Reverse stress testing
Starts with a defined adverse outcome and searches for plausible scenarios that could cause it.
How it works
Reverse stress testing starts with a specified failure such as a capital or liquidity threshold breach. It searches for plausible combinations of risk factors that produce the outcome. Constraints and a severity measure make the search meaningful and expose which drivers must occur together to reach failure.
Worked example
Suppose losses cannot exceed 15 before triggering a review, and a toy loss model is 2 times a spread shock plus 3 times a funding shock, each in defined units. With spread shock 3, a funding shock of 3 reaches 15. An optimiser can find competing combinations and rank them by a declared severity criterion.
Validation and limitations
State the failure boundary and document linkages across shocks, including timing and second-round effects. Validate the underlying loss model under extreme inputs and compare more than one plausible failure pathway. The severity ranking describes the chosen metric and does not in itself estimate occurrence probabilities.