Judge what features make Early Warning Indicator (EWI) measures sound.
The idea
An early warning indicator (EWI) signals rising liquidity vulnerability early enough for management to act, so soundness is judged by a few features. It must be forward-looking, moving before funding actually dries up, unlike a reported ratio that describes a past date. It must be measurable, with frequent reliable data and a threshold defined in advance. It must be relevant to the bank's own funding model, because a retail-deposit bank and a repo-funded dealer need different indicators. The set should combine internal signals, such as deposit run-off or collateral calls, with external ones, such as CDS spreads and market volatility. It should not all rest on one data source. Thresholds are calibrated and back-tested to balance missed signals against false alarms. The usual mistake is accepting a lagging or noisy measure as an EWI. The exam point is that a sound EWI is leading, measurable, relevant, diversified and calibrated.
A practice question
A regional bank funded mostly by insured retail deposits copies the EWI list of a large dealer bank funded by repo. Which feature of a sound EWI set is most at risk?
- Relevance to the bank's own funding model and vulnerabilities - correct
- Back-testing frequency, because the copied indicators were tested too rarelyWhy not: Back-testing concerns calibration; the problem here is that the indicators do not fit the bank's funding model, not how often they are tested.
- Independence from market data, because dealer indicators rely on market pricesWhy not: Market data are not inherently unsound; the weakness of a copied list is its fit to the business, not its data source.
- Compliance with regulatory minimum ratios, because dealers face different ratiosWhy not: Meeting minimums is not a feature of an EWI, and copying a list says nothing about ratio compliance.
A sound EWI set is relevant to the institution's own business and funding profile. A deposit-funded retail bank is exposed to deposit run-off and concentration, not repo haircuts, so a list copied from a dealer will miss its real vulnerabilities.
A harder one
A treasurer must add one indicator to the bank's EWI set. Which candidate best meets the features of a sound EWI?
- The bank's long-term credit rating, reviewed whenever a rating agency publishes a changeWhy not: Rating changes usually follow deterioration, so the indicator lags and gives little lead time.
- The bank's net income for the most recent financial yearWhy not: Annual and backward-looking: it is infrequent and fails the timeliness requirement.
- The number of risk committee meetings held in the quarterWhy not: A process count with no measurable link to emerging liquidity stress, so it has no predictive content.
- Weekly change in uninsured deposits, with amber and red thresholds set in advance - correct
A weekly, measurable deposit indicator with predefined thresholds is frequent, objective, forward-looking and relevant to funding. Ratings and annual results lag events, and a meeting count is unrelated to stress.
The hardest tier
Over 24 months a bank had 6 liquidity stress episodes. A candidate EWI signalled 8 times; 5 signals preceded a distinct episode and 3 were followed by nothing. Which pair gives the share of episodes detected and the share of signals that were false alarms?
- 62.5 percent of episodes detected; 37.5 percent of signals falseWhy not: Divides true signals by all signals (5/8), which is precision, not the share of episodes detected (5/6).
- 83.3 percent of episodes detected; 50.0 percent of signals falseWhy not: Divides false signals by the number of episodes (3/6) instead of by the number of signals (3/8).
- 83.3 percent of episodes detected; 37.5 percent of signals false - correct
- 16.7 percent of episodes detected; 37.5 percent of signals falseWhy not: Uses the missed share (1/6) as the detection rate; detection is 5/6.
Episodes detected = 5/6 = 83.3 percent (one episode was missed). False alarms = 3/8 = 37.5 percent of signals. Detection is measured against episodes, false alarms against signals, and sound calibration weighs both.
In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.
Start free →