<< Click to Display Table of Contents >> Navigation:  7EVENTS > 6.12 Loss Prediction Module > 6.12.1 Background

The module was built in collaboration with staff in the Maths department at Sussex University. Inputs to the module are Actual Losses and near misses for X previous years; Estimated Losses for High Impact risks where no actual losses have been sustained in the previous x years and manually derived Management data.

 

As we have seen, for each Business Unit the CAREweb™ system is populated with “best estimate” data about the IMPACT of each risk in the Business Unit and its PROBABILITY of occurrence given no control; these categorizations can range from Very High to Very Low. These estimates are based upon:

▪The experience of the modellers,

▪Knowledge of incidents that have occurred in the enterprise in the past,

▪Knowledge of external events.

 

For each Business Unit CAREweb™ allows for the capture, in real time, of each occurrence of a risk or a “near miss”; these occurrences are referred to as EVENTS (see above).

Events are grouped by year of occurrence thus providing a record of the number of times an individual risk occurred in a year and its total cost to the enterprise (a year for the purposes of this module is the period 1/1 – 31/12).

This event data is modelled, using Mathematical Simulation to produce an estimated range of risk impact values for the following year, to a predetermined confidence interval; the values are:

▪Upper bound,

▪Lower bound, and

▪Average.

The module can produce these estimates:

▪For the entire organisation,

▪For a particular level in the organisation hierarchy

▪For a Business Line,

▪For a Group of Entities

▪For an Entity,

▪For a risk within the Entity,

▪For groups of risks within an Entity,

 

For cases where CAREweb™ has a potentially high impact event, with a low probability and there is no event data, the prediction is based upon one “virtual event” having happened in the period being used for the simulation (i.e. 3, 4 or 5 years); this “virtual event” has a value equal to the average of the High range for the Business Unit in question in CAREweb™. The value is assumed to have occurred in the most recent year of history.

For incidents classified as “near misses” the system will either use the average of other actual losses recorded for the risk in question or, where no actual losses have been recorded, the average of the impact size taken for the particular risk taken from the Business Unit details.

To allow for changes in the enterprises’ control environment over the period during which the event data has been captured, a “Year Weighting” system is in place. To allow for modelling potential changes in the control environment in the next year a Management Control Factor can be used. This Management Control Factor will allow the system to model:

▪What is likely to happen if controls deteriorate,

▪The likely losses if controls are improved – thus allowing a “cost benefit” analysis.