Banks need better model risk management – Here’s how to build a robust framework
By Kshitij Jain, Senior Vice President, Data and Analytics at EXL
Earlier this year, the collapse of Silicon Valley Bank (SVB) shook investors worldwide. But amongst the chaos, one driving factor wasn’t discussed enough – inadequate modelling and poor model governance.
The collapse of SVB was far from the first market-rocking event that might’ve been softened with better model risk management practices. The impact of the pandemic saw huge strain on lending as many consumers were thrown into unforeseen financial circumstances – all at the same time. The upheaval held a spotlight over industry-wide issues with the collection, monitoring, and modelling of consumer data.
In the case of Silicon Valley Bank, it’s become clear that those in charge clearly didn’t anticipate the combination of interest rate and liquidity risk shocks. Negligence by a single firm had significant knock-on effects across the financial services industry, as many relied on SVB in order to provide their essential services.
Notably, only one of the board members assigned to SVB’s risk committee had any background remotely related to risk management. Without proper knowledge and training, how can boards ask the right questions around MRM practices, given the technical complexities of risk models? Worryingly, this lack of model risk management (MRM) expertise is endemic across banking boards. It is thus not surprising that regulatory guidance such PS6/23 and CP06/22 stress the importance of the board’s role in relation to MRM practices and the risks associated with data driven decisions.
At the same time, other noteworthy regulations like the US Fed’s SR 11-7, the European Central Bank’s TRIM, and, most recently, Prudential Regulation Authority’s PS6/23, all clearly state the terms of MRM and relevant governance. These present a unique opportunity for financial services to take the necessary steps to avoid the same mistakes. Financial services leaders have a duty, not just to their own organisations, but to the entire industry and the consumers they serve, to make model risk functions more robust, especially when the majority of risk decisions are increasingly driven by data.
Wondering where to begin? Here are some practical steps for building a better model risk management framework.
Ensure models are fit for purpose
To properly keep tabs on the performance of a risk model, key performance indicators (KPIs) must be developed for the model’s entire lifecycle. From the quality of input data, algorithm development, validation through traceability, model implementation, right through to third-party dependencies, goals must be set for each part of the process. And once the framework is live, checks and controls should be embedded in a constant monitoring process.
Standardisation is key
Even if model development, validation, and implementation are satisfactory from the start, a weak governance function will reduce MRM effectiveness over time. Formalising MRM standards for models and components across its lifecycle, including tollgates for audits and approvals – through an independent function and board oversight – will ensure minimum surprises and vastly smoother running of processes.
Constant monitoring and updating will help keep in step with the new regulation by making the right updates and identifying where models fall short. The regulatory landscape is still responding to issues with risk modelling highlighted by the pandemic. Financial services providers must comply with new rules while simultaneously adapting to challenges posed by today’s volatile economic environment. It may also be necessary to engage with an expert partner who can help navigate the intricacies of regulatory compliance.
Secure buy in from the board
The SVB crisis exposed the critical need for effective board oversight on banks’ MRM practices. Getting support from senior leaders is essential to implement better risk management strategies and prepare for new regulations in due time, as they can help to reprioritise the business’ activities accordingly.
Once secured, the next step is to maintain that board-level buy-in with clear and accessible reporting. Once MRM processes are effective, compliant and standardised, they can be used to give feedback to senior leadership. The simplest way to give boards better visibility for more informed decisions is to focus reporting efforts on KPIs and communicate model risks using simplified, graphical summary reports highlighting areas such as performance-against-thresholds and aberrant trends. Additionally, expert committees accountable for model risk pertaining to different stages of the model lifecycle, can guide the board with key risks.
Building a better model risk management framework will help future proof banks and other financial services organisations against any more shocks to the market. After all, we can’t control the shocks, but we can learn from past mistakes to build back better.
Why pay for news and opinions when you can get them for free?
Subscribe for free now!