Paul Cassidy, director, E-trading for SSP looks at the core trends in fraud and discusses what insurers need to consider when implementing fraud detection
Historically there has always been belief that insurance fraud is somehow a way to “get something back” from an anonymous and wealthy business entity, but this is a bit of a myth and the reality is that insurance fraud costs all of us. Are you happy giving these fraudsters £50 of your money each and every year?
We’ve all been aware of fraud in insurance for years and it’s nothing new, but historically this has been a claims issue with both fictitious and inflated claims being the main exploits.
However, fraud is an increasing challenge, with a rise in incidents of 77% in the last quarter of 2013 alone when compared to the same period of 2012. Today there is a much broader range of fraudulent activity commonly undertaken, with individual or opportunistic fraud perhaps the area that has seen the most significant shift over recent years, in part fuelled by the broader economic conditions since 2008. Today, 40% of UK motor policyholders think that the cost of their policy is too high and one in four insured motorists believe it’s acceptable to lie to their insurer in order to obtain a cheaper premium1. This is a problem that is costing the insurance industry an estimated £2.1bn2 in the UK, $2bn in Australia and $390m in South Africa. This equates to £5.2m every day in the UK.
I often talk to insurers about the 2% of their customers that cause 20% of their losses, and whilst the precise numbers may vary a little by insurer, and it’s a perhaps a bit of a simplistic notion, the evidence suggests it’s actually quite an accurate measure. I recall working with insurers a couple of years ago to analyse the historical loss ratios for a group of 250 seemingly “bad” customers that had been identified. These customers had all been found manipulating their risk data across multiple quotations on aggregator sites and the insurers were astonished to learn that this group of customers had historically made more than three times the average number of claims, and the claims outlay was in the region 450%-500% of the premiums collected.
More recently, I’ve been reviewing some early results on counter-fraud measures, which has also highlighted some potentially suspicious customer behaviours. One example I saw showed a customer performing multiple aggregator quotes, clearly disclosing and then deleting a motoring conviction and eventually incepting cover on the basis of zero convictions.
When you factor in the research commissioned by RBS that determined a customer with a single motoring conviction is 69% more likely to make a claim than a driver with no convictions you can clearly see how insurers can easily get significantly exposed in terms of collecting an inadequate premium for the real risk.
The different types of fraud
Insurance fraud typically falls in one of four areas:
Exaggerated claims fraud and false claims are well established, driven in part by a public perception that it is “ok to get something back from the system”. Legal & General research found that 29% of people surveyed in the UK felt it reasonable to exaggerate a home insurance claim. But putting aside the massive issue of application fraud for the moment, it’s the increase in organised fraud that has severely impacted the industry over the last few years.
Within this category we see criminal gangs, who are organising “crash for cash” schemes. This is where they fake or deliberately induce 1000s of accidents every year in order to submit fraudulent and exaggerated claims.
Then relatively recently we’ve seen the emergence of “ghost brokers”, who unscrupulously target young drivers and non-English speaking communities to issue misrepresented or even fake policies, cheating these people into thinking they have insurance when actually they don’t.
The best estimate of the number of “ghost broked” policies in the UK so far, has come from AXA, who suggest the number is in excess of 20,000. Ghost broking is becoming increasingly sophisticated.
I remember looking at a ghost broker’s website a year or so ago, accessible via the GumTree online classified ads site. It had been completely copied from RSA’s More Th>n website – all of the graphics, forms and text had been duplicated, and they’d even granted themselves an official looking FCA approval number!
Bringing together the behaviours and attitudes of a minority of previously “honest” customers, ghost broking and other sorts of organised crime means across the industry there is a lot to contend with. Applying an international lens, and comparing the UK market maturity with other territories, suggests we might see this issue become a growing feature in other markets as well, particularly with the increasing adoption of aggregation, which is a real enabler to some of this activity.
With insurers investing £200m per year on anti-fraud systems and resources and delivering an ROI of £900m in 20113, detecting and reducing fraud is high on the agenda, but to address this problem the insurers have to deal with a number of new challenges:
• Lack of data
• Where there is data, it is generally in large volumes and often unstructured
• Lack of ability to take immediate action due to the distributed model
• Social media integration
• Defending against organised crime, and ghost broking
• Powerful and real-time data analysis
• Visualisation of data
• Staff skills
Governments are trying to support the issue, sending out messages in an attempt to influence attitudes and behaviours, but ultimately it is the insurer who is exposed and needs to address these challenges throughout their processes, from application through to claims.
