S&P Global Ratings conducted a stress test for a hypothetical pandemic to appreciate the potential impact of a stressed morbidity event on U.S. health insurers. Our analysis found that in a moderate morbidity scenario, U.S. insurers would see a 3%-4% increase in medical claims cost, which, although not insignificant, would be manageable by most insurers. However, the impact would be far more detrimental to the credit quality of insurers in a severe morbidity stress scenario, which showed a 10%-12% increase in medical claims cost.

It helps to analyze the stress test's results by comparing its impact on insurers' medical loss ratios (MLRs). The MLR shows the percentage of incoming premiums that insurers pay out in policyholder medical claims. In 2019, the aggregate MLR for U.S. health insurers was approximately 85%, meaning 85 cents of each premium dollar received by the insurers was paid out in medical claims.

If you add our increased claims cost from morbidity stress to this 85% MLR, you end up at about 88%-89% MLR in the moderate stress scenario and 95%-97% MLR in the severe scenario. That means in both scenarios, insurers' profitability would come under pressure. The severe stress scenario in particular would result in reported losses for the year, since the MLR plus the administrative expense ratio (say, about 11%-13%) would be above 100%. A ratio above 100% means that not all medical claims and administrative expenses would be covered by annual incoming premiums. Therefore, insurers would need to dig into their capital buffers to cover the increased costs.

As for individual insurers, differences in business mix, profitability targets, and capital levels can result in differing levels of financial impact. For example, an insurer with a higher amount of self-insured group business would feel a lesser impact compared with an insurer with a higher amount of fully insured business. Similarly, insurers with higher capital buffers and higher profitability targets could be better positioned during a severe pandemic. In addition, the flexibility of health insurers to reprice their products on an annual basis does provide them with flexibility to increase premium rates to somewhat offset any meaningful financial harm from a severe morbidity event.

Chart 1

Pandemic Stress Test: Severe And Moderate Morbidity Scenarios

Our morbidity stress test involved four key steps. We estimated the following:

1) The number of individuals covered by private health insurers;

2) The medical service utilization rate in a moderate and a severe stress scenario;

3) The resulting amount of additional medical claims above the normal run rate; and

4) The MLR impact--the ratio of excess claim amounts to incoming premiums.

First, we estimated the amount of the U.S. population that was fully insured by private health insurers.  Approximately 90% of the U.S. population, or 290 million individuals, is insured in the U.S. But not everyone has insurance from a private insurance company. We estimate that about 170 million are actually fully insured by private insurers. This 170 million total, used in our analysis, excludes segments of the insured population that are either self-insured (such as self-insured employer groups) or use nonmanaged care (such as traditional Medicare).

Second, we developed a medical service utilization rate for the two stress scenarios.  We used various studies, including the U.S. Centers for Disease Control and Prevention (CDC) and U.S. Department of Health and Human Services (HHS) pandemic preparedness plans and the Society of Actuaries' influenza pandemic study, as well as research on past pandemics, to estimate the number of individuals who would be infected and what portion of those infected by the virus would need to utilize medical services.

In both scenarios, we assumed a 30% infected rate with about half of the infected not having a severe enough episode to require meaningful medical care. The key difference in the two scenarios was the intensity of the pandemic, which resulted in a higher inpatient (hospital) utilization rate of 8% (or 4 million individuals) in the severe scenario, compared with 2% (or 0.9 million) in the moderate scenario.

The inpatient utilization numbers include assumed mortality in both scenarios. We assume the deaths for the insured population would take place after or during a hospital admission, meaning that the insurer would be billed for the related inpatient utilization. We use our hypothetical mortality stress test to estimate the deaths in the two scenarios. (Please refer to our commentary "Amid Coronavirus Outbreak, S&P Global Ratings Looks At How A Hypothetical Pandemic Could Affect U.S. Life Insurers," published Feb. 14, 2020, for more information on mortality stress.)

Third, we calculated the resulting additional medical claims cost to the insurer.  The total additional claims cost was calculated by multiplying the number of individuals in outpatient and inpatient facilities by the related cost of such services. We then offset a portion of the total medical claims with possible out-of-pocket or cost-sharing measures (for example, deductibles and copays), wherein the insured pays a portion of the cost out of pocket.

To estimate the cost of an inpatient or outpatient visit, we referred to various industry surveys of inpatient and outpatient costs, including the Health Care Cost Institute's 2018 Health Care Cost and Utilization Report. For this stress test, we assumed a total cost of $500 for outpatient utilization and $20,000 for related inpatient (multiday stay) utilization.

And finally, we compared the total additional claims with the total premium revenue of the health insurers.  This provides us with the impact on the MLR from the hypothetical pandemic scenarios. As shown in chart 1, MLRs increased in both scenarios, with a far more meaningful impact in the severe scenario. One thing to note is that we are using only premiums related to fully insured, comprehensive insurance business lines. Insurers may have access to other supplemental lines of business as well as fees from self-insured, noninsurance businesses.

Table 1

Other Considerations For A Pandemic Scenario

Social factors, such as the uninsured rate and lack of universal paid sick leave, could influence medical care utilization and the containment of a pandemic.  In the case of any contagious disease, there is a clear urgency for appropriate containment and treatment. Currently, about 27 million individuals are uninsured in the U.S., and paid sick leave isn't a universal benefit. Both these factors could likely make affected individuals hesitant to seek medical care at an early stage of the sickness. Delays in accessing medical care in a pandemic can impede timely testing, effective containment, and treatment.

