Retire in Real Terms: Why a Fixed Income Is Not a Fixed Standard of Living
The central problem in retirement-income design is not simply making money last. It is making purchasing power last.
Thesis. A retirement-income plan should be specified, funded, and stress-tested in real consumption units rather than nominal currency units. Annuities remain valuable because they can insure longevity risk, but a fixed nominal lifetime payment solves only the risk of outliving the payment stream; it can leave the retiree exposed to a second, compounding risk - the possibility that the payment buys progressively less. The economically relevant retirement target is therefore a stabilized purchasing-power income floor, supported by genuinely indexed cash flows where possible and supplemented by liquid assets capable of absorbing inflation, expenditure, and market shocks.
Keywords: retirement income, annuities, inflation risk, purchasing power, real income, longevity risk, retirement planning, CPI indexation, TIPS, Social Security, sequence risk, lifetime income, pensions, asset-liability matching
There is a deceptively simple question at the centre of retirement planning: How much income will I need?
It sounds as though it should have a dollar answer. Fifty thousand dollars a year. One hundred thousand. Two hundred thousand. Whatever number fits the household’s expected retirement budget.
But that is the wrong unit of account.
The real question is: What level of consumption must the retirement system be able to finance, and for how long?
Those are not the same question. A nominal dollar amount is a number printed on a payment. A real retirement income is a claim on housing, food, energy, transport, insurance, healthcare, travel, services, and everything else a household expects to consume. Inflation separates the two.
This distinction is elementary in economics and still surprisingly easy to lose in practical retirement planning. A retiree may buy a pension or an immediate annuity that promises exactly the payment requested for life. The contract can perform perfectly in nominal terms. The insurer can send precisely the same amount every month for thirty years. Yet the economic outcome can deteriorate badly because the contract guarantees currency, not consumption.
FINRA states the problem directly: fixed-annuity payments typically do not include cost-of-living adjustments, so their purchasing power can decline with inflation; inflation-protected alternatives can be purchased, but at greater cost (Financial Industry Regulatory Authority [FINRA], 2022). That is not a minor product feature. It changes the object being insured.
A lifetime nominal annuity is insurance against longevity without income. A genuinely inflation-linked lifetime annuity is insurance against longevity without purchasing power. Those are different liabilities.
The distinction becomes increasingly important as retirement horizons lengthen. The Social Security Administration’s period life table reports remaining life expectancy at age 65 of 16.95 years for men and 19.75 years for women in the 2021 table used for the 2024 Trustees Report. More important for risk management, the distribution has a long tail: some retirees live far beyond the average (Social Security Administration [SSA], 2024). Retirement planning therefore cannot be built around the mean alone. A plan that works for seventeen years but deteriorates in years twenty-five through thirty has not solved longevity risk; it has merely postponed failure.
The correct objective is to design retirement income around stable real purchasing power.
A terminological point matters here. Economists normally use purchasing-power parity (PPP) to describe price-level or exchange-rate relationships across countries. The retirement problem is primarily intertemporal rather than cross-country. The more precise term is therefore real purchasing-power stability: specifying income in constant consumption units and allowing the nominal number of dollars, pounds, euros, or other currency units to change as the price level changes. The intuition behind “retiring in purchasing-power money” is right; the technically cleaner formulation is to retire against a real income standard rather than a nominal income target.
That one change in framing alters almost everything about retirement-income design.
The guarantee can be real or merely nominal
An annuity is attractive for a reason. Conventional investment accounts force the retiree to manage an uncertain lifespan with a finite stock of wealth. Spend too rapidly and the portfolio can be exhausted. Spend too cautiously and the retiree may die having unnecessarily suppressed consumption. A life annuity pools mortality risk: payments continue while the annuitant is alive, and the insurer can use mortality credits arising from the pooling arrangement.
