Inflation in Retirement: The 30-Year Math and How to Plan for It

The idea of sequence-of-inflation risk is laid out by Justin Fitzpatrick in a 2023 guest article at Kitces.com; we’d recommend reading it in full. Our charts and examples are built from our own calculations on public data.

What 30 Years of Inflation Does to a Dollar

A retirement is long enough for inflation to do its slow work in full. To see how much, we took Robert Shiller’s monthly price data (the consumer price index from 1913 on, which matches the government’s published series) and asked a simple question for every January from 1913 through 1993: what would one dollar of fixed income buy 30 years later?

Every window lost. All 81 thirty-year retirements ended with a dollar that bought less than it did at the start. The best start was 1920, near a postwar peak in prices that were about to fall sharply and would fall again in the Depression, and even that retiree’s dollar ended at about 82 cents. The median ended at about 39 cents. In 61 of the 81 windows, a fixed dollar lost more than half its buying power.

The worst start was 1966. A household that retired in January 1966 saw prices rise 385% by January 1996, a compound average of about 5.41% a year, which left a fixed dollar worth about 21 cents. Every start from 1964 through 1968 ended in nearly the same place, because each of those retirements ran straight through the 1970s.

A steady rate gives the same lesson more gently. At a steady 3% a year, a fixed dollar is worth about 41 cents after 30 years; at 3.6%, the average of the last 50 years, about 35 cents. That is the arithmetic behind a line we have used before, that a pension with no cost-of-living adjustment loses most of its buying power over a long retirement, and it is why inflation does so much hidden work in every retirement projection.

The chart begins in 1913 because that is when the consumer price index begins. Shiller’s data reach back to 1871 by splicing in an older index of wholesale prices, and on that measure prices fell for long stretches of the late 1800s: twelve of the 30-year windows starting between 1871 and 1883 ended with prices lower than they began, and the 1871 retiree’s dollar ended worth about $1.62. We come back to that era in the debate over where inflation goes next, because it is real evidence that sustained falling prices are possible. It is also a different measure in a different monetary world, so we keep it off the chart.

A fixed dollar is the right thing to measure because so much of a retirement is fixed in dollars: a pension without an adjustment, a fixed annuity payment, the principal of a bond or a bank balance. The spending those dollars pay for is not fixed. It rises with prices, and the gap between the two is what a plan has to fill.

Inflation Has a Sequence, Just Like Returns

Most retirement projections stopped treating investment returns as a straight line long ago. Few planners would project a retirement on a steady 7% a year, because the order of returns matters: losses in the early years, while withdrawals are being taken, do lasting damage that the same losses late in retirement do not. That is sequence-of-returns risk. Yet many projections that vary returns from year to year still run inflation as one smooth line, the same 3% this year, next year, and in year 29, in every scenario. Fitzpatrick’s point is that inflation has a sequence too, and it works the same way.

Why the order matters. Each year’s inflation raise becomes the base for every year after it. A burst early in retirement lifts the dollars withdrawn in every remaining year, and those extra dollars leave the portfolio while it still has decades to fund. The same burst arriving late lifts only the last few years. Two retirements with the same 30-year average inflation can therefore draw very different amounts from their portfolios, at very different times. Real inflation arrives in bursts, so when a burst lands matters as much as how big it is. The extra withdrawals cost real buying power only when the portfolio fails to keep pace with prices, which happens when bonds pay fixed dollars that the surprise erodes, when cash yields trail rising prices, or when stocks fall in real terms at the same time; in the 1970s, all three happened at once.

The 1966 retiree shows why. The same household that ended with 21 cents on the dollar did not lose that buying power evenly. For the first eight years, prices actually ran below an even path at the same 30-year average. Then the 1970s arrived: from January 1973 to January 1974 prices rose 9.4%, and the next year 11.8%. For the rest of the decade the yearly rise stayed above 5%, reaching 9.3% from January 1978 to January 1979, then 13.9% from January 1979 to January 1980 and another 11.8% the year after.

By 1981, prices had reached 2.74 times their 1966 level. An even path at the same 30-year average of 5.41% a year would have put them at about 2.20 times, and a steady 3% path, a common planning assumption, at 1.56 times. The actual path stayed above the even one from 1975 through 1995; twenty years in, in 1986, prices stood at about 3.45 times their start against 2.87 on the even path. Both paths end in the same place in 1996, at about 4.86 times, while a steady 3% path would have reached only about 2.43 times. What differs is when the household had to pay for it.

Same average, more dollars. Add up what the 1966 retiree had to spend each year, in dollars, to keep buying the same things, and the timing shows. Over the first ten years, the actual path needed about 3% fewer dollars than the even path at the same average. From 1976 through 1990, it needed about 16% more. Over all 30 years, it needed about 10% more, with an identical 30-year average. A projection that knows the average perfectly still misjudges both how much a retirement draws and when.

