Policy & Economic Research Council

By Patrick Walker

NCRA’s 2009 National Conference November 13th, 2009

Alternative Data Initiative Furnishing, Integrating, and Using Utility and Telecom Payment Information

2

About PERC

Founded in 2002

Non-partisan, non-profit policy institute devoted to research, public education, and outreach on public and economic policy matters

Dedicated to finding sustainable information- and market-based solutions for development challenges worldwide

3

Supporters of Alternative Data Initiative

Acxiom Corporation Ashoka Foundation Bank of America Brookings Institution CFED CFSI Consumer Data Industry

Association Equifax

Experian GE Money Hispanic National

Mortgage Association HSBC JPMorgan Chase Lexis Nexis SAS Institute Trans Union LLC

4

Access to credit crucial for asset formation (homes and small businesses) and information for risk assessment is needed to access credit

Why Care? (Consumer) 35 to 54 million Americans “unscorable” Primarily low income, immigrants, elderly, and ethnic minorities

Why Care? (Commercial) 90% of businesses are small 12.5% fail within 3 years often for reasons of credit access. Black and Latino owners face greater loan denial rates (SBA)

The Need for Alternative Data

5

Credit Crisis and Crunch Stricter lending guidelines and underwriting standards

By lenders By regulators

Reg. Z Greater need for solid evidence and measures of credit risk

and capacity Reduction of credit extended

Bank analyst estimates a reduction of $2.7 Trillion on cards alone by end of 2010

Stricter information needs for mortgage lending Consumers without credit risk and capacity information

excluded from lower cost mainstream financial services

The Need for Alternative Data

6

We identified and tested alternative data (energy and telecoms) as solution to risk assessment and credit access: ADI I showed that, in principle:

o mainstream lenders interested in using data to reach underservedo bureaus and data repositories were interested in collecting alt-datao furnishers had interest and were curious about reporting

But, chicken vs egg:o lenders not demanding alt data because it was absento data agencies not collecting because not demanded

ADI II designed to show how to move togethero lenders, furnisher, and foundations came together to testo Impact analysis to see who benefits

ADI III showed furnishers costs and benefits of reportingo Survey of utilities, telecoms, and other typical non-reporterso Survey of customerso Furnisher case studies

ADI IV begins major outreach based on learnings from ADI I,II, & III

Alt Data’s Problem and Its Promise for Asset Formation

7

Alternative Data Initiative,Phase 1

What Data Can Make a

Difference?

Released December 2006Freely Available:

www.perc.net

Qualitative Research

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

8

Assessed usefulness along 3 key dimensions

“Cash-like” vs. “Credit-like” (incentive to furnish)

Coverage (reach of data in population)

Concentration (resources needed to reach furnishers)

Why Utilities and Telecoms?

Traditional “credit-like” data

Non-traditional “cash-like” data

Energy

Water

Cable

Auto liabilityinsurance

Tuition

RentalPayments

Child care

Payment cards

Payday loans

Con

sum

ers

Util

izin

g S

ervi

ces

Concentration of Data Furnisher

HighLow

Man

yF

ew

9

Technological barriers to reporting: Complex billing cycles (footprint dependent) Legacy IT systems

Economic barriers: Compliance costs—FCRA data furnisher obligations Customer service costs from lenders scaring customers

substantial

Regulatory barriers: Three states have statutory prohibitions (CA, NJ, OH) Federal prohibitions: Section 22 of the 1996

Telecommunications Act Regulatory uncertainty at the state level

Barriers to Reporting

10

Alternative Data Initiative,Phase 2

How Much of a Difference Can the Data Make?

Released December 2006Freely Available:

www.perc.net

Quantitative Research

11

Risk profile shows promise Score distribution of thin-file sample similar to

general population Nearly 40% of Black unscoreables have scores

above 620 when energy or telecoms data added

10% of unscoreable now scoreable and thereby able to access credit 1 tradeline matters Multiple tradelines are better

Key Findings: Expanded Access &No and Thin File Not High Risk

12

Key Findings--Greater Access for All

Considerable increases in acceptance rates for a given performance level. For utilities, an increase of 6 percentage points for a 6% delinquency level.

Acceptance Rates by Targeted Delinquency Rates

13

Key Findings--Greater Observed Access

Access is not merely hypothetical but seen in the share of the thin-file population for which alt data is reported.

(“Validation sample” = no alt data)

14

Key Findings: Increased Access for Underprivileged Social Segments

Considerable lift for Blacks and Hispanics, Low-Income

Change in Acceptance Rates at 3% Delinquency Target by

0%

5%

10%

15%

20%

25%

All Asian Black Hispanic Other White

Utilities Telecoms

0%

5%

10%

15%

20%

25%

30%

<$20,000 $20,000-$29,999 $30,000-$49,999 $50,000-$99,999 $100,000+

Utilities Telecoms

Race/Ethnicity Income

15

Market Outcomes To Date

Chicken v. Egg Solved: Created Demand Lenders now testing and using alternative data for

underwriting credit Bureaus and CRAs now collecting more alternative data Analytics firms creating more scoring models

SYSTEMIC CHANGE--Pervasive reporting of alternative data could change the way banking

is done in under-served markets.

16

Alternative Data Initiative, Phase 2-3:

Can Alternative Data Harm

Consumers?

