The pension promise nobody can keep: why longevity risk has become uninsurable

The mathematics of pension provision rest on a simple bargain: pool enough lives together, estimate when they will die, and the law of large numbers will smooth out the uncertainty. For decades, that bargain held. Actuaries built mortality tables, insurers priced annuities, and defined-benefit schemes promised income for life. But the foundation has shifted beneath them. Life expectancy has risen faster and more persistently than the models predicted, and the institutions designed to absorb longevity risk are discovering that the risk has outgrown their capacity to bear it.

The problem is not merely that people are living longer. It is that they are living longer in ways that correlate across entire cohorts, violating the independence assumption on which risk pooling depends. When a pension fund miscalculates the mortality of one member, the error is noise. When it miscalculates the mortality of an entire generation, the error is systematic, and the losses compound. Longevity risk, in other words, behaves less like the idiosyncratic hazards that insurance was designed to manage and more like a slow-motion financial crisis that unfolds over decades.

Actuarial models have struggled to keep pace. Mortality improvements that were once treated as linear trends have revealed themselves to be more volatile and harder to forecast. Medical advances, behavioural shifts, and socioeconomic factors interact in ways that defy simple extrapolation. The result is a widening gap between the liabilities that pension schemes and insurers have already promised and the assets they hold to meet them. That gap represents a claim on future resources that someone, somewhere, will have to honour.

Interest rates compound the difficulty. Pension obligations are long-dated liabilities, and their present value is exquisitely sensitive to the discount rate applied. When rates were higher, the future cost of a lifetime income stream could be discounted more steeply, making promises affordable. But in an era of lower equilibrium rates, the same promise costs far more to fund today. The twin pressures of longer lives and lower returns have turned what looked like prudent commitments into burdens that strain balance sheets.

Insurers, who might once have stepped in to assume longevity risk through bulk annuity transactions, are now more cautious. The market for longevity transfer has grown, but it remains concentrated and selective. Reinsurers who backstop these transactions face their own capital constraints and are wary of accumulating exposures that could crystallise over half a century. Pricing has become more conservative, and in some segments, capacity has simply withdrawn. The institutions that were supposed to intermediate this risk are signalling, through their actions if not their words, that the risk has become too large to intermediate efficiently.

The retreat is most visible in the retail annuity market. Across several developed economies, the proportion of retirees choosing to convert pension savings into guaranteed lifetime income has declined. Part of this reflects consumer preference and the appeal of flexibility, but part reflects supply. Annuity rates, when adjusted for inflation expectations and prevailing bond yields, have become less attractive. Providers have widened margins to compensate for uncertainty, and some have withdrawn products altogether. The result is a market that offers less protection, at higher cost, to fewer people.

Defined-benefit pension schemes, once the gold standard of retirement security, are in managed decline. Corporate sponsors have closed schemes to new members, frozen accruals, and sought to transfer liabilities off their balance sheets wherever possible. The cost of maintaining these promises has become a drag on competitiveness and a source of intergenerational tension within firms. Public sector schemes, which cannot so easily walk away, face political constraints on contribution increases and benefit cuts, leaving deficits to accumulate. The structural shift from defined benefit to defined contribution is well documented, but its implications for longevity risk are less often examined.

In a defined-contribution world, longevity risk is no longer pooled. Each individual must estimate their own life expectancy and manage their drawdown accordingly. The consequences of getting that estimate wrong are asymmetric: spend too quickly and face destitution in old age; spend too cautiously and die with unspent savings that might have funded a better quality of life. This is not a risk that individuals are well equipped to assess or manage. Behavioural biases, cognitive decline, and simple ignorance conspire to make decumulation one of the hardest financial problems a household will face.

The shift also has distributional consequences. Longevity risk is not evenly spread across the population. Wealthier, better-educated individuals tend to live longer, meaning that any system which socialises longevity risk through taxation or collective provision implicitly transfers resources from shorter-lived to longer-lived groups. Conversely, a system that individualises longevity risk places the heaviest burden on those who live longest, who may also be those with the least financial resilience. There is no neutral way to allocate this risk, and every choice embeds a value judgement about fairness and solidarity.