The theory supporting annuitization is strong. Davidoff, Brown, and Diamond (2003), in a full theoretical treatment of annuity demand, show that positive annuitization remains optimal under broad conditions when annuities provide survivors with returns greater than otherwise matching conventional assets, although complete annuitization need not be optimal once markets are incomplete, liquidity matters, or the available payout trajectory does not match desired consumption. Their result is important because it prevents the inflation critique from becoming an anti-annuity argument. The problem is not that lifetime income is economically useless. The problem is that the shape and denomination of the lifetime income stream matter.
This was already central to Bodie and Pesando’s (1983) analysis of retirement-annuity design. They distinguished the certainty of nominal payments from the uncertainty of their real value. Their historical illustration was severe: during the inflationary experience they examined, a fixed nominal annuity payment that initially had real value close to its face amount lost more than half of its purchasing power over the subsequent period. Their broader point survives the historical episode that motivated it: when the price level is uncertain, a level nominal annuity transfers investment and longevity risk but leaves the retiree holding inflation risk.
The contract can therefore be “safe” in one dimension and unsafe in another.
Suppose a retiree is promised $50,000 a year for life. There is no default, no missed cheque, no market volatility in the payment, and no ambiguity about the nominal amount. If inflation averages 3%, however, the year-30 payment is still $50,000 while the price level is about 2.43 times its starting level. In retirement-year dollars, that payment has purchasing power of only about $20,599.
Nothing has gone wrong with the annuity contract. The annuity has done exactly what it promised.
The mistake was made earlier, when the retiree treated a nominal promise as if it were a real promise.
Inflation is not a one-year problem
People often reason about inflation using annual changes. Inflation is 2%, 3%, 4%, perhaps temporarily 6% or 8%. In a working-age budget, that framing is understandable because wages may adjust, employment can change, and the household has some ability to earn more.
Retirement is different. The relevant mathematics is cumulative.
Let fixed nominal retirement income be Y₀, let annual inflation be π, and let t be the number of years after retirement. Then:
Real purchasing power in year t = Y₀ ÷ (1 + π)ᵗ
Conversely, the nominal income required in year t to preserve the original purchasing power is:
Required nominal income in year t = Y₀ × (1 + π)ᵗ
These are not forecasts. They are accounting identities under constant assumed inflation rates, useful for seeing the scale of the liability.
The Federal Reserve’s longer-run inflation objective is 2% as measured by the PCE price index (Board of Governors of the Federal Reserve System, 2025). Even if inflation were held exactly at that target - an unusually clean assumption - a fixed nominal income would still lose purchasing power every year. “Price stability” in modern monetary policy does not mean a constant price level. At 2% annual inflation, the price level rises by about 81% over thirty years.
The consequences for a fixed $50,000 annual retirement income are shown below.
Figure 1. Real purchasing power of a fixed $50,000 nominal annual income. Values are expressed in retirement-year dollars. The scenarios are illustrative arithmetic, not inflation forecasts.
At 2% inflation, the $50,000 payment has the purchasing power of approximately $41,017 after ten years, $33,649 after twenty years, and $27,604 after thirty. At 3%, the corresponding figures are about $37,205, $27,684, and $20,599. At 4%, they are about $33,778, $22,819, and $15,416.
A retiree who says “I can live comfortably on $50,000” has therefore supplied an incomplete specification. The statement needs a date attached to it. Fifty thousand in today’s purchasing power is not the same consumption claim as fifty thousand thirty years from now.
The Bureau of Labor Statistics makes the underlying relation explicit: as prices increase, the purchasing power of the consumer’s dollar declines, which is why CPI measures are used to adjust income payments and other dollar values (U.S. Bureau of Labor Statistics [BLS], 2025).
The reverse calculation is equally revealing. If the retiree wants the purchasing power of $50,000 in the first retirement year to remain intact, the nominal income target must rise.
Figure 2. Nominal income required to preserve $50,000 of real annual purchasing power. The scenarios are illustrative arithmetic, not inflation forecasts.
At 3% inflation, preserving a $50,000 real standard requires approximately $67,196 of nominal income in year 10, $90,306 in year 20, and $121,363 in year 30. At 4%, the year-30 nominal requirement is about $162,170.
