Credit Scoring An Overview

1
Credit Scoring An Overview

• Jimmy Wright, FCAS
• Lawrence Cranor, ACAS

2
History

• Fair,Isaac introduced models late 80s
• Progressive as first big player (early 90s)
• Numerous Studies Confirm Correlation between
  scores and loss costs (Tillinghast)
• Initial Concerns (NAIC White Paper, 1996)
• Legislative Reactions
• VA Board of Insurance (1999)

3
What Score Is/Is NOT NCOIL Guidelines

• Insurers cannot deny, cancel, nonrenew, or base
  a policyholders renewal rate on credit data
  without considering other factors
• Consumers must be informed that their credit
  histories may be checked in connection with an
  application for insurance
• Insurers must give consumers the specific reason
  or reasons that an adverse action was taken
  against them based on credit
• Insurers may not use factors like age, income,
  and address in calculating credit-based insurance
  scores, and certain inquiries and items found on
  a consumers credit report

4
Potential Hot-Buttons

• Income/Wealth
• Disputed Items
• Insurance/Promotional Inquiries
• Current Issue Katrina-related financial
  distress (Louisiana)
• Changes due to new bankruptcy laws

5
Missouri DOI (2004)

• DOI head with an axe to grind . . .
• Zip-Code based analysis
• Concluded disparate impact on minority
  neighborhoods
• Flawed statistical analysis ignored
  distribution of scores not sufficiently refined
  for meaningful conclusions
• No impact

6
Texas

• McCombs School of Business study done at request
  of Lt. Governor (March 2003)
• DOI study done at request of legislature (2005
  with follow-up in 2006)
• Conclude BOTH explanatory value AND disparate
  impact, ultimately supports use

7
Florida

• Best example of hostile environment
• Currently legislated requirement to prove NO
  disparate impact on protected classes
• Still being litigated . . .

8
Oregon

• November 2006 Ballot Initiative to ban use of
  credit scoring in insurance pricing
• Received great deal of attention nationally from
  insurance industry (3.7 million spent to
  advertise against initiative)
• Ban overwhelmingly rejected (only 35 voted to
  approve)

9
Predictive Value

• Highly Valuable for Auto
• Across all coverages
• Even UM/UIM, Comprehensive
• Extent of impact varies widely
• For Property
• not predictive of weather losses
• highly predictive of fire/theft losses
• scores of owner predictive of tenant-occupied
  experience

10
Now Commonly Utilized

• Still not well-understood, but
• Utilized in some fashion by most major insurers
  (pricing and/or underwriting)
• Understood by much of the public as a source of
  discounts
• Adverse selection should be a real concern for
  those not yet utilizing . . .

11
Practical Considerations

• Peculiarities of Operating Jurisdiction
• Infrastructure Issues
• Political/Legal Concerns
• Disruption of Existing Book of Business

12
Variations by State

• Importance of knowing specifics of the local
  environment cannot be over-emphasized
• Guidelines vary widely, change constantly
• Seek guidance and step carefully, State
  Departments of Insurance often made very
  sensitive to these issues by legislators,
  Consumer Groups

13
Infrastructure

• Vendor relationship critical (proprietary models
  difficult to develop independently)
• Transaction costs vary
• Ability to connect may require great deal of
  development of your systems
• On-going need to revisit these scores for
  existing book of business (common NCOIL
  provision re-evaluate every three years)

14
Political/Legal Concerns

• Potential for changes in allowable use of credit
  quite real
• Potential to invite litigation
• Handling of claims for damages from those with
  errors on credit reports (recently-settled
  Allstate case illustrative)

15
Disruption of Existing Book

• Transitions always a challenge . . .
• Adverse Action??
• Predictive power most important for unseasoned
  business your own experience with a long-term
  insured may be more valuable . . .
• An individuals score WILL change over time . . .

16
Open DiscussionQuestions?