When the Insurance Intermediary Belongs to the Buyer

A company buys insurance. Its own group company places the policy. The outside insurer pays that group company a commission.

Two things this is not. It is not a captive insurer: nobody in the corporate group is underwriting anything, and the risk sits with an outside insurer exactly as it would otherwise. It is also not a conventional customer-side broker: the agency is formally appointed by the insurer even though it belongs to the customer’s group. What Japan has instead is a triangle:

The agency is appointed by the insurer. It is paid by the insurer. But it is owned by the customer.

On 11 September 2026 Japan’s Financial Services Agency opened public comment on amendments to its supervisory guidelines for insurance companies¹. Comments close on 13 October, and the first item is a reconstruction of the rules governing exactly this structure.

A structure inherited from a different insurance market

A 企業内代理店 — a corporate in-house agency — is an insurance agency with close capital or personnel ties to a company outside the insurance business, mainly handling that group’s own fire and liability cover and insurance for group employees as a welfare benefit².

It is old. The FSA’s 2024 expert panel records that such agencies have played a part in efficient insurance solicitation in Japan’s non-life market since before the war, and that as the market environment changed greatly, there are views that their role is coming to an end⁴. That is a view the panel records rather than a finding it makes, and it is worth keeping in that register.

What they actually do varies. Working-group members described agencies whose centre of gravity is employee welfare, delivered through group policies for staff, and others carrying genuine corporate risk-management work⁵.

The comparison that makes the age of the structure legible is the broker. Japan only acquired a broker regime with the 1995 Insurance Business Act, its implementing regulations following in 1996⁶ — and the FSA notes brokers remain little known and little used, with no increase in new entrants⁶. So for most of the period in which these agencies became established, Japan had no formal insurance-broker regime providing a customer-side intermediary model.

The dates in the current rules point the same way. Agencies registered on or before 31 March 1996 have had their concentration measured under a gentler formula: only fire, auto and accident insurance counted, and where several related parties existed their premiums were not added together — the ratio was computed for each and the highest one taken⁵.

This is Eveil’s reading, not the FSA’s: that line sits exactly where the modern framework begins. The market changed. The structure largely survived.

Two different realities inside one category

The four largest non-life insurers classify 9,530 of their appointed agencies as in-house — a number to hold loosely, since the FSA notes each insurer applies its own definition³. Among the largest agencies specifically — each insurer’s top 300 by premium, 736 after duplicates — 256 were in-house³.

Inside that population the FSA found two different things. In some, relative to the volume of insurance handled, very few registered insurance sales personnel are employed, or most staff also work for the parent. In others, closeness to the operating business lets the agency read the group’s risk properly, and it contributes to risk management the company could not easily replace³.

That spread is why the reform is not a ban.

Why the old rules are no longer enough

Japan constrains the structure with a ratio: the share of an agency’s premium coming from parties closely tied to it. Above 30%, the insurer must press for improvement; above 50%, insurers’ own rules generally lead to terminating the agency agreement³. The stated purposes are preventing a substantive discount or rebate of premium, and promoting the agency’s independence³.

Both softeners are going. The 1996 formula ends from April 2030. From April 2032 a closely tied party means the group as an economic entity — a parent holding more than half the voting rights². Among the 83 agencies already on the newer basis, widening that definition moves the number above the 50% line from zero to ten³.

In exchange, a way out. An agency can be exempted if its commission is built from its reported costs so that no substantive discount arises, and if it can show real capability and independence from its parent — identifying and reporting its own related-party business, antitrust compliance, and three lines of defence².

Read that exemption backwards and it is a diagnosis. It is written for agencies that do real work, and requires the commission to be demonstrably tied to that work rather than simply to premium volume.

What the commission may really represent

Here the regulator is unusually direct. Where an agency’s practical capability is insufficient, handling the group’s contracts still produces enough commission for it to survive — and that commission, the FSA writes, may amount to a substantive discount on the premium². It is put as a possibility, not a finding, and should be read that way.

The thought underneath is plain enough. If the intermediary is not doing intermediary work, money moving from the insurer to the buyer’s affiliate is a price adjustment wearing a different name.

Japan’s other exemption is worth noticing here: brokers taking their fee only from the customer fall outside the ratio rule altogether². But that is a narrower signal than it first appears. The FSA’s own survey records that in the United States, the United Kingdom and Germany, brokers may take commission from either side and usually take it from the insurer⁶. Insurer-paid intermediation is ordinary. What is distinctive here is the combination — customer ownership, insurer appointment, insurer-paid commission, and a concentration ratio built to contain the result.

The conflict does not disappear when the agency is good

This section is Eveil’s interpretation, not a position stated by the regulator.

The IAIS divides intermediaries into those acting primarily for the insurer and those acting primarily for the customer⁷. This one does not fit cleanly on either side, and the FSA’s own phrase for the problem is the unclearness of its position². Its panel goes further: despite the formal appointment, such an agency is generally seen as standing in the position of the policyholder⁴.

Two findings should travel further than the Japanese detail. A company may be better served buying less insurance and carrying more risk another way; less insurance means less commission for its own agency, so the agency has an incentive not to propose it — and the FSA notes this can arise even where the agency is genuinely capable³. And the panel concluded that even if capability improves and independence advances, the distortion caused by the unclear position remains⁴.

Competence does not dissolve the conflict. It only makes it better argued.

Eveil view

The question worth asking of an affiliated intermediary is not whether it adds value. Often it plainly does. The question is what its payment is indexed to. Where remuneration tracks premium volume, the entity sitting closest to the risk decision earns more when the group buys more cover — and that is a design choice, not a compliance failure. Japan’s own panel reaches for the same lever, noting that paying according to the role performed rather than the premium would help resolve what it calls the dual-agency problem⁴.

So the durable answers are structural rather than behavioural, which is why this reform arrives as a cost test and a carve-out rather than a prohibition. For any group with an affiliated broker, internal insurance arm or related distributor anywhere in its structure, the useful exercise is the one the FSA is about to force: write down what the intermediary does, what that work costs, and what it is paid. Where those three fail to reconcile, the gap is the answer.

Reviewing how an affiliated distribution or intermediary structure should be designed and governed? Eveil Intelligence is structured work on one organisation’s question.