The burst landed while the portfolio was struggling too. From January 1966 to January 1981, the S&P composite with dividends earned about 6.4% a year in dollars while prices rose about 6.9% a year, a loss of about 0.5% a year after inflation (Shiller’s data, before taxes and fees). From January 1973 to January 1982, stocks lost about 4.4% a year after inflation. A household drawing from its savings in those years was withdrawing more dollars to keep up with prices from a portfolio that was not keeping up with them. That combination, rising withdrawals out of falling real values, is what makes the 1966 start the year that sets the floor in much of the research behind the 4% rule and safe withdrawal rates.

No single steady rate captures that. A higher steady rate can make a projection more conservative, but it still gives one answer, smoothly spread, and it cannot show what happens when a burst lands early or lands alongside a bad market. Simulations that vary inflation help, but inflation also needs to arrive alongside the markets it actually came with: in the 1970s, high inflation and weak real stock returns came together, a pairing that a model drawing them separately can understate. That is why we lean toward a risk analysis that replays actual history month by month, inflation and returns together, rather than leaning on any one rate to stand in for that risk. A steady rate still has a job: it is the number a plan uses for its year-by-year cash flow view. It is just not the risk test.

Your Inflation Is Not the Headline Number

The consumer price index is an average across a typical urban household’s basket. A retiree’s basket is not typical, and the parts of it move at very different speeds.

Goods and services have gone separate ways. From September 2000 to September 2026, the government’s index for durable goods (cars, appliances, furniture, electronics) went essentially nowhere: it ended at about 100 on a 2000 base of 100, after falling to a low of about 83 in 2018 and jumping during the pandemic supply shock. Over the same years, services other than energy services rose to about 220, about 3.09% a year, while the all-items index rose to about 194, about 2.58% a year.

Retirement budgets lean toward services. As households age, more of the budget tends to go to things people do for them rather than things they buy once: healthcare, insurance, home maintenance, help around the house, travel. Those are the categories that have been rising faster. The cheaper television does not offset the more expensive plumber, and a retiree who has already bought the car and the furniture gets little benefit from goods getting cheaper.

Healthcare is three different questions. The first is prices. The medical care index rose about 3.18% a year from September 2000 to September 2026, faster than the 2.58% of all items, though over the last ten years it rose more slowly, about 2.4% a year against about 3.4%. The second is premiums, which can outrun prices by a wide margin: Medicare’s Part B premium has risen about 5.9% a year since 2000, as the Social Security section below shows. The third is use. Older households simply use more care, so even flat prices would mean a rising health budget. That combination, not the medical price index alone, is why a retiree’s healthcare budget tends to rise faster than prices in general. (Our article on the retirement spending smile covers the experimental price index for older Americans and what it shows about medical weight in retiree budgets.)

Some costs stop growing. A paid-off home removes the largest cost that would otherwise rise with prices for a renter, though property taxes, insurance, and maintenance still climb. A fixed-rate mortgage payment never rises, though, as Michael Kitces has pointed out at Kitces.com, the mortgage itself is not the inflation hedge; the hedge is the house it bought. In Texas, a homeowner 65 or older gets a ceiling on the school district portion of the property tax, which freezes that part of the bill in dollars, a benefit that grows with every year of inflation; our article on Texas property tax at 65 shows how much that ceiling is worth over a decade.

Real spending tends to drift down. David Blanchett’s research on actual retiree spending finds that, on average, real spending falls through retirement by about 1% a year, with spending on transportation, clothing, and household furnishings falling steadily with age while healthcare rises. When a household’s budget has a discretionary layer, our baseline lets total spending drift down about 1% a year in real terms, close to the average the research finds: the essentials hold their buying power, the decline comes mostly out of the flexible spending on top, and a small uptick in the last few years mostly stands in for long-term care costs for households without coverage. In dollars the household still spends more each year; it simply spends less in purchasing power. That drift works as a partial offset to inflation itself, since it means total spending in dollars rises more slowly than prices, and it is one reason we lean toward budgeting by category, projecting each part of the budget on its own path, an approach Kitces has written about, rather than inflating one total. The spending a household wants to keep no matter what, including its regular giving, rises with prices in full; the spending planned year by year is where the research says real spending tends to ease.

Social Security’s COLA: What It Protects and What It Misses

For most retirees, Social Security is the only lifetime income that is guaranteed to rise with prices. How it rises, and what it does not cover, is worth understanding in detail.

The mechanics. Each year’s cost-of-living adjustment compares the average of the consumer price index for urban wage earners and clerical workers (CPI-W) over July, August, and September with the same three months a year earlier. The increase takes effect with December’s benefit, which is paid in January. The 2027 adjustment is 3.6%: the third-quarter average rose from 317.265 in 2025 to 328.746 in 2026, so the raise applies to December 2026 benefits and first shows up in January 2027 checks.

The COLA lags. Because it measures last summer’s prices, the adjustment always catches up after the fact. When inflation accelerated in 2021 and 2022, retirees spent most of a year paying higher prices on benefits that had not yet been raised; consumer prices peaked at 9.1% for the 12 months ended June 2022, and the 8.7% adjustment that answered it arrived in January 2023. When inflation falls, the lag runs the other way and a retiree briefly comes out ahead.

