Free Preview

This is a preview of some of our exclusive, member only content. If you enjoy this article, please consider becoming a member.

Liability insurers frequently use exclusions to help address emerging risks that are viewed as unpredictable, intractable and potentially very large.

Executive Summary

While demand for pandemic-related insurance is growing, broad pathogen exclusions on liability insurance policies are proliferating, creating financial risks for customers and lost opportunities for carriers that could potentially write this business. Noting that liability insurance promotes good behavior and risk management as well, executives of Praedicat and Metabiota suggest that putting together fully probabilistic models of pandemic frequency and severity with exposure and loss development information from liability models reflecting business and industrial footprints provides a framework for seizing opportunities rather than shying away from them.

It comes as no surprise, then, that the emergence of SARS-CoV-2, the virus causing COVID-19, has led many insurers to add exclusions meant to remove viruses from the scope of their insuring agreements.

Simultaneously, scientific evidence indicates that climate change and other factors will result in epidemics and pandemics becoming more frequent, driven both by old threats as well as new pathogens like SARS-CoV-2. With demand for pandemic-related insurance growing and insurance options decreasing, a marketwide problem becomes evident: a coverage gap. This gap represents a financial risk to customers and a lost opportunity to carriers that could potentially write this business.

And yet the liability exclusions have proliferated.

A COVID-19 liability exclusion might be reasonable due to considerations like adverse selection, but the exclusions that are emerging from COVID-19 on liability are much broader and far more diverse. There are broad pathogen exclusions, for example, which will sweep up foodborne illness. There are exclusions for specific infectious diseases, any infectious disease, pandemic exclusions and coronavirus exclusions. In addition to differing in wording and intent, the exclusions sometimes even differ at different points in a liability tower.

The result is an opaque and chaotic coverage environment that leads to litigation and coverage gaps rather than coverage certainty and simplicity. For example, a policyholder would be wise to ask which exclusions will apply in what future pandemics or infectious disease outbreaks. An insurer might begin to ask whether pandemic liability is truly insurable.

At Praedicat and Metabiota, we firmly believe that pandemic-related liability risks are insurable. Praedicat’s SARS-CoV-2 liability scenarios, which are designed to anticipate low-probability worst-case scenarios, project tail losses comparable to a modest-sized hurricane, for example. We also believe, like hurricanes, the risk can be modeled and quantified. Between Praedicat and Metabiota, we’ve developed the theoretical and modeling foundations required to build a fully probabilistic pandemic liability insurance model by incorporating probabilities from a pandemic model with exposure and loss development from a liability model.

Advances in computational epidemiology now allow for robust probabilistic models that tell us how often epidemics may occur and how they can unfold. These models are disease-specific, capturing distinctive features of each pathogen’s epidemiology, such as how it spreads (e.g., through the respiratory route, droplets, sexual transmission, etc.), duration of the infectious period, whether asymptomatic infection is possible, and numerous other factors, including how humans could respond and control measures, such as lockdowns, that might be introduced.

For zoonotic pathogens (spread between animals and people) like novel coronaviruses, the first step is “spark modeling,” which uses machine learning trained on historical data alongside rich geospatial information on ecological, climatic and demographic factors to estimate how often—and where—a virus jumps from animal to human populations. The model then simulates the spread of disease from person to person and place to place, incorporating drivers of disease transmission (such as commuting and long-range travel), health system attributes and attempts to control disease spread (through travel restrictions, social distancing or vaccines, if available). These simulations can be run at scale, even millions of times, with varying parameter combinations that allow modelers to quantify the frequency and severity of epidemics, from mild, frequent events to rare, utterly catastrophic ones.

We also anticipate that the pandemic liability risk could be packaged up in ways that are attractive to the capital markets as they seek to diversify their portfolios.
These models can be understood and used like conventional natural catastrophe models: to estimate the contours of an event of a given probability and its associated damage alongside what an “average” year would look like. And like models in the nat-cat space, pandemic models are built using an understanding of the scientific principles underlying the hazard rather than simply using past experience as the driver of probability. This also enables modelers to account for temporal trends in the frequency and severity of pandemics (or hurricanes, for that matter), bringing insights to the industry in the process.

Unlike most nat-cat models, however, pandemic models must account for human behavior. The damage that pandemics inflict depend on countless individual choices—whether to travel, work from home, wear a facemask and other decisions that vary hugely from community to community.

Pandemic liability also results from human choices at many levels. Given an underlying event from a pandemic model, the contours of potential liability become clear. For instance, viruses like SARS-CoV-2 that spread via aerosol suggest that those who contract the virus at work are at high risk of spreading the virus to their household cohabitants. This situation leads to a risk of wrongful death lawsuits leveled at employers who do not take sufficient precautions for their employees.

The detailed business and industry modeling that underlies a latent liability catastrophe model provides the framework to simulate these effects both at the level of individual companies and portfolios of insurance policies. This detailed modeling approach also yields information to assess the risks to a portfolio given differences in fatality rates and the likelihood of permanent disability among survivors, which may include COVID-19 “long-haulers.”

The business and industrial footprint for liability will also vary by characteristics of the virus. For example, a virus that affects a younger population would have a different liability footprint than one, like SARS-CoV-2, that preferentially harms older people. In the former case, nursing home liability would likely not be material, while liability may find its way to sports teams and schools more easily.

Armed with models of pandemic frequency and severity alongside models of how a pandemic can affect commercial liability, we turn to look at the kinds of insurance products that can facilitate covering these risks.

Coverage could require review of a pandemic preparedness plan from the buyer. Underwriting would thereby encourage good public health practices.
Companies concerned about future pandemics would almost certainly benefit from tailored coverage rather than relying on insuring agreements meant to cover liabilities arising from other kinds of events. This suggests excluding them on standard policy forms and instead writing pandemic-related liabilities on a named-peril basis. Coverage could be broad to account for the high risk of cross-line casualty clash including D&O, employment, environmental and general liability. Indemnity could be expanded, perhaps on a parametric pandemic-related trigger, to include covering the costs of good proactive public health practices such as sanitization or the hiring of contact tracers by large employers. Furthermore, coverage could require review of a pandemic preparedness plan from the buyer. Underwriting would thereby encourage good public health practices.

One lesson from COVID-19 is that employers and essential industries are critical parts of our public health system during pandemics. Liability insurance helps both promote good behavior and manage risk. It supported employers during COVID-19, and it will need to be available in coming pandemics. As a coverage gap appears to be rapidly emerging, now is an opportune time to develop specialized products that will support business in the next pandemic while managing the accumulation for the insurers that write it. With sufficient uptake of these policies, we also anticipate that the risk could be packaged up in ways that are attractive to the capital markets as they seek to diversify their portfolios.

Providing certainty of coverage for unpredictable but large, quantifiable events is precisely where the insurance industry provides the most value, and there’s every reason to provide that value for future pandemics.