Vintage Sensitivities: Measuring the Quality of Originations
Signals, Winter 2001
With Dual-time Dynamics™ (DtD), consumer portfolio performance data is decomposed into three components: maturation effects, exogenous impacts, and vintage sensitivities. Vintage sensitivities provide a unique metric for originations quality, independent of the age of the account and the nature of the environment experienced during the measurement period.
A vintage is a group of accounts opened in the same time period. For example, the February 2001 vintage is the group of consumer accounts opened in February of 2001. Vintages are a segmentation approach compatible with any demographic or behavioral segmentation management may choose to define.
When Strategic Analytics' DtD decomposition is applied to performance data for a group of vintages each vintage is calibrated to nominal maturation and exogenous curves. These vintage calibrations measure the sensitivity of each vintage to the maturation and exogenous impacts also quantified by DtD. Conceptually, vintage sensitivities are like a pair of multipliers, either accentuating or diminishing the contribution of the maturation effects or exogenous impacts in explaining an individual vintage.
In a recent client project, SA studied the vintage sensitivities associated with delinquency rates. We found that originations within the last two years had low maturation sensitivities to delinquency risk. That is, these recent vintages were predicted to have low delinquency rates according to our measure of their progress along the maturation curve. However, these same apparently low-risk customers exhibited high sensitivity to exogenous impacts. In effect, they were low risk customers in good environments, but their delinquency risk would likely soar during economic downturns.
Measuring the quality of originations is a decades-old problem in consumer lending. Previously available approaches do not meaningfully compare groups of consumers through different economic conditions because the impact of the environment is not separately quantified. Similarly, such approaches fall short when comparing customers early in their lifecycle to those on the books for many years.
Traditional vintage analysis is an important practice in consumer lending, but it amounts to a visual inspection of vintage performance without identifying the components of cause and effect. The vintage sensitivity measures enabled by Dual-Time Dynamics provide a breakthrough in this area. Further application will play a key role in understanding segment behavior and in turn, higher value portfolio and acquisitions management. |
Top of page