THIS WEEK'S RMN
MARKETS · The Wealthiest Americans Now Carry the Most Flood Exposure. Models Don't Reflect That.
New Census Bureau research confirms what private flood insurers have been betting on: flood exposure is concentrating among the wealthiest Americans, and the market is moving to follow it. — Read →
MARKETS · California's Wildfire Fund Was Mispriced And Now It's Shopping for a Replacement
The state's energy industry is searching for a new financial backstop to absorb wildfire losses that have already outrun every model built to contain them. — Read →
MODELS · OpenAI Searches For Biological and Chemical Risk Model Pro as AI “Uplift” Evidence Mounts
OpenAI is recruiting a senior biosecurity policy specialist to govern how its frontier models handle biological and chemical risk requests — a hire that arrives as AI uplift of weapons-relevant knowledge is no longer a theoretical concern. — Read →
MARKETS · Texas Heads Into Hurricane Season With Funding, Model Questions Hanging and Beryl Losses Still Growing
The Texas residual market faces important decisions following legislation that restructured its funding floor — with a model advisor RFP added to the mix.— Read →
A report published Monday by Imperial College London epidemiological modelers estimates that the Bundibugyo virus disease outbreak in the Democratic Republic of the Congo may already involve 400 to 800 cases against 336 officially suspected as of May 16.
The finding arrived one day after World Health Organization Director-General Tedros Adhanom Ghebreyesus declared a public health emergency of international concern. The outbreak has since produced a confirmed U.S. citizen case, Level 4 travel advisories for DRC, Uganda, and South Sudan, and a U.S. commitment to fund up to 50 emergency response clinics in affected regions.
The current Ebola crisis comes more than a decade following the last major global outbreak of the disease, and six years following the COVID-19 pandemic.
But despite those recent economic shocks caused by global communicable disease, and advances in predictive modeling, the insurance and private market system that exists to intermediate this kind of exposure is, by any current measure, unequal to it.
Current Ebola Modeling Shows Underestimation of Risk
The Imperial College team applied two independent methods to model the current Ebola outbreak.
The first used population movement data across seven DRC-Uganda border crossings combined with two confirmed exported cases in Kampala to back-calculate total DRC case load. The second worked backward from 88 reported suspected deaths, using historical Bundibugyo case fatality ratios and what the authors describe as "the distribution of time from symptom onset to death" — a gamma distribution with a mean of 11.37 days derived from the 2012 Isiro outbreak.
Both methods converged.
The authors note that "the convergence of findings from two independent methods strengthens confidence in the conclusion of substantial under-detection and the potential for wider transmission," while cautioning that estimates rest on assumptions about transmission concentration in Ituri and Nord Kivu, population movement patterns, and epidemiological parameters "derived from past Bundibugyo virus outbreaks."
What History Says Of Costs
The 2014 West Africa outbreak (which was more than 11 times larger than all previous Ebola outbreaks combined) provides the only comparable economic baseline. According to U.S. Centers for Disease Control and Prevention (CDC) data:
- $2.2 billion in GDP was lost in Guinea, Liberia, and Sierra Leone in 2015 alone
- Sierra Leone's private sector lost half its workforce
- 881 healthcare workers were infected; 513 died
- An estimated 10,600 additional deaths resulted from disrupted treatment of HIV, tuberculosis, and malaria
- More than $3.6 billion in total international response spending was required, with the U.S. government alone committing $2.37 billion
The 2014 outbreak also cost the U.S. more than $2 billion and over 10,000 jobs tied to exports, according to a CDC retrospective published in February. The current outbreak involves a strain with no licensed vaccine or therapeutic, a containment variable 2014 did not face at scale.
What Risk Markets Address (and Don't)
Standard property and business interruption insurance policies present the first wall against possible Ebola spread and resulting precautions.
As Aon wrote during the 2014 outbreak, "most property policies, including ISO, specific insurer forms and most manuscript policies, do not cover a loss resulting from a virus." Where specialty communicable disease extensions exist, they are generally available only on large hospitality or real estate portfolio risks, carry sublimits usually under $5 million, and run on aggregate rather than per-occurrence bases.
