Since 2021, the Irving Fisher Committee on Central Bank Statistics (IFC) has established a series of workshops on “Data science in central banking”, gathering statisticians and economists from central banks, international organisations and national statistical offices (NSOs) in addition to other stakeholders. The aim has been to review the adoption of data science in the central banking community and beyond, identify the associated opportunities and challenges as well as exchange best practices. The third edition focused on how data science can effectively unlock the potential and value of data by fostering their use, reuse, access and sharing.
Against this setting, four main takeaways stand out.
First, innovative techniques can play a central role in leveraging traditional as well as emerging data sources and types, with benefits for both producers and users of economic and financial information in central banks. For producers, they offer various tools to streamline statistical processes and improve data quality, especially in terms of accuracy, frequency, timeliness and granularity. For users, they enable the analysis of large, complex and/or high-dimensional data sets to extract meaningful insights or detect new patterns.
Second, data science can support adequate and secure sharing of data sets – particularly granular ones – without disclosing sensitive information, for instance by using tools such as privacy-enhancing technologies.
Third, challenges persist when tapping into the increasing and various amounts of information available today. A prominent one is the need for robust yet costly information technology (IT) resources, such as those required to handle and share large data sets. Moreover, organisational barriers can prevent the successful integration of new data techniques into existing processes, for example, because of “silos” or a limited coordination between subject matter experts, IT and data scientists. Efficient access to and use of data may also be constrained by still limited standardisation. This is particularly the case for alternative or secondary data sources, which often lack common definitions, methodological consistency and alignment with well established standards, in turn limiting their use for compiling official statistics.
The fourth takeaway is that, fortunately, central banks as both users and producers of statistics are well equipped to address these challenges, notably thanks to their long-standing experience in big data analytics, information standards and, more broadly, data management and governance.
Looking ahead, facilitating effective data access and sharing to make the most of the information available to central banks may call for further progress in the following key areas:
- Data management and governance to ensure the quality of statistical information, including its availability, transparency and usability.
- Interoperability capabilities and statistical standards to enable effective and accurate use of data, for instance by fostering their integration, findability, access and sharing for reuse, especially by data scientists.
- Modern, metadata-driven and easily accessible (big) data platforms within organisations complemented by single access points for the public and data spaces more generally.
- Strengthened collaboration among peers and counterparts to advance data science projects, by developing IT solutions for secure and adequate data access and sharing as well as by promoting a structured exchange of experiences across interested stakeholders.