This is why retirement-income design should begin in constant dollars. The planner should decide what real standard of living must be funded first. The nominal cash flow is then an output of the inflation path, not the objective itself.
A 3% annual escalator is not the same thing as indexation
A common response is to use an annuity or pension with predetermined annual increases: perhaps 2%, 3%, or 5% a year. This is better than pretending prices never change, but it is not the same as true inflation protection.
If income grows at a fixed rate g while inflation is π, real income evolves approximately as:
Real income in year t = Y₀ × [(1 + g) ÷ (1 + π)]ᵗ
If g = π, purchasing power is preserved. If inflation persistently exceeds the contractual escalator, real income declines. If inflation is lower, real income rises. The retiree has exchanged the certainty of a flat nominal payment for a basis risk between the assumed inflation rate embedded in the escalator and the inflation actually experienced.
Soares and Warshawsky (2003) analysed this distinction directly. Their study compared nominal fixed immediate annuities, increasing annuities based on expected inflation, phased annuity purchases, and inflation-adjusted annuities. The inflation-adjusted design addressed post-retirement inflation risk most directly, while the nominal and predetermined-increase approaches retained exposure to realised inflation. Their simulations also illustrated the economic price of protection: inflation-adjusted annuities began with lower initial payments than nominal annuities in the period they studied.
That lower starting payment is not evidence that inflation protection is “worse.” It is the visible price of moving risk away from the retiree. A nominal annuity can offer a larger payment today partly because the retiree accepts the risk that tomorrow’s money buys less. An indexed annuity has to finance a payment stream that is expected to rise in nominal terms, and the insurer must hedge that liability.
Bodie and Pesando (1983) reached the same conceptual conclusion from a broader examination of annuity structures. A graduated nominal annuity is still a nominal contract. It can approximate real-income protection only to the extent that its predetermined gradient tracks realised inflation. A genuine purchasing-power annuity instead links payments to the price level itself.
This distinction should be made explicit in every retirement-income proposal. “Increasing” is not synonymous with “indexed.” “Indexed” is not synonymous with “guaranteed real consumption,” either, unless the index closely matches the retiree’s relevant cost of living.
The index itself is part of the contract
Even genuine indexation does not eliminate every form of purchasing-power risk.
Social Security provides a useful example of an indexed lifetime income stream. Benefits receive automatic cost-of-living adjustments based on changes in the CPI-W. For 2026, the Social Security Administration applied a 2.8% COLA based on the measured increase in that index (SSA, 2026). This is materially different from a fixed nominal pension because the payment adjusts with an observed price index.
Treasury Inflation-Protected Securities provide another building block. TIPS principal is adjusted with inflation using the Consumer Price Index, and the fixed coupon rate is applied to the adjusted principal; Treasury currently issues TIPS at 5-, 10-, and 30-year maturities (U.S. Department of the Treasury, Bureau of the Fiscal Service, n.d.). A properly constructed TIPS ladder can therefore match a sequence of future real cash-flow liabilities far more directly than a ladder of nominal bonds.
But indexation is always indexation to something.
A retiree does not consume “the CPI.” A household consumes a particular basket. One person may have unusually high housing costs; another may own a home outright. One may spend heavily on travel; another on healthcare and care services. Energy exposure differs. Geography differs. Tax treatment differs. A broad consumer price index is therefore an excellent general measure of inflation and a much better anchor than no indexation at all, but it is not a perfect hedge for every household’s cost structure.
This introduces household inflation basis risk. If the retiree’s personal expenditure basket rises faster than the index used to adjust income, real living standards can still fall even though the contract is technically indexed.
That does not invalidate indexation. It changes how much of the retirement budget should be treated as a hard, hedgeable liability and how much should remain flexible.
Retirement should be designed like an asset-liability problem
The most useful conceptual shift is to stop treating retirement as a pot of money from which withdrawals are made and start treating it as a long-duration liability-management problem.
