GEOGRAPHIC RATEMAKING WITH SPATIAL EMBEDDINGS
- 4 October 2021
- journal article
- research article
- Published by Cambridge University Press (CUP) in ASTIN Bulletin
- Vol. 52 (1), 1-31
- https://doi.org/10.1017/asb.2021.25
Abstract
Spatial data are a rich source of information for actuarial applications: knowledge of a risk’s location could improve an insurance company’s ratemaking, reserving or risk management processes. Relying on historical geolocated loss data is problematic for areas where it is limited or unavailable. In this paper, we construct spatial embeddings within a complex convolutional neural network representation model using external census data and use them as inputs to a simple predictive model. Compared to spatial interpolation models, our approach leads to smaller predictive bias and reduced variance in most situations. This method also enables us to generate rates in territories with no historical experience.Keywords
This publication has 25 references indexed in Scilit:
- About regression-kriging: From equations to case studiesComputers & Geosciences, 2007
- Spatial modelling of claim frequency and claim size in non-life insuranceScandinavian Actuarial Journal, 2007
- Non-life rate-making with Bayesian GAMsInsurance: Mathematics and Economics, 2004
- Tobler's First Law and Spatial AnalysisAnnals of the American Association of Geographers, 2004
- Generalisierte geoadditive Modelle zur SchadenanalyseBlätter der DGVFM, 2003
- Bayesian Premium Rating with Latent StructureScandinavian Actuarial Journal, 2002
- Geographic Premium Rating by Whittaker Spatial SmoothingASTIN Bulletin, 2001
- Premium Rating by Geographic Area Using Spatial ModelsASTIN Bulletin, 1994
- Use of Spline Functions for Premium Rating by Geographic AreaASTIN Bulletin, 1989
- A Computer Movie Simulating Urban Growth in the Detroit RegionEconomic Geography, 1970