Economic Capital Models (ECM) play an increasingly important role in the insurance industry.  Several companies have been using such models for business steering for quite a while and the use is now being extended to determine Solvency capital requirements under  several regulatory regimes (e.g. the Swiss Solvency Test and  Solvency II in Europe).

In this presentation we will discuss design criteria for a good internal model. Especially, we will advocate the separation of risk factors (modeling the external world) and exposures (the impact of the risk factors on the economic balance sheet) so that dependencies are modeled at the risk factor level and portfolio correlations are an output rather than an input to the model.  We will show in several real life examples how this can work in practice.

We will also touch on certain properties of the risk measures used to derive the capital requirements. Especially, we will elaborate on the fact that  - in contrast to VaR -   TailVaR not currency invariant which might lead to undesired phenomena in a multi currency model.

Finally, we will discuss the challenges related to the increasing requirements on model governance and  discuss features of a good validation and model change policy.