Andrew Cox, Head of Advisory - EMEA
Many insurance executives will have seen their companies incur high costs preparing for Solvency II, in particular for actuarial staff resources that build internal models. But are they seeing a good return on this investment?
The Solvency II regime places strict statistical requirements on internal models, leading to models that have become extremely detailed and complicated. Because of this, models are often slow to run, lack flexibility and are difficult to explain to non-experts.
But Solvency II also requires that these models be used in making business decisions. This is easier with models that are fast, flexible and concentrate on the issues at hand - not those that we have had to-date. How do we square this apparent circle?
The key insight is that we should be talking about an internal modeling framework.
At its heart is the current enterprise-wide model. This will be used for those decisions it is best suited to: those that require a view of the whole business, reflect the status quo rather than possible future options, and where the timing is known in advance - decisions made in the preparation of regular regulatory returns are the best examples. We call these periodic decisions.
On the other hand some decisions are dynamic. These are decisions that look at multiple options, are often time pressured and tend to focus on a single area of the business. Here we believe the optimal route is to have a suite of supporting models. In general there will be a central model with overall organizational scope used for the periodic decisions, with a series of supporting models used for dynamic decision-making. These are models that are consistent with the central model but are simplified to only what matters for the task at hand. By doing so run times are much reduced and potentially confusing, unnecessary information is excluded.
Like many good ideas, this is not really new. Investment risk is an important factor in capital models, but no one uses his capital models to decide on which corporate bonds to buy. The detailed decisions are based on investment models that are consistent with the central modeling. We are simply suggesting extending this approach more widely in the capital modeling framework. The supporting models need to be consistent with the central model, but this does not mean they need to give exactly the same answers. Indeed, a divergence in view can provide genuine insight into the validation process.