The 30% Retirement Spending Drop Is Real. Almost No One Actually Feels It.

Somewhere inside J.P. Morgan's asset management division sits a dataset most people will never see: de-identified transaction histories for more than 4.7 million households that bank with Chase, tracked month by month from January 2017 through November 2025. The researchers who built "Retirement by the Numbers," published in December 2025, were not watching balances. They were watching a single line on a chart: the recurring payroll deposit, the one that shows up every two weeks for decades, then stops.

What happens on the other side of that line is the part the popular retirement narrative gets wrong. The story most people have absorbed, from advisor blog posts, from retirement calculators, from the "70% to 80% of your working income" rule of thumb, is that spending declines gently after you retire: a long, gradual glide path down, maybe 30% lower by your mid-eighties than it was the day you stopped working. That number is real. J.P. Morgan's own data confirms it. It is also, for almost everyone who lives through it, close to fiction.

The quick answer

Average household spending among retirees does fall by more than 30% between age 60 and 85, per J.P. Morgan Asset Management's analysis of Chase account data across 4.7 million households. But that decline is a population average, not an individual's experience. The same research found that 60% of new retirees see their own annual spending swing by more than 20% in either direction within just the first three years of retirement, and 54% still see swings that large well into their seventies. The smooth downward slope shown in most retirement-planning materials is built by averaging together millions of households whose real trajectories look nothing like a slope at all. What actually happens is a decades-long automatic savings system disappearing on one specific date, with no automatic system built to replace it, right as new and often unavoidable costs, a health insurance bridge to Medicare, home and car repairs, family emergencies, become more likely, not less.

The number everyone quotes

The 30%-over-25-years figure has become the industry's shorthand because it is genuinely well supported. J.P. Morgan's data puts average annual spending at $75,630 for households age 60 to 64, falling to $51,920 for households age 90 to 94, a decline of roughly 5% to 8% every five years. The U.S. Bureau of Labor Statistics' Consumer Expenditure Survey, a separate government dataset built from household diaries and interviews rather than bank transactions, lands in the same neighborhood: households age 55 to 64 spent $84,946 on average in 2024, compared with $61,432 for households 65 and older, a gap of nearly 28%. Two datasets, built by two institutions using two different methods, one following the same population of Chase customers over time and one comparing different households at a single point in time, arrive at a strikingly similar answer. As a description of the aggregate, the decline is not in dispute.

As a description of any single household, it is close to useless, because the average obscures more than it reveals. J.P. Morgan's volatility figure, 60% of new retirees seeing year-over-year swings above 20%, means the "gentle decline" almost nobody actually lives through is a statistical artifact of blending households spending far more than expected together with households spending far less, in the same year, and calling the midpoint normal.

The BLS data offers a second layer worth pulling apart, because it shows the aggregate decline is not evenly distributed across categories either. Healthcare spending runs the opposite direction from everything else: households age 55 to 64 spent $6,711 on healthcare in 2024, rising to $7,918 for households 75 and older. Total spending falls by nearly 30% across that same rough span, while the one category retirees have the least ability to defer or negotiate rises. That crossing line, total spending down, healthcare spending up, is most of the mechanical explanation for why any retirement-spending curve eventually bends back upward. It is arithmetic on a category that does not follow the rest of the household budget down.

How the smooth story got built

The gentle-decline narrative has a specific origin, worth naming because it explains why it spread so widely. In research published through Morningstar, David Blanchett analyzed household spending data from the RAND Health and Retirement Study, tracking real, inflation-adjusted expenditures for retired households from age 60 through 90. He found spending falling by roughly 1% a year early in retirement, accelerating to closer to 2% a year through the middle stretch, then moderating back toward 1% a year in later retirement as healthcare costs partially offset the decline in everything else. That shape, decline, decline, slight upturn, became known as the "retirement spending smile," and it is a genuinely careful piece of research. Blanchett's own analysis notes that the curve stays below the zero-growth line for nearly the entire span; the popularized version of the finding, an emphasis on the late-life upturn, overstates how much healthcare actually claws back.

The smile became the default assumption baked into retirement-planning software across the industry, because it gave planners something a single number never could: a shape. Financial advisors could tell a client not just how much to save, but how spending would probably feel, year by year. That is a useful thing to be able to say to a client. It is a much less useful thing to have quietly become the working assumption a whole generation was told to expect their own retirement to follow.

The mechanism: two systems disappear on the same day

The volatility data points to something the smile curve was never built to capture, because the smile curve is a population average and the mechanism that actually drives individual experience is structural, not statistical.

For the working decades before retirement, two systems run continuously and mostly automatically. The first is the savings system: payroll deduction, employer match, automatic contribution escalation, the catch-up provisions that kick in after 50. Vanguard's "How America Saves 2026" report, drawn from roughly five million participants on its recordkeeping platform, shows the combined employee-and-employer contribution rate rising steadily with age: 9.3% for workers under 25, climbing to 13.8% for workers 55 to 64, and peaking at 14.6% for workers 65 and older, the highest of any age band in the data. The savings machine is not winding down as retirement approaches. It is running at full output right up to the exit.

