Bigbank – Adapting modelling and validation steps
“The FCG workshop was a great benefit for us, we received the necessary information in a compact and comprehensible way. FCG has focused specifically on our situation and the individual knowledge of the workshop participants, and thus presented applicable hands-on solutions. We appreciated that FCG also after the workshop supported us by answering concrete modelling questions related to our own portfolio.”
– Jüri Kuusik, Head of Credit Risk Modelling Unit
Improved data quality
Clear implementation strategies
State-of-the-art modelling steps
Bigbank is an Estonian bank with headquarters located in Tallinn. The bank operates in the Baltic countries, Finland, Sweden, and Bulgaria, but also takes deposits from Germany, Austria, and the Netherlands. Bigbank has specialized in granting loans, leasing, and deposits both to private and corporate customers. Bigbank’s vision is to become the most recommended digital financial services provider in all geographies where Bigbank operates. The mission to improve people’s lives through seamless financial services is also reflected in the credit evaluation process.
FCG has conducted workshops together with Bigbank about scorecard development where state-of-the-art modelling steps were presented, and concrete examples about the do’s and don’ts within model development were presented. Additional recommendations were made with regards to data quality checks and validation of the models.
The main driver of the assignment was to prepare Bigbank for the scorecard model development process through training and discussing mathematical and statistical topics. The workshop also discussed and highlighted business considerations during the development.
Within credit risk modelling the first step of the project is to define objectives and the scorecard role in the bank. Furthermore, there is a need to discuss how and where the implementation will take place. FCG presented different options and views to make it possible for Bigbank to decide on how the scorecard development process should be designed.
FCG covered the data selection and sampling process with recommendations for a potential scorecard development data mart covering the new default definition, including which data quality checks should be implemented. Furthermore, FCG explained how sampling could be done.
FCG focused on the scorecard model development, covering areas like risk driver selection techniques, univariate analysis, and multivariate modelling suitable for the scorecard objectives. Model (performance) evaluation criteria was covered as well.
FCG provided technical know-how and comprehensible practical examples so that Bigbank could adapt modelling and validation steps based on the workshop.
Bigbank received additional information about what needs to be done if the model is in place, therefore model validation and monitoring steps were discussed. Implementation was covered a
s well – FCG provided Bigbank with best practise examples for the implementation phase of model development.
The FCG workshop was a great benefit for us, we received the necessary information in a compact and comprehensible way. FCG has focused specifically on our situation and the individual knowledge of the workshop participants, and thus presented applicable hands-on solutions. We appreciated that FCG also after the workshop supported us by answering concrete questions specifically related to our own portfolio.Jüri Kuusik, Head of Credit Risk Modelling Unit
Although model development is very analytical and mathematical, business input and feedback should never be underestimated. The first line of defence/ local risk officers should always be part of the modelling process, so that they can contribute in the development process to secure the business perspective.Monika Nauroth, FCG