Skip to content

Location and scale in normal loss risk

Mean loss shifts both tail summaries equally; dispersion uses a different coefficient for each measure.

On this page

Read a location shift

Holding dispersion and confidence fixed, adding 5 loss units to μL adds 5 to both VaR and ES. The distribution shifts horizontally without changing its spread. A negative mean is a gain location under this loss convention.

Read a dispersion change

At 95% confidence, adding 5 units to σL adds about 8.224268 to VaR and 10.313564 to ES. The difference arises from the quantile and tail-mean multipliers. It is not a change in confidence probability.

Keep both parameters in the same loss unit

Parameter effects at fixed confidence
ChangeVaR effectES effect
ΔμLΔμLΔμL
ΔσLz ΔσLφ(z)/(1−c) ΔσL

Avoid turning sensitivity into validation

These are sensitivities of the declared normal model. They do not prove that changing a real asset changes only one parameter or leaves the model shape stable. Keep the assumption record with the comparison.

Further references