Some years bring nothing. There was no adjustment for 2010, 2011, or 2016, because prices in the measuring quarter had not risen above the last level that produced a raise. A benefit cannot fall because of the formula, but it can stand still while other costs rise.

The index is a working household’s basket. The CPI-W tracks spending by working households, not retirees. It tends to run close to the overall index, but it gives less weight than an index of older households’ spending would to the healthcare and housing costs that loom larger later in life.

Adjustments start counting at 62, whether or not you claim. Social Security applies each year’s adjustment to a worker’s benefit base beginning in the year the worker turns 62, even if benefits have not started. Someone who waits to claim does not miss the raises along the way. What waiting adds is a larger benefit, through smaller early-claiming reductions and, after full retirement age, delayed retirement credits, that then rises with every future adjustment for life, which makes a delayed benefit one of the strongest inflation defenses a household can own. That is one input in the claiming decision, not the whole of it; we model each household’s claiming choice on its own numbers, health, and goals, as we explain in our article on when to claim Social Security.

Medicare’s premium has outrun the COLA. The Part B standard premium, deducted from most retirees’ Social Security checks, was $45.50 a month in 2000. In 2026 it is $202.90, with a $283 annual deductible. That is about 5.9% a year. Over the same years, a benefit raised by every adjustment grew about 2.57% a year: a $1,000 monthly benefit in 2000 would be about $1,936 in 2026. On an index where 2000 equals 100, the premium stands at about 446 and the benefit at about 194. The premium took 4.55% of that hypothetical benefit in 2000 and 10.48% in 2026, so the check after the premium grew about 82% while the gross benefit grew about 94%.

The hold-harmless rule softens the worst years, for some. Under section 1839(f) of the Social Security Act, if the Part B premium would rise by more than a person’s Social Security raise, the premium increase is limited to the dollar amount of the raise, so the net check does not fall. The protection applies only to people whose premiums were deducted from their Social Security checks in both December and January. It does not cover people newly enrolling that year, people paying the income-related surcharge, people whose premiums are paid by Medicaid, or people on Medicare who have not yet started Social Security, including those delaying their claim, who pay the premium directly. In years of tiny or zero adjustments, that rule shifted more of the premium increase onto the people it did not protect.

The surcharge brackets are indexed, mostly. The income thresholds for Medicare’s income-related monthly adjustment amount have risen with inflation since 2020, so inflation alone does not push a household into a higher tier. The top tier, at $500,000 of income for a single filer and $750,000 for a joint return, was frozen by law through 2027 and begins adjusting in 2028. The mechanics, and the two-year lookback that makes these surcharges plannable, are in our article on IRMAA in 2026.

The Quiet Tax Inflation Collects

Inflation raises taxes without a vote in Congress. It does it in four ways, and retirees with savings tend to meet all of them.

Thresholds that never move. The tax brackets, the standard deduction, and most of the Medicare surcharge brackets adjust for inflation each year. A surprising number of other lines in the tax code do not. The income levels at which Social Security benefits become taxable, $25,000 and $34,000 for a single filer and $32,000 and $44,000 for a married couple, were set in 1983 and 1993 and have never been indexed; every year of inflation pulls more retirees into them, which is the machinery behind the Social Security tax torpedo. The net investment income tax still starts at $200,000 of income for a single filer and $250,000 for a joint return, the levels it began with in 2013. The $3,000 limit on deducting net capital losses against ordinary income has not changed since 1978. The home sale exclusion of $250,000 for a single filer and $500,000 for a married couple dates to 1997. And the 2025 tax law’s senior deduction, $6,000 per person 65 or older, phases out at 6% of modified adjusted gross income above $75,000 for a single filer and $150,000 for a joint return, with none of those figures adjusted for inflation through its last year, 2028; our article on the OBBBA senior deduction walks through how it works.

Taxes on returns that only kept pace. The tax code taxes nominal returns, not real ones, so some of the tax falls on the part of a return that only replaced lost buying power. Warren Buffett made the case in 1977 that inflation works as a tax on investors layered on top of the one Congress writes (“How Inflation Swindles the Equity Investor,” Fortune). An example of our own: for a household in the 24% federal bracket, a bond held in a taxable account has to yield about 3.95% just to keep pace with 3% inflation after federal tax, and about 4.74% to keep pace with 3.6% (before any state tax or the net investment income tax). Below those yields the bond loses buying power after tax, because the tax falls on interest that only replaced what inflation took.

Gains that are only inflation. An investment that does nothing but keep pace with 3.6% inflation doubles in dollars in about 20 years. Sell it, and the entire doubling is a taxable capital gain, even though the proceeds buy exactly what the original amount bought. The longer an asset is held, the more of its eventual gain is inflation rather than real growth. (At death, the step-up in basis erases that built-in gain for heirs, which is one reason the order in which accounts are spent and the assets that are left behind deserve planning.)

Inflation adjustments that are taxed before they are paid. Treasury inflation-protected securities add the inflation adjustment to their principal each year. In a taxable account, that adjustment is taxable as interest in the year it accrues, even though the investor does not receive it until the bond matures or is sold. That is one reason these bonds are often held inside retirement accounts. Series I savings bonds work differently: their interest is not taxed until the bond is cashed or stops earning interest after 30 years, and it is exempt from state and local income tax.