Released July 2008

Freely Available: www.perc.net

Quantitative ResearchFollow-up

17

Contra fear that alt data harms: No evidence of credit score deterioration over time

when alt data is included No evidence of alt-data scoreables becoming over-

extended and having declines in credit scores.

Evidence suggests that reporting payment data serves both as a consumer protection and a system wide protection.

Key Findings:Answering Follow-up Questions

18

Original Finding: Little Effect on Scores

Change in Credit Score with the Addition of Utility Payment Data (Statics from March 2005)

But finding is based on static information

19

Key Findings: Picture Remains Over Time

Changes in Score for Consumers with Alternative Data Over a 1-year Period, by Number of Trade Lines (from March 2005 to March 2006)

Scores increased for consumers by a larger degree than they decreased over the year-long period

20

Key Findings: Scores Go Up Over Time with New Accounts

Average Scores Increase with the Age of the Financial Account, for all and for all racial segments

March 2006 Vantage Score for Alt-Data Holders with Financial Accounts, by Age of Account

Group I New Account less

than a year old

Group II New Account 1-3

years old

Group III Alternative and 1

Traditional account over 3 years old

Average 613 637 660

Ethnicity

Asian 664 689 677

Black 573 587 602

Hispanic 627 643 628

Other 640 668 699

White 611 635 674

21

Key Findings: Scores Go Up Over Time with New Accounts

Average Scores Increase with the Age of the Financial Account, for all and for income segments

March 2006 Vantage Score for Alt-Data Holders with Financial Accounts, by Age of Account

Group I New Account less

than a year old

Group II New Account 1-3

years old

Group III Alternative and 1

Traditional account over 3 years old

Average 613 637 660

Income (000) $

<20 596 628 649 20-29 608 637 668 30-49 620 649 676 50-99 642 666 698 >99 697 695 732

22

You Score, You WinThe Consequences Of Giving Credit

Where Credit Is Due

Policy Concerns and Considerations

Quantitative Research Follow-up

23

In wake of mortgage crisis, draft FHA Modernization Act of 2007 § 257 (a):

“[P]ilot program to establish an automated process for providing alternative credit rating information for residences to be insured under this title who have insufficient credit histories for determining their creditworthiness.”

Alternative credit information may include: rent, utilities, and insurance payment histories, “and such other information as the Secretary considers

appropriate”

Proposed projects such as these to salvage homeownership can greatly benefit from opening the alt data tap

Near Term and Future Promise:What’s At Stake in Policy Questions

24

Identified first in ADI I, but feedback shows a consistent lessons: Where statutory barriers exist (direct and

indirect) must be removedo States: CA (telecom), OH (publicly owned), NJ

o Federal: Section 222 of the Telecommunications Act of 1996

Regulatory uncertaintyo More pervasive

o Requires public affirmations

Policy Change and Policy Clarity:Answering Follow-up Questions

25

Lessons from elsewhere: Retirement programs

o Opt-out systems have initial enrollment rate of 90% rising to 98% vs.

o Opt-in systems which have enrollments of 20% rising to 65% Organ donation: European countries where opt-out of

organ donor have higher rates than where citizen opts-ino Austria (opt-out) has 99.9% rate v. Germany (opt-in) which has

12% rateo Sweden (opt-out) has 86% rate v. Denmark (opt-in) which has

4.3%o France (opt-out) has 99.9% rate v. U.K. (opt-in) which has 17.2%

rate

Default Settings Matter

Policy Options:Opt-In v. Opt-Out

26

Alternative Data Initiative, Phase 3

Will the Data be Reported? Is there a

business case?

Released March 2009

Freely Available: www.infopolicy.org

Research on the Supply Side

27

ADI: Phase 3

Utility and Telecom survey Average firm benefits of fully reporting are several

times the costs Benefits include fewer delinquencies, decline in

accounts in arrears, and improved cash-flow Costs include IT upgrades, conducting new

processes, and new customer service demands All firms reported benefits > costs 10% of firms in survey currently fully report to a

bureau 89% currently refer delinquencies to a collections

agency

28

ADI: Phase 3

Customer Survey Half of customers surveyed would be more likely to

prioritize a utility or telecom payment if they knew it was fully reported

No income or ethnic group appears to be more informed as to which payments are fully reported and which are not No evidence that any of the subgroups

disproportionately use the fact that some payments are not fully reported to ease cash-flow problems

29

ADI: Phase 3

Case Studies of firms that fully report: The DTE Case

DTE Energy Serves over 3.5 million customers Includes Detroit Edison (electric) and MichCon

(natural gas) August 2006 began fully reporting to credit

bureaus

8.1% of DTE’s customers either gained a credit file or a credit score due to DTE’s decision to fully report customer payment data

Most customers that were scoreable prior to August 2006 had little or no change in their credit score when the DTE tradeline was added

30

Alternative Data Initiative, Taking Care of the Supply Side:

Inducing Reporting

ADI: Phase 4

31

ADI: Phase 4

Policymaker and Government Track Respond to questions and concerns from lawmakers,

regulators, and media Federal level State level

PUCs, NARUC NCSL, NGA, NAAG

Furnisher Education & Outreach Exhort full reporting Strategic events (USTA, RMA, EEI, AGA, CTIA) Stand alone events

Policy & Economic Research Council

100 Europa Drive, Suite 403

Chapel Hill, NC 27517

www.perc.net

Phone: (919) 338-2798