October 2020
An Investigation into the
Insurability of Pandemic Risk
1
An Investigation into the Insurability of Pandemic Risk
Kai-Uwe Schanz, Deputy Managing Director and Head of Research & Foresight,
The Geneva Association
In collaboration with
Prof. Martin Eling, Chair for Insurance Management and Director,
Institute of Insurance Economics, University of St. Gallen
Prof. Hato Schmeiser, Chair for Risk Management and Insurance and Director,
Institute of Insurance Economics, University of St. Gallen
Prof. Alexander Braun, Vice Director, Institute of Insurance Economics,
University of St. Gallen
An Investigation into the
Insurability of Pandemic Risk
2
www.genevaassociation.org
The Geneva Association
The Geneva Association was created in 1973 and is the only global association of insurance companies; our
members are insurance and reinsurance Chief Executive Officers (CEOs). Based on rigorous research conducted in
collaboration with our members, academic institutions and multilateral organisations, our mission is to identify
and investigate key trends that are likely to shape or impact the insurance industry in the future, highlighting what
is at stake for the industry; develop recommendations for the industry and for policymakers; provide a platform to
our members, policymakers, academics, multilateral and non-governmental organisations to discuss these trends
and recommendations; reach out to global opinion leaders and influential organisations to highlight the positive
contributions of insurance to better understanding risks and to building resilient and prosperous economies and
societies, and thus a more sustainable world.
October 2020
An Investigation into the Insurability of Pandemic Risk
© The Geneva Association
Published by The Geneva Association—International Association for the Study of Insurance Economics, Zurich.
Photo credits:
Cover page—World Pictures and Sdecoret / Shutterstock.com
The Geneva Association—International Association for the Study of Insurance Economics
Talstrasse 70, CH-8001 Zurich
Email: [email protected] | Tel: +41 44 200 49 00 | Fax: +41 44 200 49 99
3
An Investigation into the Insurability of Pandemic Risk
Contents
Foreword 5
1. Executive summary 6
2. Quantifying the challenge: Global protection gaps in light of COVID-19 8
2.1. The notion of protection gaps 8
2.2. COVID-19: Assessing the shortfalls 9
2.2.1. The business interruption protection gap 9
2.2.2. The health protection gap 10
2.2.3. The mortality protection gap 15
3. Getting the basics right: The insurability of pandemic risk 16
3.1. The concept of insurability 16
3.2. The criteria of insurability 17
3.3. The limits to insuring pandemic risk: A comparative and holistic view 18
3.3.1. Pandemic business interruption risk 18
3.3.2. Pandemic life and health risk 21
3.3.3. Pandemic risk compared to other catastrophic risks 23
4. Conclusions 26
References 28
4
www.genevaassociation.org
Acknowledgements
The Geneva Association is very grateful to its Board members Oliver Bäte, CEO, Allianz, and
Brian Duperreault, CEO, AIG, for their personal guidance and support as co-sponsors of this
research.
Our special thanks go to all Geneva Association member company experts whose
comments have benefited this publication: Edward Barron (AIG), Paul DiPaola (AIG), Kean
Driscoll (AIG), Andreas Funke (Allianz), Gong Xinyu (PICC), Arne Holzhausen (Allianz), Kei
Kato (Tokio Marine), Christian Kraut (Munich Re), Roman Lechner (Swiss Re), Christiane
Meyer-Gruhl (Allianz Re), Cameron Murray (Lloyd’s of London), Guillaume Ominetti
(SCOR), Gisela Plassmann (ERGO), Olivier Poissonneau (AXA), Veronica Scotti (Swiss Re)
and Lutz Wilhelmy (Swiss Re).
We are also indebted to our academic partners and contributing authors Martin Eling, Hato
Schmeiser and Alexander Braun who co-shaped section 3 of this report.
We offer our deepest empathy to the countless people, communities and businesses
who have been impacted by COVID-19. At the time of publication of this report, in late
October 2020, there have been more than 40 million cases and 1 million deaths recorded
globally. Many countries are experiencing a second or third wave of the virus.
There is a great proliferation of information and views on COVID-19, and what we read,
unfortunately, is not always rooted in facts. The insurance space is no exception.
This first report in The Geneva Association’s research series on pandemics and insurance
sets out to explore – in objective terms – the capacities of insurers to absorb pandemic-
related costs.
Encouragingly, pandemics on the scale
of COVID-19 pose no fundamental
insurability challenges for health and
life insurers, allowing them to fully play
their protection and support role to
affected people and communities.
The picture is different for property
& casualty (P&C) losses. Even those
who anticipated the scenario of a
global pandemic did not fathom the nature and scale of government decisions taken
around the world to slow infections: wide-ranging shutdown measures that brought
economies to a standstill.
From an insurance perspective, this type of government response is neither predictable
nor modellable. That is one of the reasons why pandemic risk was not included in most
business interruption policies.
Our research findings are unambiguous: the property & casualty insurance industry,
which collects USD 1.6 trillion in premiums per year for all policies – and a mere USD 30
billion for business interruption risk – is not the right vehicle for shouldering the projected
global loss in GDP for 2020 of USD 4.5 trillion.
As a consequence, governments need to involve themselves in closing the pandemic
protection gap in P&C. And insurers still have a role to play. Our second pandemics
report will explore possible solutions: innovative, public-private efforts that recognise the
enormous magnitude and unique nature of pandemic risks.
Taken together, we hope these reports will help governments and insurers think about
and agree upon feasible, effective ways to work together to better protect society from
extreme risks, such as pandemics, going forward.
Jad Ariss
Managing Director
Foreword
5
An Investigation into the Insurability of Pandemic Risk
Health and life insurers can
fully play their protection
and support role to people
and communities affected
by COVID-19.
6
www.genevaassociation.org
COVID-19 and the draconian shutdown measures adopted by many governments
to contain it have plunged the global economy into the deepest recession since the
Second World War. For the global insurance industry, too, the pandemic is a severe
loss event. Despite this massive strain, initially exacerbated by a steep decline in
capital markets, insurers worldwide promptly paid legitimate claims in all areas
where pandemic risk was intended to be covered; for example, under life, health
and event cancellation policies. In addition, also during the lockdowns, insurers
have continued to pay claims and benefits unrelated to the pandemic; for example,
in motor, liability and annuities insurance.
At the same time, COVID-19 has exposed massive protection gaps in the area
of business continuity risk. Less than 1% of the estimated USD 4.5 trillion global
pandemic-induced GDP loss for 2020 (source: The World Bank) will be covered by
business interruption insurance – a niche segment which generates annual premium
income of about USD 30 billion (less than 2% of the world’s property & casualty
insurance market), with cover generally intended for and triggered by physical
damage only.
The mismatch between economic losses and the risk-taking capacity of insurers
who offer business interruption cover, as well as past demand for pandemic
coverages, is staggering. With annual business interruption insurance premiums
of about USD 30 billion, insurers would have to collect premiums for 150 years
in order to absorb the estimated USD 4.5 trillion global output loss inflicted by
COVID-19 and its handling in 2020. Even the size of the entire global property &
casualty insurance industry (USD 1.6 trillion in premiums, according to McKinsey)
is eclipsed by the economic damage from the pandemic. In order to cover the total
cost, all property & casualty insurers worldwide would have to collect premiums
across all lines of business for almost three years, with no money left for covering
private homes and vehicles, injured workers and numerous liability exposures.
Therefore, property & casualty insurers have typically applied strict exclusions on
pandemic business continuity risk and never intended to cover it.
Insurers would have to collect business interruption
insurance premiums for 150 years in order to absorb
the estimated USD 4.5 trillion global output loss
inflicted by COVID-19 and its handling in 2020.
Existing protection gaps facing individuals and households in the areas of mortality
and healthcare risk have been much less highlighted by this pandemic due to
relatively moderate excess mortality and slightly reduced overall healthcare
expenditure.
The extent of correlation and aggregation of pandemic losses for businesses across
the globe has put the insurability of pandemic risk in the spotlight. It touches upon
1. Executive summary
7
An Investigation into the Insurability of Pandemic Risk
the pivotal question of whether pandemics are a type of
risk for which the insurance industry can play any kind of
role or if this is the type of risk where traditional insurance
products are not the solution.
It is not difficult to intuitively understand the limits to
insuring pandemic risk. The word ‘pandemic’ originates
in the ancient Greek (‘pan’ means all and ‘demos’ means
people). Pandemic-induced business continuity risk is
obviously unique given its potential to impact virtually all
policyholders simultaneously, over an extended period of
time. Applying the two most relevant customary criteria of
insurability to pandemic business interruption risk yields
the following conclusions:
First, losses are neither random nor independent. Even
though pandemics are naturally occurring phenomena,
policy decisions to lock entire economies are deliberate
and intentional. This means that expected loss amounts
and risk loadings cannot be set. There are also no historical
data for the policy responses witnessed during COVID-19.
Furthermore, the strong correlation among individual
risks renders efficient risk pooling and diversification
impossible.
Second, the maximum possible loss is not manageable
from the insurer’s solvency point of view. The
uncontrollable aggregation of losses could be ruinous to
the risk pool and, ultimately, to the insurance industry
as a whole. This in turn could lead to significantly further
financial stability risks across the wider economy.
The uncontrollable aggregation
of losses from pandemic business
interruption could be ruinous to the risk
pool and, ultimately, to the insurance
industry as a whole.
