Circular 102/8/2026: Insurance Enters the National Statistics System

IRDAI issued circular IRDAI/F&I/CIR/MISC/102/8/2026 on 7 August 2026, titled Submission of information by insurers on monthly basis for the purpose of Index of Service Production. It is addressed to all life insurers and to all general and health insurers, and it is issued under Section 14(2)(h) of the IRDA Act, 1999, the provision that lets the Authority call for information from entities it regulates.

The operative requirement for the non-life side is narrow and specific. In the circular's words:

General and Health Insurers (excluding Reinsurers) are required from July 2026 onwards to submit Gross Direct Premium, Claims Paid (Direct) and Investment Income of Policyholders and Shareholders

Three data points, monthly, per insurer. Reinsurers are outside the scope. The filing route is the NIC online data templates, and the deadline is stated as on or before the 15th day of the month succeeding the reporting month. July 2026 is the first reporting month, which puts the first filing due in August 2026.

The purpose is statistical rather than supervisory. The data feeds the Index of Service Production (ISP), the services-sector output index the Ministry of Statistics and Programme Implementation is building as a services counterpart to the Index of Industrial Production. Insurance is being wired into the national statistics system as a measured service industry.

Insurance reporting in India has historically been annual in its useful form. The claims settlement ratios that brokers and corporate buyers quote come out of annual disclosures and the IRDAI annual report. A number published once a year, twelve to eighteen months after the behaviour it describes, is a weak instrument for a decision made in a renewal cycle that runs every twelve months.

A monthly insurer-level series changes the sampling rate. If claims paid (direct) becomes visible per insurer per month, a buyer can see the shape of an insurer's payout activity within a year rather than after it, and can see a change in that shape while the renewal is still open. That is a category of evidence commercial buyers in India have never had for insurer selection.

The reason to be careful is that sampling rate is not the same as signal quality. The three fields in this circular were chosen because they are the inputs a statistical office needs to estimate output for the insurance service sector. They were not chosen to describe how an insurer treats claimants. Everything useful a buyer extracts from this series has to survive that mismatch.

The single most important limitation is in the field name. The circular requires Claims Paid (Direct), which is cash out of the door in the reporting month. It is not claims incurred, and it is not claims reported.

Claims incurred for a period equals claims paid plus the movement in outstanding reserves, including the reserve for incurred but not reported claims. That reserve movement is where an insurer's view of its own emerging loss experience lives. A monthly paid figure omits it entirely.

The practical consequences for anyone reading the series:

• Paid is lagged, and the lag varies by line. A motor own-damage claim can be paid in weeks. A liability or professional indemnity claim can be paid years after the event. A month's paid number mixes cohorts from many prior years.
• A rising paid number is ambiguous. It can mean faster settlement, a genuine deterioration in loss experience, growth in the book feeding through, or the clearing of a backlog.
• A falling paid number is equally ambiguous. It can mean a better loss year, a slower claim function, tightening acceptance, or nothing more than the timing of a few large settlements.
• Large losses distort a single month badly. One property total loss can move a mid-sized insurer's monthly paid figure by a visible margin, and the series carries no marker to say so.

The honest use of monthly paid data is directional and comparative over multiple months, never point-in-time and never single-month. A ratio built by dividing one month's claims paid by the same month's gross direct premium is not a loss ratio and does not behave like one, because the premium is written on this year's book while the payments discharge liabilities from several years of prior books.

The circular requires gross direct premium, claims paid (direct) and investment income at the entity level. There is nothing in the mandated fields that splits by line of business, by segment, or by customer type.

For a commercial buyer this is the binding constraint. A general insurer's monthly claims-paid figure is dominated by whichever segment is largest in its book, and for most Indian multi-line general insurers that is motor and retail health. A corporate buyer placing a fire insurance programme, a marine cargo open cover or a group health scheme is asking about a small and structurally different slice of the same entity.

The mechanics that break the inference:

1. Mix, not behaviour, drives the aggregate. An insurer growing motor third-party fast will show a rising paid trend that says nothing about how it handles a commercial property loss.
2. Different claim functions sit behind the same number. Retail health claims run through a third-party administrator or an in-house health desk. A large property loss runs through a surveyor appointment, a loss adjuster and often a reinsurer's claims control clause. These are separate operations with separate service records inside one legal entity.
3. The reinsurance layer is invisible. Claims paid direct is gross of reinsurance recoveries. On large commercial risks the economics and often the decision speed sit with the reinsurance panel, and none of that appears in the mandated fields.

A buyer who reads an entity-level monthly paid series as a statement about commercial claims handling has substituted a number that is available for the number that matters. That substitution is the specific failure mode to guard against here.

Set against those limits, there is a real residual use, and it is worth naming precisely so it is neither overclaimed nor discarded.

