Data-driven regulatory oversight

Contact

d-fine’s framework in collaboration with a European regulatory authority

Within Pillar I, supervisory authorities collect extensive data in a variety of frameworks at various frequencies from banks. In the EU, the Common Reporting Framework (COREP) streamlines this process by providing crucial insights into banks' capital and liquidity adequacy. A single bank, on a specific reporting date, may generate tens of thousands of data points. Additionally, regulators such as the European Banking Authority (EBA) and the European Central Bank (ECB) conduct biannual stress tests to evaluate the resilience of banks, assessing their capital ratios against macroeconomic forecasts and regulatory priorities. Targeted stress tests, like the ECB Liquidity Stress Test 2019 (LiST), are conducted occasionally.

d-fine, in collaboration with a European regulatory authority, has developed multiple use cases for leveraging data-driven regulatory oversight for stress testing purposes. Some of these methodologies and key findings are detailed in a comprehensive publication. We aim to enhance how regulators assess banks' economic and normative liquidity without the need of additional ad-hoc data requests. The proposed analyses automatically benchmark all or a cluster of supervised institutions, identifying outliers and potential systemic risks, allowing for targeted actions from the supervisory authorities.

Liquidity risk, given its short-term nature and potential for significant tail risk – illustrated by incidents such as the failures of Silicon Valley Bank and Credit Suisse – underscores the need for timely and comprehensive analysis. Thus, liquidity serves as a logical starting point for these analyses, starting with the following use cases:

Automated Economic Stress Testing Framework: Utilizes monthly AMM data based on ECB LiST 2019 parameters, while allowing flexible definition of additional stress scenarios
Automated Attribution of Changes in Quarterly NSFR Figures: Employs a sensitivity-based approach to assess variations in banks’ NSFR
Automated Analysis of Funding Concentrations: Uses network analysis techniques on AMM data (or any granular data set) to assess funding variability
Consideration of innovative measures of codependence and statistics to effectively capture complex, non-linear relationships in liquidity data

For more details and to receive the full paper, please contact our experts. We look forward to engaging discussions.