The asset behind the counterparty

David Walker's recent piece on the rating agency perspective ("UK funded re: The rating agency perspective," 3 June) captured an important shift: with CP8/26, the reinsurer's insurer financial strength rating moves to the centre of how cedants will calculate the capital they hold against funded reinsurance. Will Keen-Tomlinson of Moody's set out the economic logic plainly — the reinsurer is the cedant's counterparty, the cedant is the insured, so the IFSR is the natural anchor. For traditional reinsurance, that logic is sound. For funded reinsurance, I want to suggest, it is sound but incomplete, and the incompleteness is worth examining while the consultation is open.

The proposal works through the Counterparty Default Adjustment. Because, as the PRA acknowledges, there is no established credit rating methodology for funded reinsurance itself, the credit quality step feeding the CDA is to be based on the reinsurer's externally assigned IFS rating — with the CDA then set equal to the fundamental spread for financial corporate bonds of that credit quality step and maturity, the explicit aim being to treat reinsurer default like a corporate default on a directly held bond. The headline effect is a substantial increase in the capital a cedant holds, from roughly 2–4% of the ceded liabilities to around 10%. The direction is correct and overdue — the PRA is right that the present treatment understates the risk and has driven the market's growth, with around 15% of recent new bulk annuity business ceded this way and exposures that could climb from some £40 billion toward £110 billion over the coming decade. But the mechanism deserves scrutiny, because it determines not just how much capital is held, but against what.

The IFSR is a derived asset rating

Start with what the IFSR represents in this context, because funded re differs from traditional reinsurance in a way that matters.

Traditional reinsurance transfers a risk to a reinsurer whose ability to pay is supported by a diversified book and a broad capital base. The IFSR there captures something close to genuine entity-level financial strength. Funded reinsurance is different in kind: it transfers a defined pool of liabilities to a vehicle whose purpose is to hold the assets backing those liabilities. The reinsurer's solvency is, to a first approximation, a function of how that asset portfolio performs. The IFSR is therefore not an independent input. It is a derived rating of the asset pool, expressed at the entity level — and it carries the properties of the underlying assets rather than those of a diversified counterparty.

Two facts compound this. Most funded re counterparties of UK cedants are Bermuda entities, outside the Solvency UK perimeter; the PRA cannot apply its own capital standard to them. And the IFSR is not itself a regulatory solvency measure but a rating agency's opinion, produced on the agency's own model, informed by the entity's disclosures but distinct from any statutory capital figure. The CDA approach therefore calibrates cedant capital to an agency opinion that is standing in for a regulatory assessment the offshore structure was, in part, chosen to avoid.

That substitution is defensible — the PRA has no better cross-jurisdictional measure to hand — but it has a cost, and naming the cost is the purpose of this article. Anchoring capital to the IFSR is what one might call proxy-and-pad: take an imperfect proxy for the counterparty's strength, and hold more capital against it. The move from the existing CDA to the proposed level is a larger pad. It is not a look-through to the assets behind the proxy.

A pad on a lagging proxy

Why does the distinction matter in practice? Because the proxy lags.

Rating downgrades follow visible credit deterioration; they do not anticipate it. A framework that calibrates capital to the IFSR is therefore calibrating to a number that adjusts after asset risk has crystallised, not before. The recent experience of regulated-utility credit makes the point concretely. An investment-grade water utility, its debt long treated as predictable, matching-adjustment-eligible infrastructure exposure held as core in annuity portfolios — until the market's confidence in the regulatory settlement weakened and the credit repriced. The rating frameworks treated such exposure as stable until, fairly suddenly, they did not. A Bermuda affiliate holding a portfolio of broadly similar assets would present the same lag to an IFSR-based regime: the rating would move after the event, not before it.

The asset classes that dominate funded re collateral pools sharpen the concern. Typical Bermuda funded re portfolios skew toward structured credit, privately originated and directly placed corporate debt, and asset-backed and real-estate lending — at allocations materially heavier than the cedant would hold in its own matching-adjustment portfolio onshore. These are precisely the exposures whose values are least observable and whose behaviour in a genuine downturn is least tested. Which leads to the analytical core.

