The COVID-19 pandemic is taking a terrible toll in human life and in the livelihoods of millions the world over. As people and institutions struggle to contain the spread of the virus, the measures necessarily imposed have caused major economic disruptions. Every industry has been affected, and banking is no exception. Capital, profit-and-loss, and liquidity positions have been hit very hard. One consequence has been that banks’ models have broken down across their business. The flaws have put the reliability of these models in doubt and suggest that they cannot be trusted to help banks navigate through the crisis.
Few business leaders could have foreseen a global economic shutdown of this magnitude. The models that financial institutions depend on to run their businesses simply did not account for such a crisis. Most models are almost by necessity designed to predict a stable future. In truth, the real failure is not that banks used models which failed in this crisis, but rather that they did not have fallback plans to manage when the crisis did come.
There are a number of reasons for the failures. First, model assumptions and boundaries defined at the design stage were developed in a pre-COVID-19 world. Second, most models draw on historical data, without the access to high-frequency data that would enable recalibration. Finally, while access to the needed alternative data is theoretically possible, models would not be able to integrate the new information in an agile manner, because the systems and infrastructure on which they are built lack the necessary flexibility.
Banks are experiencing ever more model failures, and further issues can be expected with time. Financial institutions must now urgently review their model strategies. They need to develop and apply both efficient short-term actions and a long-term plan to improve model resilience. Over two prioritized time horizons, banks can carry out coordinated model adjustments to enable business continuity in the short term, while reviewing their model development and redevelopment needs and upgrading their model-risk-management (MRM) frameworks over the longer term.