Which Tools Protect, and How Well

No single tool protects a retirement from inflation in every period. Each one defends in a particular way, over a particular horizon, and with a particular cost. Here is how the main ones work.

Stocks: strong over decades, weak in the short run. Over long periods, stocks have been the most reliable way to outgrow inflation. On Shiller’s data, every 30-year period measured January to January, starting from 1871 through 1993, earned a real return of at least 3.23% a year (the weakest began in 1892), with a median of about 6.72% a year, before taxes and fees. Over shorter stretches, stocks can lose to inflation for a decade: 17 of the 143 ten-year periods starting from 1871 through 2013 ended with a negative real return, and the 1970s are the classic case. Stocks protect a retirement that is long enough to wait; they do not protect a given year’s grocery bill. That is why we lean toward holding enough in stocks for growth over a long retirement, with the exact mix decided case by case.

Treasury inflation-protected securities: precise, if held to maturity. These bonds raise their principal with the consumer price index (for all urban consumers, not seasonally adjusted, with a short lag), and the interest rate applies to the raised principal. At maturity, the government repays the inflation-adjusted principal or the original amount, whichever is higher. Bought and held to maturity, they deliver the stated real yield before taxes, regardless of what inflation does; on October 28, 2026, the 10-year real yield was 2.67% and the 30-year 3.10%. Before maturity, their prices move with real interest rates, so a fund of these bonds can lose money in a year of high inflation if real rates rise at the same time, as happened in 2022. A ladder of them, with a bond maturing each year, can match a stream of real spending; our article on the 4% rule uses one as an illustration of what a guaranteed real income costs at today’s yields.

Series I savings bonds: inflation protection in small doses. Their rate combines a fixed rate set at purchase and an inflation rate reset every six months. Purchases are limited to $10,000 per person per calendar year in electronic bonds, a bond must be held at least a year, and one cashed before five years gives up the last three months of interest. They can suit a slice of an emergency reserve; at that limit, they cannot carry much of a retirement.

Social Security: the lifetime defense most households already own. Its adjustment, its lag, and its interaction with Medicare are covered above. A delayed benefit raises the share of a household’s spending covered by income that rises with prices for life.

Pensions: the adjustment is everything. A pension with an automatic cost-of-living adjustment carries its own inflation defense, though many public plans cap the yearly increase. A pension without one is the most inflation-exposed instrument in a typical retirement plan, which is why we lean toward weighing that exposure heavily in the pension lump sum or annuity decision. The Teacher Retirement System of Texas pays no automatic adjustment, a point that matters in the choice between it and the Optional Retirement Program for Texas university faculty; our article on TRS vs. ORP runs the numbers.

Annuities with rising payments: protection you pay for up front. Some income annuities offer a payment that rises by a fixed percentage each year, such as 2% or 3%. The trade is a noticeably smaller first payment, so the rising version takes many years to pay out more in total than the level one. A fixed increase is not inflation protection in the strict sense: it protects against expected inflation, not a burst like the 1970s. Annuities tied directly to the price index have become hard to find. Our article on annuity types, costs, and taxes covers how these contracts are priced.

Cash and bonds: stability, not inflation defense. Cash holds its dollar value, which is exactly why its loss of buying power is easy to miss: the balance never falls. Short-term cash yields do reset as interest rates move, so over the long run cash has earned roughly the rate of inflation: well ahead of it through the 1980s and 1990s, when rates were high, and behind it when rates sat below inflation, as they did in the 1940s, through much of the 1970s, and again from 2009 into 2022. Ordinary bonds pay fixed interest and repay fixed principal, so an inflation surprise shrinks what they deliver. Their job in a retirement is to supply steady spending money and to cushion falls in stocks, and they do that job well. Defending against inflation is not that job.

Flexibility: the tool most plans underrate. The ability to adjust spending is itself a hedge. A household that can trim a little after a hard stretch, and take a raise after a good one, can absorb an inflation burst that would break a rigid plan. That is the logic of spending guardrails set in advance, and how inflation moves through them is the subject of a section below.

The Debate Over Where Inflation Goes From Here

Thoughtful people disagree sharply about the next few decades. One camp sees federal debt pressing inflation higher; another sees technology pushing prices down. Neither case is settled, and both deserve a fair hearing.

