Pensions in Bermuda: the cost of bringing risk home

The pensioner is waiting for a bank transfer. The insurer may be waiting for a payment from another insurer, based in Bermuda. Between them sit a reinsurance treaty, an investment portfolio and a set of collateral rights. While the chain works, it attracts little attention. When one link fails, the British insurer must keep paying the annuity for which it remains legally responsible. The International Association of Insurance Supervisors, or IAIS, explicitly states that the original insurer retains that responsibility. ^[5]

On 1 October 2026, InEvo Re announced two completed transactions with new UK insurance clients. The Bermuda reinsurer is part of Macquarie Group. Its release names neither client and discloses neither the amounts nor the treaty terms. The announcement offers an entry point into a growing market, rather than a detailed picture of these two transactions. None of the material examined establishes financial distress at InEvo. ^[1][3]

The economics become most revealing on the return journey. If an insurer must take back the risks it ceded, which assets will it recover, what will they be worth, and when can it use them? A portfolio that looks sufficient on paper may require asset sales, new currency hedges and additional capital before it can once again support the promised pensions. This contingency is central to the current British regulatory debate. ^[8]

The funding can move while the obligation stays

The first transfer often takes place between an employer’s pension scheme and an insurer. In a buy-in, the scheme purchases a policy to finance some or all of its benefits; the trustees retain their obligation to members. In a buyout, individual insurance policies take over benefit payments as part of the scheme’s transfer and winding-up process. Neither step should be confused with reinsurance subsequently purchased by the insurer. ^[14]

Funded reinsurance combines a transfer of insurance risks with a transfer of investment risks. The UK insurer, known as the cedant, pays a substantial premium, sometimes in the form of assets. The reinsurer agrees to fund the benefits covered by the treaty. The exposure can include longevity, the risk of paying pensions for longer, alongside the performance of the investments backing them. A contract covering longevity alone has different economics. ^[5]

The policyholder keeps the entitlement defined by the original insurance contract. On the cedant’s regulatory balance sheet, reinsurance creates an asset representing the amounts expected from the reinsurer. That claim has to be valued and adjusted for the risk of receiving less than promised. Some directly held investment risk has therefore become a contractual exposure to another balance sheet. ^[6]

Assumptions and method

Qualitative diagram with no flow scale or probability. Custody structures are archetypes, rather than reconstructed InEvo treaties. Asset ownership, custody and access rights depend on the contract. The text distinguishes a buy-in from a buyout.

Even the phrase “assets move to Bermuda” needs checking. Some treaties place assets in a separate trust, with the cedant as beneficiary and an independent custodian. Others use funds withheld: the cedant continues to own the assets and hold them on its balance sheet, while the contract allocates their economic performance. The reinsurer’s domicile, ownership of the securities and their custody location are different pieces of information. ^[5]

That distinction changes crisis planning. Taking over management of an existing portfolio and enforcing rights over assets held elsewhere involve different procedures. Lawyers examine access clauses, treasury teams look at payment dates, and investment staff assess the assets they may have to manage. Collateral becomes operationally useful when all three assessments produce a workable timetable. ^[8]

The economics for each party

For the insurer, reinsurance can reduce retained risks, release capital and support a more competitive quote to a pension scheme. For the reinsurer, it provides a long-duration portfolio and an opportunity to earn a return on equity from investment performance after benefits, expenses and losses. These are among the economic benefits identified by the IAIS. ^[5]

An asset manager adds another layer. It can supply long-term loans and investments that a smaller insurer might struggle to source independently, while earning management or origination fees under the relevant mandates. Those fees are distinct from the residual return earned by the reinsurer’s shareholder. Their precise allocation at InEvo is not disclosed in the announcements examined. ^[5][2]

When Macquarie announced InEvo’s first UK transaction in March 2025, it highlighted its investment platform and coverage of both pensions already in payment and deferred members. That reflects a genuine industrial task: building an investment portfolio whose cash flows can support benefits for decades. Its success then depends on asset prices, credit performance and the contractual promises. ^[2]

Regulatory capital also shapes the bargain. In its consultation of 29 April 2026, the Prudential Regulation Authority, or PRA, estimates that the average existing funded reinsurance transaction involves capital equivalent to 2–4% of the underlying annuity liabilities, compared with 11–15% for similar directly held investments. Its proposals would move the corresponding figure to around 10%. These are the PRA’s indicative estimates within the scope of its economic comparison. ^[6]

Assumptions and method

The PRA’s indicative estimates in CP8/26, paragraph 3.5, published on 29 April 2026. Capital relative to underlying annuity liabilities: currently 2–4% for an average existing transaction; around 10% under the proposals; 11–15% for similar directly held investments. The proposal primarily changes the value of the reinsurance recoverable; it does not introduce a new uniform SCR charge.

