October 14, 2011

Discriminatory Effects of

Credit Scoring on

Communities of Color

Lisa Rice and Deidre Swesnik

Prepared for the Symposium on Credit Scoring and Credit Reporting

Sponsored by Suffolk University Law School and National Consumer Law Center

June 6 and 7, 2012

Discriminatory Effects of Credit Scoring / page 1 National Fair Housing Alliance

About the National Fair Housing Alliance

Founded in 1988 and headquartered in Washington, DC, the National Fair Housing Alliance is a

consortium of more than 220 private, non-profit fair housing organizations, state and local civil

rights agencies, and individuals from throughout the United States. Through comprehensive

education, advocacy and enforcement programs, NFHA protects and promotes equal access to

apartments, houses, mortgage loans and insurance policies for all residents of the nation.

National Fair Housing Alliance

1101 Vermont Avenue, NW

Suite 710

Washington, DC 20005

(202) 898-1661

www.nationalfairhousing.org

Discriminatory Effects of Credit Scoring / page 2 National Fair Housing Alliance

TABLE OF CONTENTS

Introduction ................................................................................................................................. 3

I. The Nation’s Dual Credit Market Rooted in Discrimination ........................................ 7

A. Overt Historical Discrimination ..................................................................................... 7

B. Subprime Lending and Its Long-Term Discriminatory Effects ................................ 9

C. The Proliferation of Fringe Lenders in Communities of Color .............................. 12

II. Credit Scoring Has a Discriminatory Impact and Is Not the Best Measure

of Risk ................................................................................................................................... 13

A. Limited Scope, Quality and Transparency of Credit Information ......................... 14

B. Disparate Impact of Credit Scoring Factors ................................................................ 17

C. Existing Credit Scoring Systems Do Not Adequately Predict Risk ....................... 21

D. Risky Loan Products and Unsafe Lending Environments – Not Borrowers –

Were Clearly the Culprit................................................................................................. 23

III. Why the Federal Government and Lenders Have an Obligation to Change the

System ................................................................................................................................... 24

IV. Policy and Enforcement Solutions to Improve Credit Scoring Systems ................. 25

Conclusion .................................................................................................................................. 29

Discriminatory Effects of Credit Scoring / page 3 National Fair Housing Alliance

Introduction

Our current credit scoring systems have a disparate impact on people and communities of color.

These systems are rooted in our long history of housing discrimination and the dual credit

market that resulted from it. Moreover, many credit scoring mechanisms include factors that

do not just assess the risk characteristics of the borrower; they also reflect the riskiness of the

environment in which a consumer is utilizing credit as well as the riskiness of the types of

product a consumer uses.

Until only a few decades ago, communities and people of color explicitly were not permitted

access to low-cost government and other mainstream loans. In the 1930s the Home Owners

Loan Corporation and at least through the 1950s the Federal Housing Administration and the

Veterans Administration used blatantly discriminatory rating systems and “Residential Security

Maps” to deem communities of color high-risk. Banks, real estate agents, appraisers, and others

also perpetuated redlining and segregation in the housing markets. The passage of the federal

Fair Housing Act of 1968 improved conditions, but even up until the mid 1970s, federal

regulatory agencies refused to acknowledge their enforcement responsibilities under the Act. It

was not until civil rights groups sued the agencies that the federal government began to collect

information on the mortgage lending practices of the institutions it regulated, and to establish

and implement fair lending examination procedures.

Because of this history of racial discrimination, segregated neighborhoods formed and people of

color had limited access to affordable, sustainable credit. Instead of accessing mainstream

credit available to white borrowers and white neighborhoods, people of color were relegated

to using fringe lenders and paying much more than they would otherwise have had to. While

segregation and housing discrimination have abated somewhat, we still live in an

extraordinarily segregated society.

Access to credit is still often based on where we live rather

than our individual ability to repay that credit. As this paper will explore, people of color were

steered to subprime loans even when they qualified for prime loans, contributing to the fact that

the foreclosure crisis has hit communities of color even worse than it has hit the rest of the

country.

Credit scoring systems in use today were built upon and continue to rely upon the very dual

credit market that continues to discriminate against people of color. For example, these systems

penalize borrowers for using the type of credit disproportionately used by borrowers of color.

