Quantifying risk by analyzing the exposure and probability of loss in a customer interaction is becoming a must for any advanced risk solution. It is also a difficult burden to bear and remains a premium feature in most point solutions.
However, quantifying customer risk one factor at a time is itself a risky proposition. The reason: risk factors compound each other in a nonlinear fashion. For example if someone’s transaction has a 5 percent risk of being fraudulent and 10 percent chance of unintentional default, siloed analytical engines will treat it in one of three ways:
- Analyze risk factors independently if 10 percent credit risk and 5 percent fraud risk are both acceptable, then the transaction will go through.
- Communicate only the existence of a warning across systems if a transaction raises both a fraud and a credit flag, then it is always reviewed or denied.
- Layer independent risk premiums on the price — add an increment of cost to the customer for each potential risk.
None of the three methods accurately reflects the aggregate probability of loss (around 14.5 percent if the risk factors are independent). The third one comes closest to managing risk sensibly, but it is not available in all pricing models.
A risk model that takes a global view of all factors can use historical data to not only compound risk factors adequately on the fly but also determine which ones are independent and which are interdependent. In the above example, for instance, it can discern where fraud and credit checks caught the same signs (e.g., a clerical error in the application) or mutually confirming independent signals of risk (e.g., a bankruptcy on the credit side and an invalid mailing address on the fraud side).
Aberdeen Group is interested in your feedback on this research, please go to www.aberdeen.com/feedback.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access