As opposed to business continuity, pandemic life and
health risks are generally non-systemic and privately
insurable. Excess mortality risk is modellable based on a
wealth of historical data. In addition, increased mortality
risk is (partially) offset by reduced longevity. For health
insurers, there is a ’natural’ limit to claims given the finite
capacity of healthcare systems and temporarily reduced
expenditure on non-pandemic-related procedures. Hence,
there are generally no exclusions for pandemics or other
common causes of extreme mortality and health events.
Having said this, life and health insurers’ resilience could
1 To be discussed in-depth in our forthcoming publication Public and private solutions to pandemic risk (November 2020).
be tested by future pathogens which may be more
aggressive and lethal than COVID-19.
In addition to distinguishing between uninsurable and
insurable parts of pandemic risk, it is important to
understand the differences between pandemic and other
catastrophic risks, first and foremost, in terms of the scope
for global diversification. Pandemics are, by definition,
not diversifiable as they occur on a very wide or even
global scale (as opposed to epidemics which are more
locally concentrated). Some other risks such as terrorism
or natural catastrophes are diversifiable on a global level
and routinely transferred via re/insurance or Alternative
Risk Transfer (ART) instruments. These disasters impact
a limited number of policyholders for a limited period of
time. As COVID-19 illustrates, economic losses caused
by extreme pandemics and their handling by public
authorities are neither locally nor globally independent.
Therefore, pandemic business continuity risks are
uninsurable.
Pandemics are, by definition, not
diversifiable as they occur on a very
wide or even global scale.
Having said this, insurers are aware of the need to address
this socio-economic challenge. The industry is prepared
to explore the scope for innovative solutions and public
sector-led efforts which acknowledge the enormous
magnitude and unique nature of this particular risk.
1
8
www.genevaassociation.org
2.1. The notion of protection gaps
For re/insurers there is significant uncertainty about the ultimate claims burden
from COVID-19. According to Swiss Re 2020, the mid-point of the range of current
publicly available estimates is around USD 55 billion for all lines of business. In any
case, the insured part will be dwarfed by the economic cost of the pandemic. For
2020, the World Bank currently expects a 5.2% contraction of the global economy
(World Bank 2020), amounting to more than USD 4.5 trillion in lost output.
The share of uninsured losses in total economic losses is generally referred to as the
protection gap. A more meaningful measure, however, is the notion of the insurance
protection gap, defined as the difference between the amount of insurance that
is economically beneficial for both insureds and insurers on the one hand and
the amount of coverage actually purchased or offered, on the other. This gap is
significantly smaller than the broader protection gap, for some of the following
reasons:
• Certain risks simply defy insurability, with pandemic risk being a case in point
(see section 3.3).
• A certain level of risk retention makes economic sense to incentivise risk
prevention measures and risk-conscious behaviors.
• Insurers implement deductibles to mitigate moral hazard, i.e. a tendency on the
part of the policyholder to behave more carelessly because they have insurance
cover. Deductibles translate into lower sums insured.
• Institutional factors, such as extensive social security benefits or government
post-disaster relief, reduce the need for individuals to take out private insurance
(The Geneva Association 2018).
In reality, however, the insurance protection gap is hard to measure and highly
subjective. Each insured individual or business assesses the economic benefits of
insurance differently. Similarly, on the supply side, insurance companies differ in their
view as to what is insurable and at what minimum price (Karten 1997). Therefore the
notion of the insurance protection gap is generally replaced by the wider, easier to
quantify but less meaningful overall protection gap measure which compares covered
losses with total economic losses.
In the following section, we explore protection gaps for the three areas of pandemic
risk discussed in this report – business interruption, health and mortality. The scope
and scale of such gaps is enormous. It entails businesses (especially small vulnerable
2. Quantifying the
challenge: Global
protection gaps in
light of the pandemic
9
An Investigation into the Insurability of Pandemic Risk
ones) defaulting within weeks, households suffering
financial stress from additional out-of-pocket healthcare
expenditure or, in extremis, impoverishment if the main
breadwinner dies from COVID-19.
Against this backdrop, it is important to understand
the limits to insurability presented by the various forms
of pandemic risk. This awareness is an indispensable
foundation for subsequent discussions on pandemic risk.
2
2.2. COVID-19: Assessing the shortfalls
In the following section we explore protection gaps in the
areas of business interruption (BI), mortality and health.
These risks were selected, first, because of their relevance
to both insurers and customers during COVID-19 and,
second, especially for mortality and health, because of the
availability of meaningful data.
2.2.1. The business interruption protection gap
In the context of COVID-19, the protection gap debate
focuses on the trillions of dollars of economic losses
arising from the impact of government-mandated
lockdown measures worldwide and the share of these
losses insurers could or should absorb.
The relevant dimensions and proportions speak for
themselves: P&C insurers globally generate annual
premium income of about USD 1.6 trillion (McKinsey
2020). They would have to collect this amount of
premiums across all lines of business for almost three
years in order to cover the estimated USD 4.5 trillion
global loss in GDP (see Figure 1 for an illustration).
3
We estimate the aggregate capital base of the world’s
P&C insurance sector at a similar size of USD 1.6 trillion.
Therefore, in a global lockdown scenario, insurers’ entire
surplus would be exhausted after a few months. In light
of the probability of a major pandemic event (industry
experts put it at 30–40 years), and given the current levels
2 See section 3 of this report as well as The Geneva Association (2020).
3 Global output losses are obviously not a perfect proxy for business continuity losses. Government support measures have provided significant
relief to businesses.
4 For the U.S., The American Property Casualty Insurance Association (APCIA) estimates monthly lockdown-induced BI losses for smaller, and
arguably the most vulnerable, businesses with fewer than 100 employees at USD 255 billion to 431 billion (APCIA 2020). If fully insured, these
businesses’ losses alone would exhaust the U.S. P&C insurance industry’s entire capital in 2–3 months.
5 We estimate that 20–25% of global commercial property premiums of about USD 115 billion (source: Allianz Research, 2018 figure) reflect BI risk.
Note that this premium pot almost exclusively covers BI risks with an unequivocal physical damage trigger (e.g. a fire at a manufacturing plant).
6 Willis Towers Watson 2020 estimate insured BI losses (including event cancellations) in the U.S. at roughly USD 10–20 billion, including an
allowance to pay claims on policies where no coverage was intended, as triggered by litigation, regulation and/or legislation. The share of the U.S.
market in global commercial property premiums is about one third. Assuming a lower level of BI insurance penetration and legal uncertainty in
other parts of the world, global insured BI losses could come in at the estimated range of USD 20–40 billion.
7 Again, global output losses are not a perfect proxy for business continuity losses.
of industry capital (versus levels of exposure to pandemic
risk), this risk would impose a material solvency risk on
the sector and harm many other policyholders from other
lines of business, as well as create a potential financial
stability threat.
4
Only a tiny fraction (an estimated USD 25–30 billion
5
,
or less than 2%) of the world’s total P&C premium base
is linked to BI coverage. The lion’s share of the industry’s
premiums and capital backs private homes and vehicles,
injured workers and numerous liability exposures unrelated
to COVID-19 (Hartwig and Gordon 2020a).
Further assuming that global insured BI losses from the
pandemic could ultimately amount to USD 20–40 billion
for 2020,
6
insurance claims would cover less than 1% of
global COVID-19-induced GDP losses, translating into a
protection gap of more than 99% (based on the earlier
assumption of USD 4.5 trillion BI-related economic losses).
7
In other words, the global P&C insurance industry would
have to collect BI premiums for at least 150 years in order to
absorb the estimated global output loss from the pandemic
in 2020.
Effective private market insurance
coverage for BI losses would
necessitate rates which would likely
be unaffordable or unattractive for
commercial buyers.
Effective private market insurance coverage for BI losses
would require multiple times the current premium and
capital base of the P&C insurance industry and necessitate
rates which would likely be unaffordable or unattractive
for commercial buyers (see section 3.3). These supply-side
reasons for the private P&C insurance market’s decision
to limit exposure to pandemic risk are compounded by
demand-side barriers such as the underestimation by
10
www.genevaassociation.org
businesses and households of a pandemic’s probability
of occurrence,
8
the speed and/or extent to which a virus
spreads, the probability and/or duration of government-
imposed lockdown measures, or excessive optimism
about the ability of scientists to develop treatments and
vaccines. Demand can be further reduced by people’s
expectation that, if a truly disastrous pandemic event hits,
governments will be there to provide financial assistance
(Hartwig et al. 2020). This analysis suggests that the
overwhelming majority of pandemic BI risk will remain
uninsured by private insurers given the prohibitive amount
of premiums and capital required to offer credible and
secure insurance coverage (OECD 2020a).
In summary, in light of these supply- and demand-side
factors at work, the massive BI protection gap compares
8 PathogenRX, a BI pandemic risk product, was not widely purchased in the years preceding COVID 19; a clear indication of the demand-side
challenges highlighted above.
with a much smaller BI insurance protection gap as defined
as the difference between the amount of insurance that is
economically beneficial and feasible for both customers
and insurers, on the one hand, and the amount of coverage
actually purchased or offered, on the other.
2.2.2. The health protection gap
Even conceptually, the capture and quantification of the
healthcare funding gap is a challenging endeavour. To
a major extent, healthcare expenditure is discretionary
and depends on the quality of healthcare services. Public
healthcare services, for instance, are available in many
markets at affordable prices, but accessibility, long
average waiting times and quality are frequent issues.
Consumers seeking state-of-the-art or timely treatment
Figure 1: An illustration of the global pandemic BI protection gap
All figures are estimates.
