Taking credit risk management to a higher level
SAS enables accurate modeling and fast analysis
Record Bank recently succeeded in building a new credit risk model that features outstanding accuracy and predictive power. However, the model proved to be too complicated for implementation in the bank’s existing software environment. Management was faced with two options: simplify the model and sacrifice some of its capabilities or search for an alternative software platform able to utilize the model’s full potential. Record Bank opted for the latter.
It is now very easy for us to adjust various parameters in the credit risk model and determine how these changes affect the bank’s credit risk exposure. The feedback we get from these analyses is paving the way for even better decision making.
Credit Risk Manager, Record Bank Belgium
Record Bank is the third largest retail bank in Belgium, employing approximately 750 people in Brussels, Ghent, and Liège. The bank offers the traditional set of banking services, including savings, investments, and loans. The bank has a credit portfolio of up to 13 billion euros and carefully manages its credit exposure. “We are continuously searching for ways to make more accurate predictions and thus minimize our exposure to credit risk,” explains Jan Dodion, Credit Risk Manager at Record Bank.
Implementing a powerful, but complex credit risk model
As part of its pursuit of continuous improvement, Record Bank developed a new and much improved credit risk model. Matthias Barbé, Credit Risk Modeler at Record Bank is rightly proud of the effort put into building a very realistic and well-balanced credit risk model. “Our new model uses a very rich but complex dataset consisting of numerous correlations. It takes into account a vast set of parameters such as marital status, repayment trends, and arrear history, as well as how they affect each other, to accurately predict a full picture of credit risk.”
Unfortunately, implementing the new model into the bank’s existing software environment proved to be much too difficult. Barbé explains: “It would have taken more than six months to implement the model, and even then we would have had to make significant sacrifices and remove many vital correlations just to make it work. This was not an option for us. We decided to look for a proven software solution that would be able to accommodate such a complex model.”
Changing to a SAS environment
Record Bank turned to SAS for help. Over the years, many SAS solutions had found their way into various departments of the bank. “We already used SAS for data warehousing, analyses, and monitoring purposes,” explains Dodion. “Only for implementing the credit risk models into Thaler, our operational banking system, we were still using other software. It was quickly obvious however that this set-up was no longer able to live up to our standards. The decision to contact SAS was a no-brainer for us. And as it turned out, they had just the solution we were looking for.”
The SAS solution for credit risk modeling enabled Record Bank to implement its new and improved credit risk model, without making any concessions whatsoever. “Thanks to SAS, we can utilize our powerful new model exactly as it was conceived, with all the correlations that we had defined. The result is that we gain the full benefit from its unique predictive power,” observes Barbé.
More efficient monitoring
Not only did SAS enable the integration of the credit risk model, it also helped realize considerable efficiency gains. The previous implementation software did not feature its own business intelligence (BI) tool. As a result, Record Bank needed to rebuild the credit risk model in a separate BI environment to monitor the results of the credit risk model. Changing to SAS totally eliminated this time-consuming approach. “Our BI environment already was a SAS solution,” states Dodion. “This enabled us to make a direct link from the BI environment to our SAS credit risk model, without having to remodel anything. This saved us enormous amounts of programming.”
Thriving from scenario and impact analyses
Establishing the direct link between the credit risk model and the BI environment reinforces the continuous improvement of the model. It enables automatic adjustment of the credit risk model based on monitoring feedback. “It also allows us to execute a wide array of impact and scenario analyses,” explains Dodion. “It is now very easy for us to adjust selected parameters in the credit risk model to determine how these changes affect the bank’s credit risk exposure. This has paved the way for even better decision making.”
Getting ready for real-time credit risk calculations
Record Bank is already examining new ways to extend the possibilities of its new system. “In time, we want to extend the use of the SAS credit risk model from batch calculations to individual real-time calculations. This will provide agents in our local offices with the ability to calculate the risk of every new credit applicant based on real-time credit risk data from our extensive data warehouse. This will enable us to anticipate new events or trends in the market far more quickly than ever before and limit our credit risk exposure even further. Above all, it means staying one step ahead of the competition,” reports a very pleased Dodion.
Implement improved, but complex credit risk model
- More efficient credit risk monitoring
- Better predictions based on feedback from impact and scenario analyses
- Better and faster decision-making process
- Manage driver metadata from the beginning.
- Create more model packages directly after model development.