Replication data and code for "Dependence-Robust Spectral Capital Estimation from Noisy Dependent Marginal Data"

Published: 15 June 2026| Version 1 | DOI: 10.17632/pwnhrgrtgc.1
Contributors:
Yongjiang Li,

Description

This dataset contains processed data, parameter files, and Python scripts used to generate the simulation diagnostics and OpenFEMA National Flood Insurance Program empirical results reported in the manuscript "Dependence-Robust Spectral Capital Estimation from Noisy Dependent Marginal Data." The raw NFIP redacted claims and policies datasets are publicly available from FEMA/OpenFEMA. The included files aggregate selected state-level NFIP claims by date of loss over 2010--2024 and construct the monthly regional loss indices used in the empirical illustration. The package also includes simulation parameters, output tables, and figure data used to reproduce the reported numerical results. The reported NFIP loss index uses the claim-level coverage scale available in the public claims file and does not reconstruct active policy-in-force exposure. The stochastic simulation components use random seed 20260517.

Files

Steps to reproduce

To reproduce the results, download the full replication package and install the Python dependencies listed in the requirements file. Run the simulation scripts with random seed 20260517 to reproduce the regularization-path and effective-block diagnostics. Run the NFIP empirical script to construct the monthly regional loss-index data from the processed OpenFEMA inputs and to generate the reported dependence-stress capital tables and figures. The raw OpenFEMA NFIP claims and policies datasets are public; this package contains the processed monthly module data, metadata, parameter files, scripts, and numerical outputs used in the manuscript.

Institutions

Categories

Mathematics, Statistics

Licence