Source: The Geneva Association (based on The World Bank 2020, McKinsey 2020a and contributions from Allianz Research)
$115 billion (USD)
Global commercial property
premium volume (2018)
up to $40 billion (USD)
Global insured pandemic
BI losses (2020)
up to $30 billion (USD)
Global BI premium volume
(2019)
$4.5 trillion (USD)
Global economic losses (2020)
$1.6 trillion (USD)
Global P&C insurance
premium volume (2019)
11
An Investigation into the Insurability of Pandemic Risk
usually face significantly higher costs. In addition, the
dynamics of socio-economic variables such as ageing
populations, volatile and difficult to predict government
policies (including subsidies and tax incentives) and cost
inflation as a result of medical advancements can have a
notable impact on the cost of necessary treatment (The
Geneva Association 2019).
Despite these challenges, various parameters have been
used to gauge the size of the health protection gap. One
common proxy is out-of-pocket spending (OOPS), i.e.
the part of national health expenditure that comes from
household savings. The focus here is on the share of OOPS
that is stressful to households – that results in the need
9 Other researchers have focused on catastrophic health expenditure or the risk of impoverishment from unexpectedly high medical expenses as a
key determinant of the health protection gap (see, for example, Wagstaff et al. 2018). Every year, about 100 million people are still being pushed
into extreme poverty (defined as living on less than USD 2per day) because they have to pay for health care. And over 900 million people, around
12% of the world’s population, spend at least 10% of their household budgets on health care (WHO 2019).
to reduce discretionary spending on food or education in
order to pay medical bills. OOPS, however, fails to take
into consideration cases of non-treatment or under-
treatment due to affordability and accessibility reasons.
9
Introducing the current context of the global pandemic,
Figure 2 compares the relevance of OOPS in the G20
countries with the estimated shares of the population
at increased risk of severe COVID-19 due to underlying
health conditions (reflecting primarily demographic and
lifestyle-related peculiarities). The figure illustrates those
countries’ populations’ vulnerability to financial stress or
even catastrophe as a result of COVID-19.
In 2019, credit insurers worldwide wrote around USD 15 billion in premiums (including surety). This corresponds to about
1% of the global premium pool in P&C insurance (excluding health). Despite its relatively small size, credit insurance is
of significant economic importance. Its total exposure amounts to around USD 3.3 trillion. That is, approximately 15% of
global merchandise trade is covered by credit insurance.
This raises two questions: How can such a small industry shoulder such a large exposure? And does the fact that about
85% of global trade is not insured suggest a huge ‘protection gap’?
On the first question, credit insurance differs from most other insurance lines in one key respect: it is based on a
dynamic relationship. Trade credit insurance policies are continually updated and cross-referenced over the course
of the policy period. While policyholders can constantly adjust their needs for credit limits, notably when increasing
business with existing clients or when approaching new prospects, the credit insurer monitors its customers' business
partners throughout the year to ascertain their continued creditworthiness. In case of doubt, individual coverage limits
are reduced. This flexibility allows credit insurers to keep pricing very moderate, with premiums as a percentage of total
exposure standing well below 1%. This has to be seen against the backdrop of the annual global corporate default rate
only having dropped below 1% in two years since the Global Financial Crisis – when it rose to over 5%.
On the second question, not every business relationship requires insurance, e.g. if the customer is a long-term trusted
business partner, a government, a blue chip company or a member of the same group (about two thirds of global trade
in goods are estimated to be intra-group deliveries). This also makes the real value of credit insurance clear: supporting
customers in their expansion by opening up new international markets and business relationships. Ultimately, credit
insurance is not only about indemnifying losses incurred from a default, but also providing businesses with the support
and expertise to improve their risk management. Credit insurers offer actionable economic knowledge, making them
information providers rather than pure risk carriers.
Having said this, credit insurers are not infallible. In extreme cases, individual limits may be withdrawn completely. This
happened in the Global Financial Crisis, accelerating the downward spiral because, in the worst case, transactions without
cover were not conducted at all. As a result of these experiences, many governments acted quickly in the COVID-19
crisis and, by means of state protection shields, enabled credit insurers to maintain their limits – for the benefit of their
customers and the economy at large.
Source: Allianz Research
Box 1: Protection gaps in trade credit insurance
12
www.genevaassociation.org
Figure 2: Out-of-pocket spending as a percentage of current health expenditure (2017) and estimated share of
population at increased risk of severe COVID-19 due to underlying health conditions
10
Source: The Geneva Association (based on WHO’s health expenditure database and Clarke et al. 2020)
10 In line with WHO standards, Clarke et al. (2020) define a severe case of COVID-19 as ‘a patient with severe acute respiratory illness’: 1) fever,
2) at least one sign/symptom of respiratory disease, e.g. cough, shortness of breath and 3) requiring hospitalisation. Conditions associated
with increased risk of severe COVID-19 include the following 11 categories: 1) cardiovascular disease, including cardiovascular disease caused
by hypertension; 2) chronic kidney disease, including chronic kidney disease caused by hypertension; 3) chronic respiratory disease; 4)
chronic liver disease; 5) diabetes; 6) cancers with direct immunosuppression; 7) cancers without direct immunosuppression, but with possible
immunosuppression caused by treatment; 8) HIV/AIDS; 9) tuberculosis (excluding latent infections); 10) chronic neurological disorders; and 11)
sickle cell disorders.
Out-of-pocket
spending as %
of current health
expenditure (2017)
Proportion of
population at
increased risk and
high risk of severe
COVID-19 (2020)
70
60
50
40
30
20
10
0
India
Mexico
Russian Federation
China
Indonesia
Republic of Korea
Brazil
Italy
Australia
Turkey
United Kingdom
European Union
Saudia Arabia*
Argentina
Canada
Japan
Germany
United States of America
France
South Africa
Figure 3: Regional health protection gaps (insurance premium equivalents as a share of GDP, in %)
Source: The Geneva Association (based on Swiss Re 2019)
Emerging Asia
Advanced Asia
Advanced Europe
Latin America and the Caribbean
U.S./Canada
10.60.2 0.80.40 1.61.2 1.4
13
An Investigation into the Insurability of Pandemic Risk
Figure 2 reveals, for example, that Indians’ massive
vulnerability to health protection gaps is somewhat
offset by a low level of increased risk of severe cases of
COVID-19. Russians look particularly vulnerable with an
OOPS share of 40% compounded by a 32% risk of severe
COVID-19. It is however important to note some of the
challenges inherent in analysing this type of data. While
the U.S. appears to be in a relatively strong position when
looking at OOPS, it is clear that other variables, such
as macro-economic conditions, can quickly lead to a
deterioration in healthcare (as discussed in Case Study 1).
Figure 3 offers a different perspective on the relevance
of health protection gaps in a number of major regions.
It is based on the premium equivalents of underlying
gaps in sums assured, or protection available from public
and private schemes versus protection needed, i.e. total
healthcare expenditure. The figures understate the true
extent of health protection gaps as they do not consider
the so-called 'treatment gap' (i.e. required healthcare
services not accessed because of a lack of availability or
affordability). The figures also require an estimation of
what level of OOPS on health is stressful for households,
which depends on a country’s development status. In
advanced economies, for example, a larger share of OOPS
is part of co-insurance and deductibles (Swiss Re 2019).
Premium-based health protection gaps as a share of
GDP are, as expected, most pronounced in emerging
11 COVID-19 has led to a sharp drop in spending on other conditions, with non-urgent care cancelled and patients avoiding hospitals and clinics.
However, EIU 2020 expect spending on non-coronavirus care to recover in 2021, also driven by the expected availability of effective vaccines and
treatments.
markets, especially in Asia. In advanced Europe, the gap
is smallest, at about one-fifth the level calculated for
emerging Asia (Figure 3). In absolute premium equivalents,
‘Emerging Asia’ and the U.S./Canada exhibit the largest
health protection gaps, at USD 278 billion and 95 billion,
respectively (Swiss Re 2019).
In general, closing these protection gaps through private-
sector insurance solutions looks much more realistic than
addressing pandemic BI risk, especially as pandemic health
risk is insurable, in principle (The Geneva Association 2019
and section 3.3.2 of this report).
From a macro perspective, COVID-19 is expected to
further exacerbate health protection gaps across the
globe. Even though healthcare spending globally is
forecast to fall in 2020 (EIU 2020),
11
the hit to national
incomes (as measured by Gross Domestic Product (GDP))
is almost certain to be significantly more severe. Therefore,
the share of healthcare expenditure in total GDP is
set to further increase in 2020. Assuming a constant
relationship between total healthcare expenditure and
OOPS, health protection gaps will become even more
acute, especially in emerging countries like India where
healthcare expenditure in 2020 is projected to increase by
5% (in local currency) but GDP is forecast to contract by
at least 3% (EIU 2020).
14
www.genevaassociation.org
12 In principle, while affordability could still be a challenge, a system that requires an employer to provide cover for a period after redundancy and/or
allows the individual to take over the cover in some form could help mitigate this.
Figure 4: Regional mortality protection gaps (insurance premium equivalents as a share of GDP)
Source: The Geneva Association (based on Swiss Re 2019)
Emerging Asia
Advanced Asia
Advanced Europe
Latin America and the Caribbean
U.S./Canada
0.50.30.1 0.40.20 0.6 0.7
Based on past hospitalisations for pneumonia and other respiratory illnesses, tens of millions of Americans could
face significant out-of-pocket medical expenses for COVID-19 hospitalisations, despite the fact that many insurers
have waived cost-sharing requirements. Researchers from the Johns Hopkins Bloomberg School of Public Health
analysed out-of-pocket costs for pneumonia and other respiratory illness hospitalisations from January 2016 through
August 2019 as a potential indicator of likely COVID-19 costs. The study found that these out-of-pocket costs
were particularly high for ‘consumer-directed health plans’ which account for about 60% of employer-sponsored
health insurance plans in the U.S. Those plans typically feature lower premiums but higher deductibles compared to
standard plans, and under their terms, insurers are not required to adhere to the cost-sharing waivers.
