European Actuarial Journal

, Volume 9, Issue 2, pp 603–605 | Cite as

Discussion on ‘Multivariate modelling of multiple guarantees in motor insurance of a household’ (Pechon et al.)

  • Stanislas RothEmail author
Discussion on recent papers

The main challenge for pricing actuaries working in a non-life insurance company is to avoid adverse selection through the most possible efficient pricing, i.e. a pricing which maximizes the potential profit in a usually very competitive environment. The standard approach to do so consists in developing several multivariate risk models predicting the frequency and the severity of each of the perils covered by the insurance package to be subscribed by the policyholder, considering the individual’s characteristics and his/her claim experience over a fixed time horizon.

This product-based approach is probably not any longer the most efficient one in the context of open and large data we are now in, because:
  • from the risk assessment point of view, it does not leverage all the information already available about the client: the risk observed on the other products the client has already underwritten, or the social environment he interferes with, whereas machine learning techniques used in...


Copyright information

© EAJ Association 2019

Authors and Affiliations

  1. 1.AXAParisFrance

Personalised recommendations