Fraud remains a menace for insurers in Kenya. This is despite collective measures by the stakeholders in the industry to collectively negate the vice.
In Deloitte’s Insurance Outlook Report 2019, it is estimated that 25% of underwriting industry income in Kenya is fraudulently claimed. The biggest casualty of fraud is medical and motor vehicle classes of business.
The regulator, IRA-Kenya, set up Insurance Fraud Management Unit, and pushed local players to introduce internal detection mechanisms to mitigate fraud. The Association of Kenyan Insurers (AKI), the industry’s lobby group, has also come up with tools to address the fraud menace. This includes rolling out of virtual motor insurance certificate and Integrated Motor Insurance Data System (IMIDS), a centralised repository of motor insurance data.
Still, despite numerous measures put in place, fraud continues to have far reaching implications on Kenyan insurance companies in several ways including increased cost of doing business due to investigations, huge claims ratio and insolvency.
Industry trust among consumers also continues to abate, while many underwriters are allocating huge claims reserve ratio that would otherwise had been put to other useful investments options.
InsurTech firms are well poised to tackle fraud in the insurance sector. Through technology and large data reserves, they can put in place processes that would provide the ability to respond quickly when fraud is detected. Some of the techniques applicable include; data mining, use of artificial intelligence, and making use of big data.
Through data mining, insurtechs can determine data patterns and identify the likelihood of certain claims being more fraudulent than others. For example, by developing a machine learning tool, one can be able to examine each line of entry on claims, compare the entries against predetermined rules, for example the amount claimed and area code, then rank the claims in the order of the most likely to be fraudulent. This will highlight results that can be further investigated for higher probability of fraud.
Insurtechs are able to deploy different types of Artificial Intelligence tools to monitor and analyse, say, disgruntled employees, who are considered largest fraud catalysts within the insurance industry. Sentiment detection, a type of AI tool, could be used to analyse emotions, feelings and attitudes in written reports. The insurance company could develop a list of keywords that can be strong indicators of a sentiment.
“For example, internally one could identify warning signs of fraud by searching employee reports or emails for words that would imply that an employee is disgruntled for example, exhausted, inconsiderate etc. This would then allow management to follow up and carry out further investigations,” says Deloitte in its Insurance Outlook Report 2019.
Through AI, one is able to detect anomaly that can prove useful in deriving patterns. The analysis can determine the frequency and amount of claims for particular individuals, therefore give a warning indicator if a claim exceeds the usual amount or give an indicator when a policyholder submits more claims than usual within a particular period.
Lastly, by using big data, Insurtechs are able to get more accurate results since machine algorithms are able to learn from large amounts of data. The data can be collated from interactions on social media, smart devices such as wearables and telematics. This is then compared against pre-set rules to determine fraudulent claims.