The decline of the stress approach
In Europe, regulators expect companies to demonstrate their understanding of the full distribution of their risks and this can only be achieved through a simulation-based approach. This is a more theoretically correct approach to calculating EC, as it allows for all the interactions of each risk over many possible outcomes or simulations. A stress-based model, in contrast, only shows what losses would be at a particular point; it does not show the behavior of the tail of the loss distribution, where the real risk may lie. The survey results show that industry practice appears to be moving toward a simulation-based approach to EC computation.
This increased sophistication, however, comes at a price – a lack of transparency of the key capital drivers, unless supported by adequate management information, training, documentation and a deep understanding of the business from senior management.
The rise of curve fitting
In moving toward a simulation-based approach, insurers are favoring lite modeling techniques as opposed to full models. A lite model is a simplified representation of the full actuarial projection or asset model that can be used as a substitute in some circumstances. Examples of lite models include replicating portfolios and curve-fitting techniques.
Two or three years ago, there was no consensus about the preferred lite modeling technique. Early adopters implemented replicating portfolio approaches, but over the past 12 months curve fitting (and its variants) has emerged as the preferred approach. The increased take-up of curve fitting is probably driven by its ability to be applied to the entire risk taxonomy (both market and non-market risks), as opposed to a replicating portfolio, which typically focuses on only market risk.