The second system is less obvious and rarely discussed as a system at all: the workday itself functions as a spending governor. Eight to ten hours a day, five days a week, a person cannot spend money, because they are at a desk, on a job site, in a meeting. That constraint disappears completely and immediately the day someone retires, at the exact moment the payroll deduction that had been quietly building a habit of not touching that money also disappears completely and immediately.

Nothing automatic replaces either system. There is no payroll-style mechanism that automatically moves money the other direction, out of savings and into spending, calibrated to a household's actual needs. The one federally mandated decumulation trigger, the required minimum distribution, does not begin until age 73 or 75. Between the date the accumulation machine shuts off and the date the government eventually forces a withdrawal schedule, often a decade or more, every dollar spent requires a manual decision that forty years of automatic payroll deduction never asked a person to make.

That gap in automation would predict exactly the pattern in the data: some households, freed from both the savings deduction and the workday's spending governor at once, spend more, filling newly unstructured time with newly unstructured money. Other households do the opposite, freezing, because the balance itself has become something other than a number to be spent. The Employee Benefit Research Institute's 2020 Spending in Retirement Survey, fielded among 2,000 retirees age 62 to 75, found that only 43% planned to spend down all or a significant portion of their assets, and among the stated reasons for not doing so, fear of running out of money ranked fourth, behind concern about unforeseen future costs, a sense that spending down assets simply was not necessary, and a desire to leave money to heirs. The resistance to drawing down savings is not, for most people, a risk calculation. It is closer to an identity attachment to the number itself, the balance functioning as a scoreboard that four decades of automatic saving trained people to want to see go up, never down.

A separate data point from the same J.P. Morgan research complicates that picture rather than resolving it. Among defined-contribution plan participants, 67% withdraw their entire account balance within three years of retiring. That is not, by itself, evidence of spending; a large share of those withdrawals are rollovers into an individual retirement account rather than cash spent down. What it shows is the same structural gap from a different angle: the moment someone leaves an employer plan, they leave whatever guardrails or default paths that plan came with, and land in an account with no automatic distribution logic of any kind until the required minimum distribution rule catches up with them, often a decade later.

What the evidence shows underneath the average

The Center for Retirement Research at Boston College, in an issue brief published January 6, 2026 by researchers Manita Rao and Anqi Chen, found that 83% of households experience some unexpected expense in a given year. When these costs hit, they average $3,300 for what the researchers call "rainy day" expenses like home and car repairs, $5,700 for family-related costs such as helping an adult child or covering a funeral, and $4,100 for healthcare, dental, and related costs, a combined average of roughly $7,100 when they occur, smoothing to about $6,000 a year across a full retirement, or roughly 10% of the typical retiree's annual income. Forty-two percent of retired households do not have enough liquid cash on hand to cover even one year of these costs, and 27% still could not cover a year of them even after liquidating their 401(k) and individual retirement account (IRA) balances entirely.

That exposure is not distributed evenly. The same Boston College research found that only 33% of Black and Hispanic households, and 50% of single female or widowed households, have adequate cash reserves for a year of shock expenses, well below the already-thin overall figure. Family emergencies and caregiving obligations do not fall randomly across the population; the households facing the highest odds of an unplanned cost are frequently the same households with the least cash cushion to absorb it, which turns a general planning problem into a specific and unevenly shared one.

A significant, specific piece of that exposure is a health insurance bridge that has no automatic solution at all. Medicare eligibility begins at 65; anyone retiring earlier faces a gap, as long as three years for someone retiring at 62, that benefits consultants across the industry, including Fidelity, Merrill, Nationwide, and Mercer Advisors, put at $700 to $1,500 or more a month for COBRA, the federal law that lets someone temporarily keep their former employer's health plan after leaving a job, usually for up to 18 months. For many people retiring in their early sixties, this single bridge is the largest and least predictable line item in an entire retirement budget, and it exists specifically because it falls in the gap between the automatic system that used to provide health coverage through an employer and the automatic system, Medicare, that will eventually provide it again.

A third pressure comes from a cost most people assume would fall along with everything else: utilities. An AARP Public Policy Institute poll of 1,009 adults age 50 and older, fielded in April 2025, found that 63% had seen their electric bill increase over the prior year, with 83% expressing concern about further increases. Retirees spend more time at home than they did while working, which raises usage at the same time rates are rising, a cost that moves in the opposite direction from the story everyone has been told to expect.

And the psychological posture that makes all of this harder to plan around is showing up in real time in the newest confidence data. The Employee Benefit Research Institute and Greenwald Research's 2026 Retirement Confidence Survey, fielded in January 2026 among 2,544 Americans including 1,045 retirees, found that two in five retirees say their overall spending in retirement has been higher than they expected. In the same survey, the share of retirees who say they have adequate emergency savings fell to fewer than seven in ten, down from nearly three in four the year before, and half of all retirees say rising housing costs are already hurting their ability to get by.

