Business Intelligence for Insurance : Optimal Risk Sharing - SQLiaison


	Risk-Sharing Pool optimal cession management platform Risk Sharing Pools (RSP) have been enacted in many Canadian provinces to provide a form of reinsurance to automobile insurers: an insurer can cede a certain fraction of its book of business to the RSP. The pool then reimburses the claims made by insureds ceded to it. In return, the insurer pays to the pool between 65% and 75% of the annual premium, depending on the province. The insurer also continues to administer the policy. This mechanism prevents an insurer from being exposed alone to the entire risk of an "uninsurable" driver. The industry collectively assumes the losses of the pool: each insurer contributes in the pro-rated amount of its market share. An important question remains: which risks should be ceded to the pool in order to take maximum advantage of it? And the opposite also holds: which risks should be withdrawn from the pool because of advantageous performance yields? Risk Sharing Pool Optimization Insurers across the industry use various scoring models to identify under-rated risks, those for which we estimate that the mathematical expectation of the claims to be incurred by the insured is greater than the RSP designated percentage of the annual premium. For those risks, it can be established that ceding to the RSP is profitable. Currently, few insurers can manage their Risk Sharing Pools both reliably and profitably in large part because of their use of predictive technology that is unable to capture complex nonlinear interactions between risk factors (e.g. Generalized Linear Models). As a result, most companies do not cede the maximum premium volume allowed by the pool regulations. Insurers are still looking for models enabling them to make a more profitable use of the RSP. This goal can be reached with the FellowDSS^TM Optimal Risk Sharing module. What is FellowDSS^TM Optimal Risk Sharing? FellowDSS^TM Optimal Risk Sharing enables an insurer to better identify the most under-rated risks in order to cede them to the RSP. The system relies on scientific advances in statistics, data mining and predictive technologies (e.g. Neural Networks). FellowDSS^TM Optimal Risk Sharing is built on the most comprehensive Business Intelligence Platform for Insurance, FellowDSS^TM Foundation and benefits from the robustness and data integration capabilities inherent to FellowDSS^TM Foundation. The more historical data and rating attributes that are available to the FellowDSS^TM Foundation, the larger the number of relevant interactions the Optimal Risk Sharing module will be able to reliably identify and exploit in order to discriminate between risks to be ceded and those to be kept. In addition, users can take advantage of the FellowDSS^TM Foundation for reporting and analytics pertaining to RSP and possibly to other mission-critical aspects of the business. Implementing FellowDSS^TM Optimal Risk Sharing does not require any changes to the insurer's business processes. The aspects pertaining to ratemaking, marketing, legal, or relationships with brokers remain completely unmodified. FellowDSS^TM Optimal Risk Sharing is truly a non-invasive solution to RSP management issues. Benefits include: Increased profitability; Making it profitable to cede to the RSP a higher premium volume. Increase reliability; Proven robust predictive technology that selects risks based on the insurer's previous experience, reducing error-prone human intervention in the process. An easy way to decrease overall operating costs and increase revenues. Access to Reporting and Analytics to support other mission-critical insurance processes such as IBC reporting. Low-cost initial feasibility study using insurer's historical data to reliably estimate the financial benefits according to the insurer's specific situation and market position. Decrease your annual payment to the pool while increasing your competitor's. How does FellowDSS^TM Optimal Risk Sharing deliver these benefits? FellowDSS^TM Optimal Risk Sharing uses the differences between the insurer's ratemaking models, that can not be purely discriminative due to regulatory, marketing, and strategic aspects of the insurance business, and our model that is free from these considerations. From these differences, the insurer gets the Risk Scores enabling profitable ceding to the RSP. The pure premium is estimated by precise Modelling. The insured are sorted following the under-rating order according to the ratio: estimated profit/annual premium. The most under-rated risks are ceded to the pool to the maximum allowed. Actual annual net savings example of FellowDSS^TM Optimal Risk Sharing The plot illustrates the annual net savings that are attainable for a medium-sized insurer ($100M/yr in premium volume) under the Quebec Automobile Risk Sharing Pool. The curve shows the realized savings at year-end against the predicted savings at the beginning of the year, before having "seen" any claims data for the year. Savings made possible by FellowDSS^TM Optimal Risk Sharing according to % of ceded risks under the Quebec RSP rules.

	Claims data provided by a large North American auto insurer and exclude bodily injuries. Statistical models estimated on 1994-96. Out-of-sample test result (this graph) on 1997. Results scaled for an insurer with $100M in annual premium volume, a 15% underwriting rejection rate and a 65% loss ratio. For more information on FellowDSS^TM Optimal Risk Sharing, please contact us.