Pooled and Group Captive Structures for Indian Mid-Market: When Sharing Risk Beats Buying It in 2026

Two things happened in parallel this year that put group and pooled structures squarely on the mid-market risk manager's desk. First, the Insurance Amendment Bill of 2025, branded Sabka Bima Sabki Raksha, stayed silent on a domestic captive framework. The Rs 100 crore paid-up capital floor for an Indian insurer remains in place, and IRDAI has not carved out a lighter captive class. So the single-parent captive on Indian soil is still a non-starter for any firm that cannot park that much idle capital.

Second, the alternative side of the market kept building. IFSCA's working group on alternate risk transfer, chaired by G. Srinivasan, reported in 2025 recommending a Special Purpose Insurer (SPI) class and an insurance-linked-securities framework out of GIFT IFSC. By early 2026 IFSCA had moved toward the next stage of approval and had publicly sketched an SPI sidecar route to pull in global capital. The headline ILS ticket size being discussed (commonly cited in the order of USD 50 million minimum) tells you exactly who that channel is for, and it is not a mid-market manufacturer in Ludhiana or a logistics firm in Bhiwandi.

That gap is the point. Large conglomerates have GIFT City and offshore captives. The very small can self-insure a deductible and stop there. The mid-market, firms with a few hundred crore of turnover and a retained-risk bill that grows every renewal, sits between the two with no obvious vehicle of its own. Group and pooled captives are the structure built for that middle, and the broker who understands the conditions under which they work is selling something most of the market cannot.

A note on terms. Throughout this post, "group captive" means a captive owned by several unrelated firms who each take a share of risk and reward. "Pool" means a risk-sharing arrangement (often through a fronting insurer) where members cede agreed layers into a common account. The economics overlap, the legal form differs.

Strip away the structure and a group or pooled captive is a way to convert your premium from a sunk cost into a managed asset. In a conventional placement, the premium leaves your books and the insurer keeps the underwriting margin, the investment income on the float, and the upside of any year you do not have a loss. In a group captive, members fund a common account, retain a working layer of their own losses, share a middle layer with the group, and buy reinsurance above that. Underwriting profit and investment income on the fund stay inside the group.

The firm is buying three things. One, transparency: it sees the loss fund, the reinsurance spend, the expense load, and what is left over, rather than an opaque market premium. Two, a return of the good years: where a commercial insurer banks the profit on a clean account, a group captive returns it (net of the shared layer) to the members who earned it. Three, a seat at the underwriting table, which over time disciplines the worse-run members or pushes them out.

None of that is free. The members collectively carry the volatility of the shared layer, post collateral (usually a letter of credit) to the fronting insurer, and accept that a bad year can mean a capital call. The honest framing for a client is this: pooling trades a known, marked-up premium for a lower expected cost with real variance around it. If the firm cannot stomach the variance, or cannot fund a capital call, it should not pool, however attractive the average looks.

The total-cost-of-risk lens

The only number that matters is total cost of risk, the sum of retained losses, premium, collateral cost, and programme administration. A pool that lowers premium but raises retained-loss volatility and ties up a letter of credit has not necessarily helped. Model the total cost of risk across a five-year horizon, not a single renewal, before recommending the move.

The single biggest predictor of whether a pool works is how alike the members are. Homogeneity is not a nicety, it is the actuarial spine of the structure. When members share the same hazard profile, the loss distribution of the pool is predictable, the reinsurance is priceable, and no member is systematically subsidising another. When they do not, the structure becomes a transfer from the careful to the careless, and the careful leave first.

Think about what "alike" means in practice. A group captive for cold-chain logistics operators makes sense: similar fleet risk, similar warehouse fire load, similar cargo and transit exposures, similar claims patterns. A pool that mixes a food-processing plant, a chemicals blender, and a textiles mill does not, because the chemicals member brings a fire and liability tail the other two will end up funding. Industry-association pools sometimes get this right by construction, which is why the association route has worked for homogeneous trades.

A practical homogeneity test

Before taking a pooling idea to a client, run the members past four screens:

• Hazard class. Do members share the same dominant peril (fire, motor fleet, workers compensation, liability)? Mixed dominant perils are a warning sign.
• Severity profile. Are large losses driven by the same mechanism? A pool where one member can produce a Rs 50 crore loss and the rest top out at Rs 2 crore is unbalanced.
• Loss frequency. Members with wildly different claim frequencies will argue endlessly over the shared layer.
• Risk culture. Do members run comparable safety, maintenance and housekeeping standards? Culture predicts future losses better than past data in a young pool.

If the group fails two or more screens, the right advice is usually to narrow the membership, not to widen the structure to accommodate everyone. A smaller, tighter pool beats a larger, mixed one almost every time.

Underwriting maths gets the attention, but governance is where Indian pooling attempts most often come apart. A group captive is a small mutual insurer run by a committee of competitors who are also customers. The incentives pull in different directions: every member wants the cheapest contribution and the most generous claims treatment for its own losses, which is exactly the behaviour that bankrupts the fund.

