In Part I of my Risk-based Collections blog series, I described the differences between Judgmental and Statistical Scoring models.  In part II, I will expand upon the specific uses of these models, in particular managing existing credit lines. 

For evaluating a request for additional credit or determining the collectability of an existing customer account, statistical-based models are more prevalent than ever, and being adopted by first time users of scoring models.  Statistical-based models utilize payment behavior.  They look at your own experience with each customer to help you evaluate their future risk and use that information to manage their credit lines. The only caveat is that companies using a statistical-based model must have a sufficient number of accounts as well as the required data to build the model.

Typically, a portfolio of at least 1,500 accounts and eighteen to twenty-four months of accounts receivable data is required for an effective model. If you don’t meet these criteria, a judgmental-based model together with credit bureau data would be the best approach.  The volume of the accounts that need to be evaluated is the major driver of what methodology to use.  For example:

• Low number of existing customers and low dollars: Use bureau credit reports or scores supplemented by an analysis of customer payment patterns, i.e., do they pay on a timely basis and what is the frequency of arrears?
•  Low number of existing customers and high dollars: Use bureau credit reports or scores supplemented by an analysis of customer payment patterns and periodic requests for financial statements and bank references.
• Over 1,500 accounts: Companies with over 1,500 accounts will likely benefit from statistical scoring (or a hybrid of statistical with bureau data) versus using bureau data alone. The statistical system that uses only internal data is typically the most effective for optimal allocation of collection resources. When these companies can use the model output to drive collections prioritization, there is typically a sharp improvement in collections effectiveness.

How are you managing existing credit lines today? Are you using or thinking about using statistical-based scoring models when managing credit lines? I’d like to hear from you.