Tackling a predicted surge in fraud during the cost-of-living crisis

The world has lurched from one crisis into another, with the pandemic, inflation, supply chain problems and war in Europe. These extreme economic conditions could impact the insurance sector, in the form of attempted fraud and exaggerated claims. BAE Systems Digital Intelligence investigates the current landscape and how insurers can get smarter about fraud prevention.

Insurers are no stranger to cyber risk. The finance and insurance sector accounted for more COVID-related “cyber events” than any other in 2020, according to a report last year from the Bank for International Settlements (BIS).

Attacks can range significantly in scale and purpose: from ransomware and data theft to denial of service (DDoS). When customer data is taken there is an additional secondary risk of it being used in follow-on fraud attempts.

Cyber attacks are no longer a minor irritant. They cost global organisations trillions annually and can have a major impact on brand reputation. A recent study from Hiscox revealed that a fifth of businesses hit by an attack were almost rendered insolvent as a result. That same research claimed spending on cyber increased 60% over the past year to reach $5.3 million. But if it’s not spent in the right place, breaches and fraud will persist.

Fraud on the rise

The challenge for insurers is therefore not only to build resilience against cyber attacks but also to find better ways to mitigate mounting fraud risk. Some of this risk is coming from organised crime groups.

But increasingly we’re likely to see ordinary households engaging in criminality due to the tremendous squeeze on their finances. The Bank of England expects inflation to hit 13% over the coming months, driven by huge gas price rises, and is steadily increasing the cost of borrowing to try to cool the market. Interest rates currently stand at 1.75%, but will surely increase further over the coming months.

In response, insurers can expect a surge in three specific behaviours linked to fraud. Exaggerated or fake claims are one of the most common—households either artificially inflating the cost of existing claims or making new ones up completely. According to Aviva, the number of bogus household claims increased 45% in 2021 at an average cost of £3,645.

Another tell-tale indicator of fraud is faked ID or false information given at the point of quote. Insurers can also expect to see a rise in organised multiple claims made via different insurers.

Mining for data

Where organised criminals are involved, there are ways to better detect suspicious behaviour, if you know where to look. Sometimes a single group is responsible for multiple criminal activities, from phishing and cryptocurrency scams to car theft. Shell companies and bank accounts may be registered to linked addresses and/or names. There may be multiple attempts at travel sickness claims, crash-for-cash or other common fraud tactics. Mining data in the right way can help insurers join the dots and proactively mitigate the risk of losses.

In other cases, automated checks can help to avoid false flags. The pressure on household finances this year and next could see legitimate policyholders change address as they move to cheaper locations and switch to permanent remote working. With automated processes, insurers can seamlessly check data that is manually inputted at the point of quote, either to flag for further investigation or rule out a case. Does it match what is on file, or what’s logged in shared databases like the DVLA, MIB or Electoral Roll?

A 360-degree view

The ideal is to build as complete and comprehensive a profile of each customer as possible, without adding any additional manual steps which can introduce human error and impact staff productivity.

This 360-degree view can then be queried against any claim. Check their ID, cross-match IP addresses and devices identities, and see if bank accounts or payment services are the same as those used in previous claims. Consider even checking the time of day or night when the claim was uploaded. All of this can be automated and run within seconds.

Sometimes there are clear red flags which could indicate a strong likelihood of fraud requiring further investigation. If the claimant is in arrears with rent, mortgage or credit card payments, for example. Or if they’ve recently changed or lost jobs. But on other occasions insurers may encounter more novel behaviours for which there are no typical red flags. In these instances, machine learning tools can be leveraged to learn “normal” behaviour and then raise the alarm when anomalies are spotted.

Often there are tiny clues hidden in masses of data that human eyes might not spot. But by spotting these patterns, intelligent algorithms can find the needle in the haystack, and help insurers build an index of data which acts as a risk score. Not every high score may actually represent elevated risk.

But it will help to narrow down the number of cases sent for further review. That’s the way to maximise the productivity of human teams and minimise financial and reputational risk. Fraudulent car claims soared 170% in spring, while bogus property claims increased 25%.

To weather the coming storm, the industry will need to get smarter about how it tracks illegal activity.

David Nicholson is Senior Data Scientist at BAE Systems Digital Intelligence

Main image courtesy of iStockPhoto.com