Why Most Retirees Underspend: The Research and What Helps

This article draws on research by Sudipto Banerjee at the Employee Benefit Research Institute, by David Blanchett and Michael Finke, by Michael Kitces at Kitces.com, and by Derek Tharp and Justin Fitzpatrick, whose risk-based guardrails research is published at Kitces.com, along with our own calculations on Robert Shiller’s public market data. We’d recommend reading the original work.

What the Research Finds

In a 2018 study, Sudipto Banerjee of the Employee Benefit Research Institute followed retirees in the Health and Retirement Study, the long-running University of Michigan survey of older Americans, and measured how much of their savings outside the home they spent over time. Depending on how much they had at retirement, the median retiree had spent down between 11.8% and 27.2% of those assets in roughly the first two decades. The households that retired with the most, $500,000 or more (in 2015 dollars), spent down the smallest share. About one in three retirees had more saved 18 years into retirement than when they started. The study counted savings outside the home, including IRAs but not balances still inside employer plans such as a 401(k), in 2015 dollars, for a generation with far more pensions than today’s retirees, so it describes past retirees more reliably than future ones.

The sharpest difference was between retirees with a pension and those without one. Over the first 18 years, the median savings of retirees with a pension fell by only 4%, against 34% for retirees without one, even though the pension group started retirement with more saved. A pension check arrives every month and gets spent. Savings, for many people, are something else: a reserve to protect.

David Blanchett and Michael Finke tested that pattern directly in a 2021 paper titled Guaranteed Income: A License to Spend. Working from survey data on retirees’ wealth and spending, they found that a dollar held in investments supported roughly half the additional spending of a dollar held as guaranteed lifetime income, such as Social Security, a pension, or an annuity. Their reading was that retirees hold back for two kinds of reasons at the same time: a rational caution about how long they will live, and a behavioral comfort with spending income rather than assets. Their 2025 follow-up found the same pattern from another angle: retirees spent about 80% of their lifetime income but only about half of their other savings and income sources.

The 4% rule arrives at the same place by design. As we explain in our article on spending guardrails that measure both risks, Michael Kitces found that a retiree following the rule usually finished with more than the starting balance. In our own calculation for this article, on Shiller’s data back to 1871 with a portfolio of 60% stocks and 40% bonds, the median 4% retirement ended 30 years later with about $1.54 million, after inflation, for every $1 million it started with, and 79 of the 123 start years ended with at least the starting amount after inflation.

Ten Years In: The Room Inside a 4% Plan

The 4% rule is a durable benchmark, and the reasons it usually leaves room are the same reasons it works. William Bengen built it in 1994 to survive the worst sequence in U.S. history, the retirees of the mid to late 1960s. On our data, a 4% withdrawal raised with inflation every year survived every 30-year start year from 1871 to 1993, with the worst start year, 1966, supporting 4.02% before fees. A retiree on the first day cannot know whether they are the 1966 retiree, so the rule asks everyone to spend as if they were. Our article on what the 4% rule actually tested covers its assumptions and what it leaves out. Ten years later, far more is known.

At Intentional Living FP, we asked a simple question of the same history. Take each 4% retirement from 1871 to 1993, stop at year 10, and run the rule’s own test again on what is left: the highest starting rate that lasted 20 years in every start year in the data, which on our numbers is just under 4.92%. If that rate applied to the year 10 balance is more than the retiree is already spending, the difference is a raise the same worst-case standard would have allowed. Michael Kitces made a related point in 2015 with what he called a ratcheting 4% rule, which raises spending once the portfolio has grown well past where it started. Our re-test is a cousin of that idea, built on our own data.

In 97 of the 123 start years, about four in five, the re-test at year 10 allowed a raise. Across all 123 start years, counting no raise as zero, the median raise was about 50%. A $1,000,000 retirement that spent $40,000 a year, adjusted for inflation, had, at the median, about $1,218,000 left at year 10 in the same inflation-adjusted dollars, and just under 4.92% of that is about $59,900 a year. In 35 start years, about 28%, the re-test would have more than doubled spending and giving. In the other 26, about 21%, it allowed no raise and the original plan simply continued, which is the rule doing its job: 1906, 1937, 1966, and 1973 are among them. In the tightest case, a 1956 retirement re-tested in 1966, the raised spending would have used up the portfolio almost exactly at year 30, as a 4% withdrawal did for the 1966 retiree. Of the 97 with room for a raise, 16 had room for up to 25%, 20 for 25% to 50%, and 26 for 50% to 100%.

