Incorporating reserve value added (RVA) into reinsurance decision making for long-tail lines is a step in the right direction. However, it is not the full story, as the decision is still typically made in the context of a single accident year and usually for a single line of business in isolation. The cycle correlations clearly show that this is sub-optimal. We are encouraging our clients a step further along the sophistication and hence simplicity/complexity spectrum.
The future involves considering reinsurance not only in the context of one-year but in the context of all past-years and how, across all lines, they could develop. Unfortunately, we have not seen many capital models that accomplish this in a coherent way. There has not been much attention focused on true multi-year modeling that captures the potential for correlation across years - perhaps the distraction of the Solvency II one-year view of risk is to blame here. Capital modeling actuaries are challenged here. Parameterizing this correlation will be highly subjective and relies on limited data available from short time periods.
The complexity of the issue should not be intimidating. This is where scenario testing really comes into its own. By overlaying different reinsurance strategies on a variety of cycle and recognition scenarios there can be serious improvement in reinsurance decision making. This process helps to select a strategy that works well across the cycle and not solely for a single year in isolation. Hypothetical cycles themselves are actually a very neat way of generating the correlation. History can be replayed as it occurred, replayed with some tweaks or extreme scenarios from as yet emerging risks can be considered. Because the years and product lines are connected, reinsurance decision making across those years and lines needs to also be connected.