Facing facts: Soft fraud costs the insurance industry billions of dollars annually, specifically in premium “leakage,” a problem that results in artificially low premiums and unnecessary or over-inflated claim payouts. The NAIC (National Association of Insurance Commissioners) estimates that fraud is a problem that impacts the industry and consumers by over a hundred billion dollars per year. Again, according to the NAIC, fraud not only imposes costs on insurance companies and threatens their competitiveness and future viability, but it is also financially damaging to consumers and detrimental to the economy and society as a whole.
Stay Tuned for More
In this series of blog posts, I will cover the “why, what, and how” of underwriting fraud detection. That is, why it’s such a big problem for insurance companies, what a contemporary and technology-forward solution should look like, and then how to best implement and leverage an AI/ML-based approach for better fraud detection that incorporates speed and accuracy.
Extrapolating on Premium Leakage
Premium leakage is defined as the premium or revenue lost due to misclassification, missed exposure, changes to exposure, fraud, or the failure to recognize material facts related to the insurance premium. Premium leakage results from misapplying or missing any premium-affecting factor in the underwriting process.
Insurance underwriters evaluate the risks involved in insuring people and assets and establish a price for that risk, attempting to separate the acceptable risks from the bad ones. While this may sound like a relatively straightforward process, it is fraught with a confluence of participant motivations and fraud-related mechanisms (for both consumer and provider, both intentional and unintentional) that inevitably can work to undermine the overall symbiotic structure.
Premium leakage may occur as the result of a confluence of different factors: customers provide incomplete information, agents are not incentivized to thoroughly vet and exclude risks, and insurance companies do not have effective capabilities with which to gather and process supporting information. At the heart of premium leakage is the issue of information asymmetry in insurance contracts: the fact that the customer almost always has material information that is not disclosed, either consciously or unknowingly, to the carrier.
Striving for a Balanced Approach
Insurance companies must balance their approach to underwriting: if too aggressive, greater-than-expected claims could compromise earnings; if too conservative, they will be outpriced by competitors and lose market share.
And though premium leakage is widespread and generally acknowledged, it often falls under the remediation radar both due to its subtle, incremental, and difficult to detect nature and because of the inherent difficulty in creating the efficient processes needed to identify and correct the problem.
Diverse Constituents Complicate the Issue
Soft fraud realities in insurance are not relegated to the participation or concern of only one of the involved constituents. Each party, be it the customer, agent, or insurance company, all may have a part to play in perpetuating the larger problem.
- The consumer: The customer may fabricate information to obtain a policy that otherwise might not be granted, or to manipulate premiums to their advantage, costing underwriters to absorb added risk and payout on claims.
- Insurance agents: Agents may be guilty of lax policies in the underwriting processes, to gain new clients and near-term commissions. This is exacerbated by the desire for upsell opportunities and the potential to sell to and lock in clients to multiple policies. Similarly, agents can engage in manipulation of rating variables that influence rate calculation, underfill applications, and manipulate policies post bind to reduce premiums.
- The insurance company: Many companies may be overly aggressive in granting policies, again to meet sales and associated revenue goals, frontloading near-term results without consideration for longer-term ramifications. Insurance policy discounts such as those for affinity groups (e.g., nurses, teachers, fraternities, etc.) are a common practice. Carriers and brokers offer discounted policy premiums based on documentation submitted by agents and customers, data which may be outdated or simply inaccurate.
Fraud, particularly in the underwriting process where it assumes many dimensions, is a widespread and often acknowledged problem; however, it is just as apt to be overlooked or inadequately addressed for the reasons cited earlier. It is a dilemma with a seemingly limited path to a positive outcome.
Recognized Problem, Limited Response
Insurers do spend significant resources in auditing application, policy, and claims documents in efforts to catch and eliminate fraud, and virtually all have investigative teams devoted exclusively to this issue. However, such efforts are reactive, time-consuming, and ineffective in delivering any type of real-time detection and resolution. Manually analyzing and vetting the hundreds ̶ even thousands ̶ of documents necessary for ongoing policy evaluation and underwriting, and subsequent claims payout, just does not rise to the task.
Business rules and policies have been developed to search for and identify anomalies, duplications, and unscrupulous manipulation, but with the daunting amount of both structured and unstructured documents in the mix, such efforts are not scalable and results are often inaccurate or biased. Audit and fraud teams seldom have insights into individual agent’s behavior, further blocking the ability to resolve fraud-related behavior or practices.
So, we have looked at the “why” of insurance fraud and premium leakage. Why it is such a huge issue for insurance companies, why it is multi-dimensional and complex, and why current solutions come up short in helping to detect and remediate the problem. Stay tuned for more thoughts on what an AI/ML solution can and should look like- the true “what” that is needed to detect, catch, and combat premium leakage fraud using applicable tools that include computational analysis. Then I will tackle the “how” of getting it done.
About the Author
Max Kanaskar is CognitiveScale’s Financial Services AI Advisor. In this role, Max works with financial services organizations (including insurance companies, banks, asset managers) on their AI journey—from strategic insight into how to develop AI competencies and centers of excellence to more tactical development of AI roadmaps and delivery of AI solutions.