The study found that average OOPS for the 2016–2019 research period for these respiratory hospitalisations was
about USD 2,000 for patients with consumer-directed plans (Eisenberg et al. 2020). Another study estimates
average COVID-19-related OOPS across all employer-sponsored plans at more than USD 1,300 (Cox et al. 2020).
Those Americans who have lost or will lose employer-sponsored health insurance as a result of pandemic-related
unemployment will suffer even more financial stress,
12
with hospitalisation costs exceeding USD 20,000 for
pneumonia patients with complications and more than USD 80,000 for patients with the most serious respiratory
conditions that require ventilator support (Cox et al. 2020). More than 10 million people are estimated to lose
their employer-sponsored health insurance plans between April and December 2020 (Banthin et al. 2020).
This challenge comes on top of the well-documented fact that even before COVID-19, an estimated 87 million
U.S. adults (aged 19–64), or 45%, were inadequately insured for health (including those 23 million not insured at
all) [Collins et al. 2019].
Case Study 1: COVID-19-induced health protection gaps in the U.S.
15
An Investigation into the Insurability of Pandemic Risk
Having said this, there are COVID-19-related factors
which could actually mitigate health protection gaps. The
pandemic has catalysed the digital provision of health
services which could alleviate shortfalls associated with
access and affordability (APX/Porsche Consulting 2020).
In addition, heightened awareness of health risks and
the potential benefits of insurance could generally boost
demand for health insurance (McKinsey 2020b).
The accelerated provision of digital
health services catalysed by COVID-19
could actually mitigate health
protection gaps by addressing obstacles
to access and affordability.
2.2.3. The mortality protection gap
13
The mortality protection gap can be defined as the
difference between the amount needed to substitute
a household’s future income in the event of the main
breadwinner’s death, and the existing resources available
to repay outstanding debts and maintain the living
standards of surviving household members. Resources
available include the household’s existing financial assets,
benefits from life insurance policies and social security
payments. The mortality protection gap describes the
portion of the deceased’s regular income that cannot be
replaced by these existing resources (Swiss Re 2020b).
The pandemic is likely to have widened these gaps. Sharply
rising unemployment and eroding asset valuations reduce
available household resources, exacerbating the shortfalls.
Also, given the fiscal emergency in most countries,
the availability of social security payments is likely to
decrease.
Figure 4 offers an illustration of various regions’
vulnerability to the mortality protection gap, which is
measured in life insurance premium equivalents (Swiss Re
2019) as a share of GDP. It is most relevant in emerging
Asia, followed by Latin America and the Caribbean. It is
lowest in the U.S. and Canada.
14
For all regions, mortality
13 In addition to the heightened risk of premature death, life and health insurers are likely to be affected by longer-term changes to morbidity
patterns as a result of COVID-19.
14 Note, however, that the notional mortality protection gap in the U.S. amounts to a staggering USD 25 trillion, defined in absolute terms as the
difference between the amount needed to substitute a household’s future income in the event of the main breadwinner’s death and the existing
resources available to repay outstanding debts and maintain the living standards of surviving household members (Swiss Re 2018).
15 In absolute premium equivalents, mortality protection gaps amount to USD 129, 78 and 58 billion for Emerging Asia, Advanced Europe and the
U.S./Canada, respectively (Swiss Re 2019).
16 There are no recent comprehensive data capturing mortality protection gaps by country. Based on Swiss Re 2013, Peru’s mortality protection
gap (as defined above, i.e. in terms of the absolute notional shortfall) exceeds the country’s GDP by a factor of 2.3. The respective multiple for
Indonesia is 1.8. (Swiss Re 2020).
protection gaps are smaller than the respective shortfalls
in healthcare (see Figure 3).
15
As argued in section 3.3.2 of this report, COVID-19
pandemic mortality risk is insurable. Therefore, private-
sector insurance capital and expertise seem to be well
equipped to narrow such protection gaps.
Establishing the link to COVID-19, Figure 5 shows the level
of excess mortality for a number of countries, capturing
the period from March to mid-July 2020. Excess mortality
indicates the number of people who die from any cause
in a given region and period compared with the recent
historical average. Many Western countries, and a handful
of other nations and regions, publish such data regularly
(see FT 2020).
Figures 4 and 5 suggest that especially for Latin America,
massive regional mortality protection shortfalls are
significantly compounded by COVID-19 induced excess
deaths.
16
Figure 5: Excess mortality in % (March to mid-July 2020)
Source: FT 2020
Peru
Ecuador
Spain
Chile
UK
Italy
Belgium
France
Netherlands
Switzerland
Sweden
US
Brazil
Portugal
Austria
Denmark
Norway
Germany
Iceland
Sout Africa
7525 100500
16
www.genevaassociation.org
A pandemic creates mortality and morbidity risks that affect individuals. What affects
businesses and companies is primarily the handling of the pandemic by governments.
To prevent propagation, public authorities implement measures, such as lockdown
and stay-at-home orders, that restrict freedom of movement and the ability to work.
Such measures have major economic repercussions, simultaneously reducing supply
and contracting demand on a global scale and across a wide range of economic
sectors. This risk is the focus of the following sections.
For insurers to underwrite all of the economic losses
resulting from pandemic risk would impose a material
solvency risk on the industry and create a potential
threat to broader financial stability.
There is a broad consensus, including among governments and regulators, that it
would be ruinous for insurers to underwrite all of the economic losses resulting
from pandemic risk. As discussed in section 2.2.1., the mismatch between the
insurance sector’s current levels of capital, on the one hand, and the probability and
exposure levels of pandemic risk, on the other, would impose a material solvency
risk on the industry, as well as create a potential financial stability threat.
3.1. The concept of insurability
The insurability debate gained traction in the 1980s, based on this observation:
‘The insurance industry as a whole is increasingly confronted with risks where for
reasons of principle and capacity doubts arise as to whether they can and should
be covered. This increase of risks at the limits of insurability is due to growing
social and accumulation problems, advancing technology and concentration
of values, increased complexity and exposure of numerous risks’. (Berliner 1985).
These observations have to be seen in the context of the liability crisis in the
U.S. when, in response to the excesses of the U.S. tort system, some
re/insurers stopped insuring liability risks and the cost of cover in the U.S.
increased dramatically across all sectors as a result (Gollier 1997).
17
17 In March 1986, the front page of Time Magazine read, ‘Sorry, America. Your Insurance Has Been
Canceled’.
3. Getting the basics
right: The insurability
of pandemic risk
17
An Investigation into the Insurability of Pandemic Risk
In general, insurance markets tend to respond adversely to
major catastrophe events. Insurers may reevaluate their
estimates of the probability and severity of loss, restrict
the supply of capacity and raise the price of the (limited)
coverage they are willing to offer (Cummings 2006).
Such responses have been observed, for example, after
Hurricane Andrew in 1992, the Northridge Earthquake in
1994 and the World Trade Center terrorist attack.
18
Risks that can be insured need not be
‘legislated’; uninsurable risks, however,
have to be dealt with by nation states
(Stahel 2003).
After massive loss events in particular, uninsurability
implies that a prospective policyholder cannot buy the
coverage one reasonably needs to manage the adverse
consequences of damage resulting from an uncertain
occurrence. Specifically, this could mean three things.
First, the insurance product is not available. Second,
the insurance product is available, but the coverage
offered is insufficient. Third, the insurance product is
not affordable to certain groups because of its price
(Holsboer 1995).
Against this backdrop, the concept of insurability pivots
on ‘the ‘natural borderline’ between the market economy
and nation states: risks that can be insured need not be
‘legislated’; uninsurable risks, however, have to be dealt
with by nation states’ (Stahel 2003). This borderline
ultimately defines ‘the division of labour’ (Giarini 1995)
in risk taking between the private insurance sector and
the public sector.
3.2. The criteria of insurability
The insurability of risks is not an exact science. There
are no objective attributes which unambiguously
define a certain risk as ‘insurable’ or not. ‘Limits to
insurability cannot be defined, but only analysed’
(Berliner 1985). As a matter of fact, risks are insurable
if an insurer and an insurance buyer reach an
agreement about a specific insurance coverage and its
price, including a common understanding of what is
18 Given the vital role in the economy, major post-disaster fluctuations in the availability and price of coverage generally lead to pressure for
government intervention in insurance markets. This will be discussed in The Geneva Association (2020).
19 This is known as the ‘law of large numbers’, i.e. the larger the number of mutually independent risks in a risk pool, the lower the variance of losses
per risk (Bernstein 1996).
20 Moral hazard occurs when individuals or businesses have an incentive to increase their exposure to risk because they do not have to bear the full
costs of that behaviour, e.g. as a result of taking out insurance. Adverse selection describes a mechanism by which individuals or businesses choose
whether or not to buy insurance based on information not available to their insurer. See the seminal work by Arrow (1963).
insured and what not. From the insurer's perspective,
any decision to offer coverage also depends on
(partially) subjective elements such as the company’s
strategic objectives, risk assessment, risk aversion and
risk-taking capacity (determined by available equity
and reinsurance capacity) (Karten 1997).
