UK pension risk transfer is projected at £70 billion for 2026, and the buyout market still cannot absorb the largest defined benefit plans. A longevity swap moves mortality risk without moving assets or triggering settlement accounting, which is why WTW projects up to £20 billion of UK longevity swap volume this year. What it does not move is where the actuarial work sits.

£20B
UK longevity swap volume anticipated for 2026, up from prior years, per WTW projection
£6B
BBC Pension Scheme longevity swap with Zurich and MetLife, November 2025, covering 21,000 members
$48.7B
US pension risk transfer volume in 2025, the market into which longevity swaps are now expanding (Aon)

Key Takeaways

  • A £15 billion plan has no buyout counterparty. No UK bulk annuity insurer has executed a single transaction at that scale, and the constraint is regulatory capital and asset origination rather than a temporary shortage.
  • $48.7 billion was the entire US pension risk transfer market in 2025, so a $10 billion plan seeking full buyout is asking one carrier to absorb roughly 20 percent of annual volume in one deal.
  • The BBC Pension Scheme hedged £6 billion across 21,000 members with Zurich as counterparty and MetLife behind it, extending a £3 billion 2020 swap so that nearly all pensioner and dependant liabilities are now covered.
  • Basis risk is the unhedged residual, and on a £15 billion liability a systematic divergence of a few basis points in annual mortality improvement is tens of millions of pounds a year.
  • CMI 2024 adds roughly three months of male life expectancy at age 65 over CMI 2023, so a swap written on an older model carries a model-basis mismatch against current market pricing.

Why the Largest Plans Cannot Buy Out

LCP projects £40-55 billion of UK buy-in volume for 2026, potentially passing the 2023 record of £49.1 billion, and WTW puts total UK pension risk transfer at £70 billion. For schemes between £500 million and £5 billion those projections hold. At £15 billion they do not.

A single £15 billion buyout would require the purchasing insurer to hold Solvency II capital against the whole annuity portfolio, originate matching assets in illiquid credit, infrastructure debt and private placements to satisfy the matching adjustment, and convert an entire scheme to direct annuity payment for tens of thousands of members. No UK bulk annuity insurer has done that in one transaction, and the reason is structural rather than cyclical.

The US version of the ceiling sits at a different number. Aon put total US pension risk transfer at $48.7 billion for 2025. A $10 billion plan seeking a full buyout is asking a single carrier to take roughly 20 percent of the year's market in one deal at competitive pricing, and no carrier routinely quotes that concentration.

The BBC Pension Scheme shows what the alternative looks like at scale: £6 billion of liabilities across 21,000 members, Zurich Assurance facing the scheme and MetLife assuming the reinsurance behind it, extending a £3 billion 2020 swap with the same structure.

What Transfers, What Stays, and What It Costs to Keep

The contract settles two legs against an agreed reference population. The fixed leg is expected benefit payments derived from the mortality table and improvement scale locked at inception, paid by the counterparty to the plan. The floating leg is actual benefit payments to members in the reference population, paid by the plan to the counterparty. Settlement is quarterly or semi-annual on the netted difference: members living longer than the table projected means the counterparty pays.

What stays is the point. The investment portfolio, the investment management agreement, the funding level, the contribution schedule and the sponsor relationship all continue unchanged, and no settlement accounting is triggered under ASC 715 or FRS 102. The plan remains the benefit obligor and has hedged the cost of that obligation, not discharged it.

The cost of keeping the obligation is basis risk. Every swap is written against a reference table, typically the S-series insured pensioner tables such as S1PMA or the PNXA 08 annuitant tables with the CMI improvement model in the UK, and MP-2021 in the US. No standard table matches a specific membership. A plan concentrated in heavy manual industry experiences materially heavier mortality than S1PMA implies; a white-collar, higher-income membership experiences lighter.

That divergence does not average out, because it is a feature of the plan's demographic composition rather than noise. RGA's work on US mortality improvement documents the same finding structurally: DB pensioners are not the general population, and MP-2021 is calibrated to the national population rather than the pensioner subset. On a £15 billion liability, a systematic gap of a few basis points in annual improvement is tens of millions of pounds of unhedged exposure a year, which is why reference population selection, data cleansing and socioeconomic classification carry most of the pre-execution effort.

The residual buys something back at the eventual buyout. A plan arriving with five years of quarterly settlements showing actual mortality tracking the reference table within a defined range gives the quoting insurer observed behaviour rather than a projection, which narrows the uncertainty load carriers embed in buyout margins. Milliman's 100 Pension Funding Index closed 2025 at a $98 billion aggregate surplus and a 108.1 percent funded ratio, which is the condition under which that sequencing is available and an equity drawdown is the thing that closes it.

The Reference Model Is Not a Fixed Point Either

A swap fixes the plan's mortality assumption at execution. It does not fix the market's.

Moving from CMI 2023 to CMI 2024 raises projected life expectancy by roughly three months for males and one month for females at age 65, and that flows straight into the fixed leg of any UK swap written under the new model. Counterparties price to the current model, so a scheme holding a swap written under CMI 2021 or CMI 2022 carries a model-basis mismatch against current assumptions, narrowing only as settled data accumulates.

The two frameworks are not the same shape. CMI imposes an age-period-cohort decomposition on historical improvements, while the SOA's MIM-2021 fits history more flexibly and applies convergence periods that do not vary by age, where CMI's do. A UK swap on CMI 2024 and a US swap on MP-2021 therefore build in structurally different assumptions about how fast current improvement converges to long-term trend, and in a cross-border structure where a UK primary cedes to a US or Bermuda reinsurer, that difference has to be reconciled explicitly rather than negotiated from a menu.

Two regulatory changes move the pricing underneath it. The NAIC Longevity Risk Subgroup is finalising a C-2 RBC charge separating retained and ceded longevity exposure, targeting a year-end 2027 effective date, with counterparty factor tables for ceded risk and a retained scenario stress that raises required capital for insurers holding the exposure directly. That capital cost reaches pricing once it takes effect, so current US quotes do not yet carry it.

The IRS 2027 DB mortality table update does the same from the liability side, resetting the baseline a swap is priced against. A plan holding a swap written on an older mortality basis then has a second layer of basis risk, this time between its own actuarial assumption and the reference population it hedged to, which is the one form of basis risk the pre-execution work cannot anticipate.

Further Reading

Sources