Even fair lending defense attorneys who represent major banks readily admit that credit scoring

has a differential impact on people of color. In a recent article, attorneys at K&L Gates assert

that, “even the most basic lending standards, such as credit scores and [loan-to-value]

For example, according to 2010 Census numbers, 65 percent of individuals in large metropolitan areas

still live in areas of high segregation between whites and African-Americans. Gurian, Craig, “New maps

show segregation alive and well,” Remapping Debate, April 20, 2011.

Discriminatory Effects of Credit Scoring / page 4 National Fair Housing Alliance

requirements, ‘impact’ racial and ethnic groups differently.”

While there has been some

discussion recently by the industry about the existence of the disparate impact theory under the

Fair Housing Act and other long-established laws, disparate impact has been recognized by all

eleven circuit courts that have ruled on the matter as a legally acceptable means by which

parties can assert claims under the Fair Housing Act.

As we all look for solutions to the foreclosure crisis, lenders, regulatory agencies and policy-

makers promote tighter underwriting standards as a solution to improving the quality of loan

performance and strengthening the economy. What they mean in part, however, is requiring

higher credit scores for the best and most affordable products. This, of course, places the focus

for improving loan performance on borrowers. But many studies and analyses have

demonstrated that inappropriate loan products and their components were key factors driving

the subprime crisis. Factors including product type, presence of a yield spread premium,

distribution channel, inflated appraisals, and prepayment penalties helped significantly to

predict whether a loan would fail. Even major credit repositories and credit scoring companies,

including Vantage Score and FICO, admit that credit scores declined in predictive value leading

up to and during the foreclosure crisis. So why are some looking to increased reliance on credit

scoring as a way of originating well-performing mortgages and solving the crisis?

The use of credit scoring and its disparate impact go far beyond the lending sector, affecting

access to many other financial products and services. Credit and other scoring mechanisms are

being used by employers to evaluate job applicants, insurers to determine auto, life and

homeowners insurance, and landlords to screen tenants. Credit scoring modelers and

companies are finding even more creative ways to broaden the use of these systems. A recent

proposal in the state of Texas would use credit scores to determine utility rates.

Credit scores

Hancock, Paul; Brody, Melanie Hibbs; McDonough, Jr., David G; Malpass, Melissa S.; Shinohara, Tori

K., “Supreme Court vs. HUD: The Race to Decide ‘Impact or Intent’,” Legal Insight, K&L Gates,

November 17, 2011.

In addition, since the Fair Housing Act was amended in 1988, the U.S. Department of Housing and

Urban Development has acted in administrative proceedings and in other contexts with the full

understanding that disparate impact claims are cognizable under the Act, as has the U.S. Department of

Justice in its actions. Further, the Consumer Financial Protection Bureau recently announced that it

would utilize all tools at its disposal, including the disparate impact theory, to pursue lenders who

discriminate against consumers in violation of the Equal Credit Opportunity Act. The Bureau specifically

stated that it would use the disparate impact theory in bringing actions under ECOA. See

http://www.consumerfinance.gov/pressreleases/consumer-financial-protection-bureau-to-pursue-

discriminatory-lenders/. The Federal Reserve also recognizes disparate impact as a way to prove ECOA

claims.

Stillman, Jim, “Your Credit Score Determines the Availability of Credit . . . and the Cost,” Yahoo! Voices, June

20, 2007.

Discriminatory Effects of Credit Scoring / page 5 National Fair Housing Alliance

are even being used to determine which patients are more likely to take their medication as

prescribed.

The expanded use of scoring mechanisms has caused great consternation among consumer and

civil rights groups as well as policymakers. For example, insurance companies use credit-based

insurance scores to determine pricing. Yet, studies by the Missouri and Texas Departments of

Insurance have found that insurance scoring discriminates against low-income people and

consumers of color because of the racial and economic disparities inherent in scoring

mechanisms.

The Missouri study concluded that a consumer’s race was the single most

predictive factor determining a consumer’s insurance score and, consequently, the consumer’s

insurance premium.

