As of this writing, the new coronavirus (COVID-19) has led to more deaths in China than the 2002-2003 SARS outbreak in the country. The virus is currently considered an epidemic, with majority of the related mortality concentrated in a single region. But, concerning signs--such as human-to-human transmission and more reported cases outside China--indicate the possibility of a wider geographic spread of the disease.

For the purpose of assessing economic and credit implications of the outbreak, S&P Global Ratings assumes the outbreak will be contained by March, consistent with our recent report (see "Coronavirus To Inflict A Large, Temporary Blow To China's Economy," published Feb. 6, 2020). As the situation develops, we will update our assumptions and estimates accordingly.

The current spread of COVID-19 has no reported deaths in the U.S. and, at current levels of mortality, we expect limited to no impact on U.S. life insurers that are holding about $20 trillion of net inforce life insurance on their balance sheets. But, what could the impact of a global pandemic or an extreme mortality event be on U.S. life insurers? We undertook hypothetical (unrelated to COVID-19) extreme and moderate mortality stress tests to ascertain that potential effect on life insurers.

We estimate that an extreme pandemic event (based on the 1918 "Spanish flu" event) would result in about $52 billion of excess net mortality claims (aftertax), about 12% of the aggregate U.S. life insurance industry capital. It is important to note that this hypothetical extreme scenario is a very low-probability and high-severity event. We do not assume this scenario in our base-case insurer or sector expectations.

Our moderate pandemic event stress (based on the 1957 "Asian flu" pandemic) showed $7 billion of additional excess net mortality claims (aftertax). This represents about 2% of outstanding industry capital, and would have a limited impact on the industry's financial strength.

We view the capitalization levels for rated U.S. life insurers as strong. Our capital analysis indicates that rated U.S. life insurers have slightly above 10% capital redundancy, in aggregate, at the 'AA' level. Comparing that capital redundancy with the 12% extreme stressed impact, we expect a decline in the capitalization levels, and potential effect on credit quality of some insurers. On the other hand, a 2% effect from the moderate scenario is very manageable considering insurers' overall financial strength.

Chart 1

Stress Test: Extreme And Moderate Mortality Scenarios

There were four key steps for our mortality stress test.

The first step was to identify historical pandemic events and quantify the additional deaths based on such events (referred to as excess deaths per 1,000).

Second, we translated the excess deaths in the general population to excess insured deaths. We assume the impact would be less on the insured population. Not everyone has a life insurance policy, and socioeconomic demographics of the insured population can differ from the aggregate population.

Third, we estimated the amount of excess claims that could result from the excess insured deaths for each stress scenario.

And finally, we used the current corporate tax rate to calculate the posttax excess claims and their impact on life insurers.

Table 1

Modified 1918 pandemic for extreme scenario.  In terms of extreme mortality stress, we looked to the 1918 Spanish flu (H1N1) event. This is considered to be the deadliest pandemic of the 20th century. Related deaths are estimated to be over 40 million worldwide, including about 675,000 in the U.S. The world population at that time was estimated to be about 1.9 billion and the U.S. population about 103 million.

The H1N1 virus has been synthesized since then, but there weren't influenza vaccines back in 1918. Antibiotics needed to treat any secondary bacterial morbidity were also not easily available. We think current health care facilities are better equipped than in 1918 to reduce the mortality of such a pandemic. For these reasons, we haircut the actual deaths from the 1918 event (6.5 excess deaths per 1,000) by about 25% and used 4.9 excess deaths per 1,000 for the general population in our extreme stress scenario.

1957 pandemic for moderate scenario.  As for our moderate stress, we looked at the 1957 Asian flu (H2N2). This was also a global pandemic, with about 1.1 million deaths worldwide and 116,000 in the U.S. The 1957 pandemic was far less severe than the 1918 contagion, but it still had a larger impact than a "normal" annual seasonal flu in the U.S. We used the excess deaths from 1957 pandemic (0.7 excess deaths per 1,000) for the general population in our moderate stress scenario.

Table 2

General population-to-insured population.  Not all people have life insurance policies. The demographics of individuals with a life insurance policy can differ from the overall population. Therefore, we needed to quantify how much of the general population's excess deaths per 1,000 would relate to the insured population.

The Society of Actuaries (SOA) undertook a study in 2007 (Delphi Method) to estimate the expected impact of a pandemic on the insured population. We used the median result of this study to identify what proportion of the general population excess deaths per 1,000 can be translated to insured lives. We applied 62% for the moderate scenario and 71% for the extreme scenario. As a result, our key inputs were 3.5 excess (insured) deaths per 1,000 for the extreme scenario and 0.43 excess (insured) deaths per 1,000 for the moderate scenario (see table 1).

Calculating excess mortality claims.  By multiplying the excess insured deaths with the total life insurance policies outstanding, we get to the total number of excess insured deaths. We then multiplied the total insured deaths with average face value per policy, net of reinsurance, (assumed to be $75,000 per policy) to arrive at the excess mortality claims. This resulted in pretax excess claims of $66 billion for our extreme scenario and $8 billion for our moderate scenario.

Aftertax excess mortality claims.  For calculating aftertax excess claims, we used a 21% corporate tax rate. We realize the effective tax rate can vary by insurer, but this rate is a reasonable assumption for this test. Aftertax excess claims were $52 billion for the extreme scenario and $7 billion for the moderate scenario.

Other Considerations For The Stress Test

Age distribution.  We assumed a flat mortality 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 20-40 year age group. This is often referred to as the "W-shaped" mortality curve. 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 excess mortality distribution for our study.

Longevity offset.  We did not mitigate the mortality stress results by any reserve releases or offsets from insurers' longevity business. It's possible that, in an extreme mortality scenario, insurers would report better results in their longevity business. Although we didn't quantify the longevity offset, we recognize the benefits of having a diversified risk profile. As the industry boosts its longevity exposure through sales of retail and institution retirement products, an extreme mortality event could see some offsets from that growing line.

Mortality Risk Remains A Stable Business Line For Life Insurers

Despite some recent stagnation in improved longevity, the U.S. population has seen a long-term trend of longevity gains (see chart 2). Comparing the current life expectancy (at birth) with the pre-1918 and pre-1957 pandemics provide a stark comparison. In 2018, U.S. life expectancy was 78.7, which is extremely favorable to 48.4 in 1917 and 69.8 in 1956.

This long-term improving trend is favorable for mortality risk, and U.S. life insurers have seen good mortality results. We also look at insurers' risk management frameworks and the use of reinsurance to offset the potential impact of higher-than-expected morality events. There have been blips, where mortality trends may see isolated spikes in a certain quarter for a particular insurer, but overall we view this business as stable.

Chart 2

Stress Testing In Stressful Times

As credit ratings analysts, we often undertake stress-testing exercises to better comprehend the financial and business impacts of an 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. But there perhaps isn't a more-morbid stress testing exercise than one looking at extreme mortality. The spread of an actual pandemic has personal impacts far beyond these stressed excess mortality claims. We hope the mortality scenarios used in this stress test remain purely hypothetical.

Related Research

• Coronavirus To Inflict A Large, Temporary Blow To China's Economy, Feb. 6, 2020
• Ten Things You Need To Know About U.S. Insurers' Long-Term Care Business, Jan. 23, 2020