Using advanced fraud detection and management solutions, insurers can reduce their exposure by ensuring they really know who their customers are, by having accurate data, and pricing correctly for each risk. Their challenge is to integrate fraud solutions at the most appropriate and effective points in the carrier to consumer chain. The “front line” battleground for this problem however is undoubtedly the internet, and perhaps more specifically the price comparison sites.
Big Data is a much overused and hyped term, heralded as the solution to everything, good and bad, in any business. In insurance and fraud however it really does have a part to play, particularly in terms of claims fraud and organised crime. According to Celent4 the challenge for insurers will be to integrate external data sources, leverage a big data infrastructure, to better analyse higher volumes and a variety of structured and unstructured data. “The pace of investment in claims fraud detection systems will depend on insurers’ understanding of the importance of analysing data using modern technologies, including big data infrastructure”.
Gartner also concurs, but suggests that the focus must be on more real-time analysis of both structured and unstructured data. Gartner points to social media data integration and social network analysis to assist in identifying repeat offenders and organised fraud. Modern fraud analysis techniques using larger volumes of more varied data can reduce losses by as much as 20% by detecting suspicious behaviour early in the business process, enabling interception and special handling.
The effective use of data can support the analysis of fraud and can be sourced from many areas both internally and externally. The more data that can be blended for analysis the better, for example:
• Geocoding data – ordnance survey
• Satellite data
• DVLA - MyLicence
• Aggregator data
• Electoral roll data
• Claims data
• Cross industry insurance Fraud data – IFB, IFED, IFR
• Social media data
Geographical data is being used increasingly to fight fraud with 42% of insurance companies identifying fraud “hot spots” by analysing patterns in the data. This has led to great success in uncovering organised “cash for crash” gangs that are now being identified and prosecuted. However there is more scope to use location-based intelligence to support prosecution of fraudsters.
As we move forward, social media data is emerging as a very useful tool for insurers who are able to use geospatial analysis of social media postings. Similarly insurers are starting to use social media and other data to make connections between the policyholder and the third party, identifying if they know each other. Recent case law indicates that evidence from social media sites is being seen as a legitimate way of proving fraud.
Insurers do of course need to bear in mind local regulatory compliance when dealing with data, particularly shared data from social media sites, and it is important to be aware of data protection law and the use of the data for purposes other than for which it was collected.
The real challenge
The real challenge for all insurers however, and I mentioned this earlier, is not the availability of data and sophisticated analysis, but having the solutions that can exploit the power in the right data, in real-time at the most opportune and effective moments within the business process. The key is to identify potential fraud right at the point of application, and to achieve this, an instantaneous single view of the customer is required in order to gather and analyse the data before cover commences. Previous fraud detection technologies have largely been based upon post-inception manual checking and identity verification techniques. But if you consider a company director deliberately playing with variations of their claims and convictions on an aggregator site in order to drive their premiums down (as was the case I looked at recently) then a simple identity check on the electoral roll isn’t going to reveal anything out of the ordinary. But today, clever solutions can analyse suspicious patterns that suggest an increased probability of fraud up front and can potentially eliminate huge costs. But all of this must be achieved in careful balance so the vast majority of legitimate honest customers are not affected.
From an insurer perspective, this detection prior to inception is absolutely key, particulartly for motor policies. Under UK legislation, insurers have immediate and far reaching liability for third parties. Preventing cover in the first place saves them the exposure from staged accidents perpetrated by organised gangs. This also saves significant frictional costs arising from post-inception manual checking processes, both for themselves and potentially the broker too.
Finally, insurers will need to partner with a company that has the broadest set of skills and technologies to address the challenges of harnessing this data and making it available at the right points in the business process. From insurers to intermediaries and aggregators all the way through to consumers, the technology needs to support the journey and processing at each stage of the carrier to the consumer chain. An integrated end-to-end platform, coupled with “Big Data” concepts, and some very clever analytics can give insurers the power to precisely identify and intercept the relatively small number of customers who cause a disproportionate cost and burden for the majority. Aside from the obvious benefits for us all as consumers there’s a significant opportunity to share some of the upside with the brokers, who have frequently got a crucial role to play at the front-line interaction with customers.
About the Author
SSP is a global provider of technology systems and solutions across the entire insurance industry, using our expertise to enable our customers to transform their business and increase their profitability. SSP provides core technology solutions, distribution and trading capability, advanced analytics and solution delivery. We work with 8 of the top 10 UK insurers, 4 of the top 10 global insurers and over 40% of UK Brokers. Our unique position in the market, including the largest market share of UK e-trading, enables us to provide leading data insight and unrivalled distribution. Our knowledge, talent and technology capabilities deliver innovative results that make us the partner of choice for our customers.Follow on Twitter More content by SSP Worldwide