The arrangements that survive build hard rules before the first policy incepts. A clear contribution formula tied to each member's own loss experience, not a flat rate, so that a member who runs up claims pays for them at the next reset. An independent claims adjudication process, so no member sits in judgement of its own claim. A defined entry and exit protocol, including how a leaving member's collateral and share of the loss fund are handled, and a run-off provision for losses that develop after exit. And a written underwriting standard that lets the group decline or surcharge a member whose risk has deteriorated.

For the broker, this is the value-add the market does not price. Placing a treaty is a commodity skill. Designing a contribution formula, a claims protocol, and an exit waterfall that a dozen mid-market promoters will actually sign is consulting work, and it is defensible revenue. It also protects you: a pool that fails on governance fails publicly, and the broker who waved it through wears the reputational cost.

Fronting and regulation

Because IRDAI does not license a domestic captive, most onshore pooling runs through a fronting arrangement with a licensed insurer who issues the paper and reinsures the agreed layers back to the group's vehicle, often domiciled in GIFT IFSC or offshore. The fronting fee and collateral terms materially change the economics, so model them explicitly rather than treating fronting as a formality.

Pooling only lowers total cost of risk when the members are, collectively, better risks than the market assumes. If the commercial market is already pricing the group's risk accurately and there is no fat to recover, a captive just relocates the same expected loss while adding administration and collateral cost. So the threshold question is whether the members are currently overpaying.

Three signals suggest they are. First, a low and stable loss ratio over the trailing five years, well under what the premium implies, which means the market is keeping margin the members could retain. Second, losses concentrated in a predictable working layer, with genuine large losses rare, because that profile is cheap to reinsure and easy to fund. Third, a premium that has hardened faster than the members' own experience justifies, which is common in classes the market has decided to de-risk regardless of individual merit.

The inverse is just as important to say out loud. A group with a volatile loss history, a recent large claim, or a deteriorating trend should not pool yet. It will either fund the fund inadequately and face a capital call, or price contributions so high that the structure offers no saving. Sometimes the right professional answer is "come back after two clean years", and a broker willing to say that earns more trust than one who sells the structure regardless.

Sizing the retained layers

Work from the data outward. Set the per-member working retention at a level each member can absorb from cash flow without distress, typically aligned to an existing comfortable deductible. Set the shared layer to cover the frequency band the group can fund with reasonable confidence, and buy reinsurance above the point where the group's own capital would be threatened by a single bad year. The collateral the fronting insurer demands will key off these choices, so iterate the layers and the collateral together rather than fixing one and accepting the other.

Not every mid-market book is a pooling candidate, and a broker who pitches it indiscriminately will burn credibility. The segments where group and pooled structures earn their keep share a pattern: homogeneous members, a working-layer-heavy loss profile, and a market premium that has outrun individual experience.

Good fits in the Indian context tend to include auto-ancillary and engineering manufacturing clusters with similar fire and machinery-breakdown exposures, third-party logistics and cold-chain operators with comparable fleet and warehouse risk, and trade associations whose members run near-identical operations. Workers compensation and employee-benefit pooling also works well for clusters with similar headcount risk, because frequency is predictable and severity is capped by statute.

The segments to steer away from are mirror images. Anything with a fat catastrophe tail (large property values exposed to flood or seismic risk) belongs in the conventional or parametric market, not a small pool that one event can wipe out. Liability-heavy businesses with long-tail exposure are hard for a young pool to reserve. And firms in financial distress should never join, because a capital call lands on the weakest member at the worst time and can trigger the unravelling of the whole group.

The sharper segmentation is not by industry label but by loss shape. Two textiles mills with different fire-protection standards belong in different conversations, while a textiles mill and an engineering unit with identical, well-managed working-layer losses might pool comfortably. Lead with the loss data, not the SIC code.

Turn the analysis into a decision a client can act on. The recommendation a mid-market firm needs is not "captives are good", it is a clear yes, no, or not-yet, with the reasoning attached. Structure the advice in a sequence the promoter can follow.

Start with the total-cost-of-risk baseline: what the firm spends today on premium plus retained losses plus the cost of any collateral, averaged over five years. Then model the same five years inside a proposed pool, including the fronting fee, the collateral letter-of-credit cost, administration, and a stressed loss scenario with a capital call. If the pooled total cost of risk is not meaningfully lower even in a moderately bad year, the answer is no, and saying so is the service.

If the numbers favour pooling, the build sequence matters. Identify a homogeneous founding membership (smaller is safer). Agree governance, contribution formula, claims protocol and exit terms in writing before approaching insurers. Select a fronting insurer and negotiate the fronting fee and collateral. Decide domicile, with GIFT IFSC increasingly the natural Indian-adjacent option as IFSCA's framework matures. Then place the reinsurance and incept.

What to hand the client

• A five-year total-cost-of-risk comparison, conventional versus pooled, with a stressed-year column.
• A homogeneity assessment of the proposed members against the four screens.
• A one-page governance term sheet: contribution basis, claims adjudication, entry and exit, capital-call mechanics.
• A clear statement of what could go wrong and who pays when it does.

Deliver that and the broker has done something the placement-only competitor cannot. The goal is not to maximise structures sold, it is to put mid-market firms into pools that survive their first bad year, because those are the clients who renew, refer, and trust the next recommendation.