The relationship between insurance credit scores and race is so strong that even though the

Federal Trade Commission (FTC) used data selected by the industry in a 2007 FTC report, it

found that credit scoring discriminates against low-income people and consumers of color, and

that insurance scoring was a proxy for race.

The FTC report also confirms that, despite

growing reliance on credit-based insurance scores, scant evidence exists to prove there is a

causal relationship between a consumer’s score and auto insurance losses. Without the need to

demonstrate such a connection, insurers could theoretically use any arbitrary consumer

characteristic, such as hair color or zodiac sign, that demonstrates a correlation to a specific

outcome, to price insurance products.

This report focuses primarily on the use of credit scores by lenders, not other industries. This

report provides only an abbreviated overview of other critical issues facing consumers when it

comes to credit scoring and reporting. These issues are significant and help to demonstrate the

urgent need to reform this system. For example, credit scoring systems are based on

information obtained from consumer credit reports, even though credit reports are often rife

with errors that are difficult to correct. Credit scoring systems are also a mystery to consumers

because credit scoring companies maintain that their systems are proprietary and cannot be

revealed. These issues are covered in great detail by recent reports by Demos

and the

The FICO Medication Adherence Score will be used by insurers and medical care facilities to identify

patients who will need additional follow up services to insure they take their medication. Parker-Pope,

Tara, “Keeping Score on How You Take Your Medicine,” New York Times, June 20, 2011.

Kabler, Brent, Ph.D. et al, Insurance-Based Credit Scores: Impact on Minority and Low-Income Populations in

Missouri, State of Missouri Department of Insurance, January 2004.

Credit-Based Insurance Scores: Impacts on Consumers of Automobile Insurance, A Report to Congress by the

Federal Trade Commission, July 2007.

Fremsted, Shawn, Traub, Amy, Discrediting America: The Urgent Need to Reform the Nation’s Credit

Reporting Industry, Demos, June 2011.

Discriminatory Effects of Credit Scoring / page 6 National Fair Housing Alliance

Consumer Financial Protection Bureau

and a survey by the Consumer Federation of America

and VantageScore.

Fixing our current credit scoring system is not only a moral imperative consistent with our

national policies and beliefs about fairness and justice; it is also a legal obligation as outlined by

the federal Fair Housing Act and the Equal Credit Opportunity Act. We hope this paper will

assist with the dialogue at this conference as well as our national dialogue on how to move

forward and out of our financial and foreclosure crises.

This paper begins in Section I with a discussion of the historical discrimination that led to our

dual credit market, including subprime lending and the foreclosure crisis. Section II contains a

detailed analysis of why credit scoring has a discriminatory impact. Section III discusses the

legal obligation that the federal government and the financial industry have to promote fair

housing. Section IV offers recommendations for how to fix our broken approach to credit

scoring.

“The impact of differences between consumer- and creditor-purchased credit scores,” Report to

Congress, Consumer Financial Protection Bureau, July 19, 2011.

“New National Survey Reveals What Consumers Know and Don’t Know about Changing Credit Score

Marketplace,” Consumer Federation of America and VantageScore Solutions, February 28, 2011,

www.creditscorequiz.org.

Discriminatory Effects of Credit Scoring / page 7 National Fair Housing Alliance

I. The Nation’s Dual Credit Market Rooted in Discrimination

Credit scoring systems penalize borrowers who have anything other than mainstream, prime

loans. As described below, people and communities have been excluded from mainstream

affordable credit based on race and national origin. In the past, this was explicitly promoted by

the federal government and the private industry with discriminatory rating systems, and is

continued even today by banks like SunTrust and Wells Fargo. And it has been aided by the

blanketing of subprime loans in communities of color and fostered by continued patterns of

segregation and the dual credit market. Because many of the factors that make up credit

scoring systems rely on this dual credit market and its inherent discrimination, credit scoring

contributes to the self-perpetuating cycle of restricted access to credit that has a dramatic

disparate impact on communities of color.

A. Overt Historical Discrimination

In the not-so-distant past, government and private industry explicitly used race and national

origin in assessing borrower risk. For example, the Home Owners Loan Corporation (HOLC), a

federal agency established in 1933 in response to the foreclosure crisis associated with the

Depression, institutionalized “redlining.” HOLC utilized a discriminatory risk rating system

whereby prospective borrowers were favored if their neighborhood was deemed “new,

homogeneous, and in demand in good times and bad.”

