you saved maybe 12% of it for retirement, then a 20% reduction might be in the ballpark. The problem is that not everyone saves that much, not everyone spends everything that's left (you don't) and expenses change constantly after your retire for the rest of your life. (I wrote about this not long ago.)

    Economist, Laurence Kotlikoff refers to these as "rules of dumb."

    The 80% rule is perpetuated by Aon Consulting Company and I think Christina Benz at Morningstar covers the situation pretty well here.

    Thanks for writing!