The Financial Stability Institute released a report on supervisory technology (suptech) tools for prudential supervision and their use during the pandemic.
Suptech data analytics tools are making prudential supervision more efficient and bringing new insights on risks, areas of concern and other supervisory issues. These tools use a range of data, both qualitative and quantitative, from different sources. Tools for text analysis, text summarization and information classification allow for much faster extraction of useful insights from voluminous documents than would be manually possible. Tools for sentiment analysis, risk identification, network analysis and peer group identification provide insights from data that may not have been detected by just using traditional rule-based or statistical tools. Furthermore, tools that automate parts of the inspection process result in both more efficiencies and more significant supervisory information.
Suptech tools, particularly those for analyzing unstructured data and risk identification, proved essential during the pandemic. The migration of most on-site activities to off-site work and the various pandemic-related reports required by authorities added to the already huge pile of structured and unstructured data that supervisors must go through. In particular, suptech tools have given much needed support in the supervisory reviews of corporate governance and credit risk. NLP tools have been able to identify potential corporate governance risks from a large set of documents that may have bearing on a bank’s governance practices. Authorities also deployed risk identification and other suptech tools that help supervisors identify credit exposures that may be misclassified or under-provisioned.
The pandemic experience highlighted the need for continued improvements in IT infrastructure and data collection practices, which can support ongoing exploration of new suptech tools. A sound IT infrastructure offers an effective platform for deploying suptech tools, while improved data collection practices open further possibilities on the types of analyses supervisors can perform with suptech. Authorities are already exploring emergent tools that could have far-reaching impacts. Network analysis tools may be able to assess the likelihood that current interconnections between entities may foresee not just existing but also emerging risk concerns. Peer identification tools may result in a deeper understanding of similarities between supervised firms, perhaps making the supervisory scoping, review and analysis wrap-up process more consistent and resource-efficient. Sentiment analysis tools may provide more nuanced insights on information gathered not just from social media, but from confidential supervisory data as well. Inspection automation tools could overhaul the whole supervision process, thus profoundly affecting resource management. Authorities are also experimenting with tools that could be applied to broader themes that may shape our post-pandemic supervisory world, including climate change, cyber and the supervision of digital-only financial firms.
Implementation challenges – from lagging data science skills to governance complexities, from data quality to data integration – continue to hinder broader deployment and acceptance of suptech tools. Many of these challenges have scaled up and become pressing as more suptech tools are operationalized. Continued capacity-building through training and hiring can alleviate data science skill shortages. An emphasis on developing or updating a suptech strategy can clarify governance arrangements and soften supervisors’ resistance to new ways of doing their daily work. In addition, data quality will continue to be a limitation of data from alternative sources, such as social media. Moreover, while many authorities are already finding ways to integrate data from disparate sources, including using centralised data storage such as data lakes or the cloud, security issues remain a concern, particularly those relating to cloud storage.
It is important that deployment of suptech tools be accompanied by appropriate safeguards to mitigate potential unintended consequences. Appropriate governance frameworks have been crucial in spurring suptech development and usage, including suptech strategies and buy-in from the board and senior management. Going forward, governance structures may have to clearly address the growing and often ill-defined role of suptech tools in the organisation. As more suptech tools are deployed, supervisors may not be incentivised to exercise judgment and instead just rely on the tools to detect issues. Such over-reliance on suptech could diminish judgment-based supervision and could eventually lead to more supervisory blind spots. There is also the broader issue of whether there is still a role for on-site inspections. However, supervisory judgment is heavily informed by physically observing and assessing banks’ governance, culture and controls. Authorities have been clear that suptech tools are meant to enhance, not replace, supervisory judgment. Nevertheless, this message needs to be reinforced by introducing clear guidelines on the role of suptech in supervisory processes.