Four steps to a solid anti-fraud framework
Gilles Renouil from Women’s World Banking explains their approach to combatting fraud in microinsurance
Every year, the insurance industry spends billions of dollars to prevent and detect fraud and yet a large part of fraudulent cases remain unnoticed. So when a global insurer told Women’s World Banking that they were not participating in the tender process for our microinsurance program in Egypt, should we have been surprised? The insurer was frank in their reasoning: they did not believe that fraud could be controlled in this segment.
Our work in five low-income markets across the world has proved that fraud is challenging but can be managed.
In this article, we lay out the key lessons and recommendations for an effective fraud protection for microinsurance.
While a good anti-fraud framework should be the same in insurance and micro-insurance, there are a few points specific to this segment that are critical to consider:
· Due to poor financial education, low-income populations will want to “try” the product and submit claims that are not necessarily receivable. It does not mean however that those are intentionally forged. It is therefore important to establish a transparent filing process and draw a clear line of where fraud starts and document each fraud case thoroughly from the beginning.
· Fraud is likely to rise and evolve as first genuine claims start to be paid and the word of mouth starts to spread among clients. As a consequence, the insurer or the intermediary should first give clients the benefit of doubt while they take corrective actions in the process. They also need to clearly discuss the fraudulent cases with clients and staff and apply transparent consequences (such cancellation and penalties). In this process the communication into the communities is particularly critical, as fraudster may otherwise spread false information and damage the product reputation.
· Fraud detection requires dedication, logic and consistency. Because such high value work will hit the bottom-line with an extra cost, it is of utmost importance to design a slim yet cost-efficient framework from the very beginning. This means insisting on control quality rather than quantity, keep evidence and track of all cases, centralize knowledge and analyses.
Given an understanding of this particular market, an insurer or intermediary must next understand the operational implications of fraud detection. An institution’s “fraud equation” can be determined by answering a few simple questions:
· How much fraud are we able to catch? How many cases of fraud (in frequency terms) is the process able to identify?
· How much does the detection really save? What is the volume of fraudulent claims that can be rejected?
· What is the cost of detecting fraud? Having a fraud prevention and detection programs comes with a cost that should be commensurate with the savings generated. One challenge to measuring this cost is the fact that effective fraud prevention will also prevent a substantial part of fraudulent claims before they are filed and can be accounted for.
Based on our experience designing microinsurance schemes that are meaningful for clients while being a sustainable business for our partner institutions, Women’s World Banking has built an approach to fraud detection that can be applied across the industry:
1. Designing a comprehensive anti-fraud framework up front and refining it during a pilot phase (in other words fraud detection needs to be embedded in the process)
2. Formulating assumptions on fraud behaviors and test those through a systematic data collection and analysis (monitoring controls)
3. Using IT capabilities to support randomized check selection based on preset criteria. Track detection performance by monitoring time spent on fraud cases and calculating the hit/loss ratio.
4. Use big data techniques to expand and automate the analysis beyond the original boundaries. This is particularly promising in environments where client meta-data is available (e.g. with digital distribution channels, agent networks, product cross-selling)
Find out more: Data Analytics Project