The case for more inflation ahead. The federal government owed about $40.5 trillion on November 2, 2026, by the Treasury’s daily count. About $32.7 trillion of that is debt held by the public, owned by investors at home and abroad and by the Federal Reserve; the other $7.8 trillion or so is owed to accounts inside the government, mostly trust funds such as those for Social Security and federal employees’ retirement. The Congressional Budget Office projects debt held by the public at 101% of the economy in 2026, rising to 120% by 2036, past the record of 106% set in 1946. The argument this camp makes is that a debt fixed in dollars gets lighter when dollars buy less, so a government owing this much has reason to let inflation run and repay what it owes in cheaper dollars. Those who hold this view point to how governments have worked down debts this size. After the Second World War, the United States and Britain reduced their debts partly through growth and budget surpluses before interest, and partly through inflation with interest rates held below it; Carmen Reinhart and Belen Sbrancia estimated that negative real interest rates alone reduced debt by about 3% to 4% of the economy a year in those two countries. The parallel they cite most is Britain: its government debt peaked shortly after the war at around 270% of the economy and fell to around 50% over the next three decades, according to the country’s Office for Budget Responsibility, while whole-economy inflation averaged about 6.5% a year, budgets ran small surpluses before interest, and real interest rates were negative in 24 of those 30 years. There was no default and no hyperinflation. Savers and bondholders simply lost buying power, slowly, for a generation. The investment researcher Lyn Alden adds a mechanism worth understanding: with debt this large, raising interest rates also raises the interest the government pays to bondholders, which partly offsets the tightening the higher rates were meant to deliver.

The case against that view. Markets are not pricing it. A 10-year breakeven rate of 2.26% (on November 4, 2026) and a positive real yield of 2.67% on 10-year inflation-protected bonds (on October 28) are hard to square with financial repression, which requires interest rates held below inflation. Inflating away a debt is also harder than it sounds: a large share of today’s debt matures within a few years and is refinanced at whatever rates then prevail, and much of federal spending, Social Security’s benefits among it, rises with prices automatically, so inflation lifts the government’s costs along with its revenue. Inflation came down from its 9.1% peak in June 2022 while interest rates stayed high. The postwar inflation in the United States faded once the wartime deficits stopped. And the Federal Reserve’s stated framework commits it to a 2% goal and, since its 2025 revision, says it is prepared to act forcefully to keep longer-term inflation expectations anchored; in 2022 and 2023 it raised rates faster than at any time in four decades.

The case for falling prices. Jeff Booth makes the opposite argument in The Price of Tomorrow (2020): technology keeps driving the real cost of making things down, and left alone that abundance would push prices lower, so the inflation people see reflects the expansion of money and credit more than the cost of goods. The chart of durable goods prices above shows part of what he means: they cost about the same in 2026 as in 2000. The late 1800s show that long stretches of falling prices have happened before, and some technologists now argue that artificial intelligence and robotics could carry that kind of deflation into services and knowledge work. Skeptics answer that many economists expect slower price increases rather than falling prices, that the costs retirees feel most (housing in a particular place, the time of particular clinicians) are the least likely to fall, and that artificial intelligence’s own demand for electricity, land, and materials could push some costs up.

What Japan shows, both ways. Japan is the case both camps cite. Its consumer prices were essentially flat for more than two decades, an index of 100.08 in 1998 and 101.80 in 2020, while its gross government debt climbed from about 102% of its economy in 1998 to about 229% in 2020 on the International Monetary Fund’s current figures: a long record of very high debt without inflation. Economists also point to those same years of flat prices and weak growth as evidence that deflation has real costs of its own, including that falling prices make every debt heavier in real terms. And in 2025, with its debt still above 200% of its economy, Japan’s prices rose 3.2%. A country can live with very high debt and no inflation for a long time, and the story can still change.

Where we land. Buffett, in the same 1977 essay, offered the line we keep closest: “Forecasts usually tell us more of the forecaster than of the future.” We lean toward planning without betting on either camp. A slow, decades-long erosion like Britain’s is a realistic scenario worth weighting, and it lands hardest on income fixed in dollars and on cash. A stretch of mild or even falling prices is realistic too, and a plan built to adjust can use it. The case for a deflationary monetary asset is its own debate, which we take up in our article on bitcoin and a deflationary economy, and the way inflation shapes what money means to a family is the subject of our piece on the hidden cost of money.

How a Plan Should Assume Inflation

Every retirement projection needs an inflation assumption, and the honest first observation is that the reasonable choices span a wide range.

The candidates. The bond market’s 10-year breakeven inflation rate, the difference between ordinary and inflation-protected Treasury yields, was 2.26% on November 4, 2026, and its five-year rate starting five years from now was 2.22%. The Federal Reserve’s goal is 2% on a different measure, the personal consumption expenditures price index, which tends to run a little below the consumer price index. History gives other answers depending on the window: measured September to September on the consumer price index, inflation averaged about 2.56% a year over the last 30 years, 3.60% over the last 50, 3.00% over the last 100, and 3.16% since 1913. Including Shiller’s wholesale price era back to 1871 pulls the long-run figure down to about 2.1%, which is why any claim about “the long-run average” should name its window. And the most recent reading is higher than most of these: prices rose 3.7% in the 12 months ended September 2026 (2.6% excluding food and energy), up from about 2.4% early in the year.

History and a steady rate do different jobs. Replaying actual history, month by month, shows what inflation bursts do to a plan when they arrive alongside the markets that actually came with them. A steady rate cannot do that, however high it is set. What the steady rate does is give a plan its everyday view: next year’s budget and the cash flow a household reads at each review. Choosing that number is a judgment about risk, not a forecast.

