Paper: arXiv 2511.22272

Authors: Hansjoerg Albrecher, Jan Beirlant

Abstract

We provide a survey of how techniques developed for the modelling of extremes naturally matter in insurance, and how they need to and can be adapted for the insurance applications. Topics covered include truncation, tempering, censoring and regression techniques. The discussed techniques are illustrated on concrete data sets.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: This paper serves as a comprehensive survey and handbook chapter detailing Extreme Value Theory (EVT) applications tailored to the insurance and reinsurance sector. While the novelty is modest due to its survey nature, it combines solid mathematical derivations of heavy-tailed distributions with practical actuarial and empirical considerations.

Research Flowchart

  flowchart TD
    A[Goal: Adapt Extreme Value Theory EVT for Insurance Applications] --> B[Inputs: Insurance Loss Datasets with Heavy Tails]
    B --> C[EVT Adaptation Techniques: Truncation, Tempering, Censoring]
    C --> D[Model Extension: Integrate Covariates via Regression Techniques]
    D --> E[Outcomes: Precise Tail-Risk Quantification & Improved Capital Requirements]

Cited References (73)

  1. Parameter estimation for exponentially tempered power law distributions (2012)
  2. Alternative modelling and inference methods for claim size distributions (2020)
  3. An actuarial model based on the composite W eibull- P areto distribution (2006)
  4. Modeling with W eibull- P areto models (2012)
  5. New composite models for the D anish fire insurance data (2014)
  6. Modeling claims data with composite S toppa models (2016)
  7. Modeling and evaluating insurance losses via mixtures of E rlang distributions (2010)
  8. Maximum likelihood from incomplete data via the EM algorithm (1977)
and 65 more
  1. Fitting mixtures of E rlangs to censored and truncated data using the EM algorithm (2015)
  2. Modelling censored losses using splicing: A global fit strategy with mixed E rlang and extreme value distributions (2017)
  3. Reinventing P areto: Fits for both small and large losses (2013)
  4. Modeling loss data using composite models (2015)
  5. Composite L ognormal– P areto model with random threshold (2011)
  6. Modeling actuarial data with a composite L ognormal- P areto model (2005)
  7. Parameter estimation for the truncated P areto distribution (2006)
  8. Fitting tails affected by truncation (2017)
  9. Tail fitting for truncated and non-truncated P areto-type distributions (2016)
  10. Statistics of extremes under random censoring (2008)
  11. What are the actual costs of cyber risk events? (2019)
  12. Functional kernel estimation of the conditional extreme value index under random right censoring (2021)
  13. Multivariate fractional phase–type distributions (2020)
  14. Statistics of extremes (2015)
  15. Compound extremes in hydroclimatology: a review (2018)
  16. Weather and climate extreme events in a changing climate (2021)
  17. Generalized additive modelling of sample extremes (2005)
  18. Regression-type models for extremal dependence (2019)
  19. Trends in the extremes of environments associated with severe US thunderstorms (2021)
  20. Extreme dependence of multivariate catastrophic losses (2006)
  21. EXTREMES ON RIVER NETWORKS (2015)
  22. Sparse structures for multivariate extremes (2021)
  23. Multivariate matrix M ittag– L effler distributions (2021)
  24. Local robust estimation of P areto-type tails with random right censoring (2021)
  25. New estimators of the extreme value index under random right censoring, for heavy tailed distributions (2014)
  26. Estimation of the extreme value index in a censorship framework: Asymptotic and finite sample behavior (2019)
  27. Loss models: from data to decisions (2012)
  28. Estimating the conditional extreme value index under random right-censoring (2016)
  29. On the study of extremes with dependent random right-censoring (2019)
  30. Penalized bias reduction in extreme value estimation for censored P areto-type data, and long-tailed insurance applications (2018)
  31. Censored extreme value estimation (2023) — link
  32. Reinsurance (2010)
  33. Trimmed extreme value estimators for censored heavy-tailed data (2021)
  34. Robust estimation of the P ickands dependence function under random right censoring (2019)
  35. Extreme value estimation of the conditional risk premium in reinsurance (2021)
  36. Bias reduced tail estimation for censored P areto type distributions (2016)
  37. Risk and insurability of storm damages to residential buildings in A ustria (2012)
  38. Bayesian modeling of insurance claims for hail damage (2023) — link
  39. An extreme value model for US hail size (2017)
  40. Human influence on tropical cyclone intensity (2016)
  41. Influence of the E l N i ~n o/ S outhern Oscillation on tornado and hail frequency in the U nited S tates (2015)
  42. Index for predicting insurance claims from wind storms with an application in France (2015)
  43. Comparison of storm damage functions and their performance (2015)
  44. Informed censoring: The parametric combination of data and expert information (2024)
  45. Expert K aplan– M eier estimation (2024)
  46. Finite-sample performance of the T -and W -estimators for the P areto tail index under data truncation and censoring (2023)
  47. Heavy-tailed phase-type distributions: a unified approach (2022)
  48. Extreme severity modeling using a GLM - GPD combination: application to an excess of loss reinsurance treaty (2023)
  49. Loss modeling with many-parameter distributions (2024)
  50. Modelling extreme claims via composite models and threshold selection methods (2020)
  51. Focussed selection of the claim severity distribution (2019)
  52. Claims Modelling with Three-Component Composite Models (2023)
  53. Soft splicing model: bridging the gap between composite model and finite mixture model (2024)
  54. Modeling insurance claims via a mixture exponential model combined with peaks-over-threshold approach (2012)
  55. On composite lognormal- P areto models (2007)
  56. Multivariate generalized P areto distributions (2006)
  57. Peaks over thresholds modeling with multivariate generalized P areto distributions (2019)
  58. Statistics of extremes: theory and applications (2004)
  59. Estimation of bivariate and marginal distributions with censored data (2003)
  60. Premium rating without losses: How to estimate the loss frequency of loss-free risks (2022)
  61. Spatial Dependence Modeling of Flood Risk Using Max-Stable Processes: The Example of A ustria (2020)
  62. On flood risk pooling in E urope (2017)
  63. Tail fitting for truncated and non-truncated P areto-type distributions (2016)
  64. Conditional tail moment and reinsurance premium estimation under random right censoring (2024)
  65. Estimation of parameters and Large Quantiles Based on the $k$ Largest Observations (1978)

0 of 73 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.