CAS Ratemaking and Product Management Spring 2012 March

CAS Ratemaking and Product Management Spring 2012 – March 20 Credit Scoring John Wilson, Director Analytics 1

What are Credit-based Insurance Scores? q A numeric representation of relative insurance claim risk based on consumer credit details q Most are “bowling” scores (higher scores indicate lower risk) but some are “golf” scores q An objective, consistent, and effective tool used with other risk factors (ex. prior claims) to better estimate future claims risk and cost 2

What Data is Considered? q How long you’ve had credit established q The numbers and types of accounts you hold q Indications of recent activity, such as inquiries and newly opened accounts q The degree of utilization on accounts, and q Payment history, including timeliness as well as adverse public records or collection items 3

What’s Not Considered? q Factors such as gender, marital status, age, address, occupation, or education q Inquiries made for account review, promotional, or insurance or consumer disclosure purposes q Multiple inquiries for auto finance or mortgage finance when made within a 30 day period q Collection items designated as medical on the credit report 4

How Do They Differ From Lending Scores? q Insurance Models ≠ Financial Models q Insurance Models are developed on historical insurance losses q Financial Models are developed on bad debts or 90+ delinquencies q Insurance Scores rank order claim frequency or a similar metric q Financial Scores rank order the odds of credit “bads” q. Insurance scores are not as dependent on derogatory behavior q Financial scores are more sensitive to credit delinquencies 5

How is Their Use Regulated? q Lexis. Nexis is a Consumer Reporting Agency under the federal FCRA and state analogues q We provide disclosure and facilitate dispute resolution q Because insurance is regulated at the state level, we conform to specific state statutes, guidelines, and regulations (ex. NCOIL) q We work with state insurance departments to explain our models and try to gain approval for their use q We are not an insurance company; we don’t set rates or provide advisory services 6

Insurance Credit Score Trends • We track two different populations • Activity in the Market (drawn from our NCF transactions), and • A large retro sample (proxy for existing business) • What changes are we seeing? 7 7

National Attract Auto Score Trends – New Business 825 800 775 750 Northeast 725 Midwest 700 South 675 650 West 625 US Total 600 575 2007 • • 2008 2009 2010 2011 All regions are seeing small, gradual score improvements Western and Southern regions improved more from 2009 to 2010 8

Attribute Trends – Existing Business 0, 85 0, 80 07 1 n n 1 Ja Ju n 1 Ju n 0 Ju n Ja Overall • • 1 0, 85 n 1 0, 90 Ju 0, 90 0 0, 95 n 1 0, 95 Ju 1, 00 9 1, 00 n 0 1, 05 Ju 1, 05 n 0 1, 10 Ju 1, 10 0 1, 15 9 1, 15 8 1, 20 07 1, 20 8 DEROG PUBLIC RECORDS AA 30 SCORE Overall Attract Auto 3. 0 Scores on this large retro sample get slightly better each year Adverse Public Records are up overall, but annual increase is relatively small 9 9

Attribute Trends – Existing Business AVG ACCOUNT AGE (MONTHS) 0, 85 0, 40 0, 80 Overall n 1 Ju n 0 Ju n Ja Overall • Inquiry counts have dropped dramatically, while average trade age has increased • Revolving utilization has steadied after an initial drop; bank / consumer changes 10 10 1 0, 90 n 1 0, 50 Ju 0, 90 0 1, 00 n 1 0, 60 Ju 0, 95 9 1, 10 n 0 0, 70 Ju 1, 00 8 1, 20 n 0 0, 80 Ju 1, 05 n 1, 30 1 0, 90 0 1, 10 9 1, 40 8 1, 00 07 1, 15 1 1, 50 0 1, 10 9 1, 20 8 1, 60 07 1, 20 07 AVG DEBT BURDEN Ja # INQUIRIES