The liability is not “$1 million at age 65.” The liability is a vector of future consumption claims:-
housing and utilities in year 1;
-
food, insurance, transport, and taxes in year 7;
-
healthcare and support in year 18;
-
essential consumption if one spouse survives into year 28;
-
discretionary spending that can be reduced when markets are weak;
-
emergency liquidity throughout.
Each liability has a time, an expected real amount, an uncertainty range, and a degree of flexibility.
Once retirement is expressed that way, the asset side becomes easier to design. Different assets solve different risks.
A life annuity is unusually efficient at solving idiosyncratic longevity risk because its payments are contingent on survival. Davidoff et al. (2003) show why mortality pooling can make annuitized wealth valuable even when annuities are not actuarially fair. A TIPS ladder solves a different problem: it can provide government-backed real payments at specified maturities, but it does not pool mortality risk. A liquid diversified portfolio can support growth and flexibility, but its future value is uncertain and withdrawals are exposed to market sequence risk. Cash provides immediate liquidity but is highly exposed to inflation over long horizons.
The design problem is therefore not to choose “annuities or investments.” It is to allocate each retirement liability to the instrument best suited to it.
A robust structure can be thought of in four layers.
1. A real essential-income floor
Start with spending that should not depend on market conditions: basic housing costs, food, utilities, insurance, baseline transport, and a realistic allowance for essential healthcare and care-related expenses.
Express that amount in today’s dollars. Do not ask what nominal income will be required at age 85. Ask what today’s basket costs and require the retirement system to reproduce that purchasing power later.
Then identify existing indexed income. In a U.S. context, Social Security is the obvious starting point because it is a lifetime income stream with CPI-W-based COLAs (SSA, 2026). Other jurisdictions may have state pensions with different indexation rules. Defined-benefit pensions should be classified carefully: some are fully indexed, some partially indexed, some capped, some discretionary, and some entirely nominal.
Any residual essential-income gap can then be matched with inflation-linked assets or, where available and economically sensible, inflation-linked lifetime annuity income.
The crucial accounting rule is simple: nominal income does not get credited at face value against a real liability. A $30,000 fixed pension is not a permanent $30,000 contribution to the real income floor. Its real contribution decays with inflation and must be modelled accordingly.
2. A longevity layer
The next question is not expected lifespan but unacceptable survival states.
A retiree should not plan to the average death date. The financial consequence of dying early is typically a bequest or unused wealth. The financial consequence of living far longer than expected can be catastrophic if assets are depleted. That asymmetry is why longevity insurance exists.
Annuitization is especially useful here. It allows a retiree to transfer the tail of longevity risk rather than self-insure the entire possibility of survival to advanced ages. But the denomination matters. If the annuity is nominal, the lifetime guarantee should be entered into the model with declining real purchasing power. If it has a fixed annual increase, the model should compare the escalator with alternative inflation scenarios. If it is CPI-linked, the model should evaluate index basis, caps, lags, floors, insurer credit risk, and initial payout.
This is also where deferred lifetime income can be conceptually powerful: rather than annuitizing every dollar immediately, a retiree can preserve liquidity in early retirement while insuring against very late-life survival. The precise product and tax treatment are jurisdiction-specific, but the financial logic is general.
3. A liquid real-reserve layer
No sensible retirement plan should convert every asset into an irreversible income stream.
Davidoff et al. (2003) are particularly useful on this point. Their model shows that while annuitization can be highly valuable, market incompleteness, liquidity constraints, uninsured expenditure shocks, and a mismatch between annuity payouts and desired consumption can make complete annuitization suboptimal. A retiree may need cash for large medical costs, repairs, family transfers, relocation, or other shocks that cannot be financed by selling a life-annuity payment stream.
The retirement balance sheet therefore needs liquid reserves.
For near-term liabilities, cash and short-duration high-quality instruments provide stability. For known medium- and long-term real liabilities, inflation-linked bonds can provide stronger matching. TIPS are particularly useful in the United States because principal adjusts with CPI, allowing a retirement plan to build a ladder whose maturities correspond to future real spending needs (U.S. Department of the Treasury, Bureau of the Fiscal Service, n.d.).