Some have proposed solutions. Longevity bonds, which pay out more if a reference population lives longer than expected, could allow pension funds and insurers to hedge their exposure. But the market for such instruments remains thin, and governments have shown little appetite to issue them. Pooled annuity funds, which share mortality experience among members without locking in individual pricing, offer a middle ground between full insurance and full self-reliance. Yet take-up has been modest, and regulatory frameworks have been slow to accommodate them.

Another approach is to link pension ages and benefit levels to life expectancy, automatically adjusting the terms of the contract as longevity improves. Several countries have adopted variants of this mechanism, but it is politically contentious. Workers who have planned their retirement around a particular age may resist changes, and the distributional effects can be harsh for those in manual occupations whose healthy life expectancy has not kept pace with overall gains. Automatic adjustment may be actuarially sound, but it is not always socially acceptable.

The deeper question is whether longevity risk can be insured at all, or whether it is simply too large and too systematic to be managed through traditional risk-pooling mechanisms. Insurance works when risks are independent and losses are bounded. Longevity risk is neither. It is a slow-moving, highly correlated exposure that touches every pension fund, every insurer, and every individual simultaneously. The only entities with balance sheets large enough to absorb such risk are governments, and even they face constraints.

Governments, after all, cannot diversify away demographic risk. They can spread the cost across generations through taxation and debt, but they cannot eliminate it. The fiscal pressures of ageing populations are already visible in rising healthcare and pension expenditures, and these pressures will intensify as the baby boom generation moves through retirement. The capacity of the state to backstop private pension failures or to provide adequate safety nets is not unlimited, and the political economy of intergenerational transfers is fraught.

There is also a risk that the retreat from guaranteed income creates a new form of systemic vulnerability. If large numbers of retirees exhaust their savings earlier than expected, the demand for means-tested benefits will rise, shifting costs back onto the public sector in a less efficient and more stigmatised form. The individualisation of longevity risk may simply defer and concentrate the problem rather than solving it.

The strongest counter-argument is that individuals value flexibility and control, and that the shift away from rigid lifetime annuities reflects genuine preference rather than market failure. Many retirees prefer to retain access to their capital, to adjust spending in response to changing needs, and to leave bequests. The paternalism of compulsory annuitisation, in this view, is both inefficient and unjust. Markets have evolved to offer a wider range of products, and individuals are better served by choice than by one-size-fits-all guarantees.

Yet this argument assumes a level of financial literacy and foresight that may not be widespread. It also overlooks the externalities: individuals who underestimate their longevity and exhaust their resources impose costs on families and on the state. The case for some degree of collective provision or regulatory nudge does not rest solely on paternalism but on the recognition that private decisions have public consequences.

What is clear is that the institutional architecture built to manage longevity risk is under strain. Defined-benefit pensions are fading, annuity markets are shrinking, and the tools available to hedge longevity exposure remain underdeveloped. The risk is being pushed back onto individuals at precisely the moment when life expectancy is most uncertain and the financial environment least forgiving. This is not a temporary disruption but a structural shift, and it raises questions that go beyond actuarial technique to the heart of how societies organise security in old age.

The terms of the debate, however, remain largely technical. Pension deficits are discussed as accounting problems, annuity pricing as a matter of market dynamics, and longevity improvements as a demographic footnote. The political and ethical dimensions receive less attention. Who should bear the cost of living longer than expected? What obligations do the young owe to the old, and vice versa? How much security can a society afford, and how much risk should individuals be expected to shoulder alone?

These questions have no easy answers, but they cannot be avoided. The promise of income for life was never costless, and the institutions that made that promise are discovering that the bill is larger than anticipated. As those institutions retreat, the risk does not disappear. It simply changes hands, moving from balance sheets designed to absorb it to individuals who may lack the resources, the information, or the time horizon to manage it effectively. The consequences of that shift will unfold over decades, but the choices that shape it are being made now, often quietly and without full public scrutiny. What remains to be seen is whether the political system will confront the question of longevity risk directly, or whether it will continue to be resolved by default, one underfunded pension and one depleted savings pot at a time.

This article was produced with AI assistance and reviewed against our editorial standards.