Properties would be ranked low (and

thus judged high-risk) if they were “within such a low price or rent range as to attract an

undesirable element,” which often meant that they were located near an African-American

neighborhood.

The so-called “Residential Security Maps” used to make these classifications

labeled the lowest ranking neighborhoods “fourth grade,” and shaded them in red. According

to housing scholars William J. Collins and Robert A. Margo, “the agency’s revisions were

unprecedented. Private financial institutions incorporated the new rating system in their own

appraisals, thereby beginning the widespread institutionalization of the practice known as ‘red-

lining.’”

As discriminatory policies and practices continued to persist within the real estate

sector, private banks began to adopt the underwriting guidelines established by the federal

government in the HOLC program.

Subsequently, the HOLC risk rating system came to inform the Federal Housing Administration

(FHA) and Veterans Administration (VA) loan programs in the 1940s and 1950s. The FHA

made it possible to purchase a house with just a 10 percent down payment, as opposed to the

customary 33 percent required before its establishment. Loan terms were also extended for up

to 30 years. The VA program provided similar benefits, all while following the FHA in rating

Douglas S. Massey, “Origins of Economic Disparities: The Historical Role of Housing Segregation,” in

James H. Carr and Nandinee K. Kutty, eds., Segregation: The Rising Costs for America (New York:

Routledge, 2008), p. 69.

Ibid.

William J. Collins and Robert A. Margo, “Race and Homeownership, 1900-1900,” available at:

http://eh.net/Clio/Conferences/ASSA/Jan_00/margo.shtml.

Discriminatory Effects of Credit Scoring / page 8 National Fair Housing Alliance

properties in large part on the basis of the “stability” and “harmoniousness” of

neighborhoods.

As a result, the new benefits of a reduced down payment and better loan terms reached only

some Americans. According to FHA’s policy, “If a neighborhood is to remain stable, it is

necessary that properties shall continue to be occupied by the same racial and social classes.

Changes in social or racial occupancy contribute to neighborhood instability and the decline of

value levels.”

To implement this policy, the FHA even went so far as to recommend the use of

restrictive covenants to ensure neighborhood stability and racial homogeneity.

The notion that race had a direct impact on property values was broadly adopted by the

appraisal industry, and appraisers were trained to evaluate properties using race as a factor.

McMichael’s Appraising Manual, for example, provided the following ranking of race and

nationality by impact on real estate values (in order of preference):

1. English, Germans, Scotch

2. North Italians

3. Bohemians or Czechs

4. Poles

5. Lithuanians

6. Greeks

7. Russians, Jews (lower class)

8. South Italians

9. Negroes

10. Mexicans

Such lists remained in appraisal manuals long after the Fair Housing Act was passed in 1968.

Similar policies were employed in the insurance industry, as homeowners insurance companies

adopted policies that resulted in either the outright denial of insurance in communities of color

or the availability only of policies that provided inadequate protection at excessive costs to

consumers.

Even after passage of the Fair Housing Act, these discriminatory practices received tacit

approval from the federal banking regulatory agencies. It was not until 1976, when a coalition

of civil rights groups sued them for failing to enforce the Fair Housing Act, that the federal

banking regulatory agencies acknowledged that they had any enforcement responsibilities

Massey, “Origins of Economic Disparities,” op. cit., p. 71-72.

Frederick Babcock, Director of FHA Underwriting Division, “Techniques of Residential Location

Rating,” Journal of the American Institute of Real Estate Appraisers of the National Association of Real Estate

Boards, v. VI, n. 2 (April, 1938), p. 137.

Massey, “Origins of Economic Disparities,” op. cit., p. 71-72.

McMichael’s Appraising Manual, 4

Edition, 1951.

Discriminatory Effects of Credit Scoring / page 9 National Fair Housing Alliance

under the Act.

The settlement required the agencies to collect information on the mortgage

lending practices of the institutions they regulated, and to establish and implement fair lending

examination procedures.

