Read this article in Spanish »
Credit Bureau data modelling & scoring
A workshop dedicated to consumer credit bureau scoring was hosted as part of the Experian Decision Analytics Forum in Venice. Participants of the workshop were representatives from credit reference agency organisations from the Bahamas, Ghana, Kenya, Greece, Nigeria, Pakistan, and Saudi Arabia and the event was chaired by myself, along with Simon Harben, SVP of Global Analytics of Experian’s Decision Analytics division.
Experian provides credit bureau software solutions across different geographies, varying from well-developed economies like Germany or Turkey, to emerging countries like Pakistan or Nigeria. The credit bureau evolution roadmap comprises implementation of value-added products to streamline the value of data accumulated in the credit bureau database. One of the most effective tools in Experian’s value-added products portfolio is the credit bureau score, known as Delphi for New Business.
This is a powerful risk assessment tool, which collates all available information into a single, highly predictive analytical index for new credit applications. Delphi for New Business was developed following scorecard criteria such as the existence of previous payment defaults, the timing of these defaults, the presence of good payment performance on other accounts and overall levels of indebtedness, in the assessment of credit risk.
"A credit bureau score is strongly influenced by data availability that could vary from country to country..."
The session in Venice provided evidence of the number of different goals in risk management processes that can be achieved with credit bureau score development. In fact, Delphi for New Business is just one of the many value-added products available. Credit bureau scores can be developed to assess over-indebtedness, insolvency probability and collection success rate, and a company’s failure/bankruptcy and application fraud probability, just to mention the most important ones.
Considering the audience diversity in the workshop, it was fitting to note that a credit bureau score is strongly influenced by data availability that could vary from country to country. The ideal dataset includes positive and negative credit lines, previous application data and employment and income details. Pakistani and Saudi Arabian credit bureau representatives acknowledged that their level of sophistication would have allowed a robust/bespoke scorecard implementation. The others stated that in light of their recent establishment, they might be able to use a ‘start-up’ credit bureau score based on Experian’s know-how and expertise.
Experian’s approach to developing ‘start-up’ bureau scores was discussed in some detail. The process would be broken into different tasks such as:
Familiarisation - to understand data available, its reliability and any legal constraints;
Definition of credit bureau interface - to identify key data characteristics to be included in the generic credit bureau score;
Scorecard definition - to simulate the expected target population and assign scorecard points accordingly;
Validation - to distribute the sampled population into the appropriate individual's 'credit rating'.
The ‘start-up’ credit bureau scores technique generated a good deal of discussion within the workshop and was generally very well received. All of the workshop participants agreed that it was a key driver for a successful start-up credit bureau and a way to better address users’ expectations.
Roberto Giannantoni
Credit Bureau & Fraud Services Director
Decision Analytics
Experian
Contact us for further discussions about this article