Friday, October 13, 2017

Why a Rational Retiree Might Keep Going Back to that ATM

In particular, the presumption that a client will adhere to a deterministic spending schedule, wake up one morning, go to an ATM, and discover that the “money process” has reached zero is silly and naive.” — Moshe Milevsky[1].

Milevsky refers to periodic, constant-dollar spending from a volatile portfolio, as simulated in sustainable withdrawal rates (SWR) research.

It is with no small measure of sadness that I find myself disagreeing, at least under certain conditions, with three of my favorite retirement researchers — Moshe Milevsky, Michael Kitces . . . and me.

For Michael’s perspective, I refer to a comment he made somewhere on the Internet some time ago referring to a paper entitled, “A 4% Rule — At What Price” by Jason Scott, John Watson and William Sharpe[2]. That research showed the high cost of mitigating longevity risk by over-saving. The large amount of “fettered” assets that must remain untouched to survive long periods of poor market returns make SWR strategies economically inefficient and expensive. Wade Pfau has referred to fixed spending from a volatile portfolio as the least efficient strategy.

I couldn’t locate that comment from Kitces so I will do my best to paraphrase from my (aging) memory. The problem with the Scott analysis, I recall Kitces suggesting, is that retirees don’t “do that.” They don’t just keep spending the same amount no matter the circumstances.

That was my intuition, as well. And in fairness, we were all three mostly right, though the devil is in the details.

It was hard to imagine that rational retirees would keep spending the same amount from a portfolio that appeared to be spiraling into ruin. (Cue Richard Thaler laughing aloud[4]). But, while Michael used that intuition to question the Scott research, I have always considered it an invalidation of the constant spending model as a planning tool. (As a research tool, it taught us about sequence risk.)

A predictive model says, “If you do this (follow the model’s policies in real life) then you can expect these results. If we don’t expect people to follow those policies, then we shouldn’t expect the outcomes that the model predicts. And, clearly, most of us don’t expect rational retirees to “do that.” So, why believe the results?

Why rational retirees might keep going back to that ATM.
[Tweet this]

I still have serious problems with the model, and not just this behavioral assumption, but it took a retiree going broke in his early 80’s to help me understand that there might well be rational reasons for a retiree to keep showing up at that ATM until it spits her card back at her.

(Please note this evidence that actual retirees go broke with SWR strategies and this isn't a hypothetical outcome that occurs only in Monte Carlo simulations. Considering elder bankruptcy rates, however, the evidence seems to show that the rate of ruin is an order of magnitude less than SWR models predict. On the other hand, I suspect the rate of reductions in standards-of-living is higher.)

A retiree contacted me after depleting his savings portfolio at age 82 with a failed SWR strategy. He was down to Social Security benefits and a little home equity and hoped I could show him a way out of selling his home and living off Social Security benefits.

Sadly, neither I nor several retirement planner friends could offer a suggestion beyond a home equity loan, which didn’t work out. It’s tough to rebound after you age out of the labor market.

After our discussions and a little thought, I realized that households with substantial savings relative to their spending might well see portfolio failure looming and reduce spending. If they do, they will likely see their portfolio recover, as I showed some time ago in a series of posts entitled, Clarifying Sequence of Returns Risk. You don't go broke with variable spending from a volatile portfolio. On the other hand, the spending is, well . . . variable.

Households with limited savings, on the other hand, may well find that their non-discretionary expenses gradually overwhelm their safe level of portfolio spending. Even though the portfolio would be doomed by continued spending, they will still need to pay for groceries and housing and may have little recourse other than to keep spending and pray for a tremendous bull market. They may find themselves in a “spending trap” in which they must sustain their level of spending simply to pay non-discretionary expenses with the knowledge that doing so will most likely soon bankrupt them.

If we consider that credit card debt is a major cause of elder bankruptcy, then we can expect retirees in such a spending trap to max out their credit cards to pay bills before depleting their credit along with their portfolio. (The 2017 Retirement Confidence Survey from EBRI notes that "18 percent of all workers describe their level of debt as a major problem and another 41 percent call it a minor problem.")