The civil authority question is equally restrictive.
A WHO PHEIC declaration is "unlikely to have direct triggering effect on contingent business interruption clauses with civil authority coverage," because the relevant triggering order "is usually domestic, not the WHO."
For healthcare providers, workers' compensation would likely respond to occupational infection, but at significant cost. Marsh noted in 2014 that "full isolation protocols, for example, can cost $1,000 per hour," with long-term complications potentially requiring "kidney dialysis and other treatments."
It should be noted that six Americans in the DRC are believed to have high-risk exposures, at least one symptomatic.
Key coverage gaps in the current outbreak:
- No active World Bank pandemic cat bond — the PEF was shut down after 2020 and never renewed
- No PHEIC-linked parametric trigger instrument currently in force for filovirus
- Virus and communicable disease exclusions are standard across most commercial property forms
- Civil authority BI coverage requires domestic government closure orders, not WHO declarations
- Specialty pandemic BI coverage, where it exists, typically sublimited below $5 million aggregate
In a 2020 interview with Risk Market News, Harvard Global Health Institute Senior Fellow Olga Jonas , who previously coordinated the World Bank's operational responses to avian flu and pandemics, called the Pandemic Emergency Financing Facility "a dubious experiment because of the high uncertainties in the modeling. It's an uninsurable risk."
The PEF was shut down after 2020 and never renewed. No successor instrument exists.
Jonas told RMN in 2020 that the fundamental problem was structural.
"Any entity should always source its risk financing from its reserves first and then from borrowing," and that the PEF's parametric trigger design excluded deaths in high-income countries while relying on death counts from low-income countries with the weakest surveillance capacity — the same countries now at the center of the Bundibugyo outbreak.
The modeling uncertainty that makes Bundibugyo so dangerous epidemiologically is precisely the uncertainty that made parametric trigger design for this peril unworkable from the start.
The State Department on May 19 committed to funding up to 50 emergency response clinics through UN OCHA pooled vehicles. Government money is again filling the gap. The 2014 response ultimately required $3.6 billion in sovereign and multilateral spending to contain an outbreak that, at its worst, was generating nearly 1,000 new cases per week.
One Property at a Time — Green Shield's Brian Bastian on Wildfire's Mispricing Problem
Catastrophe modelers and reinsurance underwriters tend to think about wildfire risk the way cat models were built to handle it; at the portfolio level, one aggregate at a time. Brian Bastian thinks about it from the ground up, one property at a time, and that difference shapes everything about how Green Shield Risk Solutions approaches the market.
Bastian, who leads product at Green Shield's Property Guardian analytics platform, joined the Risky Science Podcast as the final guest in a series of conversations recorded at ClimateTech Connect in Washington this past April. The interview covers why wildfire has attracted a uniquely dense ecosystem of private analytics vendors, where the admitted market is failing, and what it would actually take to fix the garbage-in, garbage-out problem at the heart of wildfire cat modeling.
The core issue is that traditional catastrophe models default to assumptions that don't reflect ground conditions. A model that assumes zero defensible space around every structure isn't capturing what's actually there, and Bastian puts a number on it. Two hundred and fifty feet of defensible space versus none can reduce a model's estimated loss cost by 25 to 30 percent on a single property. Multiply that across hundreds of thousands of exposures and the mispricing compounds fast.
Green Shield's answer is mitigation-first underwriting. Its E&S MGA already requires mitigation as a condition of coverage and rates for it explicitly — something the admitted market wants to do but can't yet execute, constrained by regulatory timelines and the difficulty of converting mitigation credits into reinsurance pricing relief.
California's admitted carriers are retreating, the FAIR Plan is absorbing policies it wasn't designed to carry, and Colorado, Nevada, and New Mexico are following the same trajectory. The data and the analytics to support rational pricing are close but regulatory framework to let admitted carriers act on them is not.
👉 Listen to the full episode
The Risky Science Podcast is published by Risk Market News. Subscribe below or find us wherever you listen.
