The problem is well documented: fraud in the protection space has produced significant underwriting losses throughout the Indian insurance industry.
The problem is well documented: fraud in the protection space has produced significant underwriting losses throughout the Indian insurance industry.
Indiaforensic Research estimates that India’s insurance sector loses INR 300 billion (US$4.5 billion) annually due to fraud, or 8.5% of total industry revenue. The life insurance sector accounts for 86% of total insurance fraud, six times the amount of fraud in the non-life / general insurance sector.
A competitive environment only adds fuel to the fire. With insurers under pressure to develop new products to cover new risks, and with technology and business innovation creating a shifting landscape, the Indian insurance sector will remain a prime target for fraudsters. The need for solutions is overdue.
Predictive modelling has long been viewed as a promising approach to fraud management. As more data becomes available, the thinking goes, sophisticated analytics can apply learnings from past fraudulent behaviour to predict its likelihood in the future and develop models to assess fraud risk.
As of September 2016, 69É«ÇéƬ has put this thinking into practice.
69É«ÇéƬ’s new risk scoring model (RSM) employs fraud profiling as the main aspect of its rule set. This new tool, now available to insurers throughout India, works in conjunction with ), the industry’s leading e-underwriting platform.
RSM uses several variables, each weighted depending upon their impact on fraudulent behaviour risk. As individual cases come in, they are assigned a score within each variable based on historical industry claim experience. The final risk score is then calculated by applying the weighted values.
Broadly, RSM divides scores into four classes of risk. These classes have a linear co-relationship, with Risk Class A giving best mortality experience and Risk Class D the worst. Even though a lower score may warrant declining coverage in some instances, RSM does not suggest a decline decision for such risks. Rather, the model suggests a level of due diligence to be conducted for medium-to-high-risk classes to eliminate the element of fraud.
Insurance fraud is going to take place; it’s an unfortunate, inevitable part of the industry. But predictive modeling – in conjunction with other fraud detection methods – offers a new path for controlling losses from fraud while expanding services for underinsured populations.