The room grows with time. A re-test at year 5, using the worst 25-year rate of 4.35%, allowed a raise in 89 start years, with a median raise across all start years of about 21%. At year 15, using the worst 15-year rate of 6.05%, it allowed a raise in 102, with a median raise across all start years of about 97%. These figures are before taxes and fees, they apply what we know about the past rather than the future, and a real household would not wait ten years to look. What they show is the size of the cushion a fixed rule carries for most people, and how quickly it becomes visible. The re-test keeps the rule’s 30-year horizon; a household planning for a longer life would re-test at a lower rate, which shrinks the raise but usually does not erase it: at the median, re-testing for 25 more years instead of 20 still allowed a raise of about one-third.

None of this is an argument against the 4% rule, which answered the question it was asked. It is an argument for asking the question again as the years pass, and for deciding ahead of time what the answer will mean for spending and giving. The re-test is a way to measure that cushion, not the way our plans set raises, which we describe below.

Why Retirees Hold Back

Blanchett and Finke’s two kinds of reasons are a useful way to sort what we hear from retirees. Some reasons are rational responses to real uncertainty. Others are habits that served a family well while it was saving and keep running after the paychecks stop.

How long the money has to last. On the Social Security Administration’s 2023 period life table, used in the 2026 Trustees Report, a 65-year-old man has about a 24% chance of reaching 90 and a 65-year-old woman about a 35% chance. The chance that at least one of a 65-year-old couple reaches 90 is about even, at 51%, and the chance that one of them reaches 95 is about 21%. Period tables tend to understate how long healthier, higher-income households live. A plan has to pay for the long life, not the average one.

What care could cost. Here the median is modest and the tail is not. A companion EBRI study of out-of-pocket medical costs after age 70 found that the median for people who lived to 95 or beyond was about $27,000 in 2015 dollars, excluding insurance premiums, while 10% of that group spent about $172,000 or more and 5% more than $269,000. EBRI cautions that these figures are closer to a lower bound. (David Blanchett’s later analysis of the same survey, measured in 2022 dollars, put the median near $50,000 and the 95th percentile near $250,000 for retirees who died at 95.) One in three of the retirees surveyed had Medicaid coverage after 70, and nursing home costs drove much of the high end (without them, the 90th and 95th percentiles fall to nearly $96,000 and $154,000), so for a household that would not qualify for Medicaid, the tail is likely larger than these figures suggest. Economists Mariacristina De Nardi, Eric French, and John Bailey Jones found in 2010 that medical costs rising late in life go a long way toward explaining why older single retirees, the group they studied, keep saving. Planning for that tail is sensible, whether through insurance, a named reserve, or a mix, decided case by case. Holding the whole portfolio against it can be more than the risk requires.

Wanting to leave something. Many households want to help children or grandchildren, or support a church or a cause after they are gone, and that is a good reason to keep money. It works best as a stated goal with a number attached, so the rest of the plan knows what it is protecting.

A rule built for the worst case. A fixed withdrawal set to survive 1966 is, by design, lower than most other histories would have allowed. Its caution makes sense on day one, when no one knows which history is coming, and it turns into underspending later unless someone checks.

Then come the habits.

Decades of saving. People who accumulate enough to retire comfortably are usually disciplined savers, and the habit does not switch off at a retirement party. Spending from a portfolio can feel like undoing the work.

Principal and income. Richard Thaler’s research on mental accounting describes how people treat money differently depending on the label it carries. Many retirees will spend dividends and interest freely and treat any sale of shares as dipping into principal, even when the total return is the same. The label, rather than the math, often sets the spending level.

Watching a balance fall. A paycheck replaces itself; a portfolio balance does not. Daniel Kahneman and Amos Tversky’s work on loss aversion found that losses weigh more heavily than gains of the same size, and a shrinking balance registers as a loss even when it is the plan working as designed.

No signal that it is safe. Without a plan that says what can be spent, and what would change it, the cautious choice is always to spend less. No one sends a retiree a notice that the trip is affordable.

What Underspending Costs

The cost of underspending is quiet. Nothing breaks, and the money is still there at the end, which is exactly why the cost is easy to miss.