The rate interacts with the return assumption. How much a higher inflation assumption hurts a plan depends on how the plan’s returns are built. If returns are assumed as a real return plus inflation, raising the inflation rate also raises projected returns in dollars, and the damage falls mainly on the parts of the plan fixed in dollars: a pension without an adjustment, a fixed annuity, tax thresholds that never move. If returns are held fixed in dollars while inflation rises, each extra point of inflation removes a point of real return, and the plan looks far worse. Neither is wrong, but it is worth knowing which one a given projection is doing. In a historical replay, the question answers itself, because returns and inflation come from the same history.

What the choice is worth, in dollars. Take a hypothetical household spending $100,000 a year at 65, with that spending rising with prices for 30 years. At 2.56% inflation, spending at 94 reaches about $208,143, and the 30 years total about $4.43 million in dollars spent. At 3%, the final year is about $235,657 and the total about $4.76 million. At 3.6%, the last 50 years’ average, the final year is about $278,890 and the total about $5.25 million. About one point of inflation, compounded over 30 years, raises the last year’s spending by about a third. The shape of the spending can move it just as much. Apply our baseline for a budget with a discretionary layer, total spending drifting down about 1% a year in real terms (leaving aside the small late uptick): if $60,000 of the budget is essentials that hold their buying power, the $40,000 of flexible spending on top shrinks to about $14,717 in today’s dollars by 94. At 3.6% inflation, the final year then comes to about $208,379 and the 30-year total to about $4.44 million, almost exactly the flat path at 2.56%, because a drift of 1% a year does to total spending roughly what a point less inflation would. The drift would lower spending at any inflation rate, so it is not a reason to choose one rate over another; it shows that the shape of the spending deserves as much attention as the rate. It offers no help on the income side: a pension with no adjustment, or a fixed annuity payment, loses buying power at whatever rate prices actually rise. These totals are not what the household needs saved on day one, since the portfolio keeps growing while it pays them out, and in a plan whose returns rise with inflation most of the extra dollars are matched by extra growth; the gap bites where the plan holds something fixed in dollars.

Over-assuming has a cost too. A plan that assumes more inflation than arrives asks a household to spend and give less than it safely could, often in the early years when health and energy are greatest. That is not a harmless kind of caution; it is the pattern our article on why retirees underspend describes. What keeps it in proportion is how a plan is run. In a plan reviewed regularly, with guardrails set in advance, an assumption that proves too high shows up later as room for raises. An assumption that proves too low, locked into an irrevocable fixed-dollar decision such as a pension election or an annuity purchase, cannot be fixed afterward. That asymmetry is why we lean toward weighting inflation risk most heavily where a decision cannot be undone.

How Inflation Moves Through a Plan With Guardrails

In a plan with risk-based guardrails, the approach Derek Tharp and Justin Fitzpatrick have laid out at Kitces.com, inflation is not an assumption sitting off to the side. It runs through the same machinery that responds to markets, and it meets the household’s income needs along the way.

Spending is set in today’s dollars. A plan’s spending target is set in real terms, so the dollars withdrawn rise with actual prices each year. A year of high inflation raises the withdrawal; if the portfolio did not keep pace, the plan’s risk of overspending rises, and it moves toward its lower guardrail just as it would after a market decline. Inflation and markets push on the same guardrails, which is why the 1970s, when both pushed at once, test a plan harder than almost any other stretch on record.

The response tends to be modest, and it comes from flexible spending. Under the guardrail settings we lean toward starting from, a cut trims spending by a few percent, comes from the flexible part of the budget rather than the essentials, and holds only until the plan recovers enough to earn a raise. A mild stretch of inflation with decent markets runs the same machinery in reverse: the plan’s risk falls, and spending and giving rise. An inflation burst can become a modest, temporary adjustment rather than a crisis, and a stretch of mild inflation becomes room to do more.

Income that rises with prices sets how much inflation reaches the portfolio. Consider a hypothetical couple whose essential spending is $80,000 a year, with $45,000 coming from Social Security and $20,000 from a pension with no cost-of-living adjustment, leaving a $15,000 gap for the portfolio to fill. After ten years of 3.6% inflation, with Social Security’s adjustments matching it, the essentials cost about $113,943, Social Security pays about $64,093, and the pension still pays $20,000. The gap is now about $29,850, nearly double, while prices rose about 42%. Had the pension carried a full adjustment, the gap would be about $21,364. Income fixed in dollars does not only lose value itself; it hands its share of each year’s inflation to the portfolio.

The essential share sets where spending starts. Because guardrail cuts come only from flexible spending, the larger the share of the essentials a portfolio has to fund, the less room the plan has to absorb an inflation burst by adjusting, and the more conservatively spending has to start. Income that rises with prices, chiefly Social Security, covering the essentials is what gives the flexible layer room to flex. That is the intersection worth watching in an income plan: not just how much income arrives, but how much of it keeps its buying power.

Why Inflation Makes Planning an Ongoing Job

Of all the assumptions in a retirement plan, inflation is the one most certain to be wrong in some direction. A plan written once and filed away keeps assuming whatever rate it started with while actual prices go their own way, and the error compounds into every later year.

