Thursday, October 11, 2012

What is the Safe Withdrawal Rate?

Dirk Cotton's guest post this week has triggered lots of interesting discussions about the 4% rule at a variety of message boards. I think I understand Dirk's message, or at least I know my interpretation about the meaning of the article. 

In my view, the point of the article, or my own view about the matter, is not, as one commenter wrote, this:

The alternative that I believe Wade advocates is to pay a HUGE additional cost, in terms of reduced effective withdrawals, to shore up the remaining 5% - be it through simply lowering SWR, using VERY costly annuities, or dooming us to work until we're 80. Regardless of however you go about it, to go from 95% to approaching 100% will require an exponential order cost, not a linear one. "True" safe withdrawal isn't worth the cost, IMO...

I don't think anyone is saying that you have to lower the withdrawal rate to some truly safe level (i.e. zero). 

So, what is the safe withdrawal rate?  Actually, I was asked this question in the "Retirement Roundtable" in the current October 2012 issue of the Journal of Financial Planning. I answered:

 I’m not really thinking about a safe withdrawal rate any more. It’s more like the lifetime satisfaction maximizing safe withdrawal rate, or something like that, that allows for the chance of wealth depletion or of potentially making cutbacks later on, like Jonathan was saying, with having the rules in place to have a dynamic withdrawal rate and to cut spending if we did have bad market returns….
I think we really have to consider now that the bond yields are so low, and that with bonds being a big component of someone’s retirement portfolio, I wouldn’t be that comfortable going much higher than 3.5 percent at the current time.
Let me explain this a bit more, because I wish I hadn't said the last paragraph. I felt pressured to give a number for the "safe withdrawal rate" even though I think the safe withdrawal rate is not particularly relevant to a well-designed and complete retirement income strategy.

For starters, I occasionally still see references to a prediction I made in an August 2011 JFP article saying that my best guess for a sustainable withdrawal rate for a 2010 retiree is 1.8%. I've since backed off from that prediction. The regression methodology I use in that article is hurt by the fact that we've never before experienced the market conditions of recent years, and that specifically creates problems for using the regression methodology. It specifically has trouble making these sorts of out of sample predictions. The main lesson from that article is that we cannot continue to treat 4% as safe in the market environment from recent years.

An alternative I've suggested to determine a safe withdrawal rate is to design assumptions reflecting current market conditions to feed into Monte Carlo simulations, as I explained in a January 2012 JFP article. I explain about this specific application for the article's methodology in a past blog post.

As well, and this gets to what I think is the main point of Dirk's guest post, one matter which has been clear to me is that the classic "constant inflation-adjusted withdrawals until wealth is depleted" strategy of the safe withdrawal rate research is the worst-possible strategy to use. Anything that makes adjustments for the actual market performance or revised remaining life expectancy will provide an improvement. These are the conditional probabilities of which Dirk wrote. The situation changes conditional on the experienced portfolio returns and conditional on having survived another year. This is a point emphasized both in my article here, and in the article I just recently reviewed from Morningstar

My most recent evolving views about safe withdrawal rates are highlighted in my most recent article, "An Efficient Frontier for Retirement Income."  What I argue there is to forget about success rates, failure rates, safe withdrawal rates, etc. What matters from the financial side of retirement is how to best meet two objectives: to meet as well as possible one's lifestyle (or at least one' minimum) spending needs, and to preserve a buffer of financial assets to manage risks regarding unexpected expenses. Then I look at how different strategies, which include the possibility of partial annuitization, perform with regard to meeting these two objectives. There is a lifestyle spending goal in that article, but there is no safe withdrawal rate.  And anyone who complains that annuities are too expensive must understand the current reality that everything is expensive in this low yield world. The 4% rule is not magically saved from failing in current circumstances just because it would have succeeded for retirees between 1926 and 1982. 

And that is how I view, or rather don't view, the question of the safe withdrawal rate.


  1. I think the problem is that someone like me needs to know a safe withdrawal rate. I have a certain amount saved for retirement (next year) and I need about 4% at a minimum. So if things have changed that 4% will not likely last for 30 years, I need to know that (with as much certainty as possible). If things really have changed and 4% is no longer considered "safe," I need some figure to know how much I can take out.

  2. Bill,

    The trouble in that it is hard to predict the future. In that Retirement Roundtable, I am coming in with a low-side estimate. Bill Bengen still suggests 4.5%, though that includes a healthy allocation to small-cap stocks. Jonathan Guyton suggests somewhere between 4.5% and 5.5%. I said 3.5%.

    Just stay flexible, proceed with caution, and congratulations on your upcoming retirement.

  3. Wade, I think you summarized my point perfectly.

    Another point I tried to make is that the SWR probabilities are based on a flawed model. A model that tries to predict a probability of ruin using a policy requiring the retiree to continue spending even when ruin is imminent would get a C- in a college course on statistics and operations research from a kind professor, a point I confirmed with the head of that department at our local university a few years back.

    We all know that in the real world retirees will decrease spending when their portfolio balance declines precipitously. They will experience a permanent decline in standard of living, but will avoid ruin. In reality, 5% won't experience ruin. Less than 0.5% a year will, but a large percentage (not predicted by the models) may suffer from inadequate income after a market decline.

