Fraud busters and financial crime intelligence doyens at the recently held fourth annual South African insurance Crime Bureau’s (SAICB) in Midrand, Johannesburg were told there is an urgent need to address the quality of fraud models within the insurance industry, particularly on the short term side of the game.
Head of Claims at Momentum Short Term insurance, Shaun Marshall, told the insurance fraud and financial crime tackling audience that “most of us are using outdated models for fraud detection.”
The need for AI and machine learning
Fraud combating systems among smaller insurers have limited data sharing capabilities and require cash injections for systems – high-ups may not consider this top priority because of the cost implications. They however end up paying in other ways.
“Many of our fellow insurers aren’t making much. In 2018, the top ten insurers made an average underwriting profit of ten percent. The margins are not as lavish as the general public perceives. The sad reality is if you sit with a system that is outdated and you have the inability to measure, track and share data, you are paying for the new system. It’s simply under a different budget line and it’s called claims/loss ratio” warned Marshall.
The higher income insurers may have advanced systems in place - however they can’t keep up with the risk imposed by fraudulent clients who keep on finding new ways to stick their hands in the cookie jar.
“Others have systems, but they are not delivering. These individuals (fraud detectors) have sophisticated red flag systems and fraud models – whatever you want to call it. They’ve had the systems for years but a lot of them, when you speak to them say they struggle to prove the ROI to their executives. And secondly, these systems are becoming less effective. The problem is, a fraud model is like a marriage, once you stop working on it, it stops working for you. Most of us are trying to solve today’s problems with last year’s modus operandi. We need to move toward AI and machine learning.”
Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.
The numbers
According to Marshall’s expert “non actuarial calculations that make sense to a business person”, an estimated 32 percent or R19,3 billion worth of claims, of the annual R60.3 billion cost of claims for the short term insurance industry, are fraudulent per year. He got the amount from his own independent research and analysis of the short term insurance industry.
“Let’s put it in further perspective. Commercial crimes is up by 14.4 percent this year and fraud is costing our industry a potential R19.3 billion. So how much do we use as a combined industry to combat this problem? Before answering that, consider that SAICB’s revenue was R36,6 million according to their last publication for the financial year. Expressed as a percentage of our gross written premium of the industry, that means we spend 0.03 percent on our bureau to fight fraud,” he stated. He then questioned the representatives in attendance if the industry took fraud seriously.
Considerations
Innocent clients don’t want to feel like they are being criminalised by fraud detection processes and at the same time, insurers and financial crime fighters must pin down the fraudulent clients, so a strategic strutting of the fine line is always necessary. The determination is that the drive must be towards adding value to those law-abiding clients who understand the importance of quality insurance – the insurer has a responsibility to protect those clients and their money, as in safeguarding the reserves.
Analysing criminal trends
The ability to correctly analyse trends is a critical component of fraud detection and last year, reports showed that syndicates are moving towards commercial crime because the data is limited – hence the 14.4 percent spike in activity on the charts.
Head of Business Support Services at Bryte Insurance, Sedick Isaacs said South Africa is “not unique at all”, certainly one of the worst in the world but certainly not on its own.
Speaking of ways of improving the echelon, he said “Daisy analytics (AI) is a serious thing that needs to be looked at quite seriously in future. As insurers, we need to invest in it with the SAICB, as a partnership. We need to look at how we can understand the trends in fraud; that is critical. Trends will show us the red flags and areas that we need to be focusing on. There are trends that emerge that are easily identifiable, but it goes with training and making sure that there is education among our staff and all the professionals.”
“Trend analysis, claims management processes and new and improved claims handling tools all go together to provide a model that says ‘that particular claim has x percent chances of being a fraudulent claim’. Those are the kind of tool that we are going to have to work through in future because, the fact is, the fraudulent clients out there are getting cleverer about how they do things. They are also using the same type of equipment that we use” he explained.
Writers’ thoughts:
The fight against fraud is an important one. However, as Marshall pointed out, insurers should never forget their purpose. Their purpose is not to fight fraud but to provide quality service and when dealing with claims, their initiatives should not be all about rejecting claims but about protecting the consumer. Please comment below, interact with us on Twitter at @fanews_online or email me your thoughts mashoto@fanews.co.za.
Comments
Added by Stuart Collins, 18 Mar 2020Clients have the alternative view that short term insurers are on a mission to cheat them when they need to claim.
The lack of trust from both sides is a sad reflection on our beloved industry.
There will always be those companies that look for reasons to pay a claim and the others who look for reasons not to pay.
As brokers we need to know the difference. Report Abuse