Allocating Total Cost of Risk Across Group Entities in Indian Corporates 2026

A large Indian corporate group buys most of its insurance centrally, because a group programme commands better terms than a dozen subsidiaries buying separately, and the cost of that programme has to be distributed back across the entities that generate the risk. The quantity being distributed is the total cost of risk (TCOR), and getting its allocation right matters because the allocation drives behaviour, settles internal disputes, and has to satisfy the transfer-pricing rules that govern transactions between group entities.

TCOR is broader than premium. It is the sum of the premiums paid to insurers, the losses the group retains within deductibles and self-insured retentions, the cost of administering risk (the risk and insurance team, broker fees, risk-engineering spend, claims-handling cost), and the cost of risk-control and loss-prevention investment. A group that looks only at premium when it allocates cost understates the true cost of risk that each entity carries, and it sends the wrong signal: an entity whose premium is low but whose retained losses are high looks cheap on a premium-only view and expensive on a TCOR view, and the TCOR view is the one that reflects reality.

The entities across which TCOR is allocated are the subsidiaries, divisions and business units of the group, which in an Indian conglomerate can span very different risk profiles: a manufacturing subsidiary with fire and engineering exposure, a logistics arm with motor and cargo exposure, an IT-services unit with cyber and professional-indemnity exposure, a real-estate entity with property exposure. These units differ in size, in the kind of risk they generate, and in how much loss they actually produce, and a single flat allocation across them is rarely fair to all.

The allocation serves several purposes at once, and they pull in different directions. It has to recover the group's actual cost of risk and distribute it across the entities. It has to be defensible to each entity's management, who will challenge an allocation they see as unfair. It has to satisfy the transfer-pricing requirement that inter-company charges be at arm's length. And it should, ideally, incentivise the entities to manage their risk well, by making the entity that runs a worse risk or produces more losses bear more of the cost. The methods that follow trade off these purposes differently, and the right choice depends on which the group weights most.

There are four common bases for allocating TCOR across group entities, and each produces a different distribution and a different incentive. Most groups use a blend rather than a single basis.

Headcount, revenue and asset bases

The simplest bases allocate cost in proportion to a measure of the entity's size: headcount, revenue or turnover, or asset value or sum insured. These are easy to compute, easy to explain, and hard to dispute on grounds of arbitrariness, which is why groups default to them. Their weakness is that they are proxies for exposure, not measures of it. Allocating fire cost by revenue charges a high-revenue, low-hazard trading entity the same rate as a high-revenue, high-hazard manufacturing entity, which misprices the risk. Allocating by sum insured is closer for property lines because the sum insured does relate to the exposure, but it still ignores the hazard quality of the risk: two factories with the same sum insured can have very different fire risk depending on construction, protection and housekeeping.

Exposure-based allocation

A better basis allocates cost in proportion to the actual exposure each entity presents to each line of cover. For property and engineering, this means allocating by a rated exposure that reflects sum insured adjusted for the hazard of the occupancy and the quality of the risk. For liability, it means allocating by a measure of the liability exposure (turnover, payroll, the nature of the activity). For motor, by the fleet. Exposure-based allocation prices each entity closer to what its risk would cost if bought standalone, which is fairer and which gives the entity a reason to improve its risk, because a better risk earns a lower allocation. The cost is complexity: it requires the data and the rating logic to compute the exposure-adjusted share for each line and each entity.

Loss-experience allocation

The fourth basis allocates cost in proportion to the entity's actual loss experience, charging the entity that produces more losses more of the cost. This is the strongest incentive for loss control, because the entity sees its own losses reflected directly in its allocation, and it is the basis most aligned with experience rating and burning-cost analysis. The danger is volatility and the small-numbers problem: a single large loss at one entity in one year, which may be bad luck rather than bad management, can swing its allocation violently, and an entity small enough to have few losses has a loss experience too thin to be credible. Loss-experience allocation is therefore usually credibility-weighted (blended with an exposure-based expected cost in proportion to how much loss data the entity has) and smoothed over several years rather than applied raw to a single year.

Once the basis is chosen, the allocation has to be operationalised as an internal cross-charge: the central entity that holds the group programme recovers the allocated cost from each subsidiary through an inter-company charge, and the mechanics of that charge have to be clean, documented and consistent.

