Florent Lobligeois and Guillaume Magnan
Florent Lobligeois and Guillaume Magnan
Florent Lobligeois
and Guillaume Magnan
Natural hazards are a systemic risk for the insurance industry and the assesment of the stochastic losses covered by the underwritten policies is fundamental to ensure the insurance company solvability and the protection of its customers. AXA Group develops and operationnaly uses complex sophisticated modeling platorms to quantify the impact of natural events in terms of insured losses: from hazard physical-modeling to risk vulnerability and financial computations. Unfortunately, in practice, the loss assesments are strongly impacted by the model uncertainties (theoretically challenging and computationally expensive) and the quality of the input data.
A basic framework to quantify the impact of data quality on modeled loss results has been developed within AXA Group to produce pragmatic and operational loss best-estimates taking into account low-quality or unknown input data. This Data Quality Risk (DQR) framework which relies on four pillars, (i) availibility (ii) granularity (iii) completeness and (iv) reliability, is applied on a list of predefined risk drivers (risk characteristics, geocoding ...) and then loads the model results. The DQR methodology is operationally used for a financial solvability model and serves as an essential input to natural event loss assesment and (re)insurance protection purposes.