From a more theoretical perspective, Berliner 1982, in a
seminal publication, introduced a simple, yet rigorous and
comprehensive set of criteria of insurability. This approach
still shapes the academic discourse on insurability and
continues to be frequently used by practitioners to analyse
insurance markets and products. For example, Berliner’s
set of criteria has been widely applied to climate insurance,
cyber insurance and microinsurance (Biener and Eling
2012; Biener et al. 2015; Charpentier 2018; Kunreuther and
Michel-Kerjan 2004). Ultimately, these criteria ‘can (…) be
interpreted as dimensions of insurability which have to be
gone through by the professional risk carrier individually
like a checklist when assessing the insurability of a risk’
(Berliner 1985). A risk is uninsurable for a professional
carrier if at least one criterion is not satisfied.
Berliner’s framework is three-pronged, consisting
of actuarial, market and societal conditions for
insurability. The first actuarial condition requires that
risks are random and independent (i.e. accidental and
unintentional in nature) so that loss probabilities are
reliably estimable within reasonable confidence limits.
Events that are highly correlated expose insurers to
systemic risk which cannot be diversified away through
risk selection and portfolio building. Second, maximum
possible losses per event must be manageable from
the insurer’s solvency point of view, i.e. they must not
be financially ruinous. Third, average loss amounts
per event must be moderate and, with a growing
number of mutually independent risks in the insurance
pool, converge towards expected losses, allowing for
acceptable and decreasing safety loadings.
19
Fourth,
actuarial insurability necessitates a sufficiently large
number of independent exposure units (policyholders)
and loss events per annum. The size of the risk pools
has to be adequate so that insurers can calculate
loss probabilities. Fifth, insurability from an actuarial
perspective requires the absence of severe information
asymmetries (i.e. moral hazard and adverse selection),
or, at least, the possibility to mitigate them through
contract design and underwriting, for example.
20
18
www.genevaassociation.org
Conditions for insurability include that
maximum possible losses per event
must be manageable from the insurer’s
solvency point of view. They must not
be financially ruinous.
In addition to the actuarial dimension, Berliner (1982)
establishes two market-related insurability criteria. First,
insurance premiums need to cover the insurer’s cost (e.g.
claims and operating expenses, cost of capital, etc.) and,
at the same time, must be affordable to the insured.
21 22
Second, cover limits imposed by the insurer must be
acceptable to the insured, i.e. not defeat the purpose of
buying insurance.
The third dimension of insurability is the societal one and
proposes two further criteria. First, coverage must be in
accordance with public policy and societal values (e.g.
not promote criminal behaviours) and, second, comply
with the legal and regulatory restrictions governing the
operation of insurance companies and the offering of
coverage.
Figure 6: The fundamental criteria of insurability
Source: The Geneva Association (based on Berliner 1982)
21 If policyholders are risk averse, the pooling of risk makes them better off if transaction costs are low and the risks are not (fully) stochastically
dependent (Mossin 1968). In the case of stochastically dependent risks (such as pandemics) the benefits of pooling the small diversifiable risk part
may be offset by transaction costs associated with special pandemic risk schemes. This will be further explored in The Geneva Association (2020).
22 The risk loadings for pandemics are expected to be massive because of the large loss variance associated with individual events, the positive
correlation of the individual risk units in the pool and the negative correlation between pandemic risk and the insurer’s investment portfolio. These
factors are likely to translate into insurance rates where even very risk-averse policyholders would prefer to be not insured (Gründl and Schmeiser
2002). The cost of capital associated with pandemic risk will be further discussed in The Geneva Association (2020).
When exploring insurability, it is important to note
that the limits derived from Berliner’s criteria are not
set in stone. Progress in risk modelling driven by digital
technology, advanced analytics and the increased
availability of large amounts of data is a key to expanding
the boundaries of insurability. It enables insurers ‘to
more accurately quantify probabilities and underwrite
previously difficult-to-insure risks’ (Swiss Re 2017). This
will be further explored in The Geneva Association (2020).
3.3. Limits to insuring pandemic
risk – A comparative and holistic view
It is not difficult to intuitively understand the limits to
managing pandemic risk. The word ‘pandemic’ originates
in the ancient Greek (‘pan’ means all and ‘demos’ means
people). A pandemic outbreak spreads worldwide, or
at least across large regions. As witnessed during the
COVID-19 lockdown periods, pandemics have the
potential to paralyse entire countries and economies,
causing significant and simultaneous damage to virtually
all individuals and businesses. This enormous correlation
and aggregation of risks further highlights the enormous
challenges of insuring these.
3.3.1. Pandemic BI risk
Due to its systemic characteristics, the pandemics
insurability discussion pivots around commercial P&C
business and BI in particular. Pandemic-induced property
and business continuity risk is unique given its potential to
impact virtually all policyholders simultaneously, over an
extended period of time (OECD 2020a). The fundamental
mechanism of risk pooling and redistribution – spreading
the losses of the few among the many unaffected by
disaster – no longer works in the presence of systemic
risk where the destabilising effects of a pandemic ripple
through the entire economy. The simultaneous ‘losses of
the many’ cannot be diversified and mutualised across
risk pools (Van Hulle 2020; Hartwig and Gordon 2020a;
Richter and Wilson 2020).
No legal
restrictions for
coverage
Independent
and predictable
loss exposures
Manageable
maximum
possible loss
Coverage
consistent with
societal values
SOCIETAL
MARKET
ACTUARIAL
Moderate
average loss
per event
Acceptable
cover limits
Large number
of exposure
units
Acceptable
and affordable
premiums
No severe
information
asymmetries
19
An Investigation into the Insurability of Pandemic Risk
The fundamental mechanism of risk
pooling and redistribution – spreading
the losses of the few among the many
unaffected by disaster – does not work
with a systemic risk like a pandemic,
where the destabilising effects ripple
through the entire economy.
A second peculiar feature is the endogenous and political
nature of the risk which is almost entirely driven by
governments’ decisions taken before, during and after
the pandemic. Its economic severity may vary greatly
depending on:
23
• Level of preparedness of the country when the
pandemic occurs, notably in terms of availability of
masks and medical equipment (e.g. hand sanitiser,
ventilators etc.): The higher the level of preparedness,
the lower the need for implementing long lockdown
and stay-at-home orders across the board that are the
costliest in economic terms.
• Timing in terms of adopting and implementing
measures to contain the pandemic: The higher the
level of ‘denial’ from public authorities regarding the
actual risk posed by the pandemic at the onset, the
higher the ultimate economic cost to handle it.
• Decisions as to which businesses/sectors can continue
to operate, fully or partly, and which businesses/
sectors must be fully shut down.
• Timing in terms of relaxing lockdown measures: This
is to some extent a political decision; even more since
there is no clear, predictable time limit on a pandemic.
• Economic and fiscal measures adopted by
governments to dampen the economic impact of the
crisis on companies (e.g. furloughs, short-term hours,
reductions in social charges, etc.).
24
23 Special thanks to Guillaume Ominetti (SCOR SE) for contributing this thought.
24 If governments learn from this crisis and develop a ‘play book’ of responses for future pandemics, the scale of the economic loss could be
significantly reduced.
Hence there is a myriad of parameters that are driven,
or that can be changed or influenced, by governments’
actions and that will determine to a large extent the
magnitude of the economic burden of a pandemic crisis.
Covering this risk through private insurance would create
obvious moral hazard issues. Public authorities, which
decide the ways and means to handle the pandemic
and are accountable for the management of the crisis,
would ultimately not bear the full economic costs
of the decisions they take and could succumb to the
temptation to misuse private-sector capital.
A third distinguishing feature of pandemic risk is the
fact that it is very difficult to model and measure the
economic losses that are specifically linked to the
handling of a pandemic by public authorities.
Covering pandemic risk through private
insurance would create moral hazard
issues for public authorities, who would
ultimately not bear the full economic
costs of the decisions they take.
Table 1 comprehensively explores the insurability of
pandemic business continuity risk on the basis of Berliner’s
previously introduced criteria (see Figure 6).
20
www.genevaassociation.org
25 For example, changes in legislation which were unknown at the time of risk assessment and pricing or broader changes to the economic and social
environment such as the systematic increase in life expectancies.