Time and health. Spending research, including David Blanchett’s work on the retirement spending smile, finds that real spending tends to drift down through the 70s and 80s as travel and activity slow; we cover that pattern in our article on how spending changes through the decades of retirement. The years with the most energy for the family trip, the long drive to see the grandchildren, or the season spent helping a child with a new baby come first. Money held back from those years cannot buy them later.

Gifts that arrive late. When one of a couple lives into their 90s, which happens about half the time for a 65-year-old couple, children are often in their late 50s or 60s when they inherit, after the years when help with a first home, a grandchild’s tuition, or a hard season would have meant the most. Some gifts carry more value while the giver can see them land.

Giving by default. Money that is never planned for still goes somewhere. Without forethought, more of it can go to taxes, probate costs, and family disputes than the household would have chosen, and a large pre-tax account left to children who are still working must generally be emptied by the end of the tenth year after the owner’s death, with yearly withdrawals along the way if the owner died on or after the required beginning date. Every dollar is taxed at the children’s rates, a timing we explain in our article on the 10-year rule for inherited IRAs.

Margin no one chose. Keeping more than the plan needs is a fine answer when it is a decision. It is a cost when it is a default that no one has examined.

When Holding Back Is the Right Answer

Not every retiree who spends less than they could is making a mistake. A household whose budget is mostly core spending, with little room to cut if markets turn, is right to start further from the edge. A household that plans to self-insure long-term care may hold a large reserve on purpose; we lean toward planning long-term care case by case, often as its own named and modeled reserve, as we describe in our article on planning for long-term care. A household with a clear legacy goal, a family business to pass on, or uncertain health may choose to keep more margin. And some people are simply content: their life costs what it costs, and more money would not change it.

In each case the question is the same: is holding back a decision with a purpose, or a habit with no owner? Declining a raise and keeping the margin is a fine answer when it is a decision rather than a default.

What Helps Retirees Spend and Give With Confidence

Research and experience point to a handful of things that make spending and giving feel safe enough to do.

A floor that feels like a paycheck. Because retirees spend guaranteed income more readily than savings, the timing of Social Security matters for behavior as well as for the math. For some disciplined savers, claiming earlier can be the practical signal that it is all right to spend and give, even when delaying wins on paper. For a married couple whose benefits are far apart, the higher earner’s check becomes the survivor’s, which generally argues for that spouse delaying if the portfolio can carry the wait, since that check lasts as long as either spouse lives; we walk through both sides in our article on deciding when to claim Social Security. A pension can play the same role, and so can an annuity, though our default is skeptical: we generally do not lean toward annuities but are always willing to evaluate one. When we do, we lean toward judging an annuity as insurance for a named risk, sized to the gap the rest of the plan leaves once Social Security is settled on actual numbers. For many households that leaves little for an annuity to do, and for some it leaves exactly the risk an annuity is built to cover; we explain how they work in our article on annuity types, costs, and taxes.

Rules decided in advance. A plan that names, before the fact, the portfolio values at which spending would rise or fall turns a vague worry into a known line. A retiree who knows what would trigger a cut can spend and give with more confidence while the plan is above it.

Naming the margin. A legacy goal with a number, a long-term care reserve set aside as its own line for a household that self-insures, and a cash reserve held inside the investment mix tell the household what is already protected, so the rest is free to spend and give.

Values first. Writing a Statement of Financial Purpose ties spending and giving to what the household says matters most. For households already financially independent, we write it a little differently, starting from “Our wealth exists so that...”, because the question is no longer how to get there but what the wealth is for. When a client has told us for years that they want to see Italy, or take the grandchildren somewhere every summer, the plan can point to that wish and to the dollars that pay for it.

Giving with intention. We lean toward encouraging giving while living, and we like to see it, without pushing it. The trip the household pays for and takes with children and grandchildren, college help that takes stress off the parents now rather than an uncertain inheritance later, and regular gifts to church and other causes as the household is able are gifts the giver gets to see. Other gifts belong in the estate plan. Each works best as a decision rather than a default, and it starts with reflection and conversation: the household’s own thinking and, for couples, a conversation with each other; then our help turning that thinking into a plan; and, when it fits, conversations with children or others they may give to. The questions are when, why, and to whom, both now and after they are gone.

A rhythm of review. A plan revisited about twice a year can see a surplus while there is still time to use it well.

