|
9. Impact of Data Quality
Insurance executives use the output of catastrophe models
to support decisions that can have major financial
implications for their business. Like all modeling, the
quality of the output is only as good as the quality of the
input data.The key input into catastrophe models is data
about the exposure in an insurer’s portfolio.The importance
of the quality of this data cannot be underestimated,
as the completeness, accuracy, and resolution of
portfolio information has a direct impact on the magnitude
of losses output by the model. While models have
the ability to infer information to help compensate for
missing data, these assumptions may not always provide
the most accurate assessment of a company’s risk.
Better quality data does not necessarily lead to lower
modeled losses. However, it will result in more accurate
model results for better decision making. This chapter
identifies not only the key portfolio information needed,
but also the benefits of capturing the appropriate data.
9.1 Catastrophic Events
It often takes a critical catastrophic event to prompt
changes in practices for the insurance industry. For property
insurers, the key events were Hurricane Andrew in
1992 ($16.5 billion in insured loss (1)) and the Northridge
Earthquake in 1994 ($12.5 billion in insured loss (2)).These
and other large losses of the early 1990s forced many
unprepared insurance companies and reinsurers out of
business. Available capital reserves for large catastrophe
losses were depleted and there were many changes in the
industry, including a major consolidation, the arrival of
new capital and the single line catastrophe reinsurer,
exploration of alternative risk transfer mechanisms, and
increasing emphasis on regulation, security, and rating.
9.1.1 The Need for Data
For property insurers, the ability to assess risk after these
key benchmark events required significant reengineering
of their data capture processes. Aggregated data by state
or county was no longer adequate and more precise location
information with construction and occupancy detail
was required to analyze a company’s exposures accurately.
For some companies, the process of improving the
quality of the data involved changing back-office systems
and front-office data collection, and it took several years to accomplish.Today, it is standard practice for property
insurers to capture all important location information at
a high level of resolution, and to manage and demonstrate
capital adequacy through analytical models.
9.1.2 The World Trade Center Event
In 2001, the WTC attack resulted in more than $40 billion (3)
of insured loss, and has caused many companies to
reevaluate their practices of managing multi-line exposure,
including workers compensation and other areas of
potential human loss. Losses in many individual lines,
particularly workers compensation, proved unprecedented.
The potential impact of mass casualty events has
registered with management, and many companies are
looking for ways of assessing their risk from catastrophe
loss to insurance lines such as life and health.
Workers compensation writers have embarked upon
a similar process to improve the quality of the data they
collect following the WTC event. Prior to September
11, the industry standard for data capture to fulfill regulatory
reporting requirements was state aggregates by
rating class. However, as many of the workers compensation
insurers recognized the value of monitoring and
managing catastrophe risk through the property side of
their business, the process of improving data quality has
been rapid, driven by senior management and reinsurers
that have made data collection an important priority.
9.1.3 Assessing Life and Health Catastrophe Risk
If life and health insurance companies want to understand
and manage their catastrophe risk, they will need
to address the types of information they collect, and the
resolution of that data.
The assessment of catastrophe risk requires understanding
three main areas:
 How much is insured? It is important for a company to
understand how much exposure (either number of people
or limits) they have at risk. Understanding the level
of exposure is the prerequisite for modeling. Exposure
analysis should consider who is insured (how many, their
occupations) and the coverage provided (limits,
deductibles, and other financial information).
Where are the insured people located? The magnitude
of catastrophe risk varies dramatically with geographic location. For a peril capable of destroying a building and
its occupants, knowing which buildings you have insured
personnel inside is important. Capturing an accurate
street address for the company or individual will help
identify where that exposure is located. As this report
has illustrated, businesses that insure people face the
additional problem that their insureds move, and they
may need to capture both a work and home address to
better understand their exposure.
In what type of structure are the insureds located? The
strength and resilience of the buildings that insureds
occupy affects their risk. The characteristics of a structure
affects its chances of collapsing in an earthquake or
having its occupants injured in other types of events.
Building type information such as construction material
and height is useful to develop an accurate risk assessment.
While models are capable of capturing additional data
fields, the information previously listed is the most important,
and the potential impact of each is discussed below.
________________________________________________________________
(1) In 1992 dollars, $19.6 billion in 2001 inflation adjusted dollars.
(2) In 1994 dollars, $14.9 billion in 2001 inflation adjusted dollars.
(3) Insurance Information Institute estimate (September 6, 2002).
|
