Insurance: Discrimination, Biases & Fairness

The development of machine learning over the last decade has been useful in many fields to facilitate decision-making, particularly in a context where data is abundant and available, but challenging for humans to manipulate.

In the financial sector, algorithms are commonly used by high frequency traders, asset managers or hedge funds to try to predict markets’ financial evolution.

The insurance sector is no different. Insurers are increasingly using fine-grained segmentation of their policyholders or future customers to classify them into homogeneous sub-groups in terms of risk and hence customise their contract rates according to the risks taken. However, the massive use of algorithms and Artificial Intelligence (AI) tools used by actuaries to segment policyholders questions the very principle on which insurance is based, namely risk mutualisation between all policyholders.

In this context, where digital technology is increasingly used, we are faced with several issues. How should the sector’s business model evolve if individualisation is extended at the expense of mutualisation? How can insurers carry out segmentation without applying discriminatory criteria?Which biases can be avoided in algorithm-making? What about equity criteria, a notion that is both abstract and deeply rooted in our society?

In this new issue of Opinions & Debates, Arthur Charpentier, a researcher specialised in issues related to the insurance sector and massive data, has carried out a comprehensive study in an attempt to answer the issues raised by the notions of discrimination, bias and equity in insurance. In addition to the very interesting debates raised by these topics, Arthur has carried out a comprehensive review of the existing academic literature, while providing mathematical demonstrations and explanations.
Enjoy reading!