Ten per cent is therefore not a uniform capital rate that would be applied to every treaty. The proposed reform principally changes the recognised value of the reinsurance asset. A more prudent allowance reduces the resources recognised on the balance sheet. The PRA explicitly says this consultation does not propose new rules for calculating the solvency capital requirement, or SCR, associated with funded reinsurance. Asset valuation and the capital requirement remain separate. ^[6]

The simplified balance-sheet logic is straightforward. Lower the recognised value of an asset while keeping liabilities unchanged, and the surplus of assets over liabilities shrinks. The transaction’s initial benefit becomes smaller. An insurer might respond by negotiating stronger collateral, accepting a lower margin, or holding more investments directly. The commercial effect depends on the negotiation; the regulatory proposal cannot tell us the future price of an InEvo treaty.

Collateral must survive its provider

Collateral gives the cedant an identifiable resource if the reinsurer stops performing. The assets must also be separated from other creditors, accessible, and suitable for the payments ahead. Existing PRA expectations already address legal rights, investment restrictions, valuation, recapture costs and plans for managing recovered assets. ^[8]

Recapture means the insurer takes back the ceded risks. Treaty-defined triggers can activate it before formal insolvency. Waiting for a breach to be cured could allow the collateral to deteriorate; acting earlier can bring forward transition costs. A clause’s effectiveness therefore depends on the decision to exercise it as well as its wording. ^[8]

Substitution rights are especially important. A highly liquid opening portfolio may change if the contract permits different assets to replace it. The PRA expects insurers to consider the worst portfolio permitted by the contractual limits, rather than assume the opening snapshot lasts forever. It also expects recapture models to avoid relying on unrealistic collateral top-ups from an already troubled counterparty. ^[8]

Consider a wholly hypothetical model, unrelated to InEvo’s treaties. Liabilities valued at £1 billion are backed by £1.05 billion of assets, giving initial coverage of 105%. The portfolio contains £420 million of liquid assets and £630 million of less-liquid private assets. Assume valuation declines of 5% and 20%, respectively, followed by £15 million of recapture costs. Net recoverable resources fall to £888 million, leaving a £112 million gap against the original liability valuation.

Assumptions and method

Hypothetical model in £m: liabilities held at 1,000; initial assets 1,050. Portfolio A: 420 liquid and 630 private. Portfolio B: 945 liquid and 105 private. Net value after recapture = liquid assets × 0.95 + private assets × 0.80 − 15 of costs. Results: 888 and 966.75. Resource gap = 1,000 − net value: 112 and 33.25. Excludes liability responses to rates, hedges, reinsurer top-ups and additional unsecured recovery.

A second portfolio starts with exactly the same coverage, but holds 90% liquid assets and 10% private assets. Under the same category-specific shocks and the same costs, it returns £966.75 million. The gap is £33.25 million. Asset composition alone explains the difference in this experiment. The opening coverage ratio concealed it.

The model deliberately holds liability value constant. In practice, that value would also respond to interest rates and actuarial assumptions, while hedges could soften or amplify the outcome. The assumed price falls are neither forecasts nor default probabilities. The calculation excludes future collateral top-ups and any additional unsecured recovery from the reinsurer’s estate. It isolates the net value of resources after recapture; it does not predict a permanent loss for pensioners.

Full coverage can still arrive late

Liquidity asks a different question. Suppose this time that collateral value remains sufficient, but effective access takes 75 days. The insurer has £8 million of free cash and must pay £5 million every thirty days, on days 30, 60, 90, 120 and thereafter. At the second payment date, it is £2 million short. The portfolio may ultimately be recovered in full, yet a bridge is needed on day 60.

This independent simulation describes no legal or contractual timetable at InEvo. It illustrates the practical work of a recapture plan: ensuring pensions can be paid during notices, custody transfer, asset mobilisation and any rebuilding of hedges. If access were delayed until day 105, the same assumptions would require a £7 million bridge.

Assumptions and method

Independent simulation in £m: initial free cash 8; payments of 5 every 30 days, on D30, D60, D90, D120 and thereafter. Collateral access is assumed before any payment on the same day, with sufficient cash mobilisation afterwards. Peak bridge before access = max(0, 5 × number of payments before access − 8). Access on D45, D75, D105 and D135 requires bridges of 0, 2, 7 and 12. The main line stops at D75. No actual InEvo legal or contractual delay is estimated.

