Technologies Behind Underwriting

Identity theft reports, the Anti-terrorism Act, the Canadian Coalition Against Insurance Fraud’s (CCAIF) “Crime Stoppers”, the Financial Transactions and Reports Analysis Centre (FINTRAC), and adjacent market fraud cases both in Canada and the U.S. have created new databases that contain pertinent information needed by present day insurance underwriters. Integrating external data into current operations can be a nightmare. Fortunately, new solutions have been developed that combine the mountains of external data, expert analytics, and a framework to bring it all together.

The Insurance Bureau of Canada (IBC) has been collecting data from the insurance industry for decades to help determine underwriting rates and risks. The IBC’s insurance data enquiry and access (IDEA) system provides access to their data warehouse using advanced industry standard technology. The following data from the IBC and other industry sources are commonly used by insurance underwriters in the industry today:

Claims histories. Claim history information enables insurance companies to independently evaluate the potential underwriting profitability of individual personal and commercial risks and risk locations using prior claim information in the underwriting and rating process. Carriers can access their own data aggregated in their internal claims systems, or can join a claims history consortium to view claims histories from other companies.

The Investigation Services Division (ISD) of the IBC manages a claims information database, which is comprised of all types of insurance claims information. Its purpose is to allow informed underwriting and claims decisions.

CLEAR. The Canadian loss experience automobile rating (CLEAR) system was developed for rating vehicles based the premise that the vehicle specific portion of insurance rates should be based on two principal factors: the likelihood of vehicles being involved in claims; and what it will typically cost to settle each claim. CLEAR uses Canadian insurance claims experience data obtained from the official statistical agencies in Canada.

Insurance fraud. Insurance fraud has sharply increased in the last decade. The CCAIF estimates that insurance fraud costs Canadians $1.3 billion annually. Traditionally, carriers have focused their fraud fighting resources on incoming claims, but the costs associated with identification, investigation, and often times, payment recovery, have reduced or eliminated fraud savings. Carriers have now realized that the data collected on historical fraud claims and fraud trends can be applied during the underwriting process to identify potential fraud before an insurance risk is assumed.

Investigative Services Division. The ISD (which also operates as the Insurance Crime Prevention Bureau) works in cooperation with insurers, law enforcement agencies, and the CCAIF to detect and prevent insurance crime and to gather evidence in aid of prosecuting offenders and securing restitution. The ISD provides information services to insurance carriers by data analysis request, a batch query, or manual online searches.

Crime Stoppers. The CCAIF formed an alliance with Crime Stoppers in 1995 to make it possible to report insurance fraud anonymously. Crime Stoppers passes the information on to ISD and/or local police for investigation, but the data collected by Crime Stoppers can become useful in determining a previous criminal activity or intent.

OTHER DATABASES

Beyond insurance industry data, a variety of other databases should be considered when identifying underwriting risks.

Motor vehicle records. Much like its IDEA system, the IBC is working with the Ontario Ministry of Transportation (MTO) to provide improved access to driver records (MVRs) for the insurance industry. Using industry standard technology, all insurers, intermediaries, and service providers will be able to connect directly to IBC for more accurate and immediate insurance pricing for customers. To improve the accuracy of VIN data, the IBC provides a service to match numbers against the MTO registered vehicle files and perform regular VIN edits for Ontario premium transactions. Non-matches are returned to the insurers for investigation. VIN Check is used to verify whether a particular vehicle is registered at MTO before it is sent to IBC.

Address verification. A frequent activity of common-day fraudsters is to create a “mirage company” or corporation to act as a previous employer to legitimize employment history. Address analytics, available through the major credit bureaus, as well as data providers like Lexis-Nexis and Qsent, can be used to validate a current or former residential or employer address.

Confirmed fraud repository. The major credit bureaus and data aggregators are creating repositories of confirmed fraud data. Subscribers to the databases agree to deposit confirmed cases of fraud found at their company. In return, they are permitted to access the repository to identify potential fraudsters in their incoming transactions. Various types of fraud are reported through the repositories including identity theft, account takeover, fictitious identity, money laundering, asset fraud, and application data manipulation.