A credit card debt spending trap is similar. The household continues to spend from credit well past the point where they believe they can repay because it has no good alternative. We would not expect excessive credit card spending to be normal, rational behavior but if it is the best of a set of bad alternatives, we probably have to call it rational. Again, this occurs in households with little or no remaining savings.

I learned this lesson long ago when I was asked to help a lady who had run up $60,000 in credit card debt while her husband suffered a long bout of unemployment. "I didn't use the card at Nieman Marcus," she explained. "I bought groceries and clothes for my kids." Not irrational.

A similar conundrum applies to reverse mortgages. The sales pitches strongly suggest that a mortgagee can’t be forced from the home by foreclosure so long as the home remains their principal residence. In other words, just stay in the home and the reverse mortgage can’t be foreclosed. True, but notice how easily that suggestion rolls off the tongue.

A reverse mortgage borrower can lose her home through processes other than foreclosure. A retiree who chooses a reverse mortgage might find that despite her expectations when she retired, she no longer wishes to live in the home or can no longer afford its maintenance or the local cost of living. “Just” keeping the home as her principal residence might not be an attractive option.

I recently learned of a wealthy, retired corporate executive who lived in an expensive suburb of Houston. His wife developed dementia and her care bankrupted the household. Reverse mortgages were not a part of that story but it is easy to imagine that, had he borrowed one and spent most of the equity, he would choose to move to less expensive housing in a less expensive community even though that would trigger the mortgage’s repayment. While he could postpone repayment by just remaining in the home, that might not be his best option. Moving out, though it would trigger mortgage repayment, might be the rational choice.

So, returning to that “deterministic spending schedule,” I, too, prefer to believe that most people would note their deteriorating finances and reduce spending in time, but retirees with more limited resources might end up in a spending trap in which their portfolio’s death march is the best of a poor set of choices. They might also fall victim to the "boiling frog" scenario in which the deterioration is so gradual that it fails to set off trigger points in time (although the whole boiling-frog thing is fake news, according to The Atlantic.[3])

As a friend and Duke philosophy professor recently told me when I questioned people voting against their own interests, sometimes people are acting in their own interests but we just don't understand what those interests are.

This week that the Nobel Committee conferred its award to Richard Thaler seems appropriate to remind ourselves that our financial models are fairly irrelevant if we ignore the human behavior element or oversimplify it.

Sometimes a retiree may keep returning to that ATM for as long as possible because she has no better alternative. That's rational.


[1] Financial Analysts Journal : It’s Time to Retire Ruin (Probabilities), Moshe Milevsky.

[2] A 4% Rule - At What Price? by Jason S. Scott, William F. Sharpe, John G. Watson.

[3] The boiled-frog myth: stop the lying now! The Atlantic.

[4] Richard Thaler, A Giant In Economics, Awarded The Nobel Prize, Forbes.


I will participate on a panel of retirement advisers in a Twitter chat sponsored by Thomson Reuters and entitled "When Can You Retire?" on Wednesday, October 18 from 2 p.m. to 3 p.m. ET. Follow @Retirement_Cafe and @ReutersMoney to join us.

A fellow retirement planner was told at a seminar this week that "95% of retirees should get a reverse mortgage at the beginning of retirement to create income." Aside from writing a will, I can't think of any single thing that 95% of retirees should do. I am told that Mark Warshawsky estimated that 14% of 62+ households could potentially benefit. Sounds much more reasonable to me. Still others might benefit from a reverse mortgage to create an emergency fund.


  1. Hi Dirk,

    I’ve heard Meir Statman ( discuss this and Kitces has a great summary on his blog as well. There are actually many different categories of retirees.

    Understanding which one a person may be in goes a long way towards understanding their options given the cards they’ve been given to play in life.

    From Kitces blog: that discusses Meir’s paper

    “Segmenting Prospective Retirees
    The primary segments that Statman suggests in his article are the Wealthy, the Middle, and the Poor.

    The Wealthy are those who are able to earn ample income during their working years to save more than enough to fund a comfortable retirement; their retirement concerns (in Statman’s view) will revolve primarily around issues like estate taxes, and status competitions with their wealthy neighbors.