The objective is not to maximize the nominal yield on this layer. It is to reduce the probability that an essential future expense must be financed by selling risky assets after a market decline or after an unexpected inflation shock.
4. A growth and discretionary layer
Retirement can last long enough that eliminating all exposure to growth assets creates its own risk. A portfolio held entirely in nominal cash or nominal fixed income can be stable in statement value while unstable in real purchasing power.
A diversified growth portfolio can provide an imperfect but important source of long-horizon real growth. It should not, however, be described as a guaranteed inflation hedge. Bodie and Pesando’s (1983) analysis of variable annuities is instructive: asset-backed variable payments create different combinations of real-income volatility and expected return. Equity exposure can raise expected long-run resources but also increases uncertainty in the income path.
This layer is best matched to spending that is genuinely flexible: travel, gifts, discretionary purchases, larger leisure budgets, or legacy goals. In adverse markets, these expenditures can be delayed or reduced without threatening the retiree’s basic standard of living.
The result is a retirement design in which the least flexible liabilities receive the strongest real-income protection, while the most flexible liabilities bear more market risk.
That is more coherent than applying one withdrawal rule to the entire portfolio.
Inflation creates a sequence problem of its own
Retirement literature rightly focuses on sequence-of-returns risk: poor investment returns early in retirement can be disproportionately damaging when the retiree is simultaneously withdrawing from the portfolio.
Inflation has a related sequence effect.
A burst of inflation early in retirement raises the price level from which subsequent inflation compounds. Even if inflation later returns to a low rate, prices do not automatically return to their earlier level. A retiree with a fixed nominal income experiences a permanent downward shift in real purchasing power unless some other income or asset adjusts upward.
This matters because many retirement models use a single long-run inflation assumption. A model may assume 2.5% inflation every year and appear precise. Real life can produce 1%, then 7%, then 5%, then 3%, followed by lower inflation. The arithmetic average can look manageable while the price-level path creates a much larger immediate adjustment burden.
For a genuinely indexed payment, the path matters less because nominal income follows the index. For a fixed nominal payment, the path matters enormously. For a fixed-escalator payment, the damage depends on the cumulative gap between the escalator and realised inflation.
A retirement plan should therefore stress-test inflation paths, not only average inflation rates.
At minimum, I would test:-
a low and stable inflation path;
-
a central path around the long-run policy objective;
-
a persistent 3%-4% path;
-
an early-retirement inflation shock followed by normalization;
-
a household-specific expenditure path in which healthcare, insurance, or housing costs rise faster than the broad index.
The purpose is not to forecast which path will occur. It is to discover which parts of the retirement design fail under plausible alternative states.
The first-year payout is the wrong comparison
One reason nominal annuities can look compelling is that product comparisons often begin with the first payment.
Suppose Contract A pays $60,000 a year for life with no indexation, while Contract B pays $45,000 initially but rises with inflation. The nominal contract appears to “pay more.” In year one, it does. But the comparison silently assumes that a dollar in year twenty is economically identical to a dollar today.
It is not.
Soares and Warshawsky (2003) found precisely this trade-off in their historical simulation: inflation-adjusted annuities had substantially lower initial payments than nominal annuities, but they directly addressed the post-annuitization inflation exposure that fixed nominal payments retained. That is the correct economic comparison. The retiree is choosing between different state-contingent income paths, not merely different first cheques.
The relevant metric should therefore be something like expected and stressed lifetime real consumption, not initial nominal payout.
A useful annuity comparison would report, at minimum:-
first-year nominal income;
-
first-year real income;
-
indexation mechanism;
-
whether indexation tracks actual inflation or a fixed escalator;
-
any cap, floor, lag, or participation limit;
-
projected real income at years 10, 20, and 30 under several inflation scenarios;
-
survival-contingent value;
-
liquidity surrendered;
-
death-benefit or guarantee-period provisions;
-
insurer credit exposure;
-
tax treatment;
-
the cost of replacing the same real cash flows with marketable securities.
