Jonathan Guyton, CFP®, is principal of Cornerstone Wealth Advisors Inc., a holistic financial planning and wealth management firm in Edina, Minnesota. He is a researcher, mentor, author, and frequent national speaker on retirement planning and asset distribution strategies.
You always knew it was true.
Observing your retired clients told you so. Anecdotal evidence did, too. And your common sense couldn’t imagine things any other way—people spend more early in their retirement years than they do later on.
In recent years, David Blanchett of Morningstar has unveiled research confirming what Michael Stein in his 1998 book (The Prosperous Retirement: Guide to the New Reality) and Ty Bernicke in his June 2005 Journal article ("Reality Retirement Planning: A New Paradigm for an Old Science") demonstrated years ago: namely that real (inflation-adjusted) retirement spending decreases over time until, for some, health care costs cause it to rise again late in life.
Blanchett’s work rises from “nice-to-know” to “big deal” status because most empirically based safe withdrawal strategies take a static approach to spending which—unintentionally, but all the same—assumes that retirees need to increase their spending by inflation in each and every year of retirement. In reality, they don’t. Instead, retiree spending is dynamic and, of course, unique to each client. Aside from the rare and tragic cases of spending addictions, I have yet to observe a retired client who had saved and prepared for retirement at least reasonably well for whom this was not true.
Retiree Spending Is Dynamic
The negative slope to these spending declines is striking. For starters, households needing $50,000 at age 65 decrease real spending by about 15 percent by age 80, and 20 percent by age 85. For those needing $100,000, the drop is 20 percent by age 80 and nearly 30 percent by 85.
To put this in context, if the $50,000 per year spenders were to receive half their retirement income from Social Security and the remainder from portfolio withdrawals, these distributions would decrease 40 percent in real terms after 20 years for the total to fall 20 percent. Even in nominal dollars, assuming 2 percent inflation, they decrease over these two decades.
For the higher-income $100,000 per year spenders who rely on portfolio withdrawals for a bigger portion of their retirement, these distributions would also decrease in nominal terms over these two decades, assuming Social Security benefits were $40,000 with 2 percent inflation.
So it’s only 75 to 80 percent of initial retirement spending that increases each and every year with inflation. We might call this “core” spending. The not-insignificant amounts above this core level in the first 10 to 20 years of retirement are “discretionary” and can be planned for intentionally, but differently. The all-or-nothing health-related expenses of the final decade or so also require their own unique approach.
That said, the implications of this evidence are significant, no matter how you slice them. Simply put, distribution strategies that presume constant real spending do not financially describe the retirements to which clients aspire. Although interesting (and mathematically correct), they can be square-pegs-for-round-holes prescriptions that, when improperly thought through by financial planners, have potential to ultimately produce significant life regrets.
Solving the Spending Question
Consider a planner advising the following: “From everything you’ve told us, Mr. and Mrs. Jones, and everything we know and have learned from our firm’s retired clients, there will be less and less that you will want to spend money on as you move through your retirement years, even though your spending may remain quite high until well into your 70s. So even if the dollar-cost of your lifestyle continues to rise, it won’t increase as much as inflation. Now keep in mind that your Social Security benefits, which will fund X percent of your desired spending, will keep pace with inflation. To fund the other (100 minus X) percent of your initial retirement spending, you will need a nest egg of $Y based on the assumption that this income also needs to keep pace with inflation even though you won’t need anywhere near that much over time.”
That would be crazy. However, knowingly or not, this is exactly what happens when retirement plans are designed via many well-worn “safe” withdrawal strategies and/or the spending inputs we put into software programs.
There’s an important caveat here: researchers who have truly added to our profession’s robust safe-spending body of knowledge over the years have made invaluable contributions. Their work examined and solved for the constant real spending question. It is always the practitioner’s responsibility to apply research appropriately and with the utmost due care to the client situation at hand. In other words, do not apply the right answer to the wrong question.
Blanchett’s research into actual retiree spending patterns, coupled with the research around dynamic retirement distribution strategies and spending policies, means that retirees can enjoy sustainable retirement lifestyles with higher incomes and/or lower asset totals and/or earlier retirement dates than the plethora of static-approach withdrawal research would otherwise have them believe.
