Model Risk as Multiplicative Risk Factor

Insurance Companies may apply so-called internal models in order to calculate regulatory capital. In addition to the Solvency Capital Requirement inherent model risks have to be determined. Based on a real-life example the paper shows how empirical information from the validation process may be used to build up expert judgments about model risk in terms of stochastic distributions. This raises a number of questions how the distribution of model risk is aggregated to that from which the SCR is derived from. Obviously, correlation-based approaches are difficult to justify. The paper gives a detailed description of approach which mixes the distribution what is mathematical equivalent to a multiplicative approach based on conditional independent distributions. The approach is discussed on a basis of a real-life example.

Find the Q&A here: Q&A on 'Modelling a Range of Risks'