Using bureau scores during the transition to CCR
One of the challenges facing many organisations that wish to transition from a negative reporting environment to Comprehensive Credit Reporting (CCR) is how to introduce the new information without increasing the uncertainty of estimating future defaults and losses.
Internal credit scores, bureau negative scores and internal policy rules have allowed organisations to predict default and loss rates with a reasonable degree of certainty. Lenders’ extensive historical data and score monitoring instils confidence in the ongoing use of negative data and information for making the majority of the credit decisions.
Whilst CCR introduces a broad new range of data attributes, they lack the extensive historical performance that legacy negative characteristics have and that ultimately drive a lender’s decisioning systems. In a traditional operating environment, testing of attributes requires lengthy outcome periods before credit scores can confidently be placed into production. Following a traditional scorecard development trajectory means that new CCR attributes may not be able to be implemented for many months or even years.
Equifax has overcome this issue by creating scores that provide optimal weighting between old characteristics (such as defaults and enquiries) and new characteristics (such as age of account and repayment history). The scores have been developed and validated against “CCR Pilot Data” which Equifax was able to obtain prior to commencement of new CCR reporting regime. Since then, Equifax has been able to monitor the performance of these scores across the CCR contributors operating in Private Mode, including the impact of periodic data loads by various lenders.
The results have demonstrated strong score-to-odds alignment between the negative scores and the new CCR scores. This approach ensures that lenders should be able to confidently transition from a negative only score to a CCR score, even while additional data is loaded during the transitionary period.