How fast a plan goes stale. Consider a hypothetical plan written in September 2019 that assumed 3% inflation. By September 2026 it expected prices about 23.0% higher. Actual consumer prices rose about 31.2%, so a household spending exactly as the plan intended was spending about 6.7% more in dollars than the plan was built on, after only seven years, with that gap carried into every year that follows. Against a plan built at 2.5%, actual prices ran about 10.4% higher. Nothing about the household changed; the world did.

A one-time answer cannot see what comes next. Running a 4% withdrawal calculation once, or buying an annuity and calling the planning done, gives a household an answer for the world as it looked on that day. The 4% rule is a benchmark worth knowing well, and the research behind it tested exactly that case, a withdrawal that never adjusts, which is part of why it starts where it does (Bengen himself reviewed his own clients’ plans every year). An annuity can be a sound part of a plan, but buying one does not finish the planning: a level payment covers a little less of the budget every year, and the rest of the plan has to absorb the difference, as the pension example above shows. Neither answer can say what to do when a year like 2022 arrives, when the cost-of-living adjustment lags, or when the Medicare premium jumps. Those are planning questions, and they keep coming.

What a review resets. At each review, a plan can replace the inflation it assumed with the inflation that actually happened, restate the budget, the income, and the portfolio in today’s dollars, and run the risk analysis again from there. Several inputs move on their own calendar every year: Social Security announces the next year’s cost-of-living adjustment each October; Medicare’s Part B premium and surcharge brackets for the coming year are set in the fall; the IRS publishes the next year’s inflation-adjusted brackets and standard deduction, also in the fall; and the thresholds that never move fall a little further behind. A household’s own prices move too, category by category, and households often feel the change before the averages show it.

Small, early adjustments instead of large, late ones. A plan that sees inflation as it arrives can respond in small steps: a trim of a few percent after a hard stretch, a raise after a good one, a second look at a fixed-dollar decision before it is made. A plan that waits for a stale projection to fail tends to need larger adjustments, later, when there is less room. It is the same reason a one-time probability of success overstates the risk for a plan that will be reviewed, as we explain in our article on how to read a Monte Carlo probability of success. Inflation, like returns, is a moving input, and a plan is only as current as its last update.

The value is in the planning, not the plan. There are two ways to head into the backcountry. One is to stop at an outfitter, pick up supplies, and have someone circle a few spots on the map. The other is to hire a guide who walks the trail with you. A guide cannot know whether you will meet a bear on this trip, or whether rain upstream will raise the river past the usual crossing. But a guide has walked that terrain many times, knows what to do and what not to do if the bear appears, knows how the river should look at normal flow and how to judge the risk when it does not, and knows the country around the trail well enough to find a safer crossing upstream when the water is high. A good guide also remembers that it is your journey, not theirs; the job is to help you get where you are going, drawing on having helped other hikers through the same kinds of moments.

A one-time plan is the outfitter’s map: useful, and fixed on the day it was drawn. Ongoing planning is the guide. Nobody knows when the next inflation burst will arrive or how high it will run, but the questions it raises (whether to trim, what to protect, which fixed-dollar decisions to hold until the numbers settle) can be seen coming and worked through calmly when they do. The value is not the plan itself; it is the ongoing planning.

How We Think About It

Inflation is one of the few retirement risks that is nearly certain to show up; the open questions are how much and when. Here is how we lean toward handling it in our plans.

History as the risk test, and the stress test. We lean toward a risk analysis that replays actual history, inflation and markets together, month by month, so the 1966 retiree’s 1970s show up as they happened rather than as an average, alongside the stretches of low inflation and falling prices that history also holds. Rather than choose one rate and hope it is cautious enough, a plan tested this way meets every historical sequence, the worst included, and shows in advance how spending would respond to each. We pair it with retirement spending guardrails set in advance, which turn the response to an inflation surprise into a decision made in calm rather than in a crisis.

Every review starts from actual prices. We typically update retired clients’ plans about twice a year, and we lean toward having each update restate the budget at the prices that actually arrived, re-price the essentials and the income that covers them, and run the guardrails again from there. A plan that resets this way rarely drifts far from the household’s real prices. That is where most of the value in planning for inflation lies: not in the rate chosen on day one, but in the ongoing planning that notices when the trail changes.

A budget by category, with a flexible layer that moves. We lean toward letting the spending a household wants to keep no matter what, including its regular giving, rise with prices in full, while total spending drifts down about 1% a year in real terms as the flexible spending on top eases, close to the average the research finds. In the hypothetical household above, at 3.6% inflation, that drift brings final-year spending from about $278,890 down to about $208,379, which keeps a plan from holding back spending and giving today to fund a path the household is unlikely to follow. The drift is an average, and one family’s spending tends to fall in steps after events rather than smoothly, so we revisit it at every review.

The most weight where a decision cannot be undone. A pension election, an annuity purchase, the choice of when to claim Social Security: these are the moments where we lean toward pricing inflation exposure in real terms alongside the nominal numbers, across the full range of history rather than an average, because there is no guardrail to correct them later. A pension with no adjustment is the most inflation-exposed income in a typical retirement plan, and a delayed Social Security benefit is among the best defended, so we model both on the household’s own numbers rather than reaching for a rule.