Once the table is constructed this way, “highest payout” becomes a much less meaningful headline.
The correct target is a real-income floor, not a replacement-rate slogan
Retirement planning often uses a replacement rate: perhaps a percentage of final salary. This can be a useful heuristic, but it is not a liability specification.
A person earning $200,000 may need far less than 70% of salary if mortgage payments, saving contributions, commuting costs, and work expenses disappear. Another household may require more because retirement introduces travel, care, insurance, or family-support obligations. The important object is not the ratio to pre-retirement salary. It is the real cost of the desired retirement consumption basket.
A stronger process is:
Step 1: Define essential annual consumption in today’s money. Separate genuinely necessary spending from flexible spending.
Step 2: Define the planning horizon as a survival distribution, not a single age. Average life expectancy is useful, but the plan should remain viable in late-life states.
Step 3: Inventory all lifetime income and classify its inflation behaviour. Fully indexed, partially indexed, fixed-escalator, discretionary COLA, variable, or fixed nominal.
Step 4: Convert every nominal promise into scenario-dependent real income. Do not add a $40,000 nominal pension and a $30,000 indexed benefit and call the result a permanent $70,000 real floor.
Step 5: Match the essential gap using the strongest available real hedges. Depending on jurisdiction and product availability, this can include indexed public benefits, inflation-linked bonds, and real or escalating annuity income.
Step 6: Preserve liquidity for uninsurable shocks. The right annuity allocation is not automatically 100%.
Step 7: Allocate growth assets against long-horizon and discretionary liabilities. This is where market risk is most tolerable because spending can adjust.
Step 8: Stress-test the complete plan in real terms. Inflation shocks, market drawdowns, long life, death of a spouse, care costs, and tax changes should be evaluated jointly where possible.
The output should be a time series of real consumption capacity, not simply a projected account balance.
What “stabilized purchasing-power retirement” should mean
The phrase can be made operational.
Let C₀ᴱ be essential annual consumption at retirement, measured in retirement-year dollars. Let Pₜ ÷ P₀ be the cumulative price-level change. The target nominal essential income in year t is:
Target essential income in year t = C₀ᴱ × (Pₜ ÷ P₀)
A retirement system is purchasing-power stable with respect to essential consumption when the available after-tax income allocated to essentials keeps pace with that liability with high probability across the relevant survival horizon.
That definition has several advantages.
First, it makes inflation explicit rather than burying it inside an assumed discount rate.
Second, it separates the real spending target from the financial instruments used to fund it. An annuity, TIPS ladder, public pension, bond portfolio, or investment account is not the objective. Each is a mechanism for financing the objective.
Third, it allows different degrees of protection. A household might require a 95% confidence level for essential real consumption but accept much greater variability in discretionary spending.
Fourth, it makes product limitations visible. A nominal annuity can still be useful, but its contribution to the real floor declines through time unless paired with other assets. A 3% escalating annuity can be modelled against inflation states rather than assumed to be “inflation protected.” A CPI-linked annuity can be credited more directly but still adjusted for index mismatch, taxes, and insurer risk.
Fifth, it makes the cost of safety transparent. Real protection generally has an opportunity cost: lower initial income, lower expected return, reduced liquidity, or some combination. There is no reason to hide that. Retirement is an allocation of risk, not an elimination of risk.
Why full indexation is not automatically the answer to everything
The argument for thinking in real terms should not be exaggerated into the claim that every retiree should buy the maximum possible CPI-linked annuity.
Several qualifications matter.
Spending need not be constant in real terms. Some retirees deliberately spend more in early retirement when they travel and are more active, less in middle retirement, and potentially more again if care needs rise. A perfectly level real annuity may not match that desired trajectory. Davidoff et al. (2003) show more generally that when available annuity payout paths differ from preferred consumption paths, complete annuitization can be suboptimal.
