Reinsurance Data Analytics
The cost of reinsurance can be a significant part of an insurer’s expenses, particularly in the current hardening market.
One way for insurers to get on the front foot with reinsurance is to improve data quality. Reinsurers tend to penalize insurers for poor data quality, and in extreme cases, could refuse coverage. The last thing an insurer would want is additional charges or refusal of coverage due to poor data quality.
Data requirements
Typically data requirements for reinsurance are presented by grouping data on individual policies/risks. The following is a list of key data requirements and a brief explanation of each item.
- Risk and loss profiles: Group policies/risks insured together with associated claims by their sum insured values into different pre-defined levels (generally called sum insured bands).
- Large losses and catastrophe event losses: A list of historical Individual large claims above a threshold and historical claims from affected policies by catastrophe events.
- Historical experience of reinsurance coverage: Typically known as treaty statistics or treaty results generally capture historical premiums ceded to reinsurers, commissions given by reinsurers, and claim recoveries from reinsurers.
- Catastrophe exposure aggregates: Sum insured values of policies applicable to coverage for catastrophe event losses.
- Revised and projected premium income: Revisions to premium income applicable to historical reinsurance coverage periods and estimated premiums applicable to the new coverage period.
Risk and loss profiles
Generally, classes of business risk profiles are set up by the underwriting period, in which all policies incepting within a specified period are captured.
Large losses and catastrophe event losses
Reinsurers would require historical individual large claims above a certain threshold for their analysis and pricing. For example, for a non-proportional reinsurance arrangement, the threshold can be 50% of the deductible whereas for a proportional reinsurance arrangement a cash call, in which large claims above a predefined value are notified, can be the threshold. For non-proportional reinsurance arrangements, it is important to provide individual claims below the deductible as historical claims are adjusted for inflation, some of these claims can exceed the deductible making them recoverable as-if basis. For proportional reinsurance arrangements, large claims are important to determine large loss ratios.
Historical Experience of Reinsurance Coverage
The historical experience of reinsurance coverage typically known as treaty statistics or treaty results is very important to reinsurers for their analysis and pricing.
- For a Surplus reinsurance arrangement, It captures, the coverage year (Treaty year), the premium ceded to Surplus, the commission received from reinsurers, and recoveries from reinsurers with breakdown by outstanding and paid amounts. Treaty Results show final results from reinsurers’ perspective (i.e. ceded premium – commission – incurred claims).
- For non-proportional reinsurance arrangements, treaty statistics typically capture the premium income applicable (generally known as Gross Net Premium Income – GNPI), chargeable rates on GNPI, reinsurance premium amounts, adjustments to reinsurance premiums, loss recoveries and reinstatement premiums if applicable.
Catastrophe exposure aggregates
Generally, Property, Engineering and Motor classes are exposed to natural catastrophe perils. Reinsurance coverage for these and other exposed classes can have a significant bearing on reinsurers providing coverage for them. Therefore, it is paramount that insurers provide sum insured values of policies with coverage for catastrophe event losses (i.e. exposure data) to reinsurers for their analysis.
Typically, exposure data are provided by grouping exposure data from individual policies. The grouping can be at postal code, province, or other defined category levels such as Cresta Zones. However, the granularity can also be at the latitude/longitude level of risks, particularly for facultative reinsurance arrangements.
It is recommended to have exposure data by occupancy types (e.g. Residential, Industrial, Commercial, Warehouse, etc.) and coverage types (e.g. Building, Contents, Business Interruption, etc.) with breakdown by reinsurance arrangements similar to risk and loss profiles for different peril types.
Ideally, exposure data extraction should be at a specified cut-off date, in which all in-forced policies on that date are extracted. For proportional reinsurance arrangements, preferably, applicable in-forced policies should be separated by coverage year. This separation can help to apply event limits by coverage year with catastrophe risk modeling for runoff proportional reinsurance arrangements.
Revised and projected premiums
It is important to provide projected premium incomes for the new coverage year. The projections include Estimated Premium Income (EPI) ceded to proportional reinsurance arrangements and estimated GNPI applicable to non-proportional reinsurance arrangements. They should be for each class covered. In fact, these projections reflect growth assumptions and are important for risk-adjusted pricing by reinsurers. Generally, reinsurers use growth in premium as a proxy for the growth in exposure (e.g. catastrophe exposure for classes exposed to catastrophe perils).
For old coverage years, revised EPIs for proportional reinsurance arrangements and revised GNPIs for non-proportional reinsurance arrangements are required for premium and commission adjustments.