Underwriting fraud is the presentation of false or incomplete information by the customer and/or agent about a risk in order to receive a rate or product that the customer is not entitled to. This is very common and constitutes up to ten percent of premium in the Automotive insurance business.
The standard way that carriers have fought UF is through the acquisition of third party data – by matching the (hopefully) accurate and complete third party data to that provided by the customer carriers hope to highlight and ‘knock out’ fraudsters. But the theory is substantially better than the practice – even assuming that the data sources are complete and accaurate (they never are but a lot are pretty good) using data as the sole basis for knocking out an applicant is problematic. For example, what if the customer made a mistake or misunderstood the instructions? Or what if the customer has an anomalous situation like a recent move? Carriers can’t afford to exclude these common customer types simply to get at the fraudsters. You need to interact with the customer to ascertain what kind of problem you’ve got.
It’s a lot like the old ‘Cop on the Beat’ who walked his neighborhood. Say he sees someone jimmying a window in a locked house. He doesn’t shoot or immediately arrest the jimmyer – instead he walks over and asks the lad what he’s doing? Perhaps “Oh hi, officer, I lost my keys and am trying to get into my place to get my spares”. “Is that so? Well do you have any evidence that you live here?” “Sure officer, here’s my driver’s license with this address on it. And if you’ll wait a minute I’ll show you bills with my name and this address on them.” and after that the officer continues his beat. But suppose it was really a thief, how would the scenario differe “Oh hi, officer, umm i was trying to, uhh, oh look i just realized that this wasn’t my house, my house is two blocks down on the street that doesn’t have a beat cop”. In other words, the miscreant abandons his misdeed and goes to a neighborhood where the police aren’t as vigilant.
Several things about the cop are salient:
- The cop gathers lots of information with his eyes and his ears
- But the information isn’t conclusive, it simply gives the cop indications of possible misbehavior
- So the cop interacts with the subject, trying to make sense of information anomaly
- If necessary the cop requests evidence
- Separates the sheep from the goats
- Usually scares the goats off to brand x that can’t do this
- It’s the scare off that makes this so effective – the typical UW fraudster is a casual violator – trying to get a few hundred bucks of of what he feels is an outrageously high premium so he wargames the data – fiddles with it until it returns a rate that he’s happy with. He’s not prepared to be confronted – if he is, he most often just leaves.