Liquidity has value. An irreversible income stream cannot pay a large unexpected bill unless the regular payment is large enough or other liquid assets exist. Medical expenditure, home repairs, family emergencies, and relocation create reasons to retain marketable wealth.
The relevant inflation index may not match the household. CPI linkage protects against a broad price index, not every personal expenditure basket.
Indexed income usually begins lower. Inflation protection is not free. Soares and Warshawsky (2003) show the lower initial payment trade-off in their annuity comparisons. A retiree with a short expected horizon, strong bequest motive, or large liquid buffer may rationally value a different payment pattern.
Issuer risk remains. An annuity is a claim on an insurance company, subject to the issuer’s financial strength and the applicable regulatory and guaranty framework. Investor.gov and FINRA both emphasize the relevance of the insurer’s claims-paying ability (FINRA, 2022; U.S. Securities and Exchange Commission, n.d.).
Taxes operate in nominal units. The household needs after-tax real purchasing power. The tax treatment of annuity payments, retirement-account withdrawals, public pensions, and inflation-linked securities can materially change the result and varies by jurisdiction. TIPS, for example, can generate taxable inflation adjustments in taxable accounts even before maturity, which is one reason account location matters in implementation.
Growth risk and inflation risk are not identical. A growth portfolio may outperform inflation over long periods but can suffer major drawdowns at exactly the wrong time. It should not be treated as a contractual real-income hedge.
The objective is therefore not “index everything.” It is protect the right liabilities, with the right instruments, to the right degree.
A better retirement-income statement
A conventional retirement statement might say:
You have $1.8 million and projected retirement income of $110,000 a year.
That sounds informative but leaves the essential questions unanswered.
A better statement would say something like:
Your essential retirement budget is $70,000 a year in 2026 purchasing power. Indexed lifetime income covers $42,000. Fixed nominal lifetime income covers $18,000 initially but is expected to contribute progressively less in real terms. A TIPS ladder funds the remaining essential gap through age 85 under CPI-based inflation. A deferred lifetime-income component covers part of the post-85 longevity tail. Liquid reserves cover two years of essential expenditure plus a separate shock allowance. Growth assets finance discretionary spending and provide additional long-term inflation resilience. Under the central scenario, essential real consumption is fully funded. Under the adverse inflation-plus-market scenario, discretionary spending falls first while the real essential floor remains intact.
That is a retirement plan.
The first statement is a wealth report.
The distinction is not cosmetic. It changes what gets measured, what gets hedged, and what counts as failure.
The central mistake: confusing certainty of amount with certainty of value
The most dangerous feature of nominal retirement income is psychological as much as financial: the number does not move.
A $5,000 monthly payment looks stable because the digits remain $5,000. Market portfolios visibly fluctuate; a fixed annuity does not. That visual stability can make the nominal annuity feel safer than an asset whose value moves every day.
But economic safety is not numerical stillness.
If the cost of maintaining the retiree’s lifestyle rises from $5,000 to $8,000 a month while income remains at $5,000, the retirement system has experienced a 37.5% shortfall relative to the required consumption budget even though the payment never fell by a cent.
Bodie and Pesando (1983) framed this problem more than four decades ago: pensioners care about the real value of retirement income, while conventional nominal annuities expose that real value to uncertainty. The fact that the observation is old does not make it obsolete. Current FINRA guidance still identifies inflation as a drawback of fixed immediate annuities because fixed payments may buy less over time (FINRA, 2022).
The appropriate risk dashboard should therefore distinguish:-
nominal income volatility;
-
real income volatility;
-
longevity risk;
-
inflation risk;
-
market risk;
-
credit risk;
-
liquidity risk;
-
household expenditure risk.
A product can be excellent on one axis and poor on another. Calling it “guaranteed” without naming the guaranteed variable is not enough.
