This class made me realize how imperfect operational risk capital models really are. I always assumed approaches like AMA or SMA were solid and data-driven, but they each have big limitations, especially when a bank’s own loss history is thin or unreliable. It made me wonder how Meridian should judge whether CAD 40 million is actually enough, how much weight to give to the regulatory formulas versus our own view of resiliency, and how to adjust capital when real fines can be so much larger than what the models predict, like the TD AML case.
This class made me realize how imperfect operational risk capital models really are. I always assumed approaches like AMA or SMA were solid and data-driven, but they each have big limitations, especially when a bank’s own loss history is thin or unreliable. It made me wonder how Meridian should judge whether CAD 40 million is actually enough, how much weight to give to the regulatory formulas versus our own view of resiliency, and how to adjust capital when real fines can be so much larger than what the models predict, like the TD AML case.