    The Poor, by contrast, are those who are unable to earn adequate income to save enough (or at all) to fund their retirement.

    In the case of the Middle, Statman breaks the category into several sub-groups. At the top are the “steady-middle” who earn adequate incomes to steadily save throughout their working years for their retirement – perhaps not as lavishly as the wealthy, but enough to achieve a comfortable retirement. At the lower end of the Middle are the “precarious middle” who might be close to funding their retirement but are struggling to save enough.

    In turn, the precarious middle is broken into two further sub-categories: the “low-earners” who simply struggle to save enough for retirement due to marginal earning power, and the “high-spenders” who might have enough income to theoretically be able to save but spend so much that they don’t have enough for retirement after all.”

    There are different needs and issues for each of the six (6) different segments. Statman’s behavioral insights go a long ways here understanding options a person may have.

  2. Hi. Just reread all the SOR risk posts that you linked. The next to last one is difficult to read, can you put the letters in a white font as in the other posts? A few questions: do you still recommend only up to 30% stock allocation in the 5-10 years before and after retirement? I thought you were OK with closer to a 50/50 or 40/60 portfolio. Also, with the bucket strategy where you have say about 3 years of expenses liquid and maybe 3-7 years in fixed income and then the rest in equities, wouldn't it make sense to replenish the 1st bucket from equities if the market is doing well, and leave the middle bucket alone? I think this may be called (or similar to) asset liability matching?) Thanks.

    1. I actually find that there is little impact on funding retirement with equity allocations from about 30% to 60%. (There is a larger impact on short-term volatility.) I like 40% to 60% equities better around retirement date.

      I had a 40% equity allocation when I retired in 2005. Two years later the market fell over 50% and I barely noticed it.

      Gordon Irlam found that the 95% confidence interval for the optimal asset allocation ranges from 10% to 80% equities, at least in one of his studies, so I'm not sure we should overthink asset allocation.

      Moshe Milevsky showed that you can't depend on a bucket strategy to bail you out of a bear market. (It can, but it won't always.) If the bear outlasts you, you can end up having to sell equities at the market bottom. So there's a danger of overthinking buckets, too.

      For the market, "doing well" is a relative thing. If it falls 40% in the next 18 months then it is indeed doing well. If it doubles in the next year and a half then you will regret selling.

      Market timing really never works over the long term. Better to base your decisions on the notion that you can't predict the stock market, IMHO.

      I'm not sure liability-matching is what you're referring to. That's the process of matching the duration of your investments to the duration of you expenses, which I think is a good idea.

      If liability-matching is what you're trying to achieve, then what you do with the middle bucket depends on your forecasted liabilities in those years, not on stock valuations.

      Thanks for writing!

  3. I have no doubt that retirees go broke and that doing so may be completely rational in some circumstances. The question is was their SWR ever going to be sustainable? If I retire with $200,000 in savings and I take $20,000 a year out I'm probably going to go broke no matter what happens. I might not live 20 or 30 more years but I do need to eat today so spending the money isn't irrational. Thinking that I can do that for 30 years is irrational though.

  4. There is probably a correlation between whatever behavioral biases result in insufficient retirement saving (or non-robust/marginal retirement saving) and, having saved insufficiently, over-spending what savings are available. It seems likely they are two sides of the same coin. Your 82 year old friend didn't just wake up on his 82nd birthday and only then realize he had run out of money! He was on a glide path to that outcome for his entire working and retired life. Perhaps had he consulted with yourself or another planner at say age 65 or even 72, he would have been able to achieve a more optimal outcome.

    1. Actually, he did consult with another planner, who recommended a spending rate that turned out to be too high.

      It's comforting to point to people with problems and explain that their failures are due to some deficiency in their character. That's how we explain to ourselves that it could never happen to us because we aren't "like that." Comforting, but self-deluding.

      David Blanchett has shown that there is indeed a correlation between under-savers and spending in retirement, but the correlation is that they tend to spend less over time, not more.