Séminaire – Petit déjeuner – « Comportements et modèles en assurance » par Jean-Louis Rullière

Date / Heure
Date(s) - 08/11/2016
9 h 00 - 10 h 30

Auditorium Gilles Glicenstein

Ce 4ème rendez-vous des petits déjeuners thématiques de la chaire présentera deux contributions sur la relation entre comportements et modèles sur le marché de l’assurance:
Du côté de l’offre, l’assureur doit construire un modèle dont l’usage ne garantit pas pour autant la valeur et la fiabilité.
Du côté de la demande, la question porte davantage sur le dispositif plus ou moins révélateur qui permette de capter l’information sur les comportements des assurés.

Petit déjeuner d’échanges – 8 novembre 2016
Building confidence in economic capital models
Laure CABANTOUS – Faculty of management, Cass Business School, University of London

Abstract: While insurance and finance scholars have extensively studied the technical limitations of quantitative models, little scholarly attention has been given to what insurance professionals do, in practice, to deal with the whole range of issues, beyond model validation, arising from model development and use. As a result we know very little about how practitioners build and use models, and the activities they undertake in order to ensure that their models are “fit for purpose”. In this research, we address these questions by building on insights from the “practice turn” in organization studies and the sociological literature on models. Empirically, we focus on practices of model building and use among insurers, in the context of Solvency II and the development of economic capital models.  Our qualitative research is based on a set of semi-directed interviews with 22 practitioners from various insurance companies operating in the London Insurance Market (CROs, Chief Actuaries, capital modellers, consultants) and 9 experts at the Prudential Regulation Authority (PRA). Overall, in adopting a ‘sociological eye’ on economic capital models in the insurance industry, our research shows that the development and use of models is best conceptualized as a learning process whereby modellers (and their colleagues) progressively “domesticate” the model. Our research highlights the various types of practices of model development and use (such as investigating, simulating and measuring practices). It also shows how through these practices, modellers (and their colleagues) improve their understanding of the model itself, the word the model is meant to represent, and gain confidence in the model and its outputs.

The Persistence of Hypothetical Bias into Insurance Demand
Jean-Louis RULLIERE – ISFA & Laboratoire SAF, Université Lyon 1

Abstract: The primary motivation for studying the insurance demand is to gather information on consumer preferences structure. This goal is even more ambitious than it had to take into account the specific uncertainty context of insurance contracts. The very first line of research is to directly question consumers to elicit the preference structure under uncertainty. However these declarative data therefore assume that the consumer is confronted without bias into a hypothetical perspective. This approach is therefore robust to the exclusive condition that the reported data are not subject to the well known hypothetical bias.
The purpose of this contribution is to assess the extent and the persistence of this hypothetical bias inside surveys on the insurance demand. Here we proceed to the comparison with behavioral data collected through an experimental device based on monetary incentives.
The study of the hypothetical bias inside the insurance demand is more or less persistent, depending on the nature of incentives (under or not the incentive compatibility constraint), with or without framing effects and context effects; demand or supply effect; the extent of risk aversion; the presence of risk or ambiguity and whether the domain of loss or gain.

Slides de la conférence


Les réservations sont closes pour cet événement.

Filed under: Séminaires trimestriels