Category: Risk identification & assessment · Reviewed by Chrissie Anderson, Client Executive · Last reviewed
Quantitative risk assessment
Quantitative risk assessment (QRA) expresses risk in numerical terms — probability of occurrence per year, expected monetary loss, statistical confidence intervals — rather than in ordinal or qualitative categories.
Hallmarks
Inputs and outputs are continuous numbers.
Probability distributions, not point estimates, are used wherever uncertainty exists.
Outputs include central estimates and tail measures (e.g. 99.5th percentile loss).
Models can be validated against historical data and back-tested.
Methods
Frequency-severity actuarial models.
Monte Carlo simulation.
Catastrophe modelling (RMS, AIR / Verisk, CoreLogic).
Generalised linear models (GLMs) for pricing.
Bayesian inference where data is sparse.
When QRA is the right choice
The exposure is material and decisions hinge on precision.
Data exists to support distributional assumptions.
The result will be challenged by regulators, auditors or capital providers.
When QRA misleads
When data is so sparse that the model output simply reflects the analyst’s assumptions.
When false precision drives confidence beyond what the inputs justify.
When tail behaviour is extrapolated far beyond observed loss experience.
In these cases a semi-quantitative or qualitative approach, paired with sensitivity analysis, may be more honest.
References
IEC 31010:2019.
HSE (2001). Reducing Risks, Protecting People (R2P2).