Bank fraud. Individuals and businesses that are identified in the Proceeds of Crime (Money Laundering) and Terrorist Financing Act and regulations are required to report certain prescribed transactions to the Financial Transactions and Reports Analysis Centre (FINTRAC). Accessing this data can help identify potential risks in financial transactions.

Securities fraud. The North American Securities Administrators Association (NASAA) and the Canadian Securities Administrators (CSA) keep detailed records on disciplinary actions related to offenses in securities firms and by securities brokers in Canada.

Check fraud. Many government and banking industry organizations collect information related to check fraud, but new consortiums, such as the American Criminal Investigators Network (AMCRIN), have developed over the last five years in which retail and banking members share their check fraud related data for the ability to be able to search a consolidated database of check fraud.

SOLUTION FRAMEWORK

A general framework to bring all of the databases and analytics mentioned above is needed to properly assess an underwriting application. The ability to access and analyze the data without creating a data warehouse is paramount to the success of creating a risk assessment solution. The components listed below are in the market today and should be combined to create an end-to-end risk assessment system.

Virtual data warehouse. Connections to the various data sets mentioned earlier can be costly and insecure if each must be managed separately. Frameworks have been developed however, that can connect relevant data sources from public and private databases to serve specific domain needs. A central access point that has pre-configured access controls and data management creates a “virtual data warehouse” that frees the end customer from creating relationships and connections with each data provider. Additionally, creating a single secure connection to the virtual data warehouse is much easier to maintain and ensures safe data transmission.

A virtual data warehouse enhances “scalability” as all the searching and scoring are performed at the remote database: therefore, the solution scales as the database scales. This virtual data warehouse also ensures that data is as current as possible. This ability to search data without having to combine databases also solves issues of data access and ownership between organizations thereby solving privacy, political, and legal issues.

“Plug-and-play”. Solutions that utilize a “plug-and-play” architecture facilitate rapid integration of multiple analytics such as neural nets, rules engines, clustering engines, predictive modeling engine, similarity search, etc. A plug-and-play architecture provides for a lower total cost of ownership as carriers can use tools and analytics already in place and simply add advanced analytics as they are released into the market. Furthermore, the entire system does not have to be replaced when new technologies become available.

Automated risk assessment. A component that processes real-time data in the background by classifying and assigning predetermined categories to records based on business rules helps improve the productivity of the investigative staff and promotes a consistent approach to risk assessment. It is important for the risk assessment engine to carry “all decisions” so that the user can access the reasons for a particular record being placed into a particular category. This rationale allows investigators to comprehend the results of the machine-driven categorizations and provides an audit trail.

Automated data mining. Analysis engines that process thousands of records in the background to identify non-obvious relationships help investigators discover links in the data that they would not normally find. High performance processing that compares one or more input or source records against one or more target databases in real-time reduces investigators to support real-time decision analysis.

“FIELD” COMPONENTS

The following components should be integrated into the overall risk assessment framework so investigators have the same access to the data and analytics that the automated systems provide.

Case management. Web-based user interfaces for case management are a necessity as investigators are routinely out in the field. The ability for domain experts to receive alerts that are particular to their geographical area or expertise is another improvement to productivity.

Link charting. Each investigator has his or her own philosophy or methodology when conducting an investigation. Investigative paths are often spontaneous and difficult to track. An effective link charting component that ties into the risk assessment framework can allow the domain expert to further investigate records that have been referred by automated systems for more detailed analysis. A “visual link-charting tool” makes it easy for the user to find relevant connections between people, places and/or events.

Solutions exist today that provide a framework to bring remote databases together for risk assessment. Integrating advanced analytics with domain expertise provides the alert mechanisms needed to identify questionable applications as they are submitted – giving investigators more time to reach a conclusion on the underwriting risk.