Linked to the funded project “Fairness of predictive models: an application to insurance markets,” the SCOR Foundation hosted a webinar on June 18, 2025, titled “AI and Econometrics.”

The webinar was delivered by the University of Quebec’s Arthur Charpentier, who leads the project

Arthur Charpentier’s presentation explored the relationship between econometrics and machine learning by contrasting their foundational philosophies and highlighting areas of complementarity. He began with a review of least‐squares regression, illustrating how econometrics situates this method within a probabilistic framework that quantifies uncertainty and permits formal hypothesis testing. From there, he introduced penalized regression techniques to demonstrate the role of optimization in achieving both shrinkage and model parsimony — bridging naturally to logistic regression for binary outcomes. Throughout, he emphasized that many machine‐learning algorithms (e.g., support vector machines, feed‐forward neural networks) share similar loss‐minimization formulations but often lack an explicit stochastic foundation. Simulations were used to clarify practical distinctions and commonalities between these approaches. In the second half of the webinar, Charpentier addressed two unifying concepts: first, generalization error and cross‐validation as tools for assessing out‐of‐sample performance; and second, the calibration of predictive models, which raises deeper questions about epistemological criteria for model adequacy under uncertainty and risk.

Go to the project page

Arthur Charpentier, Fellow of the French Institute of Actuaries, is a full professor at UQAM, Montreal, Canada, and Université de Rennes in France. He is a member of the editorial boards of the Journal of Risk and Insurance, the ASTIN Bulletin, and Risks. He has edited Computational Actuarial Science with R (CRC), and more recently wrote Insurance, Biases, Discrimination and Fairness (Springer). He is a Louis Bachelier Fellow, and his recent work is about climate change and predictive modeling insurance. He leads the “Fairness of predictive models: an application to insurance markets” research project, which is funded by the SCOR Foundation (2023-2026).

Read the news about the SCOR Foundation Workshop - May 15, 2025