Problems with Stress Testing

Since the 2008 financial crisis, banks around the globe have been subject to stress testing. The concept of stress testing is found in engineering (can bridges withstand wind speeds of x mph?), computer science, cardiology, birthing, nuclear power plants, etc.. In banking, the idea of stress testing is whether a bank can remain insolvent under certain economic scenarios (e.g., a fall in house prices, a rise in unemployment etc) and how it would fare if a 'tail event' were to hit the banking system (e.g., the failure of a major financial institution). Central banks conduct these tests to ensure that banks have enough capital to absorb losses which may occur under any of these circumstances. Below is a Bank of England video which explains how they go about stress testing banks.

However, I have two issues with stress tests. First, stress testing makes banking stability akin to an engineering problem. But stress tests cannot pick up or model how human behaviour changes in reaction to changes in economic conditions or tail events. Second, the stress testing approach views capital as a cushion to absorb losses. As I argue in Banking in Crisis, this ex post view of bank capital is not the way our ancestors viewed bank capital - they viewed it in an ex ante sense i.e., bank capital constrains banks from taking excess risk and encourages decision-makers to act in a prudent fashion.

Popular Posts