Examine the design choices in a liquidity stress test, including its scope, scenario building, assumptions, outputs, governance and how it fits with other risk models.
The idea
A liquidity stress test is a set of design choices, and each one changes the answer. Scope fixes which legal entities, currencies, products and time horizons are covered, so that restricted liquidity in one entity is not hidden by a consolidated total. Scenario building usually combines an idiosyncratic shock, a market-wide shock and a combined case, with the combined case most severe because market-based mitigants then fail together. Assumptions cover run-off and rollover rates by counterparty type, drawdowns on commitments, collateral haircuts and management actions, and should be conservative and documented. Outputs are cash flow profiles, surplus or shortfall by horizon and the survival horizon, compared with limits. Governance means senior management and board approval, independent challenge and defined escalation to the contingency funding plan. The test is scenario based and cash based, unlike VaR, but it should use scenarios consistent with the market and credit risk models. The usual failure is a single, unreviewed run-off assumption.
A practice question
In designing the scope of a liquidity stress test for an international banking group, which choice is most appropriate?
- Cover only the consolidated group, since cash can be moved freely among entitiesWhy not: Free movement of cash is the assumption that hides restricted liquidity; scope should look at entities too.
- Cover the consolidated group and each material legal entity and currency, over several time horizons - correct
- Cover only the parent's local-currency balance sheet, because it is the largestWhy not: Excludes material foreign entities and currencies where mismatches and trapped cash can arise.
- Cover only a single overnight horizon, which is the most severeWhy not: Liquidity risk plays out over days, weeks and months; a single horizon misses the build-up of outflows.
Scope covers entities, currencies, products and horizons. Consolidated and entity-level views both matter because transfer barriers and currency mismatches can make group totals misleading.
A harder one
A risk manager runs three scenarios: an idiosyncratic shock, a market-wide shock and a combined shock. Why is the combined scenario normally the most severe for the firm's liquidity?
- Because the firm's own loss of confidence arrives just as asset sales and repo markets also deteriorate - correct
- Because market-wide shocks affect only the firm's assets and never its funding, so the effects add up simplyWhy not: Market-wide shocks hit both funding and asset values, which is why they matter.
- Because idiosyncratic shocks have no effect on funding costsWhy not: Idiosyncratic shocks, such as a downgrade, cause run-off and higher funding costs.
- Because the combined scenario is a simple average of the other twoWhy not: It is not an average; the interaction of the shocks makes it worse than either alone.
In the combined scenario the firm loses funding while the markets it would use to sell or repo assets are also stressed, so haircuts rise and mitigation weakens. It is therefore normally the most severe.
The hardest tier
A bank has a liquidity buffer of USD 1,200 million after haircuts. In its base stress, net outflows in weeks 1 to 4 are USD 400, 350, 250 and 150 million. Management then raises the run-off assumption on USD 1,000 million of uninsured deposits by 10 percentage points, all in week 1. Assuming outflows are spread evenly within each week, the survival horizon, in days, is closest to:
- 21 daysWhy not: Counts only the three full weeks that the buffer covers and ignores the part of week 4 it still funds.
- 28 daysWhy not: Counts week 4 as fully survived although cumulative outflows of 1,250 exceed the buffer of 1,200.
- 26 days - correct
- 30 daysWhy not: Leaves the base result unchanged and ignores the extra week 1 outflow of 100.
Revised cumulative outflows are 500, 850, 1,100 and 1,250. The buffer of 1,200 runs out in week 4: 100 / 150 x 7 = 4.7 days, so survival = 21 + 4.7 = 25.7, about 26 days. The base case would survive past 28 days.
In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.
Start free →