The "Studies"
Update:
The University
of Texas study that insurers say validates
the use of insurance credit scoring:
The only problem with this is that the data in the study supports the
argument that insurance credit scoring is redundant and unnecessary.
This note:
"The insurers comprising the top 70 percent of the market (in
descending order, starting with the largest companies) were then asked
to provide a random sample of new or renewing automobile policies from
the first quarter of 1998 (January 1, 1998 through March 31, 1998).
This examination period was chosen chiefly for two reasons. First,
most of the insurers from whom data were requested were not using
credit scoring at that time in rate-making or underwriting decisions, which
meant that premium data collected were not affected by credit history",
and these results:
"Chart 3 shows the distribution of scores for
policies from the standard
market insurers participating in the study. Credit scores for the
standard market
(mean=733.0) are significantly higher than the credit scores for the
non-standard market (mean=657.7)."
Clearly demonstrate that prior to insurance credit scoring,
traditional underwriting and rating factors achieved the desired
results!
Birny Birnbaum of the Center for Economic Justice takes it even
further to demonstrate exactly how the UT study fails.
"The report by the University of Texas Bureau of
Business Research on the relationship
between credit scoring and expected insurance losses suffers from
problems so severe that the
authors’ conclusions are neither credible nor reliable." His
analysis.
These studies really do exist. However
you will discover why the insurance industry is dedicated to keeping the
public from this information. .
AIA
Statement to the NAIC for Hearing on Credit-Based Insurance Scores
This paper cites several
"studies":
The
Impact of Personal Credit history on Loss Performance in Personal Lines
by James E. Monaghan, ACAS, MAAA
Commonwealth of Virginia
1999 Report of the State Corporation Commission's Bureau of Insurance on
the Use of Credit Reports in Underwriting to the State Commerce and
Labor Committee of the General Assembly of Virginia (not
available online)
The Use of Credit Related
Related Information in the Underwriting of homeowners and Automobile
Insurance in the State of Arkansas, 1996.(not available online)
The paper also refers to several studies
conducted by Fair Isaac and Tillinghast-Towers Perrin. The majority of
the "evidence" supporting the correlation claim comes almost
exclusively from these credit scoring vendors. Why these
"studies" by scoring model vendors should be discounted, click
here.
A large portion of the AIA
statement is based on The
Impact of Personal Credit history on Loss Performance in Personal Lines
by James E. Monaghan, ACAS, MAAA and incidentally, this was the only
study I found that contained any numbers at all.
This study involves
the correlation of claims filed and credit report characteristics.
They took the information of 170,000 policy holders that filed claims
within a three year period and pulled their credit reports. They found
that consumers with certain credit characteristics filed more claims
and more expensive claims. These characteristics included the number
and types of accounts a consumer holds, the number of late payments,
duration of late payments, adverse public records (e.g. civil
judgments, tax liens, bankruptcies, etc.), and the number of inquires
a consumer had on the report.
As
all accidents are not reported, there may be a major flaw in the
industry's interpretation of this study. It seems logical that what
these studies really reflect is the amount of disposable income (cash,
savings accounts, etc.) an insurance consumer had at a time of an
incident. This does not mean that they were more likely to have an
incident, just that their circumstances at the time of the incident
dictated their actions. The insurance industry rewards consumers for
settling claims without them and penalizes them with surcharges if they
do. A consumer with less disposable income, no matter what his credit
rating, will need the compensation.
In the 1998 Joint Economic
Committee Study by the 105th Congress:
"Families at the bottom end of
the income scale have very little disposable income, and every dollar
spent on premiums for auto insurance represents money that could be
spent on other essentials, such as food, shelter and health care. As
previously indicated, owning a car can be extremely important in terms
of finding and holding down a job."4.
In addition, and something the insurance
industry fails to address is that in this study:
"Data was assembled for
new policies written in a given policy year and the subsequent
three year loss. This data cannot show if long-term customers have
the same differences in loss performance. There is currently no
data publicly available that shows such relationships." (page
101 of The
Impact of Personal Credit History on Loss Performance in Personal
Lines by James E. Monaghan)
There is no evidence for the
insurance industry to use this practice on existing customers, yet
they are! Not only that, they are raising premiums and not
renewing policies for existing consumers, in many cases based solely
on this "Insurance Credit Score"! This places otherwise
"good" drivers into company subsidiaries and substandard
companies designed for the "highest" risk drivers.
Despite insurance industry claims
that the Commonwealth
of Virginia 1999 Report of the State Corporation Commission's
Bureau of Insurance on the Use of Credit Reports in Underwriting
to the State Commerce and Labor Committee of the General
Assembly of Virginia (not available online) provides concrete
evidence that claims and credit correlate, it is only
referred to as a reference.
This study had
gained importance for it's insurance scores vs. income analysis.
Learn more here.
The third study by the Arkansas
Insurance Department is really little more than a survey of
insurance companies in Arkansas. The report goes on to summarize
that:
" This means credit
underwriting appears to be very subjective and the concerns of
this office remain about the possibility of unfairly
discriminatory practices in the use of the credit reports and
financially related information. Without having clear indices of
what constitutes poor, good, or acceptable credit, the industry is
in a difficult position to defend itself in the use of these
terms."
FYI - Some of the studies
were conducted by Arthur Anderson. Are you going to trust these?
Other topics discussed in this AIA
Statement include:
- Some Benefits of Credit-Based
Insurance Scores.
