Executive Summary
Estimating retirement expenses over the entire duration of a client's retirement years is a fundamental part of retirement planning. Yet there is surprisingly little agreement from planners about the spending behaviors of clients as they go through retirement.
Some suggest that retirement spending rises as clients age, due to the accumulating impact of health care expenses. Others suggest that retirement expenditures decrease, as clients reduce their spending in areas like travel and restaurants. Still others suggest that retirement spending is relatively level and simply keeps pace with inflation, as the increases in one category (e.g., health care) offset the decreases in others (e.g., travel and restaurants) - which, notably, is also the implicit assumption of steady inflation-adjusted spending that underlies the research regarding safe withdrawal rates and how much income is sustainable from a portfolio.
So which is it? A growing cadre of research suggests that in reality, client spending probably does decrease over time... with some notable exceptions. And if you don't use a proper assumption, you may force clients to save more than is needed, or retire later than is necessary!
The inspiration for today's blog post is a conversation I had with another planner after last week's blog post regarding whether and how retired clients experience ongoing annual inflation. In our discussion, the planner highlighted a recent situation with an elderly client couple who had not materially changed their spending withdrawals in nearly a decade, and insisted that they were not really being impacted by inflation. In delving deeper, though, it became clear that in reality, the issue was not whether the client couple was being affected by inflation, but whether the couple was altering the underlying composition of their spending in the first place as they aged. In other words, the couple was experiencing inflation, but increases in the cost of things the couple buys was being offset by the fact that the couple just wasn't buying as much stuff anymore!
In point of fact, this issue was covered several years ago in the June 2005 issue of Journal of Financial Planning, in an article by planner Ty Bernicke entitled "Reality Retirement Planning: A New Paradigm for an Old Science". In his article, Bernicke drew on data from the Consumer Expenditure Survey conducted by the Bureau of Labor Statistics to analyze retiree spending patterns. The data are striking, suggesting that those in the later half of retirement (age 75+ by the BLS data) spend an average of almost 30% less than early retirees (those aged 65-74). In fact, Bernicke shows that this gap has been remarkably persistent over time; the difference is almost the same (on a relative basis) whether you look at the data from 2004, or 1984. The BLS data - along with data from the Health and Retirement Study conducted by the National Institute on Aging that show a smaller but similar decline - have also been widely examined in analysis by the Bogleheads community.
Both the Consumer Expenditure Survey and the Health and Retirement Study support the premise that health care expenses rise with age; the results are evident in both studies. However, the clear trend of the data is that in the end, those increases are not enough to offset decreases in other categories, from food to entertainment to transportation costs (e.g., as the household not only drives less and spends less on gasoline, but transitions from having two cars down to only one). The limited impact of health care cost increases is likely due in large part to the fact that in the end, Medicare puts a significant constraint on how much health care expenses can rise; for instance, as I've written in the past on this blog, even a rather affluent retired couple with $150,000/year of income likely pays only about $10,000/year in health care expenses, and between Medicare parts A, B, and D, and a Medigap policy, there's very limited exposure to further costs from there. While $10,000/year isn't cheap, it's not catastrophic to "mass affluent" clients, and aside from health care inflation adjustments to the cost of the insurance, there's very limited risk of a precipitous increase in costs (outside of long-term care needs, which can also be insured separately).
In addition, I suspect the impact of changes to the composition of retirement spending - i.e., how much is spent across various categories - is also highly dependent upon overall income and wealth levels. And the reality is that the aforementioned surveys look at the average American, who in reality has significantly less affluence than the average financial planning client. For instance, the Health and Retirement Study shows that for married couples at the bottom of the income scale (5th to 15th percentiles), spending on health care alone is 16%, and total expenditures on housing, health care, food, and transportation comprises a whopping 81% of all household spending. On the other hand, for those at the upper end of the spending levels (the 85th to 95th percentiles), health care drops to 13%, and the total expenditures on housing, health care, food, and transportation together are only 55% of household spending. At the upper wealth levels, spending on entertainment, gifts, and other durable goods purchases rises dramatically to fill the void. Yet in turn, many of these highly discretionary expenditures like entertainment only fill the void in the early years, and are the most likely to decline later with age.
In fact, a study released several years ago by Sun Life entitled "The Expense Reality" - discussed on this blog back in 2008 - showed indeed that international travel spending declines for retirees in their 70s, domestic travel expenses decline for retirees in their 80s, and a wide array of other miscellaneous expenses, from new/second business start-up expenses, to hobby expenses, decline in the later years. On the other hand, the study also showed that certain categories have "unexpected" increases; for instance, expenditures on luxury items jumps for those in their 70s, and charitable giving rises for those in their 80s. Yet these are perhaps the ultimate in purely discretionary expenses... suggesting that the reality is not only that other spending categories decline in the later years, but that clients who can afford to sustain the spending actually increase consumption on luxury items and charitable giving to fill the spending void left by other category decreases!
So what does all this mean from the planning perspective? It suggests that as a baseline, we probably should project client spending to decrease by at least 10%, if not 20% to 30%, in the later years (e.g., age 75+, or especially age 80+), on an inflation-adjusted basis. This is especially true if the client has otherwise put reasonable insurance in place for health care and long-term care. The greater the affluence of the client, and the larger the percentage of discretionary spending relative to total spending, the greater the projected spending decline in later years. Arguably, this means that clients may need less to retire that we often suggest, and/or could retire earlier; from the safe withdrawal rate perspective, it implies the initial withdrawal rate could potentially be much higher, if later spending cuts are built in up front.
On the other hand, it's equally notable that this would just be the baseline; if the portfolio is actually performing well, and/or spending has declined earlier for some reason (for instance, a change in health by one member of the couple that resulting in a travel spending decrease earlier than anticipated), it is highly probable that affluent clients may and will choose to replace some discretionary spending decreases (e.g., travel and entertainment) with other voluntary discretionary spending increases (e.g., luxury items or gifts to charities or family members), as the Sun Life study shows. But these later outflows would truly be voluntary and discretionary; they would only occur if the wealth was already there, which is why they would not be part of the baseline scenario. In other words, the plan might assume spending will decline in the later years, with a contingency that clients will simply choose to spend more if they can in fact afford to do so when the time comes.
So what do you think? Do your clients spend more, or less, in their later retirement years? Do you see greater spending decreases by your clients with higher levels of discretionary spending? Have you seen affluent clients "replace" their spending in some categories (e.g., travel and entertainment) with outflows in other categories (e.g., luxury items and gifting)? Do you project a spending decrease for your clients in their later years?