Understanding the historical context of discrimination and redlining practices is significant in

any discussion on credit scoring. Because borrowers of color could not access credit in the

mainstream market, a dual credit market developed – a market that was separate and unequal –

a market where white borrowers had ready access to more regulated, lower-cost, affordable and

sustainable credit products while borrowers of color were relegated to unregulated, higher-cost

and more unsustainable sources of credit. These fringe markets were – and in some cases still

are - the primary source of credit for communities of color.

B. Subprime Lending and Its Long-Term Discriminatory Effects

In many cases, the banking and insurance industries simply replaced their explicitly

discriminatory standards with policies and practices that were non-discriminatory on their face,

but maintained a disparate impact. (It is worth noting, as described below, that some

companies also maintained overtly racially discriminatory policies.) By setting minimum loan

values, employing tiered interest rate policies, refusing to make loans in some neighborhoods,

and offering only market value homeowners insurance in some neighborhoods, banks and

insurance companies continued to discriminate in the marketplace.

Many lenders, recognizing that borrowers of color represented a growth market, developed

initiatives to heavily target this market segment. Indeed subprime lenders (and some

subsidiaries of prime lenders) took advantage of communities that mainstream lenders

shunned. In a representative case, the St. Louis Equal Housing and Community Reinvestment

Alliance alleged that a large local bank had not made a single loan to an African-American

borrower from 2003 to 2008.

Moreover, all of the banks’ branches were located in areas with

less than two percent African-American population. Nationwide, African-Americans and

Latinos were much more likely to receive a subprime loan than their white counterparts

according to HMDA data. In both 2005 and 2006, roughly 54 percent of African-Americans and

47 percent of Latinos received subprime loans compared to approximately 17 percent of

whites.

A study conducted by the National Community Reinvestment Coalition found that

there are fewer commercial bank branches in communities of color.

Instead of targeting this market with safe, lower-cost, affordable and sustainable loans,

borrowers of color were targeted for unsustainable, higher-cost, subprime mortgages.

National Urban League et. al. v. Office of the Comptroller of the Currency, et al , 1976

Rivas, Rebecca S., “Housing Alliance calls out Midwest BankCenter for not loaning to blacks.” The St.

Louis American, October 14, 2009.

Avery, et. al., “The 2006 HMDA Data”, Federal Reserve Bulletin, December, 2007.

Are Banks on the Map? An Analysis of Bank Branch Location in Working Class and Minority Neighborhoods,

National Community Reinvestment Coalition, March 28, 2007.

Discriminatory Effects of Credit Scoring / page 10 National Fair Housing Alliance

Subprime lenders have long boasted and prided themselves on being the primary providers of

credit to African-American, Latino and other underserved groups. Countrywide, at one time

the nation’s largest lender and a major originator of subprime loans, boasted that it was the

number one lender to borrowers of color.

The Department of Justice (DOJ) recently settled an

unprecedented $335 million lawsuit with Countrywide because of its discriminatory practices,

which included steering African-American and Latino borrowers who qualified for prime loans

into subprime mortgages.

Some of the nation’s other top subprime lenders have either settled

major discrimination lawsuits or are currently defending themselves against such allegations.

These lenders include Long Beach, Ameriquest, Delta Funding, Household Finance, Associates,

Citi, and Wells Fargo.

And while banks and others continued to defend the use of credit scores as the great equalizer,

many borrowers with high credit scores received subprime mortgages even when they qualified

for prime credit. Many would-be prime consumers were instead steered into subprime and Alt-

A mortgages because of the higher short-term profits lenders could garner. For example, an

analysis conducted by First American Loan Performance found that 41 percent of subprime

loans made in 2004 went to borrowers who actually would have qualified for a prime rate

loan.

Another study, commissioned by The Wall Street Journal, revealed that in 2005, 55 percent

of subprime borrowers would have qualified for a prime loan. The Wall Street Journal analysis

also found that in 2006 that number had jumped to as high as 61 percent.

Federal Reserve

Governor Edward Gramlich noted that half of subprime borrowers had credit scores of 620 or

higher.

The recently amended lawsuit filed by the City of Baltimore against Wells Fargo provides a

glaring example of how lenders purposefully targeted African-Americans and Latinos for

higher priced mortgages in outrageously discriminatory ways.

