By Larry Jacobson, Senior Consultant, Insurance, FICO
The analyst house Ovum has recently issued a rather gloomy forecast for the insurance industry in 2013 . Economic pressures such as the eurozone crisis and high rates of unemployment are set to continue and these, combined with new industry-specific regulations like Solvency II, will create the perfect storm of business challenges.
The key to success, Ovum says, is for insurers to “optimise their core administration systems and make use of advanced analytics in order to remain competitive in the modern insurance market”. I agree that advanced analytics are essential for driving more balanced and effective risk decisions in real time and for spotting fraud more quickly and accurately. In a changeable and often volatile market environment, this ability to identify and neutralise risk at both the micro and enterprise-wide levels is even more important.
Where to start though? With every part of the business touched by technology, it can be hard to identify the top priorities. I suggest that insurers looking to build on their use of analytics, or even establish a practice, focus on identifying and combating insurance fraud, and in managing and optimising decision-making models.
Fight fraudsters more intelligently
The Insurance Fraud Bureau estimates that losses due to insurance fraud in the UK amount to around £1.9 billion per year, adding an extra 5% to the average premium. Being able to avoid some of these losses could result in a significant impact to the bottom line in an economic environment where every pound matters.
A traditional approach to dealing with insurance fraud is based on the ‘pay and chase’ model, where insurers pay a claim, then try to recoup the funds when it’s found to be fraudulent, with limited success. A more effective approach is to spot potentially fraudulent claims before anything is paid out. This is where advanced analytics come in.
Analytics such as so-called neural network models that work like the human brain can spot patterns in claims that could be indicative of fraudulent activity. By analysing transactional and relationship data, they help insurers bring to light formerly unknown fraud types and identify ongoing fraud schemes and networks.
With this in mind, one thing to look forward to in 2013 is the continued rise of Big Data as a business enabler. While many still view it as an overwhelming challenge, with the right technology in place it can become an insurer’s greatest asset. Using analytics to help filter Big Data can help ensure the right questions are asked in order to pinpoint fraud patterns while avoiding unnecessary ‘noise’.
Big Data can also be used to help an insurer assess the fraud risk of a specific claim against the bigger picture. Link analysis – a technique that examines relationships among organisations, people and transactions – can help uncover previously invisible relationships between claims. For example, a car repair garage may be handling an unusually large number of accident repairs. With link analysis, an insurer can see that despite appearances, it may not be the garage itself that is the source of the fraudulent activity, but a network of crooked ‘victims’ or vehicle owners that take their cars to the same place for repairs. When viewed individually each claim may seem legitimate, but take in the wider context and a pattern emerges. Link analysis of this sort is a data-hungry process that is actually strengthened by the broad reach of Big Data, creating better visibility of insurance fraud threats.
Follow the model approach
Insurers use predictive analytics already to help automate decisions around everything from underwriting to marketing. However, a recent survey conducted by FICO found that 64% of insurers felt they lacked the ability to rapidly deploy or update models to maximise business impact. This is often due to the fact that hundreds of models may be used in siloes across the organisation, making it very difficult to manage models and decision strategies at an enterprise level.
This lack of effective and centralised analytic models can have a detrimental effect across the board – from mismatching product offers to getting premium prices wrong. It can damage profitability, which is far from ideal in today’s economy, and can also compromise an insurer’s ability to meet Solvency II capital adequacy and risk management requirements.
Centralised model governance can both satisfy regulations and keep up models’ performance. Insurers can create a cohesive view across siloes such as geographies and lines of business to help improve data quality and availability and provide complete transparency for audit. By leveraging these technologies, insurers can by-pass the months of testing typically needed to get to optimum strategies, giving them the agility they need to keep up with today’s volatile market changes.
The U.S. presidential election was just won by the team that excelled at analysing data and developing voter-targeted strategies. Insurers should take this example and the recent Ovum report in hand and work out their own analytic strategies to not only weather the storm, but to prosper in it.
1 2013 Trends to Watch: Insurance
|
