Introduction
In 2024, regulators are increasingly vigilant about the model risks faced by financial institutions. This heightened scrutiny is driven by several high-profile failures, including Archegos Capital, FTX, Silicon Valley Bank, and Credit Suisse. These incidents have underscored the importance of robust risk management practices, prompting regulatory bodies like the UK’s Prudential Regulatory Authority (PRA) to urge banks to thoroughly assess their liquidity, credit, and counterparty risks. This blog post delves into the key concerns of regulators, providing examples and insights into how these issues are being addressed.
Liquidity, Credit, and Counterparty Risks
Regulators are particularly focused on ensuring that banks have completed adequate assessments of their liquidity, credit, and counterparty risks. This includes conducting thorough due diligence to prevent fraud and implementing effective credit risk methodologies, especially for entities with concentrated operating exposures. Examples of such concentrated exposures include crypto assets, sector-specific concentrations, and asset-liability mismatches.
Example: Archegos Capital
The collapse of Archegos Capital highlighted the risks associated with concentrated exposures and inadequate counterparty risk management. Archegos’s use of total return swaps created significant leverage, leading to massive losses when the underlying stocks plummeted. This incident prompted regulators to scrutinize the risk management practices related to leveraged trading and counterparty exposures.
Climate Change Risks in Long-Maturity Commitments
Long-term risks, such as those related to climate change, are another area of concern for regulators like the US Federal Reserve and the PRA. These risks are particularly relevant for long-maturity commitments, such as infrastructural or mortgage lending. Regulators expect banks to incorporate climate change risks into their model estimations to ensure they are prepared for future environmental challenges.
Example: Mortgage Lending
Banks involved in mortgage lending must consider the long-term impact of climate change on property values and default rates. For instance, properties in areas prone to flooding or other climate-related events may see their values decline, increasing the risk of defaults. Regulators are pushing for models that can accurately incorporate these long-term risks.
Established Model Risk Management (MRM) Practices
The broad principles of MRM have been established through the US Federal Reserve’s SR11-07 guidance and the European Union’s Targeted Review of Internal Models. Regulators now expect banks to maintain high visibility on their internal model risks, including model inventory, review lifecycle, and usage governance. They also require evidence of established control processes.
Example: Model Inventory and Lifecycle Management
Banks must maintain an up-to-date inventory of all models in use and ensure that each model undergoes regular reviews and updates. This process helps identify and mitigate potential risks before they lead to significant problems. For instance, a bank might have a credit risk model that needs recalibration due to changing market conditions. Regular reviews ensure that such models remain accurate and reliable.
Challenges from the Pandemic and Geopolitical Events
The COVID-19 pandemic and the aftermath of the invasion of Ukraine have added complexities to risk management. Regulators have extended timescales for full Basel III compliance, recognizing the challenges banks face in these unprecedented times.
Example: Oil Market Volatility
The pandemic caused significant volatility in the oil market, with West Texan Intermediate (WTI) futures trading below zero due to logistical storage issues. This extreme scenario tested the robustness of banks’ risk models, highlighting the need for timely adjustments and effective communication between model developers and stakeholders.
Influence of Machine Learning on MRM
Machine learning is transforming MRM by introducing new methodologies for model development and validation. However, these complex models also bring additional requirements for robustness and explainability.
Example: Interpretability of Machine Learning Models
As banks adopt machine learning models, they need to ensure these models are interpretable. Techniques such as local interpretable model-agnostic explanations (LIME) and Shapley additive explanations (SHAP) help in understanding model decisions. For instance, a machine learning model used for credit scoring must provide clear explanations for its decisions to satisfy regulatory requirements.
Regulatory Concerns Over Internal Models for Credit Risk
Regulators are cautious about the reliance on internal models for credit risk measurement, as these models can be complex and difficult to assess. There is a debate over whether to standardize models to simplify regulatory compliance or allow banks to use internal models that can be more tailored but harder to regulate.
Example: Standardized vs. Internal Models
If regulators decide to limit the use of internal models, banks would have to adopt standardized models, which might be simpler but less flexible. This could impact banks’ ability to innovate and manage risks effectively. However, the transparency and ease of regulation might outweigh these drawbacks, ensuring a more conservative approach to risk management.
Impact of the Fundamental Review of the Trading Book (FRTB)
The FRTB, part of Basel III, focuses on market risk estimations for banks’ trading portfolios. It sets stricter criteria for internal model use, requiring realistic estimates of historical volatility and correlations.
Example: Market Risk Management
Under FRTB, banks must demonstrate that their risk representations capture actual market profit and loss accurately. For instance, a bank trading in complex financial instruments must use models that accurately predict historical losses and are robust to market changes. This ensures that banks are adequately capitalized against potential market risks.
Conclusion
In 2023, regulators are emphasizing the importance of robust model risk management across various aspects of banking. From addressing liquidity, credit, and counterparty risks to incorporating climate change considerations and adapting to the challenges of the pandemic, banks must ensure their models are accurate, transparent, and resilient. The adoption of machine learning and the impact of regulatory changes like FRTB further complicate the landscape, requiring banks to stay agile and vigilant in their risk management practices.
