Sunday, December 13, 2015

What is Underwriting Fraud?

To answer that question we need to first define insurance underwriting which is the process of evaluating the risks of insuring a particular person or asset to set premium pricing.  Thus Underwriting fraud is the deliberate falsification, omission or obfuscation of legitimate information that would enable an insurer to understand and price the risk with the intention of reducing the price of a policy below its expected cost. Insurance fraud can be the falsification or omission of a range of different data categories from demographic (who is insured), asset (what is being insured), geographic (where the risk is being incurred), behavioral (what uses the insured is being put to) and intent (the purpose for which the policy is purchased i.e. protection against a casualty loss versus use as a platform to defraud the insurer).

When looked at this way it is clear that underwriting and underwriting fraud are activities that occur throughout the insurance life cycle from application to servicing, to claims and renewal.  Information that is false affects the initial pricing of the policy but information that changes during the life of the policy also affects the profitability of the priced policy.  Comprehending this evolving understanding of the risk enables underwriting to focus the insurer's claims and investigative units on the claims with the highest probability of fraud. This ongoing reassessment of risk based upon the latest available information is also essential for pricing at renewal.  Thus it is critical that underwriters be able to identify places where the insured is presenting false information or where his intentions are not those represented regardless of what point the policy is in the insurance life cycle.

Needless to say this has significant implications for how insurers assess and act upon underwriting fraud.  It is not sufficient to simply provide an information screening up front during pricing, it's also important to monitor the evolving risk of the policy and to ensure that other parts of the organization have the latest insights into the risk to productively focus their efforts.

Thursday, June 4, 2015

The best way to BEAT fraud is to avoid the people who are likely to Do fraud.

Imagine you're responsible for security for a sold out concert. You know that roughly 5% of the 10,000 ticket holders are going to cause problems for the rest of the audience. In the past you've managed this risk by hiring at great expense, a bunch of off duty police officers and security guards to watch all ten thousand fans, waiting for something bad to happen. If you get lucky you'll see something going down and you can pounce. Congratulations! Only 499 "problems" to go.

But what if you could identify the 500 bad actors before they got inside? What if the "problem fans" exhibited tell tale signs or patterns that you could systematically evaluate as they are queuing to get in? Then you could either turn them away at the door or you could isolate and focus on them. Doing this would prevent a lot of chaos, save money on security guards and avoid treating everyone in the audience as if they were potential "problems" - an outcome that is wholly preferable to the traditional enforcement approach.

Sadly this is a pipe dream in the physical world because the attributes that would tip you off to their criminal status are often not physically visible as they walk up. But a new company called VeracityID is solving this problem in the financial services world. Specifically, they've figured out how to identify the estimated 5% of all financial customers that commit most of the fraud. And they do so in real time as prospective customers submit their on line or agent mediated application. They've just successfully demonstrated this approach in a half million dollar paid proof of concept for the Auto Insurance subsidiary of a Fortune 50 insurer. In it VeracityID took one quarter of their book, constituting 24 million records and combined it with a wide range of external data sources.

Then, using proprietary social network analytics and boolean rules engine, they were able to identify fully 5% of the book that had either committed fraud by lying on their application or were at high risk of doing so based upon their social network risk score.

The insurance company in our example calculated that on a "same store" basis  that using the VeracityID solution would have saved them between three and five percent of revenue which is huge in insurance terms. Which is why they've agreed to pay VeracityID millions of dollars to integrate their service into their front end online and agent application process.

About the Company
VeracityID is a new Silicon valley company founded by big data analytics and technology experts from McKinsey, Deloitte, Oracle, AT Kearny and IBM with decades of experience in financial services, healthcare and other data intensive online services. The key elements of their model include:
  • Creating very targeted "apps" that address specific high value financial services problems (their first area of focus is identifying personal lines automotive insurance fraud risk).
  • Deploying each "app" in a web service SaaS Cloud model.
  • Connecting the fraud dots. by analyzing customer application data, internal company data and external databases using Social Networking Analytics and a boolean rules engine.
  • Avoiding the fraud loss by delivering decisions in real time at the point of on line or agent mediated application - before a product and price offer is made.
  • Doing so at cost benefit ratios of 10 to 20 to one.

Getting Started with VeracityID
Automotive focus.  For the next year we are focused on leveraging the work we've done by adding additional Auto Insurers in North America and the UK.  We are also interested in developing relationships with participants in adjacent spaces.

Proof Project.  The VeracityID approach is new and unique and we don't expect you to fully embrace the solution without testing it yourselves.  This is why we always start with a Paid Proof Project. You provide us with all of the relevant data for a representative segment of your business.  We then simulate the enrollment of that segment and in conjunction with you estimate the bottom line benefit of a full implementation.

Using the benefits insights from the Proof Project we set performance objectives for Veracity ID Deployment.  Deployment has two key components:  the first is technical, ensuring the integration of your data with our "app" and making sure system performance and accuracy parameters are established.  The second is configuration of the business rules and the design and routing of event and data reporting.  This typically includes a change management component to ensure that your various functional groups understand and are fully prepared to capitalize on the improvement opportunities that VeracityID uncovers.

Timing Considerations.  VeracityID allows its clients to identify and avoid fraudulent customers. As more and more industry participants utilize the VeracityID Automotive Insurance Fraud Identification "App" the number of companies absorbing the bulk of the industry fraud risk will shrink.  Being the last competitor in an industry to effectively identify it's fraudulent customers is not likely to be a viable survival strategy.