The central function (a group risk-and-insurance team, sometimes housed in a holding company or a shared-services entity) pays the premium to insurers and carries the central risk cost, and it raises a charge on each subsidiary for that subsidiary's allocated share. The charge should be raised on a clear, documented basis that the subsidiary can verify, with the allocation methodology, the inputs (the subsidiary's exposure, loss experience and size measures) and the resulting share set out, so the subsidiary's finance team can reconcile the charge to its own numbers and so the charge stands up to audit and to tax scrutiny.

The retained-loss component needs particular care. Where the group retains losses within deductibles and self-insured retentions, those retained losses have to be allocated too, and the mechanism for doing so (charging each entity for its own retained losses, or pooling retained losses centrally and allocating the pool) is a design choice with behavioural consequences. Charging each entity for its own retained losses gives the sharpest incentive but the most volatility; pooling and allocating gives stability but weakens the incentive. Many groups run an internal retention pool that absorbs the volatility of individual losses and charges entities a smoothed contribution, which is effectively an internal insurance arrangement and which, as the next sections discuss, starts to resemble a captive.

The management information the allocation depends on

The allocation is only as good as the data behind it, and the group risk function should maintain the MIS that supports it: the exposure data for each entity by line, the loss experience by entity over several years, the premium and retained-loss cost by entity, and the resulting TCOR and allocation by entity. This MIS does double duty. It supports the allocation, and it gives the group a view of where its cost of risk actually sits, which entities are improving, which are deteriorating, and where risk-control investment would most reduce the group TCOR. A group that allocates without this MIS is allocating blind, and its allocation will be both less fair and less defensible.

The cross-charge cadence should match the programme: an initial allocation at the start of the policy year based on expected cost, and a true-up after the year when the actual premium, the actual retained losses and the actual loss experience are known, so the entities ultimately bear their actual allocated cost rather than only the estimate. The true-up mechanism has to be set out in advance so that an entity facing a true-up charge after a bad loss year understands why.

Inter-company cross-charges for risk are related-party transactions, and they fall within India's transfer-pricing regime, so the allocation has to be defensible not only to the subsidiaries' management but to the tax authority. This is a constraint that a purely internal allocation exercise can overlook, and overlooking it creates tax exposure.

India's transfer-pricing rules, under the Income Tax Act, require that transactions between associated enterprises be priced at arm's length, meaning at the price that independent parties would have agreed. When the central entity charges a subsidiary for its share of the group risk cost, that charge is a related-party transaction, and the tax authority can examine whether the charge reflects an arm's-length allocation of a genuine cost that genuinely benefits the charged entity, or whether it is shifting profit between entities (and, where entities sit in different tax jurisdictions or enjoy different tax treatment, across the boundary that the transfer-pricing rules exist to police).

The requirements this places on the allocation are practical. The cost being allocated has to be a real cost (the actual premium, retained loss and administration cost), not a marked-up or invented charge. The allocation basis has to be reasonable and consistently applied, so that each entity bears a share that reflects the benefit it receives and the risk it presents, which is what an arm's-length allocation would do. The allocation has to be documented, with the methodology, the inputs and the rationale recorded, so that if the charge is questioned the group can show how it was derived. And where the central function provides a service (the risk management, the broker negotiation, the claims handling) for which it charges, the charge for that service has to meet the arm's-length standard for intra-group services, which in some cases permits a modest mark-up reflecting the value of the service and in others is a pure cost allocation.

The practical reconciliation between the risk and tax views is usually achievable, because a well-designed exposure-and-experience-based allocation of a genuine cost is close to what an arm's-length allocation looks like. The work is to document it as such, to apply it consistently, and to involve the tax function so that the allocation the risk team designs for fairness and incentive is also the allocation the tax team can defend.

The deepest tension in TCOR allocation is between fairness and incentive, and it is as much a political problem inside the group as a technical one, because every allocation creates winners and losers and the losers will object.

Fairness, in the sense each entity's management means it, is usually "charge me for my own risk and my own losses, not for someone else's." An entity that runs a good risk and produces few losses resents subsidising an entity that runs a bad risk and produces many, and it has a point: a flat or size-based allocation that ignores risk quality and loss experience does make the good risk subsidise the bad one. This is the argument for exposure-based and experience-based allocation, which charge each entity closer to its own cost.

The incentive argument runs the same way: an allocation that charges an entity for its own risk and losses gives that entity a reason to improve, because the improvement reduces its allocation, while an allocation insensitive to the entity's own behaviour gives no such reason, so the entity has no financial stake in its own loss control. A group that wants its entities to invest in risk control should allocate in a way that rewards them for doing so, which again points to exposure and experience bases.