Growth and mix change. A monthly gross direct premium series shows how fast an insurer is writing. Sharp acceleration in a soft market is worth asking about at the next renewal meeting, because rapid growth on a thin rate is one of the standard routes to a later reserving problem.

Cash-flow visibility. Claims paid, read against premium received, gives a coarse read on whether an insurer is running positive or negative operating cash flow month to month. That is a solvency-adjacent observation and it is closer to what these fields were designed to measure.

Investment income stability. The circular requires investment income of policyholders and shareholders separately from underwriting flows. A sharp change there is a balance-sheet event, not a claims event, and it belongs in a counterparty assessment rather than a service assessment.

Anomaly detection between annual filings. The strongest use is as a tripwire. If an insurer's monthly figures depart materially from their own recent pattern, that is a prompt to go and read the quarterly and annual disclosures properly. It is a reason to ask a question, not an answer.

Build any monthly-series check as a trailing twelve-month view compared against the insurer's own prior twelve months, and against peers of similar size and book composition. Never compare a single month to a single month, and never compare across insurers with materially different segment mixes.

A broker's insurer security file already has established components. The question this circular raises is where a monthly public series sits among them, and the answer is below every one of the existing components rather than alongside them.

A defensible weighting for a commercial placement:

1. Solvency filing first. The regulatory floor of 150 percent is a minimum and not a target. The trend across successive filings, and the insurer's own headroom above the floor, remain the primary financial-security evidence. Nothing in a monthly ISP submission displaces this.
2. Line-relevant claims evidence second. For the segment actually being placed, the broker's own settlement record with that insurer, turnaround times on comparable losses, surveyor appointment behaviour and repudiation history carry more weight than any published aggregate. This is described in more detail in our note on scoring a group health panel where IRDAI does not collect repudiation reasons.
3. Annual claims settlement ratio third, with its known limits. The settlement ratio counts claims, not rupees, and is dominated by high-frequency low-value retail claims. It says very little about how a commercial business interruption loss will be handled.
4. Reinsurance panel and treaty structure. On large or specialised risks the client's real security often sits behind the primary insurer.
5. Monthly ISP-derived series last, as a monitoring tripwire. Directional, entity-level, and useful mainly for spotting change between the filings that carry real detail.

The full construction of a panel file is covered in our post on insurer empanelment and security rating for brokers, and the corporate-side view in vetting insurer financial security and counterparty risk.

For insurers inside the scope, the obligation is operational and recurring rather than a one-off submission, and it starts from the July 2026 reporting month.

The points that decide whether this runs cleanly:

• A monthly close discipline that produces the three fields by the 15th. Gross direct premium and claims paid (direct) come out of different subsystems in most insurers, and investment income split between policyholders and shareholders comes out of a third. The reconciliation between them is the part that takes time.
• Consistency with the other filing tracks. These monthly figures will sit alongside quarterly and annual regulatory submissions and public disclosures. Divergence between the monthly series and the periodic filings, once both are visible, becomes a question the insurer has to answer.
• Ownership of the NIC template. The submission route is a statistical template, not the usual supervisory filing channel, so the owner of the process needs to be named rather than assumed.
• Restatement handling. Monthly figures reported early will sometimes need correction as the close firms up. An insurer should decide in advance how it flags a restatement rather than deciding it the first time one is needed.

For brokers, the relevant compliance observation is different. The circular creates a data obligation on insurers, not a disclosure right for buyers. Until and unless the underlying insurer-level figures are published in a form that identifies each insurer, the practical position for a panel review is unchanged, and any process built now should be built to accept the series if it appears rather than to depend on it.

The reasonable response to this circular is preparation rather than redesign.

1. Read the circular itself, not the coverage. The scope, the three fields and the 15th-of-the-following-month deadline are all in the text. The Hindu reported the circular on 8 August 2026 under the headline that IRDAI has asked insurers to submit premium, claims and investment data for the ISP, and secondary coverage of that kind compresses out the field definitions that decide what the data can be used for.
2. Write down what your panel file weights today. If a new public series arrives, you want a written baseline that shows what the rating rested on before it, so any later change is traceable.
3. Decide the tripwire rule in advance. Specify what magnitude of departure from an insurer's own trailing pattern would trigger a review, and specify that the trigger is a question rather than a rating action.
4. Keep asking insurers for segment-level evidence directly. The aggregate will not answer the segment question, so the bilateral request stays the primary route. Our post on day-one diligence for a newly admitted insurer sets out the questions that work when there is no public track record at all, and most of them apply here too.
5. Do not build a client-facing metric on it yet. A monthly number shown to a client acquires authority the underlying data does not support. Wait until the publication form, granularity and lag are known.