On this page
A sound operational risk review uses two directions because different people can see different failure mechanisms. Senior leaders can identify firm-wide exposures and emerging threats; people who run a process can identify the precise step and control that may fail. Reconcile those views in a common risk register, with clear ownership and evidence, rather than treating either view as a complete inventory.
What the top-down view can see
The top-down starting point is the organisation's objectives, risk appetite, key services and external environment. A board or senior risk team may see that several businesses depend on one technology provider, that a regulatory change affects more than one product, or that a strategic expansion increases operational complexity. A single processing team may know its own workflow well yet miss the cumulative exposure across the group. Senior leaders can therefore ask which shared dependencies deserve a closer process review.
This approach can use a broad risk taxonomy, discussion of exposures and vulnerabilities, scenarios, and environmental scanning. Such tools are prompts to look for causal paths. A vendor outage, for example, may delay payments, disrupt customer communication and create a backlog for reconciliation. Calling these three separate risks without examining the shared dependency may understate the concentration. The point of the high-level view is to make the connection visible and decide where detailed work is needed.
The danger is abstraction. A strategic risk label such as technology disruption has little diagnostic value until someone can identify the system, affected activity, failure mode, control and possible consequence. A heat map may show a coloured cell, but it cannot tell an operations manager whether a fallback file is tested or whether an override leaves no audit trail. A top-down exercise should end by commissioning or refining specific process investigations, not merely by producing a list of broad categories.
What the bottom-up view adds
A bottom-up review begins with the work itself. Process owners map each hand-off, data input, approval, exception and reconciliation. They ask how an error could enter, how it would be detected, who could stop it, and how long it might remain unnoticed. The Basel Committee's operational risk principles describe self-assessments, event data, near misses, control monitoring and scenario analysis as tools for identifying and assessing risk. These sources of evidence are complementary because each reveals a different part of the control environment.
A risk and control self-assessment, or RCSA, separates exposure before controls from exposure after controls. The first is inherent risk; the second is residual risk. A risk does not disappear merely because a written control exists. The reviewer needs evidence that the control is properly designed, operates at the relevant point in the workflow, and is used under normal pressure. A control that is bypassed whenever volumes rise may reduce a spreadsheet score while leaving the practical exposure high.
Event data helps test what the map predicts. A recorded loss shows that a failure happened; a near miss shows a plausible route to loss even when an independent control, a quick correction or chance prevented the financial impact. Near misses should not be dismissed as harmless merely because the booked loss is zero. The Basel principles explicitly include near misses in the event dataset. They can reveal an unstable hand-off before an expensive incident creates a clearer signal.
The limitation of bottom-up work is its local horizon. Teams may rate similar risks differently, omit a dependency that sits outside their department, or concentrate on familiar historical losses while a new product changes the threat. A common taxonomy and challenge from a risk function make findings comparable and help escalate a pattern that no one team owns. GARP's discussion of top-down and bottom-up assessment describes those distinct vantage points and the need to aggregate them consistently.
Follow an illustrative payments process
Consider an invented bank whose payment operations group handles 1,000 outgoing instructions on a typical day. This number is an illustrative planning assumption, not a claim about a real bank. The board identifies dependence on one external messaging provider as an enterprise exposure. That top-down finding tells the operations group to examine how a provider outage could affect authorisation, queued instructions, customer status messages and end-of-day reconciliation.
The process team discovers that a manual fallback file can carry instructions during an outage, but its beneficiary field is checked only after transmission. In an invented tabletop review, staff find that 40 instructions might enter the fallback route on a busy day and that a second-person check would currently occur after the point at which an incorrect instruction could leave. These figures are deliberately invented to make the sequence concrete. The important finding is the order of control and release, not the arithmetic alone.
The inherent risk is the possibility of sending a wrong payment while normal validation is unavailable. The claimed fallback check does not prevent that event because it arrives too late; it may help detect and respond after release. Residual risk remains material despite the presence of a documented control. The process team proposes moving the beneficiary check before release, records the owner and tests the altered workflow. The board-level concern becomes a specific repair with a testable completion condition.
The same exercise can run upward. Suppose several teams report similar fallback files and delayed checks. Aggregating these bottom-up findings changes the enterprise description from a single provider outage to a wider weakness in manual contingency processing. Senior management can then compare priorities across businesses and ask for a common control standard. That upward feedback is why the two directions are a loop rather than two separate reports submitted once a year.
Turn a finding into a decision
For each material risk, the reviewer should be able to state the cause, event, consequence, existing controls and remaining exposure. Without that chain, it is hard to distinguish a genuine risk from a general concern. The proposed response should address the weak link. If the issue is that a check occurs too late, more frequent reporting after the fact may improve visibility but does not prevent the payment. If the issue is that staff lack access to an alternative system, a beautifully worded procedure will not create that access.
Ratings can help compare issues, but a high or medium label should not replace evidence. Define the scale, the time horizon and the unit being rated. A severe event with low frequency can still require attention, while repeated small exceptions may indicate a deteriorating control. Where different teams use unlike scales, aggregate counts or colours may mislead the board. GARP's risk identification discussion recommends consistent rating approaches so local findings can be compared across the organisation.
When a control is improved, rerun the assessment rather than simply changing the residual rating. Verify who performs the control, what evidence it creates, when it operates and how exceptions are handled. A near miss after the change may show that the underlying route is still open. A change in product, vendor or workflow can also justify a fresh review before the normal cycle; the risk register must reflect the actual process being run.
The exam distinction
The FRM programme page places operational risk in the wider Part II context, and the methodology page offers a way to organise a study check. For this topic, classify the viewpoint before choosing a tool. A board scan of external threats and cross-business dependencies begins high. A process map, local self-assessment and analysis of near misses begin near the work. Neither source of evidence deserves automatic priority without asking what the question is trying to identify.
The common exam error is to treat one direction as superior or to confuse control existence with control effectiveness. Another is to call a risk residual before considering the controls that are actually operating. A strong answer links the strategic exposure to a concrete failure path, tests the control at that path, and feeds the observed residual risk back into enterprise decisions. That is what makes the combined view useful rather than a formal exercise in collecting labels.
Sources
- Operational risk | Basel Committee | BIS · bis.org · Retrieved
- ERM: Risk Identification and Assessment | GARP · garp.org · Retrieved
- FRM Study Materials | GARP · garp.org · Retrieved



