Longevity risk: planning for living longer than expected

The risk that gets worse the better things go

Most financial risks are things you would prefer not to happen. A market fall, a job loss, an illness. Longevity risk is different, and the difference is what makes it so badly handled.

Longevity risk is the risk of living longer than your money lasts. The bad outcome — running out — is caused by the good outcome — living a long time. You cannot hedge it by hoping for the best, because the best is the thing that triggers it.

It is also the risk with the least dramatic warning. Markets crash visibly. Longevity arrives one birthday at a time. By the point the shortfall is undeniable you are typically in your eighties, no longer able to return to work, and with the fewest remaining options of your life.

That combination — bad outcome caused by good outcome, no early signal, and no recovery path — is why longevity deserves separate treatment rather than being folded into a general "will I have enough" question.

Why life expectancy is the wrong number

Almost everyone who has ever planned for retirement has used the wrong statistic, and they used it in good faith.

Published life expectancy is usually **life expectancy at birth**, computed from a period life table: it takes the mortality rates observed across all ages in a given year and asks what would happen to a hypothetical person experiencing all of them in sequence Source^source. It is a useful summary of population health. It is a poor planning input for an individual, for three separate reasons.

Reason one: it is an average, and averages have halves

If life expectancy at a given age is 85, that means roughly half of the people at that age will live beyond 85. Planning to 85 is planning for a scenario you have roughly a fifty percent chance of exceeding. In no other area of finance would anyone accept a plan with a coin-flip failure rate and call it prudent.

The distribution matters more than its centre. Deaths are spread across a wide range of ages, with a long right tail. That tail is where the financial damage sits, and it is invisible if you only look at the mean.

Reason two: conditional life expectancy rises as you age

This is the point that surprises people most, and it is pure arithmetic rather than anything mysterious.

Life expectancy at birth is dragged down by everyone who dies young. If you have already reached 65, you are no longer in the pool that might have died at 30. Your remaining life expectancy is calculated only among people who also reached 65 — and it is therefore higher than the birth figure predicted for you.

In many countries the pattern looks roughly like this. Life expectancy at birth might be in the high seventies or low eighties. But the expected age at death for someone who has already reached 65 is typically several years higher, and for someone who has reached 80 it is higher again. Every year you survive, your expected final age moves further away from you.

The practical consequence: using the birth figure as a retirement horizon systematically understates how long the money must last, and the error compounds precisely for the people who need the money most.

Reason three: it is a population figure, not a personal one

Measured life expectancy varies substantially between countries and has risen materially over recent decades Source^source. It also varies within countries by income, education, occupation and health status. National averages blend all of that together.

Someone with good health, no smoking history, a family pattern of long lives and access to good healthcare has a materially different distribution from the national average. Someone in the opposite position has a different one again. Neither should use the national number without adjustment.

None of this means you can predict your own lifespan. You cannot. The point is not to find a more accurate personal forecast — it is to stop planning to a central estimate at all, and start planning to a conservative percentile.

Planning age is not expected age

The cleanest fix is to separate two concepts that get conflated.

**Expected age** is your best guess at how long you will live. It is a forecast.

**Planning age** is the age your money is designed to last to. It is a design decision, and it should be deliberately pessimistic — because the cost of the two errors is wildly asymmetric.

If you plan to 100 and die at 82, the consequence is that you spent somewhat less than you could have and left a larger estate. That is a real cost, and it is worth taking seriously — under-spending out of fear is a genuine failure mode that shortens the enjoyable part of a life.

If you plan to 82 and live to 96, the consequence is fourteen years of dependency, at an age with no remaining ability to earn.

These are not symmetric. Most people, offered the choice consciously, prefer the first error. Most people, planning unconsciously to an average, choose the second by default.

A stress-test ladder

Rather than picking one planning age, test several and look at how the answer changes.

1. **Plan to expected age.** Establish the baseline.
2. **Add five years.** How much does the required capital rise? If it rises modestly, your plan is robust. If it rises sharply, you are near a cliff.
3. **Add ten years.** This is roughly the range that separates a central estimate from a conservative percentile for many people.
4. **Add ten years and reduce the assumed real return by two percentage points.** Longevity risk and poor returns are not independent problems — they compound, and the combination is the realistic bad case rather than either alone.
5. **Add ten years, reduce the return, and raise healthcare costs.** This is the scenario worth knowing about, not because it is likely, but because it tells you which lever you would need to pull if it happened.

What matters is not the precise output at each step. It is the shape of the curve. A plan whose required capital rises gently as you extend the horizon is structurally different from one that rises steeply. The second one is fragile, and knowing that early is worth more than any refinement of the central estimate.

The one mechanism that genuinely transfers longevity risk

You can manage most financial risks by diversifying, hedging or self-insuring. Longevity is unusual because for an individual there is no diversification available — you have exactly one life, and it is a single draw from the distribution.

There is, however, a mechanism that works at the group level: **pooling**.

If a thousand people each set aside money and agree that those who die early leave their remaining share to those who live long, then the group as a whole faces only the average outcome, which is far more predictable than any individual's. Each person converts an unknowable individual risk into a known group average. This is the basic logic behind life annuities, defined benefit pensions and state pension systems.

The trade is explicit and worth stating honestly:

• **What you get:** income that continues for as long as you live, regardless of how long that is. The uncertainty is transferred to the pool and its provider.
• **What you give up:** access to the capital, most or all of the ability to leave it to heirs, and any upside from investing it yourself. You also take on the credit risk of the provider, which for a multi-decade promise is not trivial.
• **What determines whether it is worth it:** the terms offered, your health, your other sources of guaranteed income, your family circumstances and how much you value flexibility.

