Pioneer Award: Bank of Mongolia

The Bank of Mongolia’s big data project collects granular data from all banks in its jurisdiction, daily, using a single data model that covers statistical, regulatory and business requirements.

The new dataset includes green taxonomy, merchant category code, management gender, postcode and International Standard Industrial Classification of All Economic Activities, in addition to traditional financial data. With this initiative, the banking sector in Mongolia enters a new data era, enabling banks to develop digital loan products and improve cashflow, while providing regulators with real-time, accurate data.

“At midnight, banks run their end-of-day batch payment processing, with larger banks taking up to six hours and smaller banks taking close to half an hour to complete,” Byadran Lkhagvasuren, governor of the Bank of Mongolia, tells Central Banking. “Once this process is completed, we automatically pull data and process it, ready to be viewed at 9:00am with our morning coffee.”

In 2021, the central bank began a knowledge exchange partnership with the Bank of Korea and Sangmyung University in Seoul, which helped it develop a big data road map and a more thorough understanding of what it wanted to achieve and how to achieve it.

With the launch, 200 banking industry employees were reassigned from data preparation to other tasks, resulting in a significant cost reduction. “With multiple departments at banks involved in the data reporting process, and different departments at the central bank receiving the data at various levels, consistency between the received data and reports were difficult to maintain,” says Lkhagvasuren.

Now, 61 template-based reports covering more than 200 sheets have been consolidated into a unified data model, and the average time lag of official statistics has been reduced from 17 days to one.

“We don’t have to burden the banks with data preparation. Instead, we take all the responsibility for data preparation on ourselves,” says Lkhagvasuren. The central bank used to receive four gigabytes of data each year from the banks. Today, it estimates the platform will collect 17 terabytes of data annually.

The big data platform uses a staging database in a parallel warehouse that is kept on the central bank’s premises and transmits encrypted data through a closed-circuit connection to supervisors and banks on a shared platform. “From data generation and collection to data compilation and dissemination through our official website, application programming interface, dashboards and other sources, our whole process is fully automated, with little to no human intervention.”

The team visited banks in person to convince them of the merits of the project and sharing detailed data. “We had the opportunity to ‘grassroot’ the project from the bottom up and really make the banks believe in the project,” says Lkhagvasuren.

The Bank of Mongolia can now also conduct more comprehensive stress-testing. Previously, the central bank could only look at accumulated risk on an aggregate level. “Now, we can delve deeper into cases such as a shock in the agricultural sector,” says Lkhagvasuren. End-users can dissect, aggregate and analyse data patterns, without compromising individual privacy.

The Bank of Mongolia is now working to develop new suptech tools in three areas: creating new (early-warning) indicators and improving nowcasting models; conducting detailed stress tests on household debt in the banking sector; and creating a fraud detection alert system.