Abstract
This chapter discusses a statistical modeling strategy based on extreme value theory to describe the behavior of data far in the tails of the distributions, with a particular emphasis on large claims in property and casualty insurance and mortality at oldest ages in life insurance. Large claims generally affect liability coverages and require a separate analysis. The reason for a separate analysis of small or moderate losses (also referred to as attritional ones in Solvency 2) and large losses is that no ED distribution seems to emerge as providing an acceptable fit to both small and large claims. As a first application of extreme value theory, the “peak over threshold” strategy is presented, allowing the actuary to determine an optimal threshold separating the two types of losses. As a second application, the force of mortality at the oldest ages is studied with the help of individual ages at death above 95 for extinct cohorts born in Belgium between 1886 and 1904. The analysis supports the existence of an upper limit to human lifetime for these cohorts. Therefore, assuming that the force of mortality becomes ultimately constant, that is, that the remaining lifetime distribution tends to the Negative Exponential one as the attained age grows, appears to be a conservative strategy for the closure of life tables when dealing with life annuities.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Even if some controversy arises about her age at death, we keep here Jeanne Calment as the world record of longevity.
References
Aarssen K, de Haan L (1994) On the maximal life span of humans. Math Popul Stud 4:259–281
Bader B, Yan J, Zhang X (2017) Automated selection of r for the r largest order statistics approach with adjustment for sequential testing. Stat Comput 27:1435–1451
Bader B, Yan J, Zhang X (2018) Automated threshold selection for extreme value analysis via ordered goodness-of-fit tests with adjustment for false discovery rate. Ann Appl Stat 12:310–329
Bakar SA, Hamzah NA, Maghsoudi M, Nadarajah S (2015) Modeling loss data using composite models. Insur Math Econ 61, 146–154
Balkema A, de Haan L (1974) Residual life time at great age. Ann Probab 2:792–804
Beirlant J, Goegebeur J, Teugels J, Segers J, Waal DD, Ferro C (2005) Statistics of extremes: theory and applications. Wiley, New York
Calderin-Ojeda E, Kwok CF (2016) Modeling claims data with composite Stoppa models. Scand Actuar J 2016:817–836
Cebrian AC, Denuit M, Lambert P (2003) Generalized Pareto fit to the society of actuaries’ large claims database. N Amn Actuar J 7:18–36
Cooray K, Ananda MMA (2005) Modeling actuarial data with a composite Lognormal-Pareto model. Scand Actuar J 2005:321–334
Danielsson J, Ergun L, de Haan L, de Vries C (2016) Tail index estimation: quantile driven threshold selection. Available at SSRN: https://ssrn.com/abstract=2717478 or http://dx.doi.org/10.2139/ssrn.2717478
Einmahl JJ, Einmahl JH, de Haan L (2019) Limits to human life span through extreme value theory. J Am Stat Assoc (in press)
Embrechts P, Kluppelberg C, Mikosch T (1997) Modelling extremal events for insurance and finance. Springer, Berlin
Gbari S, Poulain M, Dal L, Denuit M (2017a) Extreme value analysis of mortality at the oldest ages: a case study based on individual ages at death. N Amn Actuar J 21:397–416
Gbari S, Poulain M, Dal L, Denuit M (2017b) Generalised Pareto modeling of older ages mortality in Belgium using extreme value techniques. ISBA Discussion Paper, UC Louvain
Gumbel EJ (1937) La Durée Extrême de la Vie Humaine. Hermann, Paris
Hong L, Martin R (2018) Dirichlet process mixture models for insurance loss data. Scandinavian Actuarial Journal 2018:545–554
MacDonald A, Scarrott CJ, Lee D, Darlow B, Reale M, Russell G (2011) A flexible extreme value mixture model. Comput Stat Data Anal 55:2137–2157
MacNeil AJ, Frey R, Embrechts P (2015) Quantitative risk management. Princeton University Press, Concepts, Techniques and Tools
Nadarajah S, Bakar S (2013) CompLognormal: an R package for composite lognormal distributions. R J 5:98–104
Nadarajah S, Bakar SA (2014) New composite models for the Danish fire insurance data. Scand Actuar J 2014:180–187
Pickands J (1975) Statistical inference using extreme order statistics. Ann Stat 3:119–131
Pigeon M, Denuit M (2011) Composite Lognormal-Pareto model with random threshold. Scand Actuar J 2011:177–192
Reiss R-D, Thomas M (1997) Statistical analysis of extreme values, with applications to insurance, finance, hydrology and other fields. Birkhauser, Basel
Rootzen H, Zholud D (2017) Human life is unlimited-but short. Extremes 20:713–728
Scarrott C, MacDonald A (2012) A review of extreme value threshold estimation and uncertainty quantification. REVSTAT-Stat J 10:33–60
Scollnik DPM (2007) On composite Lognormal-Pareto models. Scand Actuar J 20–33
Scollnik DP, Sun C (2012) Modeling with Weibull-Pareto models. N Amn Actuar J 16:260–272
Tancredi A, Anderson C, O’Hagan A (2006) Accounting for threshold uncertainty in extreme value estimation. Extremes 9:87–106
Watts K, Dupuis D, Jones B (2006) An extreme value analysis of advance age mortality data. N Amn Actuar J 10:162–178
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Denuit, M., Hainaut, D., Trufin, J. (2019). Extreme Value Models. In: Effective Statistical Learning Methods for Actuaries I. Springer Actuarial(). Springer, Cham. https://doi.org/10.1007/978-3-030-25820-7_9
Download citation
DOI: https://doi.org/10.1007/978-3-030-25820-7_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-25819-1
Online ISBN: 978-3-030-25820-7
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)