The honest complication

The framing above, a smooth industry story built on Blanchett's smile curve, undone by messier account-level reality, deserves the same scrutiny applied to the smile itself. A 2026 re-analysis by researcher Derek Tharp, circulated as a working paper and using the same underlying RAND Health and Retirement Study panel data Blanchett relied on, extended with newer statistical methods through 2021, found that the smile pattern shows up clearly when comparing different households at different ages in the same year, the standard cross-sectional method, but is not statistically detectable when the same households are tracked over time. That distinction matters. A cross-sectional comparison of, say, 70-year-old households against 80-year-old households in the same survey year can look exactly like a spending decline even if no individual household's spending actually declined that way, because the healthier and wealthier households are simply more likely to still be in the sample at 80, while lower-spending, lower-wealth households drop out through death or nonresponse. The smile, on this reading, may be measuring who survives to be surveyed, not what any actual household experiences.

That same caveat applies, in a smaller way, to J.P. Morgan's own headline number. The Chase panel runs from 2017 to 2025, eight years, which means the full 60-to-85 arc in the report is necessarily built by combining different households observed at different ages within that window, not by watching a single cohort age a full 25 years inside the data. The 30% decline is a real and carefully measured finding. It is also, like the smile before it, a population-level shape rather than a documented individual trajectory, which is exactly why the volatility number sitting next to it, that most people's actual year-to-year experience swings by 20% or more, deserves to be treated as the more load-bearing finding, not a footnote to the smoother one.

None of this means the decline curve is fabricated or the researchers behind it were careless. It means two respected, methodologically careful research efforts, twelve years apart, keep producing a smooth population-level shape that keeps failing to survive contact with within-household data, and an industry keeps adopting each new smooth shape as the planning default anyway, because a shape is easier to build software around than a distribution.

What the data actually recommends

None of this argues that the industry's savings advice was wrong. It argues that the industry's spending advice stopped one step too early, at the moment a household crosses into retirement with a fully automated accumulation system and nothing automated to take its place. The gap between those two systems, not a predictable decline curve and not a rise in healthcare costs alone, is what the volatility data is actually describing.

Retirees and the advisors working with them have a narrow set of tools that function as a manual substitute for the automatic system that disappeared, and each one works by rebuilding a piece of the structure payroll deduction used to provide for free. The first is a cash reserve sized explicitly against the Center for Retirement Research's roughly $6,000-a-year shock-expense estimate, not a generic emergency-fund rule built for a working household with a paycheck still arriving every two weeks; a reserve sized for a salaried worker's job-loss scenario answers a different question than the one a retiree actually faces. The second is a written, calendar-based withdrawal plan for the years before required minimum distributions begin at 73 or 75, reviewed on a fixed schedule the way a payroll system runs on a fixed schedule, rather than left to whatever the account balance happens to prompt in the moment. The third is a specific, budgeted dollar figure for the health insurance bridge to Medicare, priced against real COBRA quotes or marketplace plans for anyone retiring before 65, rather than folded into a vague "healthcare will cost more" assumption that never gets a number attached to it until the first bill arrives. Each of these turns an unstructured decade of manual decisions back into something closer to a system, which is precisely the thing retirement removes and never automatically rebuilds on its own.

---

This piece analyzes third-party research and government survey data cited above; it is educational and is not personalized financial advice.

Related reading: the retirement transition playbook for Medicare, Social Security, and required minimum distributions, how much to save for retirement, and run your full Money Map.

Quick answers

Does spending really go down after you retire? On average, yes, by more than 30% between age 60 and 85, according to J.P. Morgan Asset Management's analysis of 4.7 million households. But that average hides the individual reality: 60% of new retirees see their own spending swing by more than 20% year to year in either direction, and 2 in 5 retirees, per EBRI's 2026 Retirement Confidence Survey, say their actual spending has come in higher than they expected.

Why do many retirees avoid spending down their savings even when they can afford to? It is rarely pure fear of running out of money. EBRI's 2020 Spending in Retirement Survey found that concern about future unforeseen costs, a belief that spending down assets is unnecessary, and a desire to leave money to heirs all ranked ahead of fear of running out among retirees' stated reasons for not drawing down their accounts.

What expenses actually catch new retirees off guard? A health insurance bridge before Medicare eligibility at 65, commonly $700 to $1,500 or more a month via COBRA for someone retiring at 62, rising utility bills from spending more time at home, and general shock expenses like home repairs, family emergencies, and dental or medical costs, which the Center for Retirement Research found hit 83% of households in a given year, averaging around $6,000 annually.

How much cash should a retiree keep on hand for unplanned costs? Boston College's Center for Retirement Research puts typical annual shock-expense exposure at about $6,000, roughly 10% of a typical retiree's income, a more specific target than a generic emergency-fund rule built for a household still earning a paycheck.

When does the government force retirees to start spending down savings? Required minimum distributions from most retirement accounts begin at age 73 for people born between 1951 and 1959, and 75 for those born in 1960 or later, leaving a decade or more between typical retirement age and any automatic distribution requirement.