A New Method for
Collections & Recovery Models
Bruce Ratner, Ph.D.
Every stage of the credit lifecycle - in particular collections and recovery, found across many industries such as financial services, telecommunications, student lending, healthcare, and retail - depend on decision making, and profitability depends on the quality of those decisions. These decisions are based on factors such as risk and the amount owing. However, these factors and typically additional ones may not always requite the best results. For example, customers owing large amounts are often given priority because it is assumed they can carry “heavy” loads. There is an incautious order of importance that can put the customers at further risk - a point of “credit” implosion - if they receive too much attention. Customers owing smaller amounts are assumed to only carry “light” loads. There is a lost opportunity if their priority is so low that they receive no attention. The decision-making factors of collections and recovery are subtle, interactive among themselves and change in time, so intelligent analyses or machine learning methods are required to develop the optimal strategy. The purpose of this article is to discuss a new method – the GenIQ Model© – which goes beyond the traditional statistical scoring model. The GenIQ Model can accept an “injection” of real-time streams of continual performance data in its algorithmic process to update the model literally in real time, as well as uncover and operationally define the subtleties and interactions of the necessary factors to build a reliable model. Because statistical methods cannot include real-time data, and have difficulty capturing non-spurious subtleties and interactions, the GenIQ Model offers a potential to requite the best results for collections and recovery. A real case study is discussed to illustrate the collections & recovery application of the GenIQ Model.
1 800 DM STAT-1, or e-mail at email@example.com.