Developing a customer relationship has a long-standing tradition in business. Small firms and many retailers are able to relate to their customers individually. However, as organizations grow larger, marketing departments often begin to think in terms of product development instead of customer relationship development and maintenance. It is not unusual
for the sales and marketing units to focus on how fast the firm can bring a mass-appeal product to market rather than how they might better serve the needs of the individual customer. Ultimately, the difficulty is that as markets become saturated, the effectiveness of mass marketing slows or halts completely. Fortunately, advanced data mining technology enables
insurance companies to return their marketing focus to the aspects of individual customer loyalty. Creative, data-driven, scientific marketing strategies are now paving the way back to the customer relationship management of simpler, efficient economies, while on a much grander, more comprehensive scale.
Reinsurance occurs when part of the insurer’s risk is assumed by other companies in return for part of the premium fee paid by the insured. By spreading the risk, reinsurance enables an individual company to take on clients whose coverage would be too great a burden for one insurer to carry alone. Although reinsurance reduces the risk of large claim amounts for an
insurance firm, reinsurance also reduces the revenue of the firm. The goal, then, is to find the appropriate level of reinsurance. Too little reinsurance and the firm is exposed to too much risk. Too much reinsurance and the firm gives away revenue for little in return. In addition, part of the risk/revenue equation of reinsurance is the need for the firm seeking reinsurance to be aware of the credit risk exposure; that is, the counter-party to the contract—the firm providing the reinsurance—should be able to perform as contracted on claims.
Limitations of Traditional Methods of Analysis
Using traditional methods can lead to policies being reinsured when in fact their risk of a claim during the reinsurance period is minimal. For example, for most classes of general insurance, the distribution of claim amount is markedly skew, having a long tail on the right. If an insurer receives a large number of policies for a particular book of business, the total claim payment amount might be expected to be approximately normal since it is the sum of a large number of individual claims. If, on the other hand, the available data are limited, then the tail probabilities of the loss distribution play a far more significant role. In such situations the confidence level associated with the insurer’s predictions and estimates are reduced, and this diminished level of confidence represents the primary motivation for reinsurance. In such situations, an insurer will often obtain reinsurance on potential losses exceeding some established amount.
Advantages of Data Mining for Reinsurance
Data mining technology is commonly used for segmentation clarity. In the case of reinsurance, a group of paid claims would be used to model the expected claims experience of another group of policies. With more granular segmentation, analysts can expect higher levels of confidence in the model’s outcome. The selection of policies for reinsurance can be based upon the model of experienced risk and not just the traditional “long tail of a book of business.”
Exploration of Claims Distribution
For a typical book of business, an insurer might adopt the log normal distribution as the underlying loss distribution. But for situations in which the data are not warehoused properly or marketing professionals to utilize pre-established data mining models within the context of their campaign management system. Marketers are able to select from a list of such models and apply the model to selected target subsets identified using the campaign management system. The scoring code is typically executed on the selected subset within the data mining product, and the scored file8 is then returned to the marketing professional, enabling marketing to further refine their target-marketing campaigns. This form of integration is commonly referred to as “dynamic scoring” because it reflects the real-time execution of the scoring code.