Want more Insights from SAS? Subscribe to our Insights newsletter. Or check back often to get more insights on the topics you care about, including analytics, big data, data management, marketing, and risk & fraud.
The fraudometer: The ups and downs of insurance
By David Hartley, Director, Fraud and Financial Crime Practice, SAS EMEA & AP
From underwriting to claims, fraudsters have always found ways to cheat the system. The news from the Association of British Insurers (ABI) revealing the latest trends in detected insurance fraud only goes to highlight how diverse the challenge has become.
- Insurance fraud in the UK by 9 percent (worth GBP 1.32 billion, or about US$2.03 billion).
- “Slip-and-trip” fraudulent claims, with an increase of 75 percent.
- Motor insurance fraud by 12 percent.
- Property fraud, both domestic and commercial, with the latest stats revealing a decline of 29 percent from 2013.
It’s important to note that these figures only represent detected insurance fraud. So they could be indicating better detection capabilities by insurers rather than more overall fraud. My bet is that it’s both.
For this defense to remain robust, the data must be good quality. The value the business will take from the data depends on it. When data quality goes up, business value goes up.
While insurers are increasing and improving their fraud-detection capabilities, criminals are finding new ways to bypass the law and the detection measures. In other words, there is probably a rise in undetected fraud as well. And to give you an indication of the volume of fraud we’re talking about, according to the Insurance Fraud Bureau, undetected insurance fraud has previously been around GBP 2 billion, versus GBP 1 billion in detected fraud.
In a webinar I took part in, we talked about the potential for integrating application fraud checks into price comparison sites and for bringing new business on. This addition of a front line of defense reminded me of an existing and important group of people: the claims-handling staff. You know that they’re trained to spot instances of fraud, yet they can’t be your only means of protection.
Going up: Better use of data to eliminate insurance fraud
Insurers are increasingly turning to data analytics to deliver additional lines of defense. For this defense to remain robust, the data must be good quality. The value the business will take from the data depends on it. When data quality goes up, business value goes up.
This is no easy task. It requires integrating data silos, resolving individual entities, improving data quality and harnessing unstructured text. There are nuances to consider, such as being careful not to “over-clean” the data. This is because “David Hartley the fraudster” may have purposefully used the name “Dave Hartley” on some occasions, but on other occasions intentionally provided the name “David Harley” to throw off detection. If this data were to be cleaned, the discrepancy would go undetected as linked to fraud.
Perhaps an incident is reported with a specific postcode for some, and just a street name for others, simply because claimants don’t have all the information available to them. Your system needs to be able to recognize this and act (or not, as the case may be) accordingly. Context is an important part of data analytics as well as data gathering.
As the ABI stats revealed, fraudsters move on to new types of fraud all the time, so in your analysis, be careful not to include data from more than three years ago, to avoid throwing up too many anomalies. In fact, this approach is in line with the fact that many fraudsters will limit the time they target a specific company or type of fraud in order to avoid detection.
Rethinking your first line of defense
In the future, it is to your benefit to take a more data-driven approach that will contribute to the first line of defense, for example, by preventing the creation of uncleansed data. Then it’s time to ask yourself: How can your first line of defense be improved by learning from the insights your data offers? Are there patterns of behavior that you can’t spot manually?
Insights can be many and varied. Your data analytics teams need to understand as much as possible about the claims process, customer (and fraudster) behavior, the top fraud trends and the nuances of ensuring clean data. The more they learn about claims departments, the more the process will yield results, benefiting the rest of the business.
While the regularity of different types of fraud will continue to go up and down year by year, what must remain constant is your business focus on data. As long as the data quality goes up, so will the value of that data to your business when it’s analyzed.
David Hartley is a 14-year SAS veteran based in the UK working across Europe and Asia. He is responsible for the direction and development of specific analytics solutions that address insurance fraud detection and prevention. Prior to SAS, Hartley worked in the UK insurance industry helping establish a major direct writer and a leading bancassurer.