Building reliability in risk

Banca Mediolanum uses SAS® Viya® to develop high-performing, reliable credit scoring models

Risk management is no longer an abstract function to develop risk models when needed and meet regulatory requirements. It is a strategic business asset used in credit scoring.

At Banca Mediolanum, which has more than 2.2 million customers, risk management is cross-functional and affects each department. The evolution of credit scoring models represents a significant shift at the bank. This path has led to the adoption of sophisticated, advanced analytical solutions based on artificial intelligence technologies, such as machine learning, to increase the reliability of credit-scoring models.

Technological innovation provides a smooth transition

When the European Banking Authority changed their definition of default to include stricter criteria and change the way banks and financial intermediaries classify clients, the bank had to find a way to adapt.

“Banca Mediolanum has used machine learning and deep learning techniques well before the new regulations were put in place,” said Stefano Biondi, the bank’s chief risk officer. “This allowed us to consolidate our use of advanced technologies in model validation.”

Not only has the bank’s investment on new technology allowed for a smooth adaptation, but it allowed them to exceed the threshold of 30 billion euros in assets in their last balance sheet. The profit they have gained from adopting reliable, advanced analytics solutions makes it clear why they chose the path they did. It provides a strong support for needed business improvements as well as a faster time-to-market.

A successful pairing of the best models and the latest technological advancements

SAS software and solutions have been an integral part of Banca Mediolanum’s successes. The constant innovation SAS supports enables continuous improvement of the bank’s technological platform.

For example, to obtain the best analytics to face current and future challenges, the bank knew credit scoring improvements was the next step. Using SAS Viya, the bank was able to develop several models simultaneously and verify their validity during development. They also used machine and deep learning techniques in the development phase – both techniques are critical for scoring models in the digital lending domain. The technology SAS provided ultimately strengthened their credit scoring and improved their service.

Transparency and openness, a fundamental shift

“The platform makes available traditional algorithms and the most advanced machine learning algorithms as well as neural networks for sophisticated deep learning," says Fabrizio Manstretta, head of the bank’s credit risk management efforts. “A fundamental aspect for ensuring clarity in communications with supervisory authorities is model transparency and the ability to easily understand analysis results. SAS allows for all of these crucial components and more.”

Integration is ensured by an open platform –­ which allows integration with open-source codes, in-memory operations and development in container as a service. These features enable a high-performance platform that is accessible online. “We have seen large improvements not only in terms of faster development and release, but in terms of reliability and security,” Manstretta said.

Implementing the open platform has been a migration path toward an evolved platform. SAS Consulting and technology partners were crucial considering the technical component of the project, the functional and process skills as well as the collaboration with the internal team.

Next steps: extend the application of advanced models in business processes

“At the beginning, we started with the needs of risk management, but it was a catalyst for innovation for other business units,” Biondi said. They soon realized the power of the advanced analytics laboratory and are working on various projects in model development. This includes the creation of new credit products, as well as exploring the potential of open banking and PSD2 by using the most advanced big data analysis capabilities. Biondi concludes that their advanced analytics laboratory, thanks to SAS Viya, has sufficiently prepared Banca Mediolanum for the future and they are excited for upcoming opportunities to enhance their services.