Private assets can make sense against long-dated liabilities. Holding them allows an insurer to earn compensation for accepting illiquidity. The difficulty arises when a forced timetable requires sales or the replacement of hedges whose continuity depended on the reinsurer. Valuation risk and cash needs can then reinforce one another. These are precisely the constraints the PRA expects cedants to address in their recapture planning. ^[8][15]

Dollar assets introduce another check. An investment can be sound and pay on schedule while backing a sterling pension. The cedant needs to know who holds the currency hedge, how it transfers and which margin calls it could generate. None of those questions is answered by the reinsurer’s domicile. ^[8][9]

Returning assets may receive different regulatory treatment

Under UK rules, an insurer can obtain permission to use the matching adjustment. Subject to conditions, this recognises its ability to hold investments whose cash flows match the promised benefits. After allowing for retained risks, it raises the discount rate used to value eligible liabilities and reduces their present value. The aim is to prevent every temporary change in market prices from forcing an excessive response by a long-term investor. ^[9]

Recovered collateral must meet the requirements applicable to the receiving insurer. A promise of full collateralisation does not establish whether the assets will fit its existing permissions. The cedant might have to rebuild the portfolio or recognise a larger liability value. That creates another reason to negotiate collateral content before signing. ^[8][16]

A third teaching model isolates the valuation effect. Assume £60 million is paid at the end of each year for 25 years. Discounted at 4.5%, the stream is worth about £889.7 million today. At 4%, it is worth £937.3 million, an increase of £47.6 million. The payment schedule and its nominal total remain unchanged: £1.5 billion over 25 years.

Assumptions and method

Independent model: £60m at each year-end for 25 years, totalling £1.5bn nominal. Present value = sum of 60 / (1 + r)^t for t from 1 to 25. Constant rates of 4.5% and 4% give 889.6925 and 937.3248 million; difference 47.6323 million. Excludes mortality, inflation, expenses and regulatory yield curves. The 50 basis points are illustrative; this does not calculate a treaty’s regulatory matching adjustment and is not added to the other models.

The example uses a constant discount rate, with no mortality, inflation, expenses or regulatory term structure. Fifty basis points is an assumption, not an estimate of a treaty’s matching adjustment. The extra liability value is not an immediate £47.6 million payment to a pensioner. It shows why a balance sheet can require more resources even when the next pension payment is unchanged. It must not be added to the first model’s collateral shortfall as though the two calculations described one transaction.

The British stress test also offers reassurance

Supervisors have already tested actual portfolios. In the funded reinsurance component of the 2025 Life Insurance Stress Test, the PRA required relevant insurers to recapture arrangements with their most adverse counterparty after the core financial shock. The reference exposures were those held on 31 December 2024. Only firms with material funded reinsurance exposures took part in this component. ^[9]

The simulated recapture covers £12.3 billion of liabilities, roughly half of those firms’ aggregate funded reinsurance exposure. For the same subset, aggregate SCR coverage falls from 154% after the core stress to 144% after recapture. The PRA reports an approximately 6% increase in required capital and a roughly 1% decline in eligible own funds. Much of the ratio’s movement therefore comes from risks returning to the balance sheet. ^[9]

Assumptions and method

LIST 2025 Annex 2, published on 17 November 2025. Reference exposures at 31 December 2024; subset of insurers with material funded reinsurance exposure. The scenario recaptures £12.3bn of liabilities. Eligible own funds / SCR: 154% after the core stress, then 144% after recapture. Changes of approximately +6% in SCR and −1% in own funds are rounded and cannot produce an exact decomposition. Year-end 2024 collateral composition; excludes future substitution and a currency shock.

SCR coverage compares eligible own funds with the solvency capital requirement. It does not mean that 144% of future pensions have been funded. Likewise, £12.3 billion is the amount of liabilities recaptured in a scenario, not assets lost. The result supports the conclusion that participants were resilient at the tested exposure levels. It provides a useful check against treating every recapture as a failure. ^[9]

That reassuring outcome has specific conditions. Most recovered collateral was assessed as eligible for the matching adjustment. The exercise used the end-2024 investment mix, without future deterioration through substitution. It imposed no currency shock, despite a substantial share of collateral being dollar-denominated. The PRA itself sets out these limits. They matter when extrapolating to larger future transactions or to less-liquid portfolios. ^[9]

Bermuda has a supervisor, and InEvo has a rating

Jurisdiction and contract quality deserve separate analysis. The Bermuda Monetary Authority, or BMA, supervises local reinsurers. Its 2025 stress exercise, published in September that year, found adequate sector capital buffers and resilience for a majority of firms before corrective actions. That is the Bermuda supervisor’s conclusion within its exercise’s scope and assumptions. It is not a guarantee for an individual treaty. ^[10]