The countervailing considerations are real, though. Pure experience-based allocation is volatile and penalises bad luck as if it were bad management, which entities rightly resent in a different direction. A small entity's loss experience is not credible enough to allocate on. And some risks are genuinely group risks (a group D&O programme, a group cyber tower) whose cost is hard to attribute to individual entities and which it is fairer to allocate on a broad basis. So the fair allocation is not the maximally entity-specific one; it is the one that charges entities for what is genuinely theirs, smooths the volatility and the small-numbers noise, and allocates genuinely shared cost on a reasonable broad basis.

Making the allocation accepted, not just correct

An allocation that is technically sound but imposed without explanation will be resisted, so the group has to make the allocation accepted as well as correct. That means publishing the methodology so entities understand how their charge is derived, showing each entity the inputs that drive its allocation (its exposure, its losses, its size) so it can see the link between its own risk and its own cost, and giving entities a route to improve their allocation by improving their risk, so the allocation is experienced as something they can influence rather than a tax imposed from the centre. The group should also be willing to explain and, where justified, adjust: an entity that can show its allocation rests on stale or wrong data has a legitimate grievance, and a group that refuses to engage breeds resentment that undermines the whole exercise. The aim is an allocation the entities accept as fair and can act on, because that is what turns the allocation from an accounting exercise into a driver of better group-wide risk management.

When a group forms a captive insurer, the TCOR allocation problem changes shape, because the captive becomes the internal mechanism through which risk is financed and cost is allocated, and it brings a structure and a discipline that an informal internal allocation lacks.

A group captive (whether established in GIFT City under the IFSCA regime, or offshore in a centre such as Singapore or Bermuda) insures the group's risk and charges each group entity a premium for the cover the captive provides. That premium is the allocation: the captive's underwriting of each entity's risk produces a premium for that entity, and the premium is set on an exposure and experience basis as an insurer would set it, which gives the allocation an external, insurance-grounded logic rather than an internal accounting one. The captive can run the retention pool formally, absorbing the volatility of individual losses across the group and charging each entity a premium that reflects its risk, smoothed and credibility-weighted, which is exactly the blend the earlier sections described, now operated as an insurance arrangement rather than an internal cross-charge.

The captive sharpens the transfer-pricing position in some ways and complicates it in others. The premium an entity pays the captive is a related-party transaction subject to transfer pricing, and it has to be at arm's length, meaning the captive has to price the risk as an independent insurer would, with actuarial support for the premiums, so the captive premium is defensible as arm's length. Done properly this is cleaner than an informal cross-charge, because the captive premium has an insurance-pricing rationale the tax authority recognises. Done improperly (a captive premium set to shift profit rather than to price risk) it is a transfer-pricing problem the authority is alert to, so captive premiums need actuarial and transfer-pricing support.

The captive also changes the incentive structure for the better. Because each entity pays the captive a risk-based premium and the captive's results flow back to the group, the entities' loss control directly affects the captive's results and so the group's bottom line, which aligns the entities, the captive and the group around reducing losses. A captive that prices each entity's risk, retains the entities' good and bad experience, and returns its surplus to the group makes the cost of risk visible and attributable in a way that drives behaviour, which is one of the strategic reasons groups form captives in the first place, beyond the risk-financing and tax-efficiency reasons.

Whether a group should form a captive to run its TCOR allocation depends on the same factors that drive any captive decision: the group's premium volume, its risk profile, its appetite for retention, and the cost of running the captive against the benefit. For a large group already allocating a substantial TCOR across many entities, the captive often formalises and improves an allocation the group is already doing informally, which is why the captive decision and the TCOR allocation question are closely linked.

Doing any of this well, the exposure rating, the experience analysis, the captive pricing, the allocation that entities will accept, rests on understanding what the group's covers actually provide and how each line's cost is built, so the allocation reflects the real structure of the risk rather than a crude proxy. Sarvada gives commercial insurance brokers and corporate risk teams structured, searchable access to insurer policy wordings, so the people designing a group's cost-of-risk allocation can see how each cover responds, where the retentions and sub-limits sit, and how the lines differ, grounding the allocation in the real terms of the programme. Request Access to bring that wording-level detail into your group's total-cost-of-risk and allocation work.