Emmanuel Randle Research & practice

Independent technical work

Projects

Catastrophe-risk modeling projects by Emmanuel Randle covering earthquake damage, insurance, reinsurance, and parametric risk transfer.

Completed / v2.0.2

Seismic Correlation and Insurance Loss

A reproducible earthquake catastrophe-risk workflow that connects seismic sources to structural and nonstructural damage, insured portfolio loss, reinsurance, and parametric basis risk.

470 buildings in Seaside, Oregon
$384.24 million replacement value
2,000,000 simulated years / 13 notebooks

Read the full case study View source & results
Modeled gross insured aggregate and occurrence exceedance curves for independent and spatially correlated cases
Comparing portfolio tails while keeping the event catalog and financial assumptions fixed.

The question

How does spatial dependence in ground motion change the risk retained by an insurer, the reinsurance needed to transfer that risk, and the shortfalls left by parametric protection?

The model uses the same event catalog, portfolio, policy terms, and paired random streams across three dependence cases. This allows the comparison to focus on the spatial-dependence assumption.

What the project covers

  • USGS rupture-level annual rates and stochastic event simulation.
  • Paired ground-motion fields and structural and nonstructural damage.
  • Ground-up, gross insured, uninsured, ceded, and retained losses.
  • AAL, AEP, OEP, PML, and TVaR-based tail comparisons.
  • Occurrence and annual aggregate reinsurance sensitivity.
  • Parametric trigger calibration, held-out evaluation, and basis risk.
  • Reproducibility checks and paired bootstrap uncertainty.

This is a portfolio demonstration using synthetic exposure and financial terms. Results are conditional on the selected hazard, damage, and contract assumptions. The case study documents the limitations.