On 20 August 2026, AM Best assigned InEvo an A- financial strength rating, with a stable outlook. The agency highlighted balance-sheet strength and capital support from Macquarie, while assessing the new entrant’s business profile as limited. This independent opinion provides a counterweight to an alarmist reading. The support described by the agency does not by itself establish an unconditional parent guarantee of every obligation under every treaty. ^[3]

InEvo advertises ring-fenced structures, automatic novation mechanisms and transparent reporting. Those are potentially useful protections. Without the signed treaties, it is impossible to establish which features apply to the two October transactions, exactly what activates them, or how quickly assets become usable. ^[4][1]

The issue also extends beyond Bermuda. In the Netherlands, DNB requires prior consent for asset-intensive reinsurance contracts that permit assets to be held outside the European Economic Area. The assessment includes what a contract may allow during its lifetime, including through retrocession. An EEA reinsurer can therefore be part of a chain whose assets end up elsewhere. Relevant geography follows the investments and the legal rights. ^[12]

The significance of 30 September

The British consultation draws an important time boundary. The PRA proposes to apply the new calculation methods from 1 July 2027, while excluding arrangements under which all covered risks were fully transferred by 30 September 2026. Some intragroup arrangements and temporary reinsurance preceding a legal business transfer would also receive conditional exceptions. ^[6][11]

InEvo’s announcement is dated 1 October. It would be tempting to infer the regulatory treatment of its two treaties from that date. The communication date cannot establish their effective risk-transfer date, which was not disclosed. As of 11 October 2026, this research had not located a final policy text replacing CP8/26’s proposals. The article distinguishes the proposal from supervisory expectations already in force. ^[1][6][8]

Assumptions and method

Documentary timeline, intervals not to scale. CP8/26 published on 29 April 2026; all risks fully transferred by 30 September 2026 for the proposed saving provision; InEvo announcement on 1 October 2026; proposed implementation on 1 July 2027. The treatment of the two transactions cannot be determined without their effective transfer dates. Research closed on 11 October 2026 found no final policy replacing CP8/26.

Collateral quality would enter the proposed calculation. After legal access and segregation conditions are met, the PRA would permit up to three upward credit notches for specified protections: adequate opening coverage, matching cash flows and matching-adjustment eligibility, and collateral credit quality superior to the counterparty’s. Several tests use the worst portfolio contractually allowed. A restriction on investment composition can therefore have an explicit economic value. ^[11][16]

Stronger restrictions also limit the reinsurer’s freedom to seek investment returns. The treaty price must distribute that trade-off: an advantage for the cedant today, against the quality of protection retained for tomorrow. The lowest opening quote may be less attractive after allowing for standby liquidity, the capital needed on recapture and the operational resources that must be maintained. Quantifying the comparison requires the treaties and a complete balance sheet; InEvo’s public announcements do not provide them.

The pension must continue without the partner

The evidence establishes a transformation of risk, with genuine potential benefits: diversified capital, access to investments and greater capacity to fund long-term promises. It also reveals a common exposure when several insurers delegate to counterparties invested in similar credit markets. In a shared shock, the need to recover collateral can coincide with falling asset values and reduced market capacity to absorb sales. That is the concentration scenario the PRA is addressing. ^[5][6]

For beneficiaries, the contractual obligation remains with their direct insurer. The UK Financial Services Compensation Scheme can protect benefits under eligible long-term insurance contracts at 100%, without an upper cap, where its conditions are met. This conditional protection deserves mention. It does not remove the need for insurers and supervisors to plan payment continuity. ^[13]

A useful treaty analysis starts with a simple exercise: remove the reinsurer from the payment chain, retain every pension due date, and verify the receipts needed to meet it. Are the assets there? Do the legal rights make them usable? Will the portfolio still support benefits under the cedant’s prudential rules? The value of the return journey rests on those answers.

Life insurers, retirement savings and private credit in Bermuda follows insurers and asset managers within this financial chain. The guide Reading private-credit risk examines credit quality, valuation and liquidity in the assets that may return to an insurer’s balance sheet.

Sources and method

Research closed on 11 October 2026. Supervisory figures retain their original periods and scopes. The three hypothetical models use current pounds and remain independent; formulas and limitations appear beneath the charts. None describes an InEvo treaty. CP8/26 and its appendices are proposals, distinct from the already applicable SS5/24. Company announcements are attributed to their issuers; the contracts, amounts and effective transfer dates of the two October transactions are not disclosed in the documents reviewed.