Collections behavioral models

April 2, 2009 by Jeff Bernstein

Behavioral scoring is one of the most important tools that allow collections managementand account management groups to evaluate accounts in an efficient and cost-effective manner. Although behavioral models are developed in a similar manner as new applicant models, there are several key differences that make behavioral models a better choice for many account management applications and collections workflow systems:

By using only internal master file data as opposed to external credit bureau data, for example, accounts can be regularly evaluated without incremental cost. The most common practices are to score accounts on a weekly or monthly basis, which allows for quick strategic responses to a customer’s change in behavior. Frequent evaluations can result in automated or manual actions such as the acceleration or deceleration of collections efforts, adjusting credit limits and changing terms and conditions.

The performance definitions of behavioral scores are very specific to each strategy and task, and it is typically not advised to use models in applications for which they were not designed. For example, a new applicant model definition of “bad” may be a high probability of charge off during the initial term of a line of credit. For collectionsstrategy, a more appropriate bad definition might be the likelihood of an account rolling to the next delinquency bucket, regardless of the age of the account.

Behavioral models also have a much shorter outcome period of three to four months versus new applicant models that forecast over one to two years. Since behaviors with one creditor can typically be recognized more quickly than with all lending institutions associated with a particular debtor, behavioral models provide a unique and timely evaluation of the ongoing risk once the account is already on the books.