Table 1: Criteria of insurability of pandemic business continuity risk
Highly problematic
Problematic
Less problematic
Source: The Geneva Association and University of St. Gallen, Institute of Insurance Economics
Insurability criteria Comments Assessment
1
Randomness and
independence
of loss occurrence
Losses are neither random nor independent
• Policy decisions to lock entire economies are deliberate and intentional. This means that
loss amounts and risk loadings cannot be set
• There are no historical data for the policy responses witnessed during COVID-19
• The strong interrelations among individual risks render efficient risk pooling impossible
2
Maximum possible
loss
The maximum possible loss is not manageable for the insurer
• The uncontrollable aggregation of losses could be ruinous to the risk pool
3
Average loss per
event (severity)
It is very difficult to keep the average loss amount per event at a moderate level
• The average loss for pandemic risk needs to be managed to an accepted level by
cover limits and exclusions, as adopted after previous pandemics
• In light of current political discussions and stakeholder expectations, the broader
acceptability of cover limits post COVID-19 is questionable
4
Exposure units
The number of independently exposed policyholders (exposure units) is too small
• As the economy as a whole is affected simultaneously by a pandemic, insurers
cannot build risk pools that are large enough and that diversify the losses. The law
of large numbers does not work
5
Information
asymmetries
Information asymmetries limit insurability
• Insurers are likely to face higher demand from exposed sectors (adverse selection)
and have to expect less risk-conscious behaviours (moral hazard)
• The mitigation potential (e.g. through contract wordings) is limited
6
Insurance premiums
Insurance premiums are not economically viable
• As pandemics threaten most, if not all, members of the risk pool at the same time,
the probability of loss (in addition to severity) is very high
7
Cover limits
Cover limits present challenges of complexity
• Non-physical trigger definitions create complexity (compared with clearly
describable property damage events) which can be problematic for both the
insurer and the insured
8
Public policy
Pandemic risk coverage should be in the public interest
• Issues could arise from certain government interventions (e.g. compulsory
insurance requirements)
9
Legal restrictions
Pandemic risk coverage should be compliant with existing legal and regulatory restrictions
• There might be a ‘risk of change’
25
or a ‘warlike’ scenario of the public sector
‘taking over’ and rewriting the rules that underpinned pricing and risk assessment
during ‘peace times’
21
An Investigation into the Insurability of Pandemic Risk
3.3.2. Pandemic life and health risk
This paper so far pivoted around P&C exposures
and business interruption in particular. Prior to the
unprecedented experience of wide-ranging lockdown
measures, the most obvious and probable source of
major insured pandemic losses was (term) life insurance
(individual and group policies) (CRO Forum 2007).
In contrast with P&C policies, there are generally no
exclusions for pandemics or other common causes for
extreme mortality events such as terrorist attacks or
natural disasters (NAIC 2020; Kraut and Richter 2015).
26 The reasoning for permanent life insurance which offers both death benefits and a savings portion is slightly more nuanced. It offers guaranteed,
not contingent, benefits, and as such the timing element to loss is more relevant than the contingency of loss. Considering the key issue is when
the loss will occur, not if it will occur, life insurers price in some level of volatility in the mortality and may also hold some extra capital for
pandemic stress scenarios.
Unlike the P&C exposures of
pandemic risk, such as business
interruption, life exposures, including
excess mortality, can be modelled.
There are therefore generally no
policy exclusions for pandemics and
other extreme mortality events.
The two main reasons behind the different treatment
of P&C and (term) life exposures by insurers is data
availability and the ability to model exposures.
26
For
life insurers, excess mortality (mortality above what
would normally be expected over a specific span of
time) associated with diseases is well-documented and
Politicians all over the world have compared COVID-19 to a warlike challenge. From an insurance perspective,
too, there are analogies between wars and pandemics. Both represent fundamentally cataclysmic, correlated and
incalculable risks which insurance as a risk transfer mechanism was never intended to cover. Both also come with
harsh restrictions and strong measures that are outside the scope of ordinary law and taken for national security
matters due to force majeure circumstances. These features, which explain why most insurance policies have a
war exclusion clause specifically excluding coverage for acts of war, would also justify that insurance contracts do
not cover economic losses arising from measures taken by public authorities to handle a pandemic.
Similar to a nation-wide economic lockdown due
to pandemic risk, there is no identifiable maximum
possible loss in a war scenario. It could cause a
catastrophic amount of damage that would be likely to
wipe out any insurance company liable to cover such
damages (Fitzsimmons 2004). Another parallel is that
exposures depend on mandatory government actions
which are impossible to model and to predict. As a
result, insurers are unable to calculate premiums for
both risks.
In summary, similar to pandemic business continuity
risk, war risk defies the criteria of insurability introduced before. Losses are neither random nor independent. A
maximum possible loss is impossible to establish. The average loss per event is very difficult to contain through
cover limits and exclusions. The law of large numbers does not work in the absence of a sufficient number of
independent exposure units. Insurance premiums covering the insurer’s cost of capital would be unattractive or
even unaffordable to customers.
Box 2: Analogies between pandemics and wars
Pandemics, like wars, represent
cataclysmic, correlated and
incalculable risks which insurance
contracts were not meant to cover;
not least because the insurance
premiums would be unattractive or
even unaffordable for customers.
22
www.genevaassociation.org
researched by epidemiologists around the world (CRO
Forum 2007).
27
For P&C insurers, however, modeling
pandemic risk is virtually impossible as it is driven as
much by subjective decisions of countless government
officials on national, regional or local levels as by
epidemiology (Hartwig and Gordon 2020b). In addition,
average mortality rates among life policyholders are
usually significantly lower than in the population as a
whole, mainly on the back of medical underwriting in
individual life insurance business (CRO Forum 2007).
Also, due to higher than expected mortality rates,
annuities may offer a natural hedge to the mortality
shock caused by a pandemic (Cox and Lin 2007; GCAE
2006), especially as many life insurers focus on annuities
rather than pure risk (term life) products and, as such,
tend to be more concerned about life expectancy
increases than sudden jumps in mortality.
28
29
27 The effects on group life business might be different as, in general, less underwriting has taken place and the state of health of the individuals in
the portfolio is less well known.
28 The full impact on life insurers may take time to develop. For example, increases in suicides tend to be correlated with extended higher levels of
unemployment.
29 It is important to emphasise that pandemic risk can be covered by life insurers offering broad, needs-based cover against hospitalisation costs or
death as the uncertainty in the pandemic element is small relative to the total coverage. However, this does not mean that they could necessarily
offer specific pandemic cover which might cause a number of elements in Table 2 to move from amber to red.
30 The impact of the Spanish flu on the insurance sector was limited due to the demographics of infections and deaths (Richter and Wilson 2020).
31 A 50% drop in surplus – presumably an existential threat to the industry – and maintaining all other assumptions of the severe pandemic scenario
would require a general population excess mortality of 13% or 4.3 million excess deaths (based on a U.S. population of 330 million in 2019). The
excess death toll from COVID-19 in the U.S. passed the threshold of 200,000 in September 2020.
32 Stracke and Heinen (2006) come to similar conclusions for the German insurance market.
The underwriting losses of life insurers
from COVID-19, while significant, are
expected to remain manageable.
Against this backdrop, the underwriting losses of life
insurers from COVID-19, while significant, are expected
to remain manageable. It is too early to estimate ultimate
losses but the standard pandemic scenario typically used by
insurers, regulators and industry observers may provide an
indication. Based on an excess death ratio of 1.5 per 1,000,
in the U.S., this would imply about 500,000 deaths (larger
than current official estimates of the potential death toll).
In such a scenario, underwriting losses would be about
USD 15 billion, or about 3% of pre-shock industry capital
(Kirti and Mu 2020). Having said this, life insurers’ exposure
could be significantly more severe in the case of a pandemic
involving a more lethal virus (see Box 3).
The Spanish influenza pandemic of 1918–19 is the most virulent on record, with an estimated global death toll of 25–50
million people, or 2–4% of the world’s population at that time (CRO Forum 2007). In the U.S., 675,000 excess deaths
from the flu were recorded between September 1918 and April 1919, corresponding to an excess mortality of 6.5‰
(Toole 2007).
Toole 2007 also analyzes the impact of a severe pandemic on the scale of the Spanish flu on today’s U.S. life insurance
industry. The scenario is based on 1.9 million excess deaths or an excess mortality of 6.5‰ for the general population
and 5‰ for the insured population.
30
It disregards, however, the potential mitigating impact of medical and other
interventions that were unavailable a hundred years ago. Under such an extreme scenario, the U.S. life insurance industry
would lose about 25% of its surplus.
31
Despite this massive hit, only a very small number of U.S. life insurers would face
an increased risk of insolvency. The industry as a whole could weather even a severe pandemic similar to the Spanish flu
(Toole 2007).
32
A wide range of circumstances must be taken into account when estimating the consequences of a disease outbreak
like the Spanish flu for today’s world: Medical care and technology have progressed dramatically, with the availability of
antibiotics, vaccines and anti-viral drugs. In addition, global surveillance and early-warning systems (e.g. by the WHO)
have been established. Furthermore, the socio-economic environment is markedly different from a hundred years ago,
with much improved hygiene conditions, nutrition and health status.
Risk factors to watch, however, include potential shortages in drugs, delays in data exchange among public authorities,
the increased prevalence of chronic diseases and the particular vulnerability of developing countries. Also, today’s degree
of urbanisation and global connectivity need to be taken into consideration (CRO Forum 2007).
Box 3: An extreme pandemic scenario on the scale of the Spanish flu
23
An Investigation into the Insurability of Pandemic Risk
For health insurers, too, pandemic risk poses no
fundamental insurability challenges. Some saturation
effects can be expected in the event of a large-scale
pandemic as healthcare provision capacities are limited,
e.g. a hospital bed can only be allocated once at any given
time (CRO Forum 2007; GCAE 2006). However, a virus
could cause more people to become chronically ill, with
negative consequences for health, long-term care and
occupational disability insurance.
Pandemic risk poses no fundamental
insurability challenges for health
insurers.
So far, the impact of the COVID-19 pandemic on private
health insurance has been relatively modest. In many
countries, for health insurance companies, the decline in
medical care for non-COVID conditions and routine or
elective procedures has more than offset the impact from
COVID-19 claims. However, as for life insurers, a more
severe pandemic could result in major insured losses,
with one estimate putting potential U.S. health insurance
losses at more than USD 30 billion (Dunks 2006).
In summary, pandemic life and health risks are privately
insurable in the context of COVID-19. Excess mortality
risk is modellable based on a wealth of historical data. In
addition, increased mortality risk is (partially) offset by
reduced longevity. For health insurers, there is a ‘natural’
limit to claims given the finite capacity of healthcare
systems and temporarily reduced expenditure for non-
pandemic-related procedures.