- High Correlation to Risk.
- Questions of Causation (You must
see this! Notice how 'morale' hazard becomes a 'moral' hazard!)
- Resources on Link between
Insurance Risk and Credit-Based Insurance Scores.
- Laws, Regulation and Market
Imperatives - Consumer Protections in Place.
- State Law Prohibiting Unfair
Discrimination.
- FCRA (Fair Credit Reporting Act)
Ensuring Proper Communication with Consumers.
- Quality of Credit Bureau
Information.
- Extraordinary Circumstances
(Yeah, right!)
- Individual Rating and Risk
Classification Issues
- Consumer Education
The industry
answers "Why is there a correlation"?
When the question, "Why is there
a correlation?" is introduced, it is disregarded as
unnecessary. The industry claims that as long as they have the numbers,
they don't need causation. The AIA claims:
"Credit appears to measure
personal responsibility, which extends from the financial realm into
other areas of an individuals life such as his risk under a personal
automobile insurance policy. Credit may also be an indicator of
whether an insurance claim will be filed, whether an insurance claim
will be inflated, and whether fraud will be committed. For these and
other reason, the relevance of credit as a factor in underwriting and
rating personal lines of insurance seems to be established.
Furthermore, credit is used heavily in our society, which is one of
the reasons it can hardly be considered obscure".
As to the correlation of the
numbers, Brenda J. Cude of the University of Georgia,
explains:
"A statistical correlation is
just a number. You put two sets of numbers in the computer and you
come out with this third number that's a statistical correlation and
it tells you if there is a relationship between thee two numbers and
if so, how strong it is and it's positive meaning if one number goes
up the other goes up too, or negative meaning if one goes up and the
other goes down. It's a simple number, it just tells you that these
two numbers are related. Well, I could show you a correlation
that says something called the SOB index perfectly selects who wins
the Super Bowl and the SOB index is the number of performances of the
symphony, opera, and ballet in each team's metropolitan area. So
according to that all we need to do is schedule a few more performance
of the Atlanta Symphony Orchestra and the Falcon's record will go up.
I really don't think that is going to happen. These two numbers are
correlated; it doesn't mean that they mean anything."
Texas Insurance Commissioner Jose
Montemayor:
"There is a high correlation
between a rooster crowing and the sun rising, but that does not mean
that one drives the other." How true.
What do the
critics say?
David "Birny" Birnbaum
of The Center for Economic Justice is the insurance consumer's strongest
advocate against the use of Insurance Credit Scoring. He has provided
valuable information through his testimony and a follow up paper: Extended
Comments of Birny Birnbaum for
the Florida Insurance Commissioner’s Task
Force on Credit Scoring,
January 23, 2002. This excerpt is from the final page of his paper and
sums it up:
"Insurers claim that any
prohibition or limitation of their use of credit scoring will cause
insurers to write less business."
"These claims should be viewed with great skepticism. The logic
of the insurer argument is that any restrictions on their underwriting
or rating practices will limit insurance availability. Yet, in 1994,
after being sued by the National Fair Housing Alliance for their use
of age and value of the home in underwriting property insurance, State
Farm and Allstate agreed to stop using these guidelines – and
admitted that they would write more business in poor and minority
communities as a result."
"The insurer claims that
credit scoring benefits more consumers than it harms is also
suspect."
"First, the insurers provided no
substantial evidence to indicate whether most consumers get a rate
increase or a rate decrease from insurers' use of credit scoring.
Second, and more important, what happens today is not necessarily what
will happen tomorrow. Insurers are in a position, when they introduce
credit scoring, to cause the majority of consumers to get a lower
rate. In fact, insurers have a strong interest in limiting the initial
impact of credit scoring on their policyholders. However, there is
nothing to prevent future iterations of rate plans and credit scoring
to cause most consumers to pay higher premiums that in the absence of
credit scoring."
"The bottom line is that there is
no answer to the question of whether a majority of consumers benefit
or lose from the use of credit scoring because the use of credit
scoring simply redistributes premium among consumers and that
redistribution can change over time. In contrast, we can say that the
introduction of a rating factor that provides a discount for engaging
in some loss prevention activity (theft prevention device, wind
resistant construction, etc.) lowers premiums for some consumers
without raising them for others so that consumers as a whole benefit
and claim costs drop.
Finally, the insurers will engage in
the scare tactic of claiming that any prohibition or limitation on
credit scoring will cause good drivers to subsidize bad drivers. This
is a red herring argument that seek to obfuscate the fact that credit
scoring simply redistributes premium from some consumers to others and
does so in a manner that seems arbitrary and penalizes the poorest
members of society -- for being poor."
Other topics discussed in this paper
include:
- Why there are serious questions about
the alleged correlation between consumer credit information and risk
of loss.
- How consumer credit characteristics
are related to consumer income and age.
- Why insurers are so committed to
using credit scoring.
- Why simple correlation is not
sufficient to justify the use of credit scoring for underwriting or
rating.
- Why credit scoring violates
principles of risk classification.
- Why insurers’ use of credit scoring
is inherently unfair to many consumers.
- Why the failure of insurance
regulators and/or state legislatures to limit insurers’ use of
credit scoring for underwriting and rating is effectively
deregulation of private passenger automobile and residential
property insurance rates.
Do you think "Insurance Credit
Scoring" is wrong? Get Involved! |
