Why insurance regulators need to require sensitivity settings of internal models for their approval

Abstract

According to the Solvency II directive, insurers can use internal models for solvency assessment, but regulators must approve these models. Sensitivity analysis is a crucial part of the approval process. However, the directive lacks clarity on the required sensitivity analysis. Various techniques exist in literature to assess the impact of model assumptions on output, each revealing different aspects of model behaviour. In this letter, we suggest a minimum standard for regulators to ensure model quality. We propose complementary sensitivity settings for internal model development, governance, and approval. Implementing these settings enhances the explainability of approved models and their reliability.