Pandemic Risk Modeling: Lessons Learned
How the actuarial profession has adapted its approach to pandemic risk modeling after recent experience.
Modeling Pandemic Risk
Pandemic risk poses unique challenges for actuarial modeling because pandemics are rare, highly correlated events that affect multiple lines of business simultaneously. Unlike most insured perils, a pandemic is not confined to a geographic region or a single line of coverage. It can increase mortality claims in life insurance, drive up healthcare utilization and costs in health insurance, generate workers' compensation claims, trigger business interruption losses, and cause event cancellation payouts, all at the same time. This cross-line correlation makes pandemic risk fundamentally different from most risks that actuaries model.
The rarity of pandemics means that historical data provides limited guidance for estimating frequency and severity. Major pandemics with significant mortality impact have occurred roughly two to four times per century over the past several hundred years, but each event has unique characteristics in terms of pathogen type, transmission dynamics, population vulnerability, and public health response. This makes traditional actuarial experience-based methods insufficient on their own, requiring supplementation with epidemiological models, scenario analysis, and expert judgment.
Epidemiological Models for Actuaries
Epidemiological compartmental models provide the foundation for projecting disease spread. The SIR model divides the population into three compartments: Susceptible (S), Infected (I), and Recovered (R). The model's dynamics are governed by the transmission rate (beta) and the recovery rate (gamma). The basic reproduction number, R0, equals beta divided by gamma and represents the average number of secondary infections produced by a single infected individual in a fully susceptible population. When R0 exceeds 1, the disease can spread exponentially in the early phase.
The SEIR model adds an Exposed (E) compartment for individuals who have been infected but are not yet infectious, capturing the incubation period. More sophisticated models add compartments for hospitalized individuals, deaths, vaccinated individuals, and different age groups. These extensions allow more realistic projections but require additional parameters that may be uncertain early in a pandemic.
For actuarial applications, the key outputs from epidemiological models include the total infection rate (what percentage of the population ultimately becomes infected), the peak infection rate (which determines healthcare system strain), the infection fatality rate (the proportion of infected individuals who die), and the timeline of the epidemic curve. Actuaries translate these epidemiological outputs into insurance-relevant metrics: excess mortality rates, excess morbidity rates, healthcare utilization changes, and economic disruption estimates.
Insurance Line Impacts
Life insurance is directly affected by pandemic mortality. The severity of impact depends on the age distribution of mortality (a pandemic that primarily kills working-age adults has a larger life insurance impact than one affecting the elderly, because life insurance in-force concentrates in working-age populations), the total excess mortality rate, and the duration of elevated mortality. Actuaries model pandemic mortality as a shock to the baseline mortality table, typically expressed as an additive or multiplicative adjustment applied over the pandemic's duration.
Health insurance faces increased claims from both direct treatment of the disease and delayed or deferred care for other conditions. Hospital admissions, intensive care utilization, and pharmaceutical costs all surge during the acute phase. The behavioral response to a pandemic (deferred elective procedures, reduced preventive care visits, increased telehealth utilization) can temporarily reduce some categories of claims while creating a pent-up demand effect that increases claims in subsequent periods once the acute phase subsides.
Workers' compensation exposure depends on whether the disease is classified as an occupational illness and which industries are affected. Healthcare workers, first responders, and essential workers in close-contact settings face elevated exposure. Business interruption insurance is affected when government-mandated closures or supply chain disruptions prevent businesses from operating. Event cancellation and travel insurance policies face surges in claims when large gatherings are prohibited or travel restrictions are imposed.
Scenario Analysis Approach
Given the limitations of probabilistic modeling for rare, complex events like pandemics, scenario analysis is particularly valuable. A well-designed set of pandemic scenarios spans a range of severity levels and explores different transmission characteristics, public health responses, and economic impacts. A typical scenario set might include a mild pandemic (similar in severity to the 2009 H1N1 influenza), a moderate pandemic (similar to the 1957 or 1968 influenza pandemics), and a severe pandemic (approaching 1918 influenza severity).
Each scenario should specify key parameters including the pathogen's transmission rate and infection fatality rate, the age distribution of mortality, the duration and shape of the epidemic curve, the public health response (social distancing measures, vaccination timeline), and the economic impact (GDP contraction, financial market effects, unemployment). Actuaries then project insurance impacts under each scenario, estimating claims, reserves, and capital requirements. The correlation between insurance losses and investment losses is particularly important, as a severe pandemic may simultaneously increase claims and reduce asset values.
Lessons from Recent Experience
Recent pandemic experience highlighted several areas where actuarial models needed improvement. Business interruption coverage exclusions proved ambiguous in many policy forms, leading to extensive litigation over whether virus-related closures constituted "direct physical loss or damage" to property. This experience prompted insurers to clarify policy language and develop explicit pandemic exclusions or, alternatively, specific pandemic coverage endorsements with defined terms and limits.
Mortality models underestimated the potential severity of a novel pathogen, particularly the possibility of high mortality among specific demographic groups. The correlation between mortality and morbidity impacts across lines of business was stronger than many enterprise risk models assumed. The economic and behavioral impacts of public health interventions, including lockdowns, remote work transitions, and changes in consumer behavior, were not captured in existing models and had cascading effects across multiple insurance lines.
The profession has responded by developing more sophisticated pandemic scenarios, improving policy language clarity, building better correlation structures into enterprise risk models, and creating dedicated pandemic risk reserves. Many companies have added pandemic-specific stress tests to their ORSA processes and capital planning frameworks.
Building Better Pandemic Preparedness
Actuaries can strengthen their organization's pandemic preparedness through several concrete actions. First, develop and maintain a library of pandemic scenarios with clearly defined assumptions and regularly updated loss estimates. Second, review policy language across all lines of business to identify potential coverage ambiguities that could create unexpected liabilities during a pandemic. Third, build correlation structures into enterprise risk models that explicitly capture the cross-line dependencies triggered by a pandemic event. Fourth, establish data collection protocols that can be activated quickly when a pandemic begins, ensuring that emerging experience data is captured and analyzed in near-real-time.
Monitoring emerging infectious disease risks is also important. Actuaries should stay informed about developments in epidemiology, public health surveillance, and pandemic preparedness policy. Understanding the characteristics of potential pandemic pathogens (novel influenza strains, coronaviruses, and other respiratory viruses with pandemic potential) helps actuaries develop more realistic scenarios and respond more quickly when a new threat emerges.
Study Relevance
Pandemic risk modeling appears in exam syllabi for both the SOA and CAS, particularly in enterprise risk management, health insurance, and life insurance contexts. Candidates should understand the basic structure of SIR and SEIR models, the concept of R0, the multi-line nature of pandemic insurance impacts, and the role of scenario analysis in managing low-frequency, high-severity risks. The ability to integrate epidemiological concepts with actuarial modeling frameworks is an increasingly valued skill in the profession.