Who pays for the rating

There is a further property of the IFSR that a capital framework should not pass over: who produces it, and on what commercial terms. Insurer financial strength ratings are assigned by credit rating agencies under the issuer-pays model — the rated entity, or its group, commissions and pays for the rating. The party whose strength is being assessed is the agency's paying client, and can in principle seek a preliminary view from more than one agency before awarding the mandate. The dynamic this creates — ratings shopping — is not a fringe concern: it was the central structural finding of the post-2008 inquiries into the agencies' role, and it is the reason rating-agency regulation was rebuilt on both sides of the Atlantic. Competition between agencies, which in most markets improves the product, can here compete down the stringency of the rating, because the buyer prefers a high one.

The relevance to funded re is uncomfortably specific. The 2008 failures were concentrated not in plain corporate bonds — which have long histories, market prices and many observers — but in structured, model-valued products with short histories, where the agency's model was the main thing standing between the buyer and the risk. That is precisely the profile of a funded re collateral pool. So the framework rests insurer capital on issuer-paid ratings of structured, model-valued assets — the very combination the post-crisis reforms were intended to make us wary of. Where the pool is also originated by the reinsurer's affiliated manager, the agency is rating assets whose values are set by the group paying for the rating, and the conflict compounds.

This is not to dismiss the IFSR. Insurer strength ratings are a more holistic and conservative exercise than structured-finance ratings ever were, the agencies are now supervised for conflicts and methodology by the relevant authorities, and insurance ratings did not fail in 2008 as structured products did. The agencies' expertise is real and their methodologies public. The narrower point stands: a number produced under issuer-pays incentives, on opaque and often related-party assets, carries known reliability limits — and CP8/26 asks it to bear regulatory-capital weight, which is a heavier load than the rating was built to carry.

The cycle that has not happened

Much of the asset base now sitting behind funded re structures, in its current form, has been originated and held through a single, benign phase of the credit cycle — a long period of low rates, rising asset values and abundant liquidity. The relevant question is not whether these assets have performed. It is whether they have been tested. "Has not defaulted" and "has been tested through a cycle" are different statements, and the gap between them is where prudential risk accumulates unobserved.

An IFSR built on through-the-cycle assumptions, applied to asset classes that have not themselves been through a cycle in their present structured-and-private form, inherits an estimation problem it cannot signal. The rating is an extrapolation dressed as a measurement. This is not a criticism of the agencies, who are explicit about their methodologies; it is an observation about what any rating can and cannot encode when the underlying has no cyclical history. The CDA, in leaning on that rating, leans on the same extrapolation.

Related-party origination

The structure of much UK funded re compounds all of the above. Where a single asset-management group originates the loans, structures the reinsurance, holds the assets in the offshore vehicle, and earns fees across each of those layers, the credit assessment of the asset pool is not independent of the parties whose economics turn on that assessment. The IFSR-centred framework treats the reinsurer as a standalone entity with an independent credit profile. But for several of the dominant funded re counterparties, the reinsurer is operationally and economically inseparable from its asset-management parent. An assessment of counterparty strength that does not engage with that inseparability is measuring something narrower than the risk. CP8/26, in its current scope, does not address it.

Does the collateral limb close the gap?

In fairness, the proposal does reach toward the assets in one place. CP8/26 allows credit where collateral controls are effective, where collateral need not be rebalanced or transformed on recapture, and — most directly — where the collateral is of higher credit quality than the reinsurer, judged by the weighted-average rating factor of the worst-case collateral portfolio. So it would be wrong to say the asset pool is invisible to the regime; a credit-enhancing collateral limb exists.