Based on the previous sections Table 2 offers a
comparative and illustrative summary assessment of
barriers to insurability of pandemic BI, mortality and
health risks.
As for pandemic BI risks, six of the nine insurability
criteria proposed by Berliner are deemed to present
insurmountable barriers to insurability and, as mentioned
before, a risk is uninsurable for professional risk carriers
if at least one criterion is not satisfied (Berliner 1985).
Most importantly, the almost perfect correlation and
uncontrollable accumulation of losses make the risk
uninsurable.
Barriers to insuring pandemic mortality risk, however, are
generally manageable and of relatively minor relevance,
such as the average loss size per event, the acceptability
and complexity of cover limits, the alignment with
public policy objectives and the compliance with existing
legal frameworks. For health insurers, too, barriers to
insurability appear to be surmountable, with no real
identifiable ‘game stopper’.
3.3.3. Pandemic risk compared to
other catastrophic risks
Table 3 compares the insurability of pandemics, wars,
nuclear accidents, cyber events, terrorist attacks and natural
catastrophes, outlining parallels and differences for these
types of low-frequency/high-severity risks.
Table 2: An illustrative summary assessment of obstacles to insuring pandemic risk
Prohibitively high barrier to insurability Manageable barrier to insurability Insignificant barrier to insurability
Source: The Geneva Association
Insurability criteria
Business
interruption
Mortality Health
Randomness/independence of loss occurrence
Maximum possible loss
Average loss per event
Number of exposure units
Information asymmetries
Insurance premiums
Cover limits
Public policy
Legal restrictions
24
www.genevaassociation.org
33 Projected economic losses from conceivable cyber viruses can be as massive as from natural viruses such COVID-19. In fact, while the economic
damage from lockdown measures was somewhat mitigated by digitally-enabled remote working, there would be no ‘safety net’ to fall back on in
the event of a wide-ranging IT outage.
34 One can argue that different forms of the handling of pandemic risk by public authorities introduce a ‘man-made’ component.
35 There are no fixed definitions of ‘extremely rare’ or ‘extremely severe’ events. Any assessment should reflect the relevant context. For the sake of
this paper, we consider an event as extremely rare if the return period is more than 25 years. We consider severity as extreme if the economic loss
is larger than 1% of GDP.
Table 3: A comparison of various types of extreme events
Source: University of St. Gallen, Institute of Insurance Economics
Type Pandemic War Nuclear Cyber
33
Terror NatCat
Examples
(loss amount)
COVID 19
(Economic loss
might be > 5% of
global GDP)
World War I
and II
(Economic loss
> 15% of global
GDP
Fukushima
(USD 214 billion
economic loss,
USD 36 billion
insured loss)
WannaCry
(USD 8 billion
economic loss,
insured loss
insignificant)
9/11
(economic loss
USD 80 billion,
USD 40 billion
insured loss)
Katrina
(USD 164 billion
economic loss,
USD 76 billion
insured loss)
Time horizon
Months
(potentially
longer)
Years Weeks
(potentially
longer)
Days/weeks Days Days
Risk origin
Natural (with
exceptions,
e.g. bio-logical
weapons)
34
Man-made Man-made Man-made
(with exceptions,
e.g. solar storm)
Man-made Natural
Frequency
35
Extremely rare
(SARS, COVID)
Small wars
common;
significant events
(world wars)
extremely rare
Extremely rare
(Chernobyl,
Fukushima)
Small losses
are of high
frequency (e.g.
data breaches);
Blackout
scenarios are
extremely rare
Low/rare, but not
extremely rare
Low, with
upwards trend
(climate change);
extreme events
rather rare
Severity
Extremely high
(particularly
for business
interruption
losses)
Extremely high High Small losses
common (hacker
attacks, data
breaches);
blackout scenario
high severity
High High, with
upwards trend
(climate change)
Measurability of
risks
Difficult,
especially
when driven
by government
decisions, but
possible to some
extent
(life & health)
Difficult, but
possible to some
extent
Difficult, but
possible to some
extent
Small losses
measurable
(hacker attacks,
data breaches);
Blackout scenario
not measurable
(only in the
context of
scenarios)
Daily terror
measurable,
e.g. political risk
indices; extreme
events difficult to
measure
(e.g. 9/11)
Yes, but only
for the more
regular disaster
scenarios; little
historical data
on most extreme
events
Independence vs.
dependence
of risks
Independence
not given for
global scenarios
Independence
not given for
global events
Local
dependence
Small losses
rather
independent;
blackout scenario
might lead to
closely correlated
losses
Can exhibit
dependencies,
but not as
extreme as
pandemic
Regional
dependencies;
very few events
with global
dependence (e.g.
sun flares and
mass ejections)
Standardisation
of risk
Definition: yes;
coverage: no
Definition: yes;
coverage: no
Definition: yes;
coverage limited
(nuclear pools)
Definition: no;
coverage: no
Definition: yes;
coverage limited
(terror pools)
Definition:
yes; coverage
relatively
standardised
(e.g. CAT bonds)
25
An Investigation into the Insurability of Pandemic Risk
The main difference between these catastrophic risks is
the scope for global diversification. Pandemics are, by
definition, not diversifiable as they occur on a very wide
or even global scale (as opposed to epidemics which are
more locally concentrated). Some other risks, such as
terrorism or natural catastrophes, are diversifiable on a
global level and routinely transferred via re/insurance
or Alternative Risk Transfer (ART) instruments. These
disasters impact a limited number of policyholders for a
limited period of time. Using BI as an example, physical
losses associated with a hurricane are largely a coastal
phenomenon which typically dissipate over a span of
36 In principle, such losses may, at least partially, be globally diversifiable through the capital markets. However, COVID-19 has cast doubt on this
hypothesis given its clear correlation with the capital markets.
hours. In contrast, pandemic-induced BI losses can impact
virtually all policyholders, irrespective of location and
nearly simultaneously, with losses continuing over months
or even years (Hartwig et al. 2020).
In order to further illustrate the unique nature of pandemic
BI risk, we refer to the classification of catastrophic
risks introduced by Cummins (2006) (see Figure 7). As
COVID-19 has shown, business continuity losses caused
by extreme pandemics are neither locally nor globally
independent. Therefore, they are uninsurable.
36
Figure 7: A classification of catastrophic risks
Source: The Geneva Association (based on Cummins 2006)
Geographical scope
Approach to risk transfer
Globally insurable
• Locally dependent, but globally
independent low-frequency/
high-severity risks
• Example: A major hurricane with
globally re/insured losses of
several tens of billions USD
Globally diversifiable
• Very low-frequency/very high-
severity risks
• Example: A 1-in-100 year
California earthquake, with
insured losses exceeding USD
100 billion, partically backed by
capital markets
Globally undiversifiable
• Cataclysmic events with a major
impact on global capital markets
• Example: COVID-19
Locally insurable
• Independent risks with moderate
standards deviation
• Example: Personal automobile
insurance
26
www.genevaassociation.org
4. Conclusions
The uncontrollable aggregation and correlation
elements of pandemic risk defy insurability in the
commercial insurance space.
1. Pandemic risk is a multi-faceted phenomenon. In the commercial insurance
arena, it exhibits systemic elements of uncontrollable aggregation and
correlation which defy insurability. On the other hand, it has non-systemic
characteristics for which private-sector life and health insurance solutions and
the necessary risk appetite and absorption capacity could exist.
2. Based on a thorough analysis of existing research and a review of available
premium and loss data and estimates, we have shown that pandemic BI risk
associated with nationwide government-mandated lockdowns is uninsurable
for the private P&C insurance industry. It violates all essential criteria of
insurability; first and foremost, the criterion of manageable correlation and loss
magnitude. The latter exceeds the risk-taking capacity of global BI insurers by
a factor of more than 100. The amount of capital needed to offer meaningful
and secure insurance coverage would be prohibitively high given the endemic
lack of historical data for this unique combination of (random) viral and (non-
random) political risks (i.e. decisions taken by public authorities), thwarting the
ability of insurers to model the frequency and severity of losses and calculate
27
An Investigation into the Insurability of Pandemic Risk
premiums. Therefore, pandemic risks should be excluded from commercial P&C
insurance policies.
3. In contrast, life- and health-related pandemic risks are generally non-systemic
in nature and covered by most mortality- and morbidity-based policies,
at affordable prices and with wide availability. Life and health insurers are
able to model pandemic risk and price it accordingly. Even though existing
protection gaps may have been exacerbated by the pandemic, they seem to be
addressable based on the risk appetite, capacity and expertise of the private-
sector. With COVID-19, life and health insurers underwent their ‘pandemic
baptism of fire’. Having said this, future pandemics could turn out to be more
aggressive and lethal than COVID-19.
4. Our analysis suggests that public and private-sector decision-makers
should resist the temptation to measure pandemic risk by a single yardstick.
It rather requires a clear differentiation between uninsurable and insurable
variations as well as a careful distinction from other catastrophic risks such
as natural disasters, cyber and terrorism, with different local and global
insurability and diversifiability characteristics. Systemic pandemic economic
and business continuity risk cannot be treated in the same way as other
(catastrophic) risks. Government and society must accept this distinction
when setting their expectations for the role of the insurance industry in
addressing this issue in future.
37
Public and private-sector decision-makers should resist
the temptation to measure pandemic risk by a single
yardstick.