Enough growth for a long retirement, and the rest for stability. Stocks are the long-run defense against inflation, and bonds and cash are the short-run source of spending money; we lean toward enough of the first for a long retirement, with the mix decided case by case.

Why we lean this way. The historical evidence above is consistent: inflation has eroded a fixed dollar in every 30-year retirement since 1913, it arrives in bursts that a steady rate cannot show, and a retiree’s own basket tilts toward the services and healthcare that have risen fastest. Weighting that risk, without forecasting it, protects the decisions that cannot be revisited while leaving the rest of the plan free to adjust.

The honest trade-offs. History holds only the sequences that actually happened. The next 30 years could bring inflation worse than anything in the U.S. record since 1913, as other countries’ histories show is possible, or a long stretch as mild as the bond market’s 2.26% breakeven and the Federal Reserve’s 2% goal suggest, below the last 30 years’ average of 2.56%. A plan whose spending is tested against every historical sequence, the 1970s included, carries some margin for a stretch like that, so if inflation stays low it will look cautious, and our reviews are designed to turn that caution into room to spend and give more as the years pass. Blanchett has also argued that because real spending tends to fall and Social Security is indexed, most retirees do not need a perfect inflation hedge, and for many households he is right. A household whose income is mostly Social Security and a pension with a full adjustment has far less inflation exposure than one living on a portfolio and a fixed pension, and a household whose budget is almost all essentials cannot count on the discretionary drift. So the weighting is a lean, applied case by case, not a formula.

Common Questions

What inflation rate should I assume for retirement planning?

We lean toward not relying on any single rate: testing a plan against actual history, month by month and the 1970s included, shows what inflation bursts do in a way no steady number can. Where one number is needed, the reasonable candidates span a range: the bond market’s 10-year breakeven rate was 2.26% in early November 2026, inflation averaged about 2.56% a year over the last 30 years, 3.60% over the last 50, and 3.00% over the last 100, and prices rose 3.7% in the 12 months ended September 2026. Whichever is chosen, a steady rate cannot show the risk of inflation arriving in a burst, so it works best alongside a historical test rather than in place of one.

Does Social Security keep up with inflation?

Mostly, with a lag and a few gaps. Each year’s cost-of-living adjustment follows the consumer price index for urban wage earners and clerical workers from the prior summer, takes effect with December’s benefit, and is paid in January; the 2027 adjustment is 3.6%. There was no adjustment for 2010, 2011, or 2016, and the index reflects a working household’s basket rather than a retiree’s. The larger gap is Medicare: the Part B premium has risen about 5.9% a year since 2000, against about 2.57% a year for a benefit raised by every adjustment, so the check after the premium has grown more slowly than the benefit itself.

Do retirees face higher inflation than everyone else?

Often, because of what they buy. Since 2000, prices of durable goods have barely moved while services other than energy have risen about 3.09% a year, and retiree budgets lean toward services such as healthcare, insurance, and help at home. Healthcare adds premiums that can outpace prices and use that rises with age. Offsetting that, some costs stop growing, such as a paid-off home, and research finds that real spending tends to drift down through retirement, mostly in discretionary categories.

What protects a retirement from inflation?

Several tools, each over a different horizon. In U.S. market data since 1871, stocks outgrew inflation over every 30-year period measured January to January, before taxes and fees, but they have lost to it over a full decade. Treasury inflation-protected securities held to maturity deliver a known real yield, and Series I savings bonds offer protection in small amounts. Social Security and pensions with automatic adjustments rise with prices for life, while pensions without one, fixed annuities, cash, and ordinary bonds do not. Flexible spending, the ability to trim a little after hard stretches and take raises after good ones, is a hedge of its own.

Will the federal debt cause high inflation?

Nobody knows, and the evidence cuts both ways. In early November 2026 the federal government owed about $40.5 trillion, about $32.7 trillion of it held by the public (investors at home and abroad, and the Federal Reserve); the Congressional Budget Office projects that publicly held debt will rise from 101% of the economy in 2026 to 120% in 2036; and after the Second World War both the United States and Britain worked down large debts partly through inflation and interest rates held below it. Against that, the bond market expects inflation near 2.26% a year over the next decade, real interest rates are positive, much of today’s debt is refinanced at new rates within a few years, and Japan has carried far higher debt without sustained inflation. We do not forecast it; we lean toward weighting the risk of a slow erosion most heavily in decisions fixed in dollars, and keeping the rest of a plan flexible.

How often should a retirement plan be updated for inflation?

We lean toward updating a retired household’s plan about twice a year, each time restating its budget and income at the prices that actually arrived. The gap builds quickly: by September 2026, actual prices had run about 6.7% above what a hypothetical plan written in September 2019 at 3% a year assumed, and that gap carries into every later year. Each fall also brings a new Social Security cost-of-living adjustment, a new Medicare Part B premium, and new inflation-adjusted tax brackets, so a plan left alone for a few years is working from old numbers. A one-time calculation, whether a 4% withdrawal rate or an annuity quote, answers for the day it was run; ongoing planning keeps answering as the numbers change.