Optimizing reinsurance can save millions
Choosing a reinsurance program is one of the hardest decisions an insurer makes. Even small improvements in how a program is structured can have a large effect on profitability and capital.
Strategic questions first
Reinsurance optimization should be part of a broader company strategy. It starts with questions such as:
- Which lines of business will we grow next year, and which are the most profitable?
- How will our claims distribution change, given the sales plan?
- How much capital do we need to hold, and how much can reinsurance replace?
- What is our risk appetite? Are we aiming for maximum expected profit, or greater security?
Technical questions
Once the strategy is clear, the program still has to be designed. If an insurer combines proportional and non-proportional cover, where should each layer start and end? For example, should a quota share apply up to $1 million, with excess of loss cover above that? And why $1 million rather than another figure?
Layer limits work best when they come from analysis rather than habit or round numbers.
Building the models
Actuaries model the expected number and size of claims, using past data on policies, claim frequency and claim severity, together with the future sales plan. These models produce probability distributions of outcomes for each possible program.
Choosing reinsurers and structures
Each class of business can be covered by several layers, perhaps a quota share with one reinsurer, a surplus treaty with another and excess of loss cover with a third. Optimization compares the outcomes of feasible combinations. Reinsurers' financial strength and their willingness to work as long-term partners matter as well as price.
What "optimal" means
The optimal program depends on the insurer's goals. It might deliver the highest expected profit while keeping the probability of ruin acceptably low. It might buy the most protection for a given ceded premium, or a given level of protection for the least ceded premium. Dedicated modeling software can help by testing many combinations quickly.