37 To be further explored in The Geneva Association (2020).
28
www.genevaassociation.org
AIR. 2018. The AIR pandemic model. https://www.air-worldwide.com/siteassets/
Publications/ Brochures/documents/AIR-Pandemic-Model
APCIA. 2020. APCIA: Insurance perspective on COVID-19.
APX/Porsche Consulting. 2020. The rise of digital health.
Arrow, K.J. 1963. Uncertainty and the welfare economics of medical care. American
Economic Review 53: 941–973.
Banthin, J., M. Simpson, M. Buettgens, L. Blumberg, and R. Wang. 2020. Changes in
health insurance coverage due to the COVID-19 recession: Preliminary estimates using
microsimulation. The Urban Institute and The Robert Wood Johnson Foundation.
Bernstein, P.L. 1996. Against the gods. Wiley.
Berliner, B. 1982. Limits of insurability of risks. Prentice-Hall.
Berliner, B. 1985. Large risks and limits of insurability. The Geneva Papers on Risk and
Insurance—Issues and Practice 10: 313–329.
Biener, C., and M. Eling. 2012. Insurability in microinsurance markets: An analysis of
problems and potential solutions. The Geneva Papers on Risk and Insurance—Issues
and Practice 37: 77–107.
Biener, C., M. Eling, A. Matt, and J. Wirfs. 2015. Insurability of cyber risks – An
empirical analysis. University of St. Gallen Working Papers on Risk Management and
Insurance No. 151.
Charpentier, A. 2008. Insurability of climate risks. The Geneva Papers on Risk and
Insurance—Issues and Practice 33 (1): 91–109.
Clarke, A., M. Jit, and C. Warren-Gash. 2020. Global, regional, and national
estimates of the population at increased risk of severe COVID-19 due to underlying
health conditions. The Lancet 8 (8): E1003–E1017.
Collins, S.R., H.K. Bhupal, and M.M. Doty. 2019. Health insurance coverage eight
years after the ACA: Fewer uninsured Americans and shorter coverage gaps, but more
underinsured. The Commonwealth Fund.
References
29
An Investigation into the Insurability of Pandemic Risk
Cox, C., R. Kamal, and D. McDermott. 2020. How have
healthcare utilization and spending changed so far during
the coronavirus pandemic? Petersen-KFF Health System
Tracker.
Cox, S., and Y. Lin. 2007. Natural hedging of life and annuity
mortality risk. North American Actuarial Journal 11 (3).
CRO Forum. 2007. Influenza pandemics. Emerging Risks
Initiative – Position Paper.
Cummins, J.D. 2006. Should the government provide
insurance for catastrophes? Federal Reserve Bank of St.
Louis Review 88 (4): 337−379.
Dunks, P.J. 2006. When...not if – The impact of a
catastrophe on private health insurers. Health Watch.
23–27. https://www.soa.org/globalassets/assets/library/
newsletters/health-watch-newsletter/2006/january/
hsn0601.pdf
Economist Intelligence Unit. 2020. COVID-19 – The impact
on healthcare expenditure.
Eisenberg, M., C. Barry, C. Schilling, and A. Kennedy-
Hendricks. 2020. Financial risks for COVID-19-like
respiratory hospitalizations in consumer-directed health
plans. American Journal of Preventive Medicine 59 (3):
445–448.
Financial Times. 2020. Coronavirus tracked: The latest
figures as countries fight COVID-19 resurgence. https://
www.ft.com/content/a2901ce8-5eb7-4633-b89c-
cbdf5b386938
Fitzsimmons, A. 2004. Terrorism – Maximising the
insurability of a catastrophe risk. Annals of Air and Space
Law XXIX: 67.
GCAE. 2006. Actuarial reflections on pandemic risk and its
consequences.
Giarini, O. 1995. Insurability and the economic relevance of
insurance: A historical economic perspective. The Geneva
Papers on Risk and Insurance—Issues and Practice 20:
419–422.
Gollier, C. 1997. About the insurability of catastrophic
risks. The Geneva Papers on Risk and Insurance—Issues and
Practice 22: 177–186.
Gründl, H., and H. Schmeiser. 2002. Pricing double-trigger
reinsurance contracts: Financial versus actuarial approach.
Journal of Risk and Insurance 69 (4): 449–468.
Hartwig, R., and R. Gordon. 2020a. Uninsurability of mass
market business continuity risks from viral pandemics.
Hartwig, R., and R. Gordon. 2020b. Communicable disease
exclusions: Maintaining stability in property casualty
insurance markets amid a global pandemic.
Hartwig, R., G. Niehaus, and J. Qiu. 2020. Insurance for
economic losses caused by pandemics. University of South
Carolina Working Paper.
Holsboer, J.H. 1995. Insurability and uninsurability: An
introduction. The Geneva Papers on Risk and Insurance—
Issues and Practice 20: 407–413.
Jazabkowski, P., R. Bednarek, and K. Chalkias. 2018.
Between state and market: Protection gap entities and
catastrophic risk.
Karten, W. 1997. How to expand the limits to insurability.
The Geneva Papers on Risk and Insurance—Issues and
Practice 22: 515–522.
30
www.genevaassociation.org
Kirti, D., and Y.S. Mu. 2020. The impact of COVID-19 on
insurers. Vox.
Kousky, C., B. Collier, A. Braun, and H. Kunreuther. 2020.
Preserving small business in the aftermath of the coronavirus
catastrophe. http://www.brinknews.com/LcQ
Kraut, G., and A. Richter. 2015. Insurance regulation and
life catastrophe risk: Treatment of life catastrophe risk
under the SCR standard formula of Solvency II and the
necessity of partial internal models. The Geneva Papers on
Risk and Insurance—Issues and Practice 40: 256–278.
Kunreuther, H., and E. Michel-Kerjan. 2004. Insurability
of (mega)-terrorism risk: Challenges and perspectives.
Report prepared for the OECD Task Force on Terrorism
Insurance, Paris.
McKinsey. 2020a. State of property & casualty insurance
2020.
McKinsey. 2020b. What insurers can learn from China’s
continuing COVID-19 recovery.
Mossin, J. 1968. Aspects of rational insurance purchasing.
Journal of Political Economy 76: 533–568.
NAIC. 2020. COVID-19 and insurance.
OECD. 2020a. Responding to the COVID-19 and pandemic
protection gap in insurance.
OECD. 2020b. Initial assessment of insurance coverage and
gaps for tackling COVID-19 impacts.
Poundrik, S. 2011. Disaster risk management in East Asia
and the Pacific. World Bank/Global Facility for Disaster
Reduction and Recovery Issues Paper.
Richter, A., and T. Wilson. 2020. COVID-19: Implications for
insurer risk management and the insurability of pandemic
risk. The Geneva Risk and Insurance Review 45: 171–199.
Stahel, W.R. 2003. The role of insurability and insurance.
The Geneva Papers on Risk and Insurance—Issues and
Practice 28 (3): 374–381.
Stracke, A., and W. Heinen. 2006. SOA – Influenza
pandemic: The impact on an insured lives life insurance
portfolio. https://www.soa.org/library/newsletters/the-
actuary-magazine/2006/june/pub-influenza-the-impact-
on-an-insured-lives-life-insurance-portfolio
Swiss Re. 2013. The mortality protection gap in 2013.
Swiss Re. 2015. The essential guide to reinsurance. https://
www.swissre.com/Library/the-essential-guide-to-
reinsurance.html
Swiss Re 2017. Commercial insurance – Innovating to
expand the scope of insurability. Sigma No. 5.
Swiss Re 2018. Life underinsurance in the US: Bridging the
USD 25 trillion mortality protection gap.
Swiss Re 2019. Indexing resilience – A primer for insurance
markets and economies. Sigma No. 5.
Swiss Re 2020a. World insurance: Riding out the 2020
pandemic storm. Sigma No. 4.
Swiss Re 2020b. Closing Asia’s mortality protection gap.
The Geneva Association. 2018. Understanding and
addressing global insurance protection gaps. Author: Kai-
Uwe Schanz. April.
The Geneva Association. 2019. Healthcare in emerging
markets – Exploring the protection gaps. Author: Kai-Uwe
Schanz. March.
The Geneva Association. 2020. Public and private risk
solutions to pandemic risk. Author: Kai-Uwe Schanz.
Forthcoming.
Toole, J. 2007. Potential impact of influenza pandemic on
the US life insurance industry.
Van Hulle, K. 2020. Pandemics and insurability.
Wagstaff, A., G. Flores, M.-F. Smitz, J. Hsu, K. Chepynoga,
and P. Eozenou. 2018. Progress on impoverishing health
spending in 122 countries: A retrospective observational
study. The Lancet 6 (2): E180–E192.
Willis Towers Watson 2020. Scenario analysis of the
COVID-19 pandemic.
World Bank. 2020. Global economic prospects.
World Health Organization. 2019. Universal health
coverage (UHC). https://www.who.int/news-room/fact-
sheets/detail/universal-health-coverage-(uhc)
iii
An Investigation into the Insurability of Pandemic Risk
This first report in The Geneva Association’s research series on pandemics and insurance
explores, in number terms, the capacities of insurers to absorb pandemic-related costs.
Encouragingly, research findings indicate that pandemics on the scale of, and similarly lethal
to, COVID-19 pose no fundamental insurability challenges for health and life insurers. In the
commercial insurance arena, however, the uncontrollable aggregation and correlation elements
of pandemic risk defy insurability.
The Geneva Association
International Association for the Study of Insurance Economics
Talstrasse 70, CH-8001 Zurich
Tel: +41 44 200 49 00 | Fax: +41 44 200 49 99
