C5.3 Events, near misses and root cause

ASQ RCA and 8D [14]; BCBS PSMOR [17]

The third component requires that every loss event and near miss enter one base with its cause in a fixed taxonomy; that the investigation perform a barrier analysis naming, from the register, the control that should have prevented the event and the one that should have detected it, and recording why each did not; that the corrective action be systemic in the sense of the eight disciplines, changing the system rather than the instance; and that a recurrence chart be kept for each closed case, so that closure is demonstrated rather than declared. The obligation to record near misses beside losses is principle 6 of the operational-risk principles [17]; the method is the root cause toolbox of the quality tradition [14].

This is the bridge chapter 12 called the one finance never built. The loss base was constructed for capital, by event type and business line, and the control register lived in another team with another taxonomy; nothing required the two to be joined, and so control effectiveness remained the opinion of the control owner. Barrier analysis joins them by asking two questions of every event, and the second, which detection control should have caught this and did not, is the uncomfortable one because it names a monitoring control that was reported as effective the week before. The component also closes the loop to L4: when a barrier is found to have failed, the control's measured effectiveness is written down, its C4.3 ratings are revisited with the previous values kept, and the expected exposure of every risk that relied on it changes.

Two instruments from the same toolbox complete the component. The recurrence chart plots the event rate after the fix and keeps the case open until the rate has demonstrably stayed down, so that “closed” means “has not come back” rather than “action marked complete”; the framework makes it a field of the incident object rather than a habit of good investigators. The Pareto view turns a year of events into the three causes that explain most of the cost, and it is how the base speaks back to the layers above it: to L3, where the likelihood bands are recalibrated under C5.5, and to L1, where a cause that keeps appearing is a reason to revisit the appetite position for its category. Without the chart, the base is a filing system; without the Pareto, it is a filing system nobody reads.