Group & Segment Risk Distributions

The individual risk residual distributions are the foundations for building a risk distribution for a segment of the organisation's business. This segment could be for a product group, a customer segment, a department, a subsidiary or for a Group as a whole.

A target segment of the Group might have several risks attached to it. Some of these clearly belong to the target segment and to it alone. Others may attach to part of the entity that affects the target segment - for example a risk that attaches to the Group's legal processes would impact on all the Group's segments that rely on legal processes. The proportion of this overhead risk that should attach to the target segment can be estimated directly or can be calculated by reference to some relevant statistic, such as for instance in this case the share of total legal time used.

Whereas for individual risks we have derived risk distribution curves, for groupings of risks we need to construct a segment risk distribution based on the underlying individual operational risk distributions.

The segment expected loss might be simply achieved by adding the expected losses of all the underlying distributions (weighted by the appropriate share where a risk covers more than the business in question). But the process is more complicated when it comes to estimating segment unexpected losses to high confidence levels.

For example if a segment has two underlying risks, A and B, and A and B's unexpected loss to a 1 in 100 year or 99% confidence level were $10m and $20m respectively, this does not mean that the 1 in 100 year confidence level unexpected loss for the segment is $30m. The likelihood of both extreme 1 in 100 year events occurring in the same year is generally much less likely than 1 in 100.

On the other hand the group's unexpected loss is more than $20m, the unexpected loss of the larger underlying risk. We need a way of aggregating the effects of individual risk distributions into a risk distribution for a segment that represents some part of the organisation that management wish to analyse. A Monte Carlo simulation answers this need.

To Monte Carlo simulations

Last updated:16/5/07

Jump to... Top | Risk Management Home | Austega Home