Captives, self-funding and multinational pooling

What a funding decision actually decides

An employer with people in fifteen countries is running fifteen insurance contracts, fifteen renewal negotiations and fifteen sets of claims data that do not add up. The recurring proposal, from brokers and from finance directors alike, is to stop buying insurance for risk that is not really risky — routine outpatient care, diagnostics, repeat prescriptions — and to retain it instead, either directly or through a captive.

The proposal is often sound. It is also routinely oversold, because the pitch conflates three things that move independently: who funds predictable claims, who carries the catastrophic tail, and whose paper the local policy is written on. A funding change can move the first. It usually cannot move the second cheaply, and it cannot move the third at all where the law requires a locally licensed insurer.

This guide describes the spectrum honestly, explains why the admitted-insurance requirement constrains everything above it, and sets out the questions worth putting to a broker before any of it is agreed. The plan-design side of an employer scheme is covered in the existing guides to corporate IPMI and group private medical insurance for employers.

The funding spectrum

Read across the rows rather than down the first column. The first three are all insurance; what changes is how directly the employer's own claims history feeds its own cost, and how quickly. The last two change who owns the risk, and with it who owns the volatility, the reserving and the capital.

The honest summary of the whole spectrum is that predictable, high-frequency, low-severity claims are the part worth retaining, because insuring them mostly means paying someone else's expense loading to process them. Low-frequency, high-severity claims — a neonatal intensive care admission, a complex oncology course, an air evacuation from a remote site — are the part worth insuring, because a single one can exceed a year's expected spend for an entire country's headcount. Every structure above is a different way of drawing that line.

What does not move

The catastrophic tail. Retaining it is not a saving; it is a decision to hold volatility on the balance sheet. Captives typically buy reinsurance for exactly the same reason insurers do, so the tail generally ends up with a third-party carrier however the structure is drawn.

Network access. Direct-settlement arrangements, negotiated tariffs and provider directories are assets of the insurer or administrator, not of the funding vehicle. A self-funded programme still runs on somebody's network, and access to it is bought rather than created. Whether an employee can walk into a particular hospital and not pay upfront is a network question, addressed in provider networks and checking your hospital and how an IPMI claim works.

The local admitted-insurance requirement. This is the constraint that shapes everything else, and it deserves its own section.

Why admitted insurance constrains all of this

Wherever health insurance is compulsory, the mandate is almost always drawn as an obligation to hold a policy from an insurer licensed in that jurisdiction — and in several markets writing local risk on non-admitted paper is prohibited outright. Dubai requires both a UAE insurance licence and a health insurance permit from the health authority. Saudi Arabia requires a Saudi-registered cooperative insurer and restricts foreign reinsurance as well. Qatar confines the mandatory cover to insurers licensed under Qatari law and registered with the Ministry of Public Health. Germany requires an insurer licensed to do business in Germany, and confines substitutive health insurance to domestic business conducted on life-insurance principles.

None of those requirements care how the risk is funded behind the scenes. They care whose name is on the policy. So a self-funded or captive programme reaching into those countries needs a locally admitted insurer to issue the policy, which then cedes the risk to the captive or reinsurance panel — the same fronting arrangement that appears in individual international cover, for the same reason. The mechanics, and the point that Saudi restrictions narrow the fronting route, are set out in admitted versus non-admitted insurance and who carries the risk.

The practical consequence for a global programme is that it will almost never be one structure. It will be a captive or self-funded core in the countries where retention is permitted and worthwhile, fronted local policies where the mandate requires admitted paper, and fully insured contracts in the small markets where neither the headcount nor the claims volume justifies anything else.

Multinational pooling

Pooling addresses a different problem: not who funds the risk, but why fifteen good years in fourteen countries are wiped out by one bad year in the fifteenth.

Under a pooling arrangement, an employer's separate country contracts are combined for experience purposes within an insurer's international network. The local contracts remain local — issued by local carriers, on local terms, satisfying local mandates — and the pooling operates on the accounting above them. Where the combined experience across the participating contracts is favourable, a dividend can be returned to the employer; where it is not, the loss is normally carried forward against future years rather than settled in cash.

The published detail on eligibility is thin. Minimum numbers of participating contracts, minimum lives per contract and the treatment of loss carry-forwards are set network by network and are largely absent from public sources, so any figure quoted to you should be treated as that network's own criterion and confirmed in writing. What can be said structurally is that pooling rewards scale and stability, that it does nothing for a country where the local contract itself is uncompetitive, and that a dividend is a return of favourable experience rather than a discount negotiated in advance.

The group-size thresholds that are published

Pooling thresholds are not published. Underwriting thresholds are, and they shape what any structure can offer members.

Two boundaries do real work here. The first is around three lives, below which group terms are generally not available at all. The second is around ten, where medical history disregarded becomes available and members stop being individually assessed — the single largest benefit change in any of these programmes, and the one with the sharpest consequence when it ends, described in medical history disregarded and the group cover cliff and leaving your employer's medical scheme.

The carrier can change with the structure

A live example, from a policy wording supplied by a client and dated May 2026: under one international brand, group schemes of six or more lives are underwritten by a French mutual insurer supervised by the ACPR, while individual policies are underwritten by a Portuguese insurance company supervised by the ASF. The brand, the administrator and the policy wording are common to both. The risk carrier, its jurisdiction, its supervisor and therefore the member's complaints route are not, and the wording states that the actual insurer is confirmed on the certificate of insurance rather than in the wording itself.

That is worth internalising before any structural change. Moving a population between segments — from individual to group, from one group size band to another, from insured to fronted — can move the risk carrier without changing anything the member can see. The due diligence question is not which brand is on the card but which entity is on the certificate, and who supervises it. Insurer solvency ratings and due diligence covers how to check.

Governance and cost containment

The levers that actually move medical spend are the same whatever the funding structure, and they belong in the programme design rather than in the funding decision.

Bupa Global, in its own published pricing guidance, describes its cost-control approach in four terms: settling directly with a network of providers, service partners in 50 different countries, pre-authorisation of in-patient treatment, and an evidence-based funding policy. Those are the standard levers, stated plainly by a market leader, and each one is a governance decision rather than a financing one.

Pre-authorisation determines what is approved before it is incurred. Direct-settlement networks determine the tariff paid and remove the member from the payment chain. An evidence-based funding policy determines what is eligible at all, and is the reason claims for unproven or experimental treatment are declined. Claims analytics determine whether any of it is working — and this is the one area where self-funding genuinely helps, because a self-funded employer typically has better access to its own claims data than a fully insured one does.

None of that reduces medical inflation itself, which is driven by the cost, frequency and mix of treatment across the markets a programme touches. That is the subject of how IPMI premiums are calculated and why your premium increased.

The questions to ask, and the advice this is not

Before agreeing any change of funding structure, get written answers to five questions. In which countries would the arrangement require locally admitted paper, and who fronts it? Where does the catastrophic tail sit, at what attachment point, and with which reinsurer? Which entity carries the risk for each segment of the population, and which regulator supervises it? What network and administration services are being bought, from whom, and on what terms if the funding arrangement later changes? And what claims data will the employer actually receive, in what format, and how quickly?

Captive and self-funding structures raise corporate tax, transfer pricing, premium tax and regulatory capital questions in every jurisdiction they touch — premium taxes alone are covered in outline in insurance premium tax on medical cover. Those questions have entity-specific answers and require specialist advice in each relevant country. Nothing here is tax, legal or regulatory advice, and the regimes described change frequently; confirm the current position before relying on any of it.