Jonas Osman AbdelghafourQuantica Risk Modelling
All articles

Longevity Risk in Pensions: Why It Matters and How to Manage It | Jonas Osman Abdelghafour

July 23, 2026
Longevity Risk in Pensions: Why It Matters and How to Manage It | Jonas Osman Abdelghafour

Longevity risk quietly erodes pension funding. Jonas Osman Abdelghafour, actuary and risk expert, explains how it is measured, why mortality tables mislead, and the de-risking options that actually work.

# Longevity Risk: The Pension Problem That Compounds Quietly

*By Jonas Osman Abdelghafour — Actuary & Risk Expert*

Longevity risk is the risk that people live longer than expected — and that someone has promised to pay them for every one of those extra years. For a defined benefit pension scheme or an annuity writer, it is one of the least dramatic and most persistent risks on the balance sheet. There is no single bad year, no headline event. Just a slow, compounding drift between what was assumed and what actually happens.

That quietness is exactly what makes it dangerous. A one-year increase in life expectancy at retirement typically adds somewhere in the region of 3–5% to the value of pension liabilities, depending on the scheme's maturity and benefit structure. Miss the trend for a decade and the funding gap can be material — and unlike an equity drawdown, it does not recover.

## Why mortality tables mislead

Every pension valuation rests on a mortality assumption, usually built in two parts: a base table describing how likely members are to die at each age today, and an improvement assumption describing how mortality rates will fall in the future.

Both parts can go wrong, but in different ways.

Base tables go wrong through *basis risk*: standard tables describe an average population, and your scheme is not average. Mortality varies sharply by occupation, pension size, postcode, and lifestyle. A scheme of former manual workers and a scheme of retired executives can have life expectancies several years apart, even in the same country and the same industry. Using an off-the-shelf table without adjusting for scheme-specific experience — or without a credibility-weighted blend where experience data is thin — bakes in a systematic error from day one.

Improvement assumptions go wrong through genuine uncertainty about the future. The twentieth century delivered decades of steady mortality improvement, driven by declining smoking rates, cardiovascular medicine, and public health gains. Then, in many developed countries, improvements slowed sharply in the 2010s. Anyone who extrapolated the 2000s trend forward overstated liabilities; anyone who now extrapolates the recent slowdown may be understating them. Improvement is not a physical constant. It responds to medical innovation, health spending, inequality, and — as the pandemic reminded everyone — shocks.

The honest position is that no one knows the long-term trajectory. That is precisely why longevity is a risk to be managed rather than an assumption to be set once and filed away.

## Measuring the exposure

Before managing longevity risk, it pays to quantify it properly. Three approaches, in increasing order of sophistication:

**Sensitivity testing.** Rerun the valuation with life expectancy shifted by one year, or with the improvement assumption moved between published variants. This is cheap, transparent, and answers the trustee-level question: "how much does this matter?"

**Stochastic mortality models.** Models in the Lee-Carter and Cairns-Blake-Dowd families treat mortality improvement as a stochastic process and generate a distribution of outcomes rather than a single path. They are the standard tool for setting longevity stress calibrations and for pricing longevity transactions. Their limitation is worth stating plainly: they extrapolate historical patterns, so they capture trend uncertainty well and structural breaks poorly.

**Scenario analysis.** Complement the models with narrative scenarios — a breakthrough in cancer treatment, widespread adoption of effective anti-obesity drugs, a sustained deterioration in health system capacity. Scenarios will not give you a probability, but they force a conversation about which developments would actually hurt, and how much warning you would get.

A well-run scheme uses all three: sensitivities for communication, stochastic models for calibration, scenarios for imagination.

## The de-risking menu

Pension de-risking has matured into a genuine market, and longevity risk can now be transferred in several ways. The main options, in rough order of completeness:

**Buy-out.** An insurer takes on the full liability — longevity, investment, inflation, everything — and the scheme's obligation is extinguished. This is the endgame for most closed defined benefit schemes, but it requires being well funded on the insurer's pricing basis, which is typically more prudent than the scheme's own.

**Buy-in.** The scheme buys a bulk annuity as an asset: the insurer pays the scheme exactly what the scheme pays covered members. Longevity and investment risk on the covered block are hedged, but the liability stays on the scheme's books. Buy-ins are often used as stepping stones, insuring the pensioner population first while deferred members remain uninsured.

**Longevity swap.** The scheme keeps its assets and pays a fixed leg based on expected mortality; the counterparty pays the floating leg based on actual survival. This isolates longevity risk specifically — useful for large schemes that want to retain their investment strategy. The trade-offs are collateral requirements, counterparty risk, and the fact that pricing embeds the reinsurer's margin on precisely the risk you are offloading.

Which route makes sense depends on funding level, scheme maturity, sponsor covenant, and appetite for residual risk. A poorly funded scheme cannot afford a buy-out and may get better long-term outcomes running longevity risk while closing the funding gap. A well-funded, mature scheme holding longevity risk it is not being rewarded for should ask why.

One practical warning: de-risking transactions are priced off the scheme's data. Incomplete member records, unresolved spouse benefits, and stale mortality experience all translate directly into pricing prudence — which is to say, cost. Cleaning data before approaching the market routinely pays for itself.

## What good management looks like

For schemes and insurers not ready to transact, longevity risk management is mostly discipline: conduct regular experience analyses rather than reusing last valuation's basis; document why the improvement assumption was chosen and what evidence would change it; monitor the gap between assumed and actual deaths as a standing item, not a triennial surprise; and understand the concentration of liability — often a minority of members, typically those with the largest pensions and the longest life expectancies, carry a disproportionate share of the longevity exposure.

## The takeaway

Longevity risk is slow, systematic, and unrewarded — schemes are generally not paid a premium for bearing it, unlike equity or credit risk. It hides inside mortality assumptions that are easy to set carelessly and expensive to get wrong, and it compounds for decades before anyone notices. The tools to measure it are mature, the market to transfer it is deep, and the schemes that handle it well share one habit: they treat mortality as an evolving risk to be monitored and managed, not a static assumption to be filed. The worst outcome is not choosing the wrong hedge — it is discovering, twenty years in, that no one was watching.

---

*Jonas Osman Abdelghafour is an actuary and risk expert who helps insurers, pension schemes, and corporates quantify and manage complex risks — from mortality and longevity analysis to capital modeling and de-risking strategy. If your organization is weighing its longevity exposure or preparing for a risk transfer transaction, contact@jonasosman.org to discuss how he can help.*