But notice what that limb actually measures. It is, again, a rating of the collateral — the same issuer-paid, lagging, often internally-assigned rating apparatus, now applied to the pool rather than the entity — expressed as a single worst-case average rating factor. An average rating factor says nothing about concentration in correlated asset classes, nothing about whether those classes have been cycle-tested, and nothing about related-party origination. It is a second layer of rating, not a look-through. The collateral enters the calculation as a rating adjustment to the counterparty charge, not as a direct capital charge on the asset risk itself. The gap is narrower than it would be without the limb; it is not closed.

The padding is held against the wrong failure mode

Here the threads converge, and here is the point I would most want the consultation to engage.

The proposed capital sits on the cedant's own Solvency UK balance sheet, against the reinsurance recoverable. It is counterparty-default capital, calibrated to the IFSR, held by the onshore party that did not originate the asset risk and cannot directly control it. Against an idiosyncratic reinsurer failure — one counterparty stumbling for reasons specific to it — that capital is real and welcome protection.

But the tail that should worry a prudential regulator is not idiosyncratic. It is correlated: a broad repricing of the asset classes concentrated across funded re pools, hitting many vehicles at once. In that scenario the IFSR-based pad behaves poorly precisely when it is needed. The reinsurer's rating falls after the deterioration is visible. The cedant's recoverable impairs as the event unfolds. And the recapture machinery — the cedant pulling assets and liabilities back onto its own balance sheet on a downgrade or covenant trigger — fires exactly when those assets are worth least and the cedant's own position is most stressed. A buffer sized to a normal-times default probability is not sized to that correlated unwind. The PRA's own 2025 Life Insurance Stress Test made the point with figures: recapturing the exposures of a single largest counterparty — some £12.3 billion of liabilities — would cut SCR coverage by around ten percentage points and reduce industry surplus capital by roughly £3 billion, and that on the relatively small exposures held at year-end 2024. The regulator has seen the dynamic; the concern is what it looks like at £110 billion rather than £40 billion.

This is what I mean by proxy-and-pad addressing the wrong failure mode. The regime moves more capital onto the cedant and calibrates it to the counterparty symptom. The asset risk that drives the symptom — and the wrong-way recapture dynamics that risk sets off — remains substantially where it was, in lightly supervised vehicles, behind a lagging rating, in asset classes that have not been cycle-tested, often assessed by parties not independent of them.

What a more complete answer might involve

None of this argues against CP8/26's direction. The capital tightening is broadly correct, and the IFSR has the practical merits Keen-Tomlinson identifies — it is standardised, externally produced and cross-jurisdictional, which is no small thing when the alternative is no measure at all. The argument is narrower and, I hope, constructive: the framework as scoped prices the counterparty layer while leaving the asset layer that drives counterparty solvency largely untouched, and a more complete regime would need to reach through the rating to the pool behind it.

Several routes exist, and the consultation is the place to weigh them. A degree of look-through capital treatment, charging the cedant by reference to the composition of the ceded collateral pool rather than solely the counterparty rating — extending the logic of the credit-enhancing collateral limb from an average rating factor toward genuine asset-class detail. Explicit concentration limits on asset classes within ceded pools, mirroring the kind of limits applied to directly held matching-adjustment portfolios. A downgrade and recapture overlay that sizes capital to the correlated-stress scenario rather than the idiosyncratic one — the PRA has already identified concentrated collateral and uncaptured downgrade risk as concerns, so the analytical groundwork is partly laid. And explicit recognition, in the counterparty assessment itself, of related-party origination where the reinsurer and the asset originator are one economic group — a feature the proposed intra-group exclusion engages only for the narrow case of mirror portfolios with no group-level surplus creation.

Each has costs and trade-offs, and reasonable people will weigh them differently. But the prior question — whether the asset risk behind the counterparty is being priced at all, or only proxied — deserves to be on the table before the CDA approach settles into rules. Funded reinsurance in its current UK form is not quite the instrument the traditional reinsurance frameworks were built around. The regulatory apparatus is adapting; the adaptation, on the evidence of CP8/26, is real but not yet complete.