This article was first published on May 26, 2020.
Editor's note: Oliver Wyman is monitoring the COVID-19 events in real time, and we have compiled resources to help our clients and the industries they serve. Please continue to monitor the Responding To Coronavirus Hub for updates.
As economies around the world gradually come to terms with the COVID-19 pandemic and consider reopening plans, businesses will continue to grapple with the uncertainty and lagging impacts of the pandemic for the foreseeable future.
The pandemic has introduced new and unprecedented challenges for financial institutions across the model life cycle — in development, validation, and monitoring. And COVID-19 will continue to drive a major surge in necessary model management efforts over the coming months to address the expected performance deterioration of existing models.
Our new paper, Model Management In The COVID-19 Era, provides the critical actions to take now and helps financial institutions develop a more resilient model management program to guide 2020 (and beyond) activities.
Financial institutions need to understand the impact of the pandemic on model management and start acting now.
Senior stakeholders will need to tackle a host of difficult questions to guide these efforts, leveraging existing frameworks where possible, while considering alternate solutions that might have previously seemed unacceptable or nuanced approaches that might have previously seemed suboptimal. Financial institutions will need to appropriately preempt the significant surge of model management efforts during and coming out of the pandemic by implementing six immediate actions.
CHALLENGES OF MODEL MANAGEMENT
The COVID-19 pandemic has put large parts of the global economy on “pause” and greatly disrupted the way other parts operate on a day-to-day basis. While financial institutions have grown more resilient through the accumulated experience of previous crises such as the global financial crisis, there are many aspects of the COVID-19 challenge that are unique in nature.
The rapid onset and magnitude of COVID-19 presents a range of challenges across the entire model management life cycle. Model developers and validators across the first and second lines of defense will need to confront these difficult issues while working against time pressure and resource constraints. We see challenges across each stage in the model life cycle including model performance/model monitoring, model (re-)development, and model (re-)validation.
GUIDING PRINCIPLES FOR MODEL MANAGEMENT
SIX IMMEDIATE ACTIONS TO TAKE NOW
There are six immediate actions for financial institutions to take to address the challenge and guide upcoming model management activities. Please see the "Model Management In The COVID-19 Era" PDF version for a comprehensive breakdown of each of these actions.
1Develop a set of guiding principles
Model developers and validators should jointly align on a set of carefully thoughtout principles to provide guidance to involved teams and drive consistency in model management decisions during the COVID-19 pandemic.
2Establish a tiering framework to the model inventory
Financial institutions should tier the impacted models from the overall model inventory based on expected COVID-19 impact. The tiering will be critical to effectively control the increased model management workload and to concentrate on the most material areas of increased model risk.
3Adopt a more nuanced interpretation of model performance results
Model developers and validators should align on how to interpret model performance and monitoring results, and work closely together to understand areas of concern and implications of monitoring exceptions for in-use models. Adopting a more nuanced interpretation of model monitoring results will help manage the expected large volume of exceptions.
4Create standardized decision trees to guide model treatment during the pandemic
Given the expected surge of model monitoring exceptions, these decision trees will provide templates to guide the decision-making process and support convergence on model management outcomes.
5Revisit the approach to compensating controls needed during the pandemic
A key step is to assess the full range of compensating controls available for use, including, but not limited to, overlays, assumption changes, adjustments, use of challenger models (where available), and the use of benchmarking/reference points.
6Convene a “Model Management SWAT team” of experienced stakeholders to provide oversight and guide triage decisions
Given the challenges that will need to be thoughtfully addressed over the coming months, this team can help the efficiency and effectiveness of the approach through improved issue management, better decision making, prioritization, and more reporting on the situation to senior management and the board.
MODEL MANAGEMENT SWAT TEAM
The extraordinary and rapidly evolving nature of the COVID-19 pandemic poses a host of new and unprecedented challenges for model management practices across the financial services industry. Most notably, the pandemic will cause a significant surge in model management activity over the coming months. While the need for increased effort and resources will likely be more acute starting towards the end of 2020, the time for financial institutions to prepare is now.
A critical step for senior executives is to take tangible actions upfront to ensure that institutions are focusing on the right set of problems and effectively managing the increased workload. The six actions outlined in our paper will help financial institutions better deal with the situation and develop a model management plan — including timelines and resources, that guides the remainder of 2020 (and beyond) activities. Financial institutions able to implement these six actions will be in a position to best monitor and address model management issues as we navigate through the pandemic. These actions will help financial institutions reduce the operational burden by focusing on the most material model management questions, and emerge with a stronger and more resilient model management program.