The practical rule: retire on purchasing power, not on currency
The simplest rule is also the most important:
Do not decide whether you can retire by asking whether your nominal income is high enough today. Decide whether your portfolio of lifetime income, inflation-linked assets, liquid reserves, and growth assets can sustain the required real consumption floor across the plausible length of retirement.
That means quoting retirement goals in today’s dollars and allowing future nominal income to float upward with the price level. It means treating non-indexed pensions and annuities as declining real assets. It means recognizing that a fixed annual escalator is a forecast embedded in a contract, not full inflation insurance. It means distinguishing mortality credits from inflation protection rather than assuming that an annuity automatically provides both. And it means preserving enough liquidity that the pursuit of lifetime income does not create a new vulnerability to expenditure shocks.
For most retirees, the best solution is unlikely to be a single product. It is more likely to be a layered balance sheet:-
indexed public or private lifetime income for part of the floor;
-
real bonds or equivalent inflation-linked assets for dated essential liabilities;
-
annuitization for the longevity tail where it is efficiently priced and appropriately indexed;
-
liquid reserves for shocks;
-
diversified growth assets for flexible spending, future real growth, and legacy goals.
The percentage in each layer will differ by household. The measurement principle should not.
The retirement income number that matters is not the number printed on the cheque.
It is what the cheque can still buy when the retiree is 75, 85, or 95.
A successful retirement plan is therefore not one that guarantees the same money for life. It is one that has been designed so that life does not quietly become unaffordable while the money remains the same.
References
Board of Governors of the Federal Reserve System. (2025). Statement on longer-run goals and monetary policy strategy. https://www.federalreserve.gov/monetarypolicy/monetary-policy-strategy-tools-and-communications-statement-on-longer-run-goals-monetary-policy-strategy-2025.htm
Bodie, Z., & Pesando, J. E. (1983). Retirement annuity design in an inflationary climate. In Z. Bodie & J. B. Shoven (Eds.), Financial aspects of the United States pension system (pp. 291-324). University of Chicago Press. https://www.nber.org/books-and-chapters/financial-aspects-united-states-pension-system/retirement-annuity-design-inflationary-climate
Davidoff, T., Brown, J. R., & Diamond, P. A. (2003). Annuities and individual welfare (Working Paper 2003-11). Center for Retirement Research at Boston College. https://crr.bc.edu/wp-content/uploads/2003/05/wp_2003-11.pdf
Financial Industry Regulatory Authority. (2022, July 14). Immediate annuities: Money now and for the rest of your life ... for a price. https://www.finra.org/investors/insights/immediate-annuities-money-now-and-rest-your-life
Financial Industry Regulatory Authority. (n.d.). Annuities. https://www.finra.org/investors/investing/investment-products/annuities
Social Security Administration. (2024). Actuarial life table: 2021 period life table, as used in the 2024 Trustees Report. https://www.ssa.gov/oact/STATS/table4c6_2021_TR2024.html
Social Security Administration. (2026). Latest cost-of-living adjustment. https://www.ssa.gov/OACT/COLA/latestCOLA.html
Soares, C., & Warshawsky, M. (2003). Annuity risk: Volatility and inflation exposure in payments from immediate life annuities (Research Paper No. 2003-01). U.S. Department of the Treasury. https://home.treasury.gov/system/files/226/annuity-risk.pdf
U.S. Bureau of Labor Statistics. (2025). Consumer Price Index frequently asked questions. https://www.bls.gov/cpi/questions-and-answers.htm
U.S. Department of the Treasury, Bureau of the Fiscal Service. (n.d.). Treasury Inflation-Protected Securities (TIPS). TreasuryDirect. https://www.treasurydirect.gov/marketable-securities/tips/
U.S. Securities and Exchange Commission. (n.d.). Annuities. Investor.gov. https://www.investor.gov/introduction-investing/investing-basics/investment-products/annuities
This article is for general educational purposes and is not individualized investment, tax, insurance, or legal advice. Retirement-income design depends on jurisdiction, tax status, household spending, health, insurer terms, public benefits, and individual risk preferences. ETc etc blah blah…