How We Think About It

We lean toward treating underspending as a planning risk alongside overspending, not an afterthought. We lean toward risk-based guardrails, built on the probability-of-success research Derek Tharp and Justin Fitzpatrick published at Kitces.com, and toward measuring two risks in every plan, in the vocabulary Fitzpatrick has described: the risk that spending will need to come down, and the risk that the household is living well below what its resources allow. Our article on what a probability of success really means explains why a very high score can be a sign of room to spend and give.

In the settings we generally start from, adjusted plan by plan, spending starts at a level with about a 20% risk of overspending and an 80% risk of underspending, which keeps a cushion. When the plan’s risk of underspending reaches 100%, meaning no historical scenario would have run short at the current level, spending rises all the way back to the starting risk level, a raise large enough to plan around: a trip, a project, a gift. When the risk of overspending reaches 75%, spending comes down by a small step, typically after the plan has stayed past that line for two quarters (a waiting period Matthew Jarvis has described), and with warning beforehand. Underneath sits a hard floor at the household’s core spending, using the core and adaptive split Michael Kitces has described. As an illustration, a couple whose core spending is $80,000 a year and whose plan supports $110,000 can see which $30,000 is adaptive, what portfolio value would bring a raise, and what value would bring a cut, all in the plan document and in conversation. The legacy number and any long-term care reserve are named lines, so the couple can see what is already protected before deciding what to spend and give.

We lean this way because several lines of evidence point the same way. Retirees as a group spend less than they can, a fixed rule built for the worst case leaves most households with more than they needed, and on history the cushion is usually visible within five to ten years. A plan that checks about twice a year and acts on lines agreed in advance lets a household spend and give more in the years when it matters most, without taking on the risk that a loose sense of being probably fine would carry.

The trade-offs are real. Spending that rises can also fall, and a household that cannot tolerate any cut may prefer to start lower, or to decline raises and let good years build a larger reserve. Some households are right to hold more back on purpose, for long-term care, a legacy, or a business. And no plan removes the feeling that spending savings undoes a lifetime of work. What a plan can do is show, in dollars, what the household can afford and what it would take to change course. Our primary job is to help clients articulate what is important to them in life and use their money for it efficiently and effectively, which sometimes means encouraging the trip and the gift, and sometimes means honoring the choice to keep the margin.

Common Questions

Why do retirees spend less than they can afford?

Many retirees are afraid to spend their savings, and two kinds of reasons work at once. Some are rational: uncertainty about how long the money must last (about even odds that one of a 65-year-old couple reaches 90), the small chance of very large care costs, and the wish to leave something behind. Others are habits: decades of saving, treating the portfolio as principal not to be touched, and the discomfort of watching a balance fall. Research by David Blanchett and Michael Finke found that a dollar held in investments supports roughly half the additional spending of a dollar held as guaranteed lifetime income. We lean toward plans that name what is protected and set spending lines in advance, so a household can see what it can spend and give.

How much do retirees actually spend down their savings?

Less than most people expect. An Employee Benefit Research Institute study of the Health and Retirement Study found the median retiree had spent down only 11.8% to 27.2% of non-housing savings in roughly the first two decades, and about one in three had more saved 18 years in. The median savings of retirees with pensions fell far less over 18 years (4%) than those of retirees without them (34%). On our calculation with U.S. market data, ten years into a 4% retirement, a retiree could have raised spending and giving in about four in five historical start years, under the same worst-case standard the rule used at the start.

Is it bad to have money left over at the end of retirement?

Not when it is chosen. A named legacy goal, a long-term care reserve, or a deliberate cushion are good reasons to keep money. It becomes a cost when it is a default: trips not taken while health allowed, gifts that arrive after they would have helped most, and money lost to taxes, probate, or family disputes that a plan could have avoided. We lean toward making what is left a decision, with the legacy number and reserves named in the plan and the rest available to spend and give.

How can retirees feel more comfortable spending their savings?

What helps most is structure. A floor of income that feels like a paycheck, such as Social Security timed with spending behavior and, for couples, the survivor’s income in mind; a plan that sets, in advance, the portfolio values at which spending would rise or fall; a legacy number and, where the household self-insures, a long-term care reserve, each named as its own line; spending and giving tied to stated values; and a review about twice a year. We lean toward risk-based guardrails for this: they start with a cushion, raise spending fully when the plan clearly allows it, and cut in small steps with warning, so a household can spend and give with confidence while it is above the line.