First Consulting helps a large bank with risk monitoring

09 October 2020 Consultancy.eu 3 min. read

First Consulting has assisted a large bank in the Netherlands with improving the risk monitoring of its leveraged portfolio. Each client is monitored at a centralised, structured location. By providing visual insights into historical data, the bank can provide a better estimation of the risk level of new loans.

The structuring, execution and management of event-driven loans is monitored by one department within the bank. In light of the current uncertain outlook and facing increasing pressure from regulators (e.g. ECB and FED), the department wants to gain more control on the financial position of its customers in order to minimise risks and curb losses. 

The team was restricted by an outdated system, which made monitoring a time-consuming, complex, and inefficient process. Following a design sprint, the department decided to replace the old tool with a new application and dashboards. The new application and all the monitored information will enable the bank to create a model, which predicts before and after the execution of a loan whether it can be paid within the agreed terms.

On the back of First Consulting's experience with complex Mendix applications, PowerBI implementations and data migration, First Consulting was called in to design, build and implement the digital solution. 

Approach

The approach consisted of three phases, each lasting to the tune of three months. During each phase, an Agile working method was used, allowing functionality to be developed at a rapid pace in short iterations. This approach enabled the consultants to effectively manage the changing needs and feedback from stakeholders within the bank.

Approach for better risk monitoring of the leveraged portfolioBy involving the team involved in the project with regular demos, they were able to quickly master the new system and the new functionalities and train new team members themselves.

Phase 1 - Centralise
In the first phase, the First Consulting team developed the Minimal Viable Product (MVP) in collaboration with the department. This MVP consisted of a Mendix application and a PowerBI dashboard. The application stores all relevant historical data, which is then displayed on the client dashboard using slice-and-dice visualisations. As a result, the status of the customer can be analysed year-on-year. This allows the department to monitor clients/loans more effectively.

Phase 2 - Visualise
In the second phase, new functionalities were added to the application, such as client-specific financial warnings. Customers facing difficulties are immediately identified by the system and users are quickly alerted. Moreover, a second dashboard was built, aimed at enabling managers to monitor the overall status of the portfolio.

As a result, the bank can quickly make year-on-year performance comparisons of the whole portfolio and determine when (timely) actions need to be taken. Both dashboards can also be used before executing new loans, by analysing the status of similar customers (e.g. industry peers).

Phase 3 - Automate
The main objective of the third phase was to integrate the application with source systems, to ensure that the necessary information is automatically loaded. Furthermore, new functionalities have been added that simplify the monitoring process. For example, reports related to special situations, such as Covid-19, can be written and exported with the click of a button. As a result, the team can spend more time on the primary risk management process. 

Result and next steps

Thanks to the new application and dashboards, monitoring has become a more efficient process with many steps automated. The department can monitor individual clients and the whole portfolio and achieve better, data-driven estimations on the risk level of new loans. Moreover, regulatory audits are now easily conducted. 

The next step is dedicated to further automation of the input of customer and market data using robotic process automation (RPA). Scanning files using document recognition technology contributes to this further automation. Ultimately, this will contribute to a more accurate and effective model.