A casualty actuary might be forgiven for thinking that illness and disease are what those “other” actuaries worry about.

Though risk of illness is usually considered the province of the life-health actuary, a session at the 2017 CAS Annual Meeting in Anaheim, California, showed how epidemics can affect property-casualty risks. The session also described how to approach modeling those exposures.

Speakers intoned that, if done right, the modeling could drive new insurance products. These developments could narrow the insurance gap — the chasm between what is insured and what could be insured.

Milliman actuary Cody Webb, FCAS, began by demonstrating how big the insurance gap is, particularly in developing nations. He explained that the spectrum of losses ranges from minuscule (loss of a single strand of hair) to catastrophic (sudden, instant death) and can affect a single person or every entity in the universe across eons. But the insurable losses share some traits, Webb said, including:

• a large number of similar exposures.
• a definite loss, driven by some sort of accident.
• the ability to create an affordable premium to reimburse after such a loss.
• the ability to accurately quantify the amount of loss sustained. This is the most important shared trait.

In showing a chart of property-casualty insurance as a percentage of GDP — with the wealthier countries better insured than others — Webb noted that insurance companies need to “quantify and develop products that meet all criteria of insurability.” (See chart below.)

Penetration of the non-life insurance industry, 2015.

Direct gross premiums as a percentage of GDP.

Cathine Lam, ACAS, an actuarial associate at Metabiota, continued the session with examples demonstrating how epidemics and pandemics involve property-casualty exposures. She pointed to what happened to a Dallas, Texas, hospital during the 2014 Ebola outbreak and described how the Zika virus befell Miami in 2016.

Ebola in Dallas

The deadly outbreak (28,000 cases/11,000 deaths worldwide) originated in West Africa. One man who contracted the disease only displayed symptoms after he traveled to Texas. He went to a hospital and died two weeks later. In the meantime, he infected two individuals.

The property-casualty exposures include the following:

• Business interruption. When potential patients learned of possible Ebola exposure, the number of emergency room visits was cut in half. Additionally, the number of patients per day fell by 22 percent and net revenues declined by 25 percent ($12 million). After the U.S. Centers for Disease Control (CDC) said the hospital was Ebola-free, revenues returned to normal.
• Malpractice. The victim’s family sued the hospital. A nurse also sued, alleging that the staff had inadequate training to deal with Ebola victims and that she had suffered a loss of privacy, becoming known as “the Ebola nurse.”

And the exposures extended beyond the hospital. A bridal shop that the nurse had visited became known as “the Ebola shop.” It closed.

Zika in Miami

The 2016 Zika outbreak was the widest ever, Lam said. The virus is generally not dangerous to adults, but a pregnant woman can pass it to her fetus, causing severe birth defects.

The first case in Miami occurred in the Wynwood neighborhood, where the CDC had issued a six-week travel advisory. The advisory was later lifted, but at its height, airline travel to the Miami area fell by 17 percent. Revenues from hotel taxes dropped about five percent and the majority of all businesses reported a decrease in revenue of at least 20 percent.

The disease being modeled has to originate somewhere. From that point of origin, the amount that the disease spreads depends on the originating state’s ability to combat it. This is reflected in what Lam calls a “Preparedness Index.”

Keeping these examples in mind, it takes little imagination to construct an insurance product that would respond to an epidemic. Pricing that product, though, would be a challenge. Epidemics are fortunately rare, Lam noted, so there will never be much historical data from which to project. The alternative was to build a model. Such a model, Lam said, would require a solid scientific foundation. It would be a multidisciplinary exercise, involving knowledge of how diseases spread and how economies are affected, among other things. The result would provide a granular look at how the spread of a disease would affect property-casualty exposures.

At a high level, the procedure she described resembles catastrophe modeling, translated to the world of disease. As with a catastrophe model, one models the exposure, creates a catalog of events, and then uses the information to inform pricing and capital decisions.

The disease spread model, Lam said, would be a large scale computational model that shows how a disease would move across the world. The model would show how people progress from susceptible (within the exposure area) to exposed.

Some of the exposed become infected. Then they are asymptomatic for a time before becoming fully symptomatic. They then either recover or die. The length of time at each stage varies by disease.

An additional consideration in the model is the impact that travel has on increasing the potential for the disease to spread. Lam discussed how government policy and efficacy was used to incorporate this into the model.

The disease being modeled has to originate somewhere. From that point of origin, the amount that the disease spreads depends on the originating state’s ability to combat it. This is reflected in what Lam calls a “Preparedness Index,” a metric that describes the level of access to vaccines, drugs and physicians. It also accounts for how capably a government can respond to epidemics, from informing people about the outbreak to establishing quarantine areas, and more.

Ultimately, the goal is to establish an event catalog— a collection of scientifically plausible, hypothetical events. Higher probability events are more frequent in the catalog. Analysis of the catalog will produce exceedance probabilities (the probability that losses will exceed a given loss amount). To revisit the catastrophe modeling comparison, the disease spread model exceedance probabilities similarly play a large part in informing property-casualty insurers’ capital and pricing decisions.

Lam gave examples of how such modeling could be used. During a 2015 outbreak of Middle East Respiratory Syndrome (MERS) in South Korea, a policy offered up to $4,500 for medical costs if a person traveling to the area contracted MERS (and $91,000 if they died within 20 days of diagnosis).

But the insurance had a fairly narrow scope; it only insured people traveling to the region. It did nothing for people in the infected region.

Hope for a solution comes from the African Risk Capacity, an agency established by the African Union to help member countries prepare and respond to extreme events, of which epidemics is one. Modeling could help them prioritize the risks they face and respond efficiently, mitigating property-casualty exposure along the way.

Lam envisioned a number of other insurance products, including:

• Travel insurance with an epidemic-driven trigger (e.g., the number of cases in a country exceeding a predetermined threshold, or issuance of a government travel alert).
• Business interruption insurance for hospitals. In addition to covering lost revenue, a policy could extend to cover absenteeism, which rises during an epidemic. She cited a survey stating that if an avian flu struck, 42 percent of nurses said they might show up or might not. Fifteen percent said they would not show up, even if they would lose their jobs.

Absenteeism is a “major concern during an outbreak,” Lam said. “When there is inadequate care for the patient, it just makes the outbreak worse.”

James P. Lynch, FCAS, is chief actuary and director of research for the Insurance Information Institute. He serves on the CAS Board of Directors.