Cost-sharing could be higher or lower based on the timing of the pandemic.  Out-of-pocket spending, especially for employer-based plans, has gradually increased. The first hurdle to cross--before the insurer starts paying the medical costs--is the deductible. As per the Peterson-KFF Health System Tracker, the average person in employer- or group-based coverage reached their deductible in May in 2019. So if the pandemic spreads earlier in the year, the insured will bear more of the cost, whereas if it spreads later in the year, the insurance company will cover more of the cost.

Chart 2

It is also worth noting that governments and private insurers may take steps to waive some of the cost-sharing requirements. Whether cost requirements are enforced or waived may have a meaningful impact in a pandemic event.

Infection rates and utilization rates may vary depending on the reproduction rate and severity of the virus.  Not all pandemics spread the same. The reproduction rate can differ, which influences the population infection, or "attack," rate. For the purpose of this exercise, we assumed a 30% infection rate, which is similar to estimates of the 1918 Spanish influenza pandemic. If infection rates were different, the impact on insured and insurer could be materially different. For example, if the infection rate was 10 percentage points higher or lower (compared with our 30% assumption), it would result in a 33% higher or lower medical cost in the stress scenarios.

But what portion of the 30% actually uses medical services? We looked at the CDC and HHS pandemic preparedness plans, as well as the actual disease burden during a flu season (see table 2). Over the past 10 years (2010-2019), we found that on average half of the individuals who show symptoms of the seasonal flu do not require any medical or hospital visits. For the other half who require medical treatment for the seasonal flu, a little less than 2%, on average, end up in a hospital setting. This was the basis of our moderate stress scenario, while the utilization severity used in the HHS preparedness plan, which seems to align with historic pandemics, was the basis for our severe stress scenario.

Table 2

Age distribution plays a role in medical cost trends.  We assumed a flat morbidity distribution for this stress test. However, pandemics could affect various age groups differently.

Generally for a seasonal flu in the U.S., the most-affected people are the very young and very old, with limited mortality in the middle-age group. This is referred to as "U-shaped" mortality, which was also the case for the 1957 pandemic. But epidemiologists have noted that the Spanish flu was quite unique in that there was a spike in the 20-to-40-year age group. This is often referred to as a "W-shaped" mortality curve.

On the other hand, initial reports of COVID-19 indicate that its impact may be more focused on the older population. We have no way of knowing how exactly the next pandemic could affect different ages, and an age distribution of the insured population wasn't easily available. Therefore, we maintained a flat morbidity distribution for our study.

Hospital capacity also plays a significant role in insurer costs.  For the purpose of this stress test, we assumed hospitals will have capacity to admit and treat individuals who need inpatient care in both stress scenarios.

However, hospital capacity may be limited in a pandemic. According to the American Hospital Assn., there are 924,107 staffed beds in the U.S. Assuming 65% of those beds are in use, there would be 323,437 beds available at a given point in time. That would be well short of the required hospitalization needs (see table 1) in both our moderate and severe pandemic stress test scenarios.

As a reminder, the hospital utilization highlighted in table 1 does not include the uninsured or self-insured groups, which, if included in this analysis, would only increase the need for inpatient facilities. For example, if we used a similar infection rate and severity assumptions for the entire U.S. population (irrespective of insured status), there would be about 1.7 million individuals in need of inpatient services in the moderate morbidity scenario and almost 8 million in the severe scenario. We do recognize that not all affected individuals will need beds on the same day or even in the same week, and hospitalization needs would be spread over a period of time. But given the number of people potentially needing inpatient care during a wave of the pandemic, it is likely that additional medical capacity would be needed.

It is, of course, possible that additional beds could be made available in a hospital setting, with hospitals expanding their resources or delaying any elective care. But despite hospitals adapting the best they can, government support may still be needed to add health care facilities in a pandemic scenario. We may see more government intervention, especially to develop new sites to house patients outside of a traditional hospital setting. For example, with COVID-19, some individuals have been temporarily housed on an Air Force base after being transported from a cruise ship or an international location.

If care is outside of a traditional facility, who pays for the medical bill? Would the government pick up the tab, or would there be a way to still bill the insurer? The impact on the insurer will be significantly less in both the severe and moderate hypothetical scenarios if the insurer doesn't need to cover treatments outside of a traditional facility.

Stress Testing In Stressful Times

We aren't epidemiologists. We don't have the ability to predict the spread of a novel virus. This hypothetical pandemic stress test should not be seen as reflecting our expectations for COVID-19. We often undertake stress-testing exercises, similar to this one, to better comprehend the financial and business impacts of expected or hypothetical events. Such testing results provide us and the users of our ratings with better insights into the credit quality of the sector. The spread of an actual pandemic has a personal impact far beyond these stressed medical claims costs for insurers. We hope the morbidity scenarios used in this stress test remain purely hypothetical.

Related Research

• Amid Coronavirus Outbreak, S&P Global Ratings Looks At How A Hypothetical Pandemic Could Affect U.S. Life Insurers, Feb. 14, 2020
• U.S. Health Insurer Outlook: Stability In 2020 Will Give Way To Major Change Over The Next Decade, Jan. 15, 2020