Two affidavits filed by former

Wells Fargo employees revealed that Wells Fargo:

 Specifically targeted African-American communities for subprime loans but did

not do so in white communities;

Morgenson, Gretchen, “Inside the Countrywide Lending Spree,” New York Times, August 26, 2007. See

also “Countrywide Nation’s No. 1 Lender in Emerging Markets”, Reported by AllBusiness, available at:

http://www.allbusiness.com/personal-finance/real-estate-mortgage-loans/285920-1.html.

“Justice Department Reaches $335 Million Settlement to Resolve Allegations of Lending Discrimination

by Countrywide Financial Corporation”. Available at http://www.justice.gov/usao/cac/countrywide.html.

Klein, Ezra, “Digging into finance’s pay dirt: The risky business of payday loans and more,”

Washington Post, July25, 2010.

“Subprime Debacle Traps Even Very Creditworthy,” Wall Street Journal, December 3, 2007.

Kirchhoff and Block, “Subprime Loan Market Grows Despite Troubles”, USA Today, December 14, 2004.

(At the time of his statement, a score of 620 qualified a borrower for a prime loan.)

Mayor and City Council of Baltimore v. Wells Fargo Bank, N.A. and Wells Fargo Financial Leasing, Inc. Third

Amended Complaint for Declaratory and Injunctive Relief and Damages, October 21, 2010.

Discriminatory Effects of Credit Scoring / page 11 National Fair Housing Alliance

 Targeted African-American churches for the purpose of selling subprime loans.

Employees of color were tapped to make presentations to the churches. A white

employee was told she could only attend the presentations at African-American

churches if she “carried someone’s bag;”

 Used derogatory language to refer to African-American consumers. African-

Americans were referred to as “mud people” and “niggers.” And employees

referred to loans in African-American neighborhoods as “ghetto loans.” And

they referred to Prince George’s County as the “subprime capital” of Maryland.

Comparatively, Wells employees felt that predominately white counties like

Howard County, Maryland were bad places for subprime mortgages;

 Gave employees substantial financial incentives for steering borrowers who

actually qualified for prime mortgages into the subprime market.

America’s separate and unequal financial system is a direct result of the bias perpetuated by

both the private and public sectors. Here are some statistics that demonstrate our dual financial

system:

 African-American and Latino homebuyers “face a statistically significant risk of

receiving less favorable treatment than comparable whites when they ask

mortgage lending institutions about financing options;”

 The denial rate for first lien mortgages for African-American borrowers was 2.5

times higher than the rate for Non-Hispanic white borrowers in 2010.

 In 2008, African-Americans were 2.63 times more likely and Hispanics more than

two times more likely than their white counterparts to receive a higher-priced

loan.

 Even higher-income African-Americans and Latinos received a disproportionate

share of subprime loans. According to one study that analyzed more than

177,000 subprime loans, borrowers of color were more than 30 percent more

likely to receive a higher-rate loan than white borrowers, even after accounting

for differences in creditworthiness.

 Borrowers residing in zip codes whose population is at least 50 percent non-

white were 35 percent more likely to receive loans with prepayment penalties

Turner, et al. All Other Things Being Equal: A Paired Testing Study of Mortgage Lending Institutions. The

Urban Institute, 2002.

Robert B. Avery, Neil Bhutta, Kenneth P. Brevoort, and Glenn B. Canner. The Mortgage Market in 2010:

Highlights from the Data Reported under the Home Mortgage Disclosure Act; Table 19, February, 2011.

Avery, et. al; “The 2008 HMDA Data: The Mortgage Market during a Turbulent Year.” Federal Reserve

Bulletin, October 12, 2009.

See Bocian, D. G., K. S. Ernst, and W. Li, Unfair Lending: The Effect of Race and Ethnicity on the Price of Subprime

Mortgages, Center for Responsible Lending, May 2006, p. 3.

Discriminatory Effects of Credit Scoring / page 12 National Fair Housing Alliance

than financially similar borrowers in zip codes where non-whites make up less

than 10 percent of the population.

It follows, then, that borrowers of color are disproportionately represented in foreclosure claims

and that communities of color experience higher foreclosure rates than the general population.

