Fast catastrophe bond valuation with neural-network surrogates

Julian Sester and Huansang Xu

Papers from arXiv.org

Abstract: Catastrophe bonds are increasingly important risk-transfer securities, but structural pricing is too slow for real-time valuation, screening, and sensitivity analysis. We develop a neural-network surrogate for the pricing operator of a compound-Poisson catastrophe bond model. Training labels are generated by Monte Carlo simulation with importance sampling, so the network learns variance-reduced structural prices rather than sparse market quotes which are difficult to extrapolate reliably. Across Gamma and Lognormal severity specifications, the selected networks attain very small absolute approximation error on the stated training domain. After training, 1000 contracts are priced in about 0.03--0.04 seconds, compared with tens to hundreds of seconds for Monte Carlo with importance sampling and many hours for a partial integro-differential equation benchmark. The result is a fast in-domain structural valuation engine that also produces economically interpretable sensitivities to catastrophe intensity, attachment threshold, and interest rates.

Date: 2025-09, Revised 2026-07
New Economics Papers: this item is included in nep-big, nep-cmp and nep-rmg
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://arxiv.org/pdf/2509.25899 Latest version (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:arx:papers:2509.25899

Access Statistics for this paper

More papers in Papers from arXiv.org
Bibliographic data for series maintained by arXiv administrators ().