Data analytics are spoiling auto fraudsters’ fun

You might have as many scams as you do car models. The ol’ flash-for-cash is one—where they blink you and try to coax you to make a turn, and then the creep accelerates and plows into you. Rinse and repeat. The fraudster goes from place to place, hits different insurers, changes their name slightly, and it’s been difficult for any single company to spot the pattern.

The fraudsters had one single advantage they could always count on: the silo thinking and practices of the insurance industry. Insurers only relied on their own data analytics and used them for their own investigations, and while they might have relied on the additional resources of say, the IBC or the cops, it was pretty much a closed shop. For the industry to pool data seems to be a no-brainer, yet a combination of Canada’s unique insurance regs, as well as its industry culture, put the damper on this until recently. Meanwhile, in the States, they’ve had the McCarran– Ferguson Act, which has allowed insurers to compare data and risks since 1945.

Now that’s going to change. It’s been a slow train coming, with various pilot programs tested and the nine insurers behind the initiative all getting on-board and agreeing on what should be done. The non-profit consortium only incorporated back in 2013. But at last CANATICS is due to “go live” with its data analytics tool some time this month.

President and CEO Ben Kosic says the magic “of the pooling that we do, is that we can see across time and we can see across insurers, and with the sophisticated technology, we can understand that Jack Smith over here and John Smith over here and James Smith at this other insurer are likely the same individual, and therefore we can start to see repetitive and statistically unusual behaviour.”

Such things start to look anomalous, says Kosic, who’s quick to underscore the point that an anomaly in itself doesn’t prove fraud, that it “doesn’t say so-and-so is guilty and this person is innocent. All we do is we say there’s anomalies in the data that would suggest somebody should look at this. And it’s a lead, right?” It’s up to the insurer to pursue an alert and see if a claim needs investigating. In fact, under the service agreement with member insurers, Kosic says, insurers are obligated to investigate before denying a claim.

He’s understandably coy when the conversation turns to the intricacies of his company’s methods. “And I’d love to share all of the different quirks and habits that we’re onto, but I won’t.”

He will say that the insurance industry wanted to be rock-solid in its approach over handling various privacy requirements. It’s meant working extensively with FSCO, IPC, the privacy commissioner of Ontario, as well as the OPCC. “When you speak to the lay person on the street, everybody’s like, ‘Of course, why aren’t you doing this? Bring my premiums down, get rid of the bad guys.’ However, we also have to ensure that we’re compliant with all the various statues and laws, and we absolutely are.”

Kosic says there’s no such thing as one kind of auto insurance fraudster in Ontario. “They have schemes crossing geographies, they have schemes crossing lines of business, they have schemes— that’s what they’re doing. So we do ourselves a disservice when we try to pigeonhole them that way.”

And new schemes are emerging all the time that he says prompt adjustments in the technology. It’s an ongoing process in which “we’re generating some statistics for the investigators to sit with us and figure out what changes do we make. And we do this for two reasons. One, obviously we want to give them the best possible alert, but the other is we’re trying to do from a privacy perspective what we call ‘minimizing false positives.’ So even though we’re not ruling on saying this is fraud and that’s not, we’re just giving leads, we still want those to be the best possible leads.”

For now, CANATICS is only dealing with auto insurance in Ontario, but he says “we’re looking very strongly at moving to Alberta in 2016. And of course this would require us to do all the same things: work with the fiscal equivalent, work with the privacy folks and make sure that they’re comfortable, that the consent language changes, and then go into Alberta.” As a choice, he says it’s the easiest growth option, given that Alberta represents $2.7 billion in direct written premiums for the current nine members.

Fraudsters can bump, they can flash, but it’s getting harder to hide.

__________________________________________________________________________
Copyright 2015 Rogers Publishing Ltd. This article first appeared in the April 2015 edition of Canadian Insurance Top Broker magazine

This story was originally published by Canadian Insurance Top Broker.