118 KS Klein
If ubiquitous underinsurance is not primarily driven either by demand
surge or homeowner choice, then that suggests a new hypothesis: the likely,
primary cause of ubiquitous, unintended underinsurance is the estimating
tool. Coverage limits are not plucked out of the sky. Rather, insurers estimate rebuild costs using tools in the United States called ‘component cost
estimators’ and in Australia called ‘elemental estimating calculators.’
These estimating tools essentially are big data analytics at their finest.
Data sets of millions of construction projects and price lists are broken
down into individual labour and materials line items, sorted by location
and date. These data sets are updated at least quarterly, sometimes more
frequently, to capture localised construction cost trends. An insured house
is identified as to its elements or components, down to screws and bolts, and
then an estimate is built up for the price of building that precise house in
that precise location. And, as alluded to above, because of the prevalence of
demand surge, other data streams also are involved.
What appears to be happening is that the cost estimating tools informing coverage limits are more often than not simply under-estimating reconstruction costs. Of some frustration is that this is a testable hypothesis, but
it has not been tested. There is a moment in time when one knows with
precision the cost of building (or rebuilding) a home. That is the moment
that construction is actually completed on that home. And at that moment,
the home usually is insured. So, at that moment, the estimating tools are
deployed. The insurer may not know the actual construction costs of every
newly constructed home the insurer insures. But at a minimum, if the home
is one that was rebuilt after a total loss, and the rebuild was adjusted under
the same insurer that now will insure the new home, the insurer has or had
access to the actual cost of building that specific home, which, in turn,
means the insurer has a data point allowing it to compare at the same point
in time the actual cost of constructing a specific home and the estimated
cost to build the same home. A large insurer has access to lots of these data
points. Which means an insurer can construct a mature data set from which
an insurer can know for its insureds and its estimating tools the frequency
of inaccuracy, and the average depth of inaccuracy.
All of this raises two questions: (1) Why would an insurer be open to
selling less insurance than a homeowner wishes to buy? (2) Why would an
insurer be incurious about how well its estimating tools are working? Each
answer is intertwined with the other.
The legal landscape of insurance is complicated, in part because there
is not consensus on what precisely insurance is. Is insurance a quasi- public utility to have maximum risk spreading, or is it a variant of a personal
security product or is it an ordinary contract? Is the relationship between
insurer and insured arms-length, adhesive, or fiduciary? Should insurance
markets be free market structures, lightly regulated, or highly regulated? Is
insurance a luxury or a necessity? For an insurer, a complex legal and conceptual landscape creates a set of market incentives that may at first glance
If ubiquitous underinsurance is not primarily driven either by demand
surge or homeowner choice, then that suggests a new hypothesis: the likely,
primary cause of ubiquitous, unintended underinsurance is the estimating
tool. Coverage limits are not plucked out of the sky. Rather, insurers estimate rebuild costs using tools in the United States called ‘component cost
estimators’ and in Australia called ‘elemental estimating calculators.’
These estimating tools essentially are big data analytics at their finest.
Data sets of millions of construction projects and price lists are broken
down into individual labour and materials line items, sorted by location
and date. These data sets are updated at least quarterly, sometimes more
frequently, to capture localised construction cost trends. An insured house
is identified as to its elements or components, down to screws and bolts, and
then an estimate is built up for the price of building that precise house in
that precise location. And, as alluded to above, because of the prevalence of
demand surge, other data streams also are involved.
What appears to be happening is that the cost estimating tools informing coverage limits are more often than not simply under-estimating reconstruction costs. Of some frustration is that this is a testable hypothesis, but
it has not been tested. There is a moment in time when one knows with
precision the cost of building (or rebuilding) a home. That is the moment
that construction is actually completed on that home. And at that moment,
the home usually is insured. So, at that moment, the estimating tools are
deployed. The insurer may not know the actual construction costs of every
newly constructed home the insurer insures. But at a minimum, if the home
is one that was rebuilt after a total loss, and the rebuild was adjusted under
the same insurer that now will insure the new home, the insurer has or had
access to the actual cost of building that specific home, which, in turn,
means the insurer has a data point allowing it to compare at the same point
in time the actual cost of constructing a specific home and the estimated
cost to build the same home. A large insurer has access to lots of these data
points. Which means an insurer can construct a mature data set from which
an insurer can know for its insureds and its estimating tools the frequency
of inaccuracy, and the average depth of inaccuracy.
All of this raises two questions: (1) Why would an insurer be open to
selling less insurance than a homeowner wishes to buy? (2) Why would an
insurer be incurious about how well its estimating tools are working? Each
answer is intertwined with the other.
The legal landscape of insurance is complicated, in part because there
is not consensus on what precisely insurance is. Is insurance a quasi- public utility to have maximum risk spreading, or is it a variant of a personal
security product or is it an ordinary contract? Is the relationship between
insurer and insured arms-length, adhesive, or fiduciary? Should insurance
markets be free market structures, lightly regulated, or highly regulated? Is
insurance a luxury or a necessity? For an insurer, a complex legal and conceptual landscape creates a set of market incentives that may at first glance