    But your new blog makes an even better point. People look at SWR's as if they are a predictable 4.5% (or 3.5%, or 1.8%. . .) at any point in the future, regardless of the economic situation at that time. TIPs ladders and annuities present the same problem. We can't predict what interest rates will be when we retire several years in the future, so we cannot predict a safe withdrawal rate (I don't believe there is such a thing), or the payout for annuities, or for TIPs ladders. Therefore, we can't know how much money we will need to save for retirement.

    My hero, William J. Bernstein, narrowed down the correct amount as follows in The Investors Manifesto, "save as much as you can, start as early as you can, and do not ever stop."

  4. Wade -

    The thing that really irks me about this whole topic is that the 1.8% figure has lurked in the back crevices of my mind, causing 4:00am worry, ever since I heard it published by a guy with three letters next to his name….And now to see him just sort of casually say in comment #2, “The trouble in that it is hard to predict the future”….followed by what appears to be kind of an off-the-cuff “answer” when pressed of: Maybe 3.5%??....Argh!

    The other takeaway as I bounce around the internet on related topics is that there really does appear to be some basis for the argument that the SWR answer may legitimately change based on current valuations….Perhaps Bengen’s “lowest-common-denominator” approach accounts for this, but the guy retiring at the peak of the 2000 bubble admittedly has a different risk profile than the gal retiring in March 2009….And I guess this is ultimately why no one can ever definitively just give us an “answer”….


    1. Thank you. I'm not sure that the SWR rate changes based on current valuations, but the point I tried to argue is that the best guess we can make about what the sustainable withdrawal will end up being is related to current valuations, and the relationship was rather close in the past. However, in recent years I worry whether that close relationship can hold, since market valuations and now bond yields are out of the range of historical experience. That's what caused me to de-emphasize the specific predictions such as 1.8%

  5. Perhaps to put it simply: If you must have a "safe withdrawal rate" that does not change with market conditions, the only option is an inflation-adjusted annuity. Otherwise, you need to have a strategy that takes market performance into account. The bright side is that doing so may increase the withdrawals as well as decrease them.

    1. Yes, this is a good way to put it. Thanks.

  6. Annuities! Yes from an academic point of view it seems like a viable solution.
    But how much would you invest in the stock of one company, and how is putting a large amount with an insurance company any different? (Some states do have a back-up fund to cover insurance company failures but not to 100%.)
    I bought a SPIA a few years ago suggested by one of the largest mutual fund companies. I did my homework, it seemed fairly safe and inflation indexed too. From the largest insurance company in the world.... AIG! Where would I be now without the much critisized bail-out? Fred

    1. You would be fine without the bailout. AIG's annuity unit was well capitalized and the reserves belonged to the policyholders, not the company. To the best of my knowledge, only one U.S. annuity company has failed in the past 75 years; that was before the current regulatory scheme and the policyholders there (even those without state insurance) lost about one to two years income and then again received payments as promised.


    2. Right.

      Joe Tomlinson wrote an excellent column about the safety of annuity providers. I would suggest it for further background, including a discussion of AIG:

  7. SWR is like penis size/envy. Those at the extremes inherently know where they stand. Well endowed = low withdrawal worries. Short changed = high withdrawal rate...lots of anxiety. Those in the middle of the statistical distribution will always be looking for someone to tell them that they're doing fine. "It's OK fella, your withdrawal rate is LIKELY to be fine."


    1. Good simile. This is also related to a point that annuities may best serve those in the middle ground. With a low withdrawal rate, there's no need to worry.

  8. Annuities don't pay much right now, either. About 3.8% payout at age 65 with inflation protection. And the expected returns from a TIPs ladder (see Sharpe "4% Withdrawal Rate -- At What Cost"),which have historically had a real return of 2%, are now about zero.

    Wade's point is that when all of these expected returns and interest rates are low, as they are now, all future income streams are going to be expensive for TIPs, annuities AND SWR. SWR can't stay at 4.5% while the rest of the financial world's returns slip toward nothing.

    SWR is a crap-shoot. It's a market bet. If you're looking for a withdrawal rate to bank on, there isn't one, unless you know what the market is going to do over the next 30 years.

    TIPs ladders and annuities are far safer, but in today's economic situation, they just don't guarantee much return.

    Sometimes even the best choices aren't very good.

  9. So I'm confused. Any person off the street can throw up his hands and say the market is too hard to predict and therefore its a crapshoot, buy an annuity, etc, etc....

    But somehow when people who do this stuff for a living say the same thing, they are under the misperception that they are making some profound intellectual statement....Well guess what guys, if this is the best you got, its time to move on to a different profession because you are not helping.

  10. Jim Otar gives another view of Safe Withdrawal Rates in his book _Unveiling the Retirement Myth_ ( Simplifying quite a bit, you (1) figure out what you need to withdraw from your savings to cover expenses (be it 1%, 10%, or whatever), and then (2) discover whether you are lucky or not. If you are lucky, your withdrawal rate will be safe (and it doesn't much matter what rate you are doing); if you are unlucky, your savings will run dry (and again, it doesn't much matter what rate you are doing). Within about four years you'll have a good idea whether you are lucky or not.

