Simply stated, underwriting is the process of determining what loss exposure will be insured for what amount of insurance at what price and under what conditions. That can be challenging because of the emergence of direct Internet sales, the complexity of data needed, and the competition to attract and retain customers who are a good insurance risk.
Underwriters are responsible for defining enterprise risk policy around acceptable physical, moral, attitudinal and legal hazards. Fraudulent policyholder behavior is linked to moral and attitudinal hazards and has significant impact on severity and frequency results:
- Underwriting rating errors caused by application misrepresentation has led to an estimated $15.4 billion a year in premium leakage cost.
- Property and Casualty Insurance fraud is estimated to cost insurers between $33 and $80 billion.]
So, fraud matters. And avoiding risks that show a propensity for fraud should be foremost in the enterprise strategy. Underwriting is the most effective point in the insurance lifecycle to address fraud.
Analytics helps underwriters be successful
The difficulty with applying analytics to the underwriting process is that there’s a vast amount of disparate data throughout the organization. But in-depth analytics of internal and external information is the best way to combat adverse risk selection.
For assessing new business, this means you have to identify risk profiles that should be rejected outright or sent to experienced underwriters for further review. And this has to be done in real time. But only 40 percent of carriers collect and analyze their data for fraud detection in the underwriting process.
Claim loss and policy history are critical data elements. From this data you can create predictive models associated with detailed, high scoring attributes to identify adverse selection and premium leakage – by geography, business type, applicant or policyholder and insured asset.
You can use advanced analytical methods to aggregate, match and analyze both structured and unstructured data. These should include:
- Predicative analytics
- Anomaly detection
- Business rules
- Visual intelligence
Another reason for real-time
Having access to real-time data and information is the necessary attribute for a fraud detection methodology. With the incredible growth of online insurance sales and service, collecting and quickly analyzing underwriting data is becoming more and more important: The online channel makes it extremely easy for those with fraudulent intent to hide information or create paper policies for organized accident schemes. And, it would be very easy for an applicant to get the cheapest premium by simply changing demographic information before anything is submitted to the carrier.
The bottom line is that focusing powerful analytical tools on improving risk selection is necessary for underwriting’s essential task: developing and maintaining profitability.