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