The use of automated scoring is now being
applied to people who have been left out, and that may mean new,
profitable opportunities for property/casualty insurers who write
personal lines.
At press time, an application called LIFT® (the
“Loss Improvement Forecasting Tool”) is being used by a regional
carrier to help assess risks posed by people who do not have credit
scores, and a national carrier is scheduled to launch a pilot
program using LIFT®.
LIFT® provides a supplement to credit-based
insurance scoring that utilizes an applicant’s record of writing
checks and using subprime credit to identify fiscally responsible
individuals likely to be good insurance risks.
“We’ve shown that this data is beneficial to
consumers who don’t use credit,” says Bill Wilson, chief
operating office of Convergence Data, Chicago (www.getlift.com),
the company that developed LIFT®.
“For companies, we can help them segment
people they have no other information on.”
As the LIFT pilot is underway, another Chicago
firm will be completing a “proof of concept” demonstration for a
“pattern recognition system” designed to use non-traditional
underwriting information to score individuals as insurance risks.
The Chicago-based Urban Insurance Partners
Institute (www.uipi.org) is
project manager for the effort to create a system that will search
databases of legal and financial transactions to seek patterns that
suggest individuals may be attractive insurance risks. The project
is supported by the Center for Financial Services Innovation (www.cfsinnovation.com)
At the core of both initiatives is the
accumulation of data on economic behavior not previously used in
underwriting loans and insurance.
“There’s a rapid ramp-up in non-traditional
data,” says Suzanne Reade, UIPI president.
Until recently, companies had little, if any,
information on the buying patterns of lower income people who did
not use traditional credit.
In recent years, however, firms have built
databases of information on payments for rent, utilities, and other
household expenditures, plus information on previously untracked use
of “subprime” lending markets.
PRBC, Annapolis, Md. (www.prbc.com), has
developed a “Bill Payment ScoreSM” (PBSSM) that, like a credit
score, assigns points depending on the type and number of bill
payment accounts a consumer has, and the consumer’s record of
paying them.
“With our service, if people don’t have
traditional credit, they can still show they are fiscally
responsible,” says Michael Nathans, PRBC’s founder and CEO.
“For many people who don’t have a
traditional credit history, it’s a substitute for one,” he says.
“For people who have had credit problems in the past, it’s a way
to show they’ve turned the corner.”
Additional information is also being provided
through relatively new forms of payment used by “unbanked”
people, such as stored value cards.
With a stored value card, a cardholder purchases
the card and then deposits money into it. It then functions like a
debit card, allowing the holder to draw upon the deposited amount.
A stored value card gives the holder access to
the “plastic” that is now essential for renting cars or hotel
rooms or carrying out many other transactions.
Stored value cards also integrate individuals
who once paid bills on a cash-only basis into the electronic
payments system, where their payment behavior can be tracked.
Accumulation of bill payment information is
still in its relative infancy, however, and many
vendors--particularly landlords--do not regularly report payments to
a bureau like PRBC. Therefore, many individuals will not have a
complete or adequate profile on themselves.
In such cases, consumers can take the initiative
to establish their own bill-paying score, says Nathans. In contrast
to the traditional credit bureaus, PRBC allows consumers to report
their own payments, which are then verified through third party
services.
There’s a cost to manual verification, says
Nathans--$18 for verifying rent payments and $8 for verifying other
types of payments--but mortgage lenders frequently pay those
charges.
PRBC’s bill payment score was recently
incorporated into LIFT®, the new scoring model specifically
designed for personal lines insurance underwriting.
LIFT® also incorporates data from TeleCheck,
the world’s leading check verification company, and Teletrack, a
database of records from merchants and creditors, including rental
companies, cable television franchises, personal finance companies,
and subprime lenders.
“The results of TeleCheck’s transaction
evaluation process correlates directly to insurance risk,” reads
the Convergence Data Web site. “As would be expected, individuals
with no bad checks and good risk profiles have lower loss ratios
than those with the presence of bad checks.”
“Every data test so far has proven that this
data is predictive [of insurance loss ratios],” says Wilson.
PRBC and LIFT® pride themselves on being
supplements to traditional credit bureaus in offering consumer
scoring based on models that use new types of data.
Yet, while their criteria is non-traditional,
their methodology is similar to that of the standard credit bureaus,
in that an individual’s payment history is measured and scored
against a predetermined scale.
The pattern recognition system being developed
by UIPI and Pattern Recognition Systems (PRS), Chicago, seeks to
take the use of non-traditional information even further. That
system is being designed to search databases to identify new
patterns of consumer behavior that correlate to insurance risk.
As PRS explains it, companies typically use
regression-based forecasting to determine when to award loans or
insurance coverage. Under regression-based methods, information
about each applicant is entered into an equation, and applicants are
approved or denied depending on how their scores compare with a
criterion value.
According to PRS, pattern recognition systems
harness the enormous power of modern computers to analyze the
interaction between variables and the combination of factors that
significantly predict a customer’s behavior.
“Customer databases become their memory of
past experiences,” PRS writes on its Web site (www.patrec.com).
“When a new situation arises, our systems search that memory and,
using pattern-finding techniques, find examples of similar
situations.”
“In essence, the applicant is processed by
asking, ‘Who have I seen like this before?’ and ‘What type of
credit risks have those people been?’ The process [then] selects a
special subset of prior applicants and determines an appropriate
weighting factor for each one.”
Pattern recognition relies on the ability of a
system to recognize how new variables that may not have been
considered in the past can dramatically change how a consumer is
scored, says Canh Tran, PRS president.
For example, he says, three reports of rejection
for credit may produce a negative profile for an applicant. But, he
adds, if you combine them with reports that a person has just
ordered new checks and has a satisfactory bill payment history, a
new, more complete profile emerges that suggests the person is
starting to build or rebuild a financial profile.
In that case, a small number of credit
rejections is not necessarily predictive of a poor risk for either
lending or insurance, according to Tran.
The ability to discern and analyze different
patterns of consumer behavior is critical to enabling carriers to
write more business in urban areas, he adds.
“Urban markets are more heterogeneous than
suburban markets,” he says. “Urban markets have a wider range of
people and behaviors than you find in the suburbs.”
“People with different behaviors may still
have good success profiles.”
When facing calls to write more business in
urban markets, “insurance companies are faced with customer
segments for which they have incomplete information,” says Reade
at the UIPI.
“Companies are reluctant to embrace new market
segments if they cannot rate and price accurately,” she says. “This
pattern recognition initiative is designed to increase the appetite
for a wider array of risk.”
Companies that want to learn more or participate
in the project can contact Suzanne Reade at 773-880-8780.
Giving credit 
