By Thomas Chesbrough, Max Seybold, & Stephen Brobst

Part 2: Mining the Insurance Data Warehouse
by Max Seybold

Advancement in the insurance industry has been hampered by a lack of investment industry-wide in data warehousing and data mining technology. In their absence, product designers must work with limited detailed data about policyholders and claim histories. Instead, they rely on data summarizations (actuarial tables summarize data), supplemented by intuitive "hunches," to guide the development of new policy products, creative rating structures and appropriate pricing.

In addition to constraining new product design, the lack of data warehousing in the insurance industry also impairs the successful marketing of what few new policy products are introduced. Furthermore, these policy products, which are based on summarized information and intuition, potentially increase risk for the agent, the carrier and the reinsurer.

Finally, the lack of detailed information about policyholders and claims has prevented the insurance industry from following the lead of its "cousins" in the financial industry, such as mortgage banking, which have developed mechanisms for securitizing products after they are sold and selling product-backed securities in a secondary market.

The potential benefits of data warehousing and data mining for the insurance industry are similar in many respects to the benefits realized in other industries where data warehousing is in widespread use. The insurance data warehouse, once established, might contain keys for improving business operations and for designing new products by identifying correlations and patterns in the data that could not be otherwise deduced or intuited in the absence of data mining analysis.

The uniqueness of insurance data warehouses derives from the predisposition of the industry to wait for technologies to establish themselves and to become productized before they are adopted. By their nature, data warehouses and data mining technologies will never be "off-the-shelf" commodities.

Assuming that visionary insurance industry leaders develop an appreciation of the potential of the technology, the data warehouse can deliver a wide range of benefits to the company. For one, data warehousing will enable insurers to create better policy types and to provide product bundles that better target specific customer classes. Information from data mining will enable these new products to be fielded without creating undue risk to the insurer.

Data warehousing may also enable the creation of better pricing strategies (within the constraints of state laws) for products. Pricing reflects a risk rating for most property and casualty policies. By basing prices on a more exacting analysis of risk potentials--a function made possible through data mining--insurers can price products in a manner that attracts better customers while reducing their own loss exposures.

Ultimately, data warehousing will set the stage for the securitization of risk portfolios. Based on the experience of the mortgage banking industry, there is significant potential through data warehousing for the criteria-based definition of insurance policy-backed securities that could be sold in aftermarket to reinsurers and private investors. Forward-looking property and casualty companies that deploy data warehousing and data mining strategies first will be in the best position to capitalize on this potential revenue stream.