What if automation could help efficiently and accurately estimate risk reserves?
Prime Meridian Direct
Prime Meridian Direct (PMD) wanted to automate its Outstanding Claims Reserve (OCR) process through custom and traditional modeling methods.
First, we developed a custom OCR model to incorporate specific claim process information and line-level claim information to ensure accurate reserve setting.
Second, we implemented and automated various traditional actuarial methods, including the Basic Chain Ladder (BCL) and Bornheutter Ferguson (BF). These custom, traditional methods enabled efficient real-time reserve estimation. Previously, PMD received estimates three months after the fact from their underwriter’s actuarial team.
We also built stochastic risk-reserving models tailored to PMD’s business. These enabled efficient and accurate confidence estimation of risk reserves by development period. The advantage of customising these models rather than buying off-the-shelf software was that the model assumptions could be tailored to PMD’s business.
Also, custom-built models make scenario modeling capabilities and what-if testing now possible. PMD had the freedom to explore various cases and scenarios to ensure that a) the business processes were healthy and b) the reserves were set accurately every month.
The stochastic reserving models, traditional reserving models, and custom reserving models made use of various mathematical techniques, including:
All models are automated and set up to run on serverless Azure functions on a schedule as pre-defined by PMD