    I want better, and accept that it is expensive.

  11. You guys have let fear overtake you.

    Fact: US Businesses have returned ~9.5% over the last century. (Including: Earnings growth, Dividends, Valuation).

    Bonds, although terrible at the moment, have consistently returned more than inflation over the last 100 years.

    Inflation has averaged about 3% over the last century in the US.

    Do the math for a "Balanced" portfolio of above ingredients and returns average 4 or 5%, above-inflation. This is just math, a fact.

    So while "sequence of returns" matter, etc, etc, its not that big of a leap to accept that if you're willing to dip into your principle from time-to-time, withdrawing up to 4% is an extremely safe bet if the definition of success is simply to not run out of money. (I think Bengen even "proved" this deterministically)

    So let's forget the nonsense of 1.8% WR's....(Unless of course, you're working as an annuity salesman).

    1. Try that same calculation with other countries' stock markets.

      There is no reason to expect the last century's results in the US to continue in the future, especially considering the vast changes in the political and economic landscape here.

  12. I agree that if Obama wins the current election, we will have a further setback. But he is only one man, his term will eventually end and the US will revert back to its mean (5% of world population generating 25% of worlds Revenue).

    But if you feel compelled to expand your investments beyond the US, I agree, you are playing by a different set of rules...I just don't know why one would choose to do so unless they have insights that lead them to....And in that case, as long as their insights are right, they should be ok...

  13. Kudos to Wade as he continues to present some of the best thinking in the retirement planning industry. I have come to appreciate those around us who offer ways to think differently about the issues surrounding retirement funding toward the ends of greater certainty and quality of life.

    This article is a significant advance in thinking about two questions he proposes to be solved simultaneously; (1) to meet as well as possible one’s lifestyle spending needs and (2) to preserve a buffer of financial assets to manage risks regarding unexpected expenses. He says “forget about success rates, failure rates, safe withdrawal rates, etc.,” and remain focused on these two questions. I hope he continues this line of research and writing as it seems to present a realistic alternative for retirees as they think about a sustainable retirement.

    However, Wade’s response to Bill Comb’s comment may be found less than informative and helpful, much like advice given to retirees for years: work longer, save more, or die early. This advice is often given when the negative impact of a SWR comes down on a retiree. The retiree questions their advisor, “What now?” Often this question comes at a time when alternatives may be severely constrained by low or negative market returns. The retiree alone will experience the full weight.

    Bill, I’d offer an approach which conforms to some of the research Wade has published. First, take a deep dive into your budget. Seek to understand in detail what it costs you to live, i.e. your basic living expenses, now, and as best as you can determine, in retirement. These are expenses that will not go away in retirement, like utilities, housing, clothing, food – and any other expenses you consider to be basic living necessities. Once these are categorized as base expenses, develop sources of net cash flow monthly that are simultaneously stable, secure and sustainable. These are not expenses that you can depend on the market and its statistical vagaries to fund for you.

    Next, seek to understand your unique unknowns. How can one understand the unknown? This is the second question that Wade seeks to address with a buffer of financial assets to manage risks regarding unexpected expenses and is essentially a question of risk mitigation – therefore a question of insurance. Not insurance products necessarily, but a question of insurance – a critical distinction. Wade also mentioned in his response to you that “it is hard” to predict the future”. I would respectfully press Wade on degree – it is not hard to predict, it is impossible to predict the future for the individual. But with this impossibility to predict comes choices that each individual must make to mitigate risk of the unexpected that could wreck an individual retirement plan.

    After decisions are put in place for basic expenses and contingencies, there are lifestyle questions. What would you like to do in retirement should you have the resources to do so? Define these and invest accordingly, with understanding that these expenses are discretionary and can be delayed or deferred if circumstances warrant. I think these discretionary expenses confirm to what Wade refers to when he advises you to stay flexible. Discretionary expenses are lifestyle necessities, but decisions can remain flexible since, by definition, they are not absolutely necessary to live.

    Finally, consider leaving a legacy if funding permits. If your resources permit this, leaving a legacy may be something to consider but not at the expense of basic living expenses, mitigating contingencies and lifestyle necessities.

    Bill, with Wade, I congratulate you on your forthcoming retirement. May you have health and happiness. I encourage you to stop looking for a single number – it just does not exist in the form you wish it to exist. Rather, keep doing your retirement homework by thinking in structured, systematic ways about what you need and develop your own sense of how to best move forward.

    1. Prof. Grubbs,

      Thank you for the very thoughtful response. You are indeed right. I was stuck in the "safe withdrawal rate" mindset when answering Bill's question.

      Prof. Grubbs along with financial planner Jason Branning created a framework called "Modern Retirement Theory," which I think makes a lot of sense. It is what he is describing here. Very simply, it is about prioritizing one's financial liabilities and then matching assets to those liabilities in a way keeps the risk levels appropriate.

      Prof. Grubbs, may I re-run your comment as a guest post on my blog?

      Best wishes, Wade