But what about the possible increased health-related spending we must also plan for later in retirees’ lives? Obviously, there’s long-term care insurance. There is also the possibility of earmarking a pool of assets (including home equity) as a just-in-case self-insuring mechanism. And because virtually all “safe” withdrawal strategies end up with a significant terminal value, retirees could spend down their nest egg in the final decade or so of life. These approaches are not new.
So how might we connect this empirical spending research with the evidence we have on sustainable withdrawal strategies? Although the purpose here is not to attempt an exhaustive answer to this question, there are at least several possible frameworks.
4 Ways to Connect Research with Evidence
Approach 1. If you use retirement software, model retiree spending so that it reflects the evidence and patterns around actual retiree spending, rather than a simplistic constant real-spending assumption. The results will be very different.
Approach 2. Go ahead and use a constant real-spending-based strategy. However, calibrate it to fund (along with Social Security and/or pension benefits) a lesser amount such as 75 to 85 percent of intended initial retirement spending. Of course, those assets alone won’t be sufficient. Your clients will need additional assets from which to draw during their first 20 years or so of retirement to fund items beyond “core” lifetime expenses. Remember that it takes less capital to fund an inflation-adjusted dollar for 15 to 20 years than over the entirety of retirement. And any Social Security claiming strategies where full benefits do not commence until post-retirement require their own distinct asset pool to provide an income “bridge” until benefits are claimed (see my June 2015 Journal column "Bridges to Social Security" for more on that).
Approach 3. Use one of the empirically based, dynamic withdrawal strategies that more closely mirrors actual retiree spending. These dynamic strategies/spending policies allow higher initial withdrawal rates (and therefore require less capital to fund a given initial spending level) because their adjustment triggers prescribe small withdrawal “cuts” in the face of unfavorable returns. If reductions are triggered at all, they produce a distribution pattern that, in conjunction with Social Security benefits, lines up well with actual retiree spending patterns.
In fact, it would take an unusually poor investment climate (worse than that encountered by someone retiring in 2000 or 2008) to generate cash flows that fall below the patterns revealed in Blanchett’s work. For example, a retiree household using such a dynamic distribution strategy and a 5 percent initial withdrawal rate would require 20 percent less capital at retirement than one employing a static strategy with a 4 percent initial withdrawal rate that sought to fund the same distribution amount.
Approach 4. Use a combination of Approaches 2 and 3. Because we’re neither wise enough nor foolish enough to know in advance how much of our clients’ initial spending is “core” (increasing with inflation) and how much is “discretionary” (declining over time), this is what we do at Cornerstone (see my February 2013 Journal column, "When an Ounce of Discretion(ary) is Worth a Pound of Core").
Indeed, many clients have distinct core income portfolios that follow empirically based dynamic distribution policies, as well as distinct discretionary portfolios where they make all the decision rules. This also leads to distinctly invested portfolios; because a portfolio designed for sustainable lifetime income will have a different distribution pattern than one that is used more aggressively in retirement’s early years.
We regularly converse with clients about this core-discretionary balance. These are rich conversations and an empowering implementation strategy for clients that encourage thoughtful ongoing prioritization and make clear the safe/unsafe spending lines—“nudges” and “framing.” Indeed, it was these kinds of conversations that years ago persuaded us that such an approach may be well-received. To say that’s been the case would be an understatement.
For example, a household retired with an initial pre-tax annual spending goal of $120,000 and immediately began receiving $40,000 in Social Security benefits. Through conversations, we decided that $20,000 or more of this would go for discretionary items; an amount they realized would decline over time and did not reasonably believe they needed to fund beyond 12 to 15 years. Therefore, $60,000 of their core spending needs would be funded by portfolio distributions.
Using the 5 percent initial withdrawal rate supported by most dynamic (though, not static) strategies, this would require $1.2 million. Another $250,000 to $300,000 would serve as the discretionary fund for the $20,000 beyond their yearly core amount. In total, we would want to see these clients have about $1.5 million in portfolio assets. Compare this to the $1.6 million a planner would prescribe in assuming the entire $80,000 income needed to be funded with a dynamic distribution policy employing the same 5 percent withdrawal rate. Or the $2 million they’d be told under a static, constant, real-income approach with a 4 percent withdrawal rate. That difference, possibly the difference between “you can” and “you can’t” retire, can’t be measured in dollars.
It’s quite a difference that Blanchett and the retirement spending researchers who came before him have made. Isn’t that always the case when planners truly understand their clients’ goals and can also integrate them into the advice they give? That’s planning done well.