Indicators must buy decision time
An early-warning indicator is valuable only if it changes the timing or quality of action. A metric that turns red after liquidity has already left may describe the crisis accurately but does not help manage it.
Design should begin with the institution's vulnerabilities and available actions. Indicators should then be selected for their ability to reveal deterioration before a key funding, collateral or operational constraint is reached.
Use a balanced indicator set
Firm-specific indicators may include deposit flows, concentration, rollover, collateral utilisation, intraday positions and contingent outflows. Market indicators may include spreads, equity moves and funding conditions. Operational indicators should cover payment disruption and data quality because a liquidity event can be amplified by an inability to see or move cash.
No single metric should dominate without context. A composite view can help, but its construction must not conceal a severe signal through averaging.
Calibrate thresholds to actions
Thresholds should correspond to defined escalation and management actions. The calibration question is not simply what level is statistically unusual; it is how much time remains to execute an action before the exposure becomes difficult to reverse.
Different indicators may require different persistence rules. A sharp intraday outflow may demand immediate review, while a gradual deterioration in concentration may justify a staged response.
Test the operating process
Institutions should test whether data arrives on time, whether owners understand escalation and whether committees can act outside the normal reporting cycle. A tabletop exercise can expose gaps that dashboard design alone cannot reveal.
Post-event review should assess false positives, missed signals, action timeliness and data latency. This creates a learning system rather than a static traffic-light report.