Hit by several fraud-related losses every year, the insurance industry is now deploying data analytics to detect policies filed under false information
One of the biggest challenges that the insurance sector faces is that of fake claims. According to industry experts, the insurance industry is annually hit by fake claims to the tune of 20-22%. The latest amendment in Section 45 of the Insurance laws Amendment Ordinance means that no life insurer can repudiate a claim from a policy holder three years after the issuance. This means that a policy can’t be canceled on the grounds of miss-statement of facts—even fraudulent claims are protected.
Unique Initiative to Fight Insurance Fraud
Sunder Krishnan, Chief Risk Officer, Reliance Life Insurance, decided to take up the issue of ‘fake claims’ head-on. With many years of experience in handling claims in the insurance industry, he was in a position to understand the general strategy adopted by the filers of false claims.
“I was heading claims temporarily, post the exit of the COO a few years back. While having got the chance to work in the department, I came to know about the factors that generally lead to false claims. My conclusion was, we will have to go after the fraudsters from the very beginning. I designed an investigation process based on certain parameters—location, occupation, age, products, etc. A risk matrix was prepared based on this criteria. We embarked on an initiative known as Post Issuance Risk Verification (PIRV),” says Krishnan.
According to Krishnan, this was the first of it’s kind initiative in the life insurance industry. They were planning to take the potentially fake policy holder by surprise. Under this system, the suspect policy holder would be detected by the help of Information Technology and after that the investigator from Reliance Life would do a reality check—verifying the details of the policy holder by visiting him at the location.
Usually the insurers begin their investigation after the claim has been filed, but by then the fake claimant is prepared to prove himself innocent. A surprise check is important as it leads to the fraudster being taken by surprise. It should be noted that filing of fake claims is in essence a team effort—the doctor, pathological labs, lawyer, hospital all work in tandem to perpetrate the fraud. At times, surveyors, investigators and agents too get involved in the shady deals.
However, the possibility of Krishnan’s pro-active strategy of going after the fraudsters, having an impact on the honest customer could not be ignored. The drive could lead to the cancellation of the account of a genuine policy holder. To ensure that such mistakes did not happen, Krishnan used cutting edge IT to evaluate the customers. The company is now using SAS data analytics to develop information about the locations from where frequent bad claims are generally generated.
“The total investment on PIRV is Rs. 50 lakhs—this includes the cost incurred on investigation. But we have able to save a significant sum by cancelling the fraudulent policies upfront and not waiting for the claims to be filed. The total savings in the first year is about Rs. 20 crore. We follow the IRDA regulation while cancelling any policy in which the customer has given false information. We do our homework and collect evidence against the potential fraudster,” says Krishnan. He is hopeful that in the second year, PIRV will lead to savings of close to Rs. 50 crore.
Hunting for Bad Claims
The maximum number of fake claims are being filed by organised syndicates operating out of selected pockets in the country. The first priority is to identify the locations from where they are operating, and this must be followed by the verification of their details like occupation, age, product, etc. Krishnan says, “SAS data analytics helps us mine the data and it throws up the list of the most likely negative locations, from where customers who plan to file fake claims in future are operating.”
The risk teams across the country constantly engage with the underwriting team. The underwriting department is responsible for checks and verifications of the details in the customer forms filled by the agent. They escalate forms with suspicious details back to the central retail team. The suspicious cases are coloured in red, amber, green, based on the risk matrix, which is linked to the parameters of location, age, occupation, product.
“We go after the red and amber cases. We don’t go after the green cases,” says Krishnan. Earlier the colour coding of the suspicious cases was being done manually, but now the process has been automated.
Every month, the investigators are finding 4-5 cases of Insurance policies of people who are dead and yet have policies named after them. Cases of impersonation are also being discovered. These cases are forwarded to the ethics committee and after the assessment, appropriate action based on the disciplinary matrix is taken.
Recently the system detected a fraudulent policy, which was worth more than 10-15 lakhs. The business type was not mentioned and the other parameters were also unconvincing. The location also seemed suspicious. “On the basis of the sum assured, location, area, occupation, we picked up that case and sent the investigator who found out that the person in whose name the policy was taken was dead even before the policy had been purchased,” says Krishnan.
The fraudsters were planning to pad the company with a claim, but as the investigators took them by surprise, they could not place their claim and the fraudulent policy was canceled. Now the underwriting processes have been tightened in all the negative locations. “We have put cameras in the pathological labs to capture images of all the insurance applicants to check if the person who has come for the check is the same person who applied for the policy,” says Krishnan.
The Way Forward
Having successfully deployed PIRV to detect insurance fraud, Krishnan now wants to take the system to the next level. He is in favour of the Insurance industry putting up a collective fight. “I have organised joint meetings on fraud prevention. We have brought this to the regulator’s notice to look into this area. We have also done a lot of spade work for putting in common systems for the Industry so that the system gives lag and lead indicators,” says Krishnan.
The lag indicator is based on past data. For instance, the data on disproportionate claims from certain locations, which is higher then the LIC table. Lead indicator is predictive in nature, based on analytics systems, which predict the persistency issue about who is more likely not to pay the claims next year.
If you have an interesting article / experience / case study to share, please get in touch with us at editors@expresscomputeronline.com