According to the Coalition Against Insurance Fraud, fraud in insurance steals approximately USD 80 BN across all lines of insurance in North America alone. This translates into nearly USD 400-700 per year in increased premiums for each American family.
Fraud in Insurance has been around in since the inception of the industry. Of late, however, modus operandi has become more sophisticated and the fraudsters have been able to rapidly adapt to changing anti fraud measures.
Why is Insurance Fraud so hard to catch?
- Ideally, fraud should be controlled at the proposal stage and not at claims stage. Hence there is a need to strengthen the underwriting processes. But most insurers are still saddled with the legacy systems and their limitations which make it difficult to identify frauds to a great extent at point of sale / log in stage / underwriting stage.
- The burden of the fraud is eventually borne by the larger policy base in form of increased premium, longer policy issuance cycles, longer claims cycles etc… just to name a few. This in turn adversely impacts the overall customer experience as every customer is looked at with the same lens.
- Increasing frauds impact the financial viability of the company by paying out ingenuine and fraudulent claims.
- Due to increasing frauds, current process of claims management involves claim investigations through outsourced vendors which takes time to unearth hard frauds. It also burdens the rosk assessment team as they have to investigate all claims however big or small with the same approach. Therefore, claim processing has become time consuming.
- With increasing fraud in the industry, assessing whether a customer is genuine or not is a big challenge. In the quest of establishing fraud, the claim assessors at times, lose sight of straight through and genuine claims. This delays the claim settlement of genuine claims leading to negative customer experience and thereby defeating the basic purpose of insurance.
- Insurers lack robust systems in place that caters to identifying and controlling fraud. In the current scenario, insurers are reactive and finds solution once the fraud has happened but to combat with the rising frauds, insurers need to be proactive in detecting frauds much earlier even before it happens or identifying immediately once it happens so that right actions can be taken at the right time
When does Fraud Happen?
Fraud would mostly happen at two points in the policy life cycle:
- New Policy Issuance
- Claims Stage
Earlier, insurance frauds were restricted to suppression of material facts, or “padding,” or inflating actual claims of damages which could lead to higher renewal premiums or refusal to issue a policy. In recent years, the pattern of frauds have changed, and insurers are witnessing greater number of hard frauds. E.G., policy taken on a person who has died before policy issuance; insurance cover for a non-existent entity; policy taken intentionally on a critical ill life; multi insurance frauds i.e. over insuring than eligible by taking policies from several other insurers at the same time; willful wrong categorization of damages (e.g. claiming flood damages as theft, or fire damages) etc…
At the claims stage, early claims fraud is what is keeping a lot of insurers on tenterhooks. ‘Early claims’ are defined as claims that come within 2-3 years of policy issuance. While claims in themselves impact the insurers overall profitability, carriers are usually prepared for them. Based on previous years’ experience, carriers mark out reserves that would be used for any claim event. Early claims, on the other hand, throw a spanner in the works for carriers. These are claims that were not anticipated so early in the policy lifecycle.
Proportion of early claims fraud are typically of the order of 1% or less of the policy portfolio.
According to the Insurance Information Institute, Healthcare, workers compensation and auto insurance, are most vulnerable to insurance fraud.
Using analytics to tackle Early Claims Fraud
The type of fraud at each stage varies and requires a different anti-fraud approach. Most insurance carriers use a flag based / business rules driven anomaly detection approach as the first fraud filter. However, this approach has a higher probability of throwing out false positives, which can impact the overall customer experience. Another challenge with this approach is that the frequency of refreshing these business rules may not be often enough. As per a report by the Insurance Fraud organization most insurers (34%) refresh their automated red flags/business rules annually.
Predictive modeling can help insurers identify potential frauds with greater accuracy and confidence. Fraud detection models work with many internal and external data sources to send out alerts in real time. Traditional data sources such as CRM, claims database, policy administration systems, etc… are now supplemented with external sources as credit bureau data, social media, etc.
Predictive models are self learning. After the initial training and run, they require minimum handholding except for tuning the models to adapt to newer anti fraud techniques. These models track nearly hundreds of variables to identify potential fraudulent cases. These variables could range from the customers demographics data and policy information to macro trends of their segment and specific geographic indicators.
When does Analytics come into play?
Insurers large and small have dedicated teams to counter fraud. However, fraudsters have consistently managed to stay ahead of the curve by exploiting some or the other loophole within the system. While it may not be possible to address all revenue leakages due to fraud, advanced analytics and predictive modelling can stem the flow to a substantial extent. Analytics models at the time of new policy issuance can help categorize proposals as either low risk or high risk or something in between. The high risk proposals are given additional scrutiny and the low risk proposals can be processed straight through.
Likewise, at the claims stage, the low risk claims can be processed straight through and the high risk claims can be investigated in detail. This optimizes the effort and time of the team and improves the overall customer experience as well