Best Insurance

A volcano erupts in Iceland and cancels air travel throughout
Europe and across the Atlantic for several days in 2010,
causing an estimated $2 billion in business interruption
losses. An earthquake, tsunami, and nuclear disaster strike
Japan in 2011, causing an estimated $5 billion in global
business interruption losses. These and many more recent
natural disasters, most notably the 2011 Thai floods and
last autumn’s Hurricane Sandy, have resulted in significant
supply chain disruptions.
Natural catastrophe (or “nat cat”) modeling effectively
began in the aftermath of Hurricane Andrew, which struck
Southern Florida in 1992. Since that time, both insurance
companies and their insureds have had the benefit of risk
models to guide decision-making on properties at-risk of
nat cats such as hurricanes, earthquakes and tornadoes.
Over the past twenty years, nat cat models have continued
to develop and become institutionalized at insurance
companies the world over.
Similar to the growth of “nat cat” models following
Hurricane Andrew, we believe that the insurance industry
will start to focus on modeling contingent business
interruption after the significant losses following the storms
referenced above.
There is currently a limited amount of insurance available
to cover this risk. According to the third edition of the
Dictionary of Insurance Terms, the “Contingent Business
Interruption Form [provides] coverage for loss in the net
earnings of a business if a supplier business, subcontractor,
key customer, or manufacturer doing business with the
insured business cannot continue to operate because of
damage or destruction. For example, a specialty hot dog
stand noted for its great buns cannot sell its product if the
bakery supplier of hot dog buns burns down. In instances
where a business is heavily dependent on its suppliers or
subcontractors, interruption of the flow of material from the
supplier usually results in a substantial loss” (p. 100).
One of the reasons for limited contingent business
interruption (or “CBI”) capacity is that insurance companies
do not have the benefit of a risk model that can inform CBI
underwriting, pricing and risk management. The industry
has recognized the need for such a model for some time but,
despite several notable attempts, no-one has yet been able
to produce one that adequately quantifies the dynamics of
CBI risk. The reasons for this are fairly straightforward:
First, nat cat models are anchored to and focused on specific
regional perils such as Florida wind events or a California
earthquake, etc., while a CBI model would have to be far
broader in scope. For example, the supply chain effects of
the Thai floods were far broader than many industrial firms
and insurance companies originally anticipated.
Second, the exposures that nat cat models quantify are
generally very well defined; for example, the location of a
property to a peril (often segmented down to the zip code
level), as well as the property’s physical dimensions and the
quality of its construction, are fairly easy to discern, and
thus can be subjected to intensive engineering analysis. In
contrast, CBI exposures are less well-defined and therefore
obfuscate classical analytical techniques such as statistical
and engineering analyses.
This is not to suggest that nat cat models are in any way
perfect. They are not. All models are simplifications of
reality that are designed to facilitate decision-making.
As a result of that simplification, they are subject to error
(known practically as “model miss”). The fact that nat cat
models sometimes miss the mark is effectively why property
underwriting is a business rather than a science. And in that
business, nat models help to facilitate a much more informed
view of property underwriting and risk management than
would be available without them

Because they are simplifications of
reality, models are not perfect. However,
they do help facilitate a much more
informed view of underwriting and risk
management than would be available
without them.