A recent study released by the Center for Responsible Lending reveals that a home owned by

an African-American family is 76 percent more likely to go into foreclosure that a home owned

by a white family.

The Center for Responsible Lending estimates that African-American and

Latino communities will lose $194 billion and $177 billion respectively in housing wealth as a

result of the foreclosure crisis including the resulting depreciation of living near foreclosed

properties.

These high rates of foreclosure caused by discriminatory practices have resulted in thousands of

bank-owned (also known as real estate-owned or REO) properties in communities of color. A

recent undercover investigation by NFHA and its members shows that discrimination by the

banks continues even after foreclosure.

The investigation found striking disparities in the

maintenance and marketing of foreclosed properties in white areas compared those in

neighborhoods of color. Investigators used 39 different factors to evaluate the maintenance and

marketing of REO properties, subtracting points for broken windows and doors, water damage,

overgrown lawns, no “for sale” sign, trash on the property, and other deficits. Overall, REO

properties in communities of color were 42 percent more likely to have more than 15

maintenance problems than properties in white neighborhoods. NFHA has since filed housing

discrimination complaints against Wells Fargo and U.S. Bancorp for disparities in the

maintenance and marketing of REO properties.

C. The Proliferation of Fringe Lenders in Communities of Color

As described above, fringe lenders – including payday lenders and check cashers – have

historically been a primary source of credit for underserved borrowers and are highly

concentrated in communities of color. One analysis revealed that there were more payday

lender outlets in the country than all McDonalds and Burger King restaurants combined.

These fringe lenders saturate predominantly African-American and Latino neighborhoods. A

study of fringe lenders in California found that payday lenders were nearly eight times as

concentrated in neighborhoods with the largest shares of African Americans and Latinos as

Bocian, D.G. and R. Zhai, Borrowers in Higher Minority Areas More Likely to Receive Prepayment Penalties on

Subprime Loans, Center for Responsible Lending, January 2005.

Bocian, et. al., Foreclosures by Race and Ethnicity: The Demographics of a Crisis, The Center for Responsible

Lending, June, 2010.

Ibid.

The Banks Are Back, Our Neighborhoods Are Not: Discrimination in the Maintenance and Marketing of REO

Properties, National Fair Housing Alliance, April 4, 2012.

Klein, Ezra “Digging into finance’s pay dirt: The risky business of payday loans and more,”

Washington Post, July25, 2010.

Discriminatory Effects of Credit Scoring / page 13 National Fair Housing Alliance

compared to white neighborhoods, draining nearly $247 million in fees per year from these

communities.

The study includes several maps of communities throughout California

showing this pattern. Below is a map of Los Angeles depicting the heavy concentration of

payday lenders in African-American and Latino communities. Conversely, there are few

mainstream bank facilities in predominantly African-American and Latino communities.

Map: Center for Responsible Lending

Conversely, there are few mainstream bank facilities in predominantly African-American and

Latino communities. Borrowers who are targeted by fringe lenders and shunned by

mainstream financial institutions are susceptible to volatile credit markets. Consumers who

access credit from fringe lenders will undoubtedly have lower credit scores because the

products these institutions peddle have abusive terms that carry higher delinquency and

default rates.

II. Credit Scoring Has a Discriminatory Impact and Is Not the Best Measure of Risk

Have a mortgage from a finance company? Your credit will likely be lower than if you had

gotten the loan from a depository lending institution. Lose that same home to foreclosure

because you can no longer make the inflated payments? Your credit score just went down

again.

Li, et. al. Predatory Profiling: The Role of Race and Ethnicity in the Location of Pay Day Lenders in California,

Center for Responsible Lending, March 26, 2009.

Discriminatory Effects of Credit Scoring / page 14 National Fair Housing Alliance

As described above, people of color were disproportionately steered to subprime loans and

targeted by fringe lenders. Because credit scoring systems and other automated valuation

systems are promoted as a great equalizer and a non-discriminatory way of measuring credit

risk, one might then think that credit scores would not rely on discriminatory assumptions to

measure risk. In fact, that is exactly what the systems do in some instances. For example, some

scoring mechanisms assume that a borrower who received a loan from a finance company is a

worse credit risk than one who got a loan