Pooled arrangements are not available or attractive in every market, and their treatment varies enormously by jurisdiction. Many people in the Gulf have limited access to local annuity markets and may have partial entitlements in a home country instead. The relevant point is structural rather than product-specific: **pooling is the only mechanism that actually removes individual longevity risk. Everything else manages it.**

Self-insurance and what it costs

The alternative to pooling is holding enough capital to survive the long tail yourself. That works, but it is expensive in a specific way: you must hold capital sized for a scenario that probably will not happen, and the capital held against that scenario is unavailable for anything else.

Roughly speaking, self-insuring against the tail means funding to a high percentile rather than the median, and the extra capital required for that is the implicit premium you pay for keeping control and preserving a bequest. It is a legitimate choice. It is just not a free one, and people who dismiss annuitisation as "poor value" often have not priced what self-insurance costs them.

Floor and upside

The most useful structural idea in this area is to stop asking "how much do I need" and start asking "which spending is non-negotiable".

**The floor** is the income that covers costs you cannot cut: housing, food, utilities, basic healthcare, insurance. This spending must continue for as long as you live, which means it should ideally be matched with income that also continues for as long as you live — pooled income, state pension, or a very conservative capital reserve explicitly ring-fenced for the purpose.

**The upside** is everything above the floor: travel, gifts, discretionary spending, supporting family. This can be funded from a portfolio that fluctuates, because if it has a bad decade you can spend less without a crisis.

The reason this matters for longevity specifically: once the floor is secured for life, longevity risk stops being an existential threat and becomes a question of comfort. Living to 98 with a secured floor means a longer period of modest living. Living to 98 without one means dependency.

Building the floor first also changes how much risk the rest of the money should take. A portfolio funding discretionary spending can tolerate volatility that a portfolio funding rent cannot.

The expatriate complication

For someone working outside their home country, the floor question is unusually difficult, and it deserves naming.

• Home-country state pension entitlements may be partial, because contribution years were missed while working abroad.
• Employment-linked end-of-service entitlements are typically lump sums rather than lifetime income, which means they do not address longevity at all — a lump sum can be exhausted, a lifetime income cannot.
• The country of eventual residence in old age may not be the country of employment, which introduces currency risk on a liability that runs for decades.
• Access to healthcare systems in later life may depend on residency status, contribution history or visa category, none of which is guaranteed to persist after employment ends.

The practical response is to map, explicitly and in writing, which sources of income will still exist at 75, at 85 and at 95, in which currency, and under which residency assumption. Most people discover that the list at 85 is much shorter than the list at 65, and that discovery is more useful than any calculation.

Six late-life spending categories that behave differently

General inflation is not the right assumption for every part of a retirement budget. These six behave distinctively and are worth modelling separately.

• **Healthcare.** Tends to rise in real terms and to grow as a share of spending with age. It is also the category with the widest variance between individuals.
• **Long-term care.** Costs for assistance with daily living are a distinct component of late-life spending and vary enormously depending on setting and country Source^source. The distribution is highly skewed: most people need little or none, a minority need a great deal for many years. That skew is why it resembles an insurable risk rather than a budget line.
• **Housing.** Often stable or falling if a home is owned outright, but rising sharply if renting, and potentially replaced entirely by care costs later.
• **Transport.** Usually declines with age as driving and commuting reduce, then can partially reverse if assisted transport becomes necessary.
• **Travel and discretionary.** Typically front-loaded in early retirement and declining thereafter, which offsets part of the healthcare increase.
• **Support for others.** Frequently omitted from projections entirely, and frequently significant — adult children, elderly parents and extended family obligations do not stop at retirement.

The composite of these is why real retirement spending in many households follows a curve rather than a straight line: relatively high early, lower in the middle, and rising again late. Assuming flat real spending for thirty years is a simplification, and it is not always a conservative one.

Common mistakes worth avoiding

• **Using life expectancy at birth as the planning horizon.** Use remaining life expectancy at your current age, then add a margin for the percentile you actually want to fund.
• **Planning to a single age for a couple.** For two people, the relevant horizon is when the _second_ person dies, which is materially later than either individual expectation. Joint planning horizons are longer than single ones, sometimes by several years.
• **Treating a lump sum as if it solved the problem.** Capital without a spending rule is not income. The longevity question is not "how big is the balance" but "how long can it sustain the required withdrawal".
• **Assuming you will simply spend less if things go badly.** Some spending is genuinely flexible and some is not. Assuming flexibility that does not exist is how plans fail quietly.
• **Assuming you can keep working.** Ability to work in later life is itself uncertain and correlates with health, which correlates with the very scenarios that stress the plan.
• **Ignoring the second-order effect on the survivor.** When one member of a couple dies, some pension income may reduce or stop while household costs do not fall proportionately. The survivor's position is often worse than the couple's, and often unexamined.

What to take away

Longevity risk is not a reason for anxiety. It is a reason for specificity.

Three things change a plan materially:

1. **Replace the average with a percentile.** Decide explicitly how conservative you want the planning age to be, and know that planning to the average is a coin flip.
2. **Separate the floor from the upside.** Secure lifetime income for spending you cannot cut, and let the rest carry the risk. This single structural change converts a potentially existential problem into a manageable one.
3. **Stress the plan in combination, not in isolation.** Long life alone is survivable. Long life plus poor returns plus high healthcare costs is the scenario that breaks plans, and it is the one worth knowing the shape of in advance.

None of this is advice about specific products, and this article does not recommend any. It is a framework for asking better questions of whatever arrangements are available to you, in whichever jurisdiction you will actually grow old in. The single most common error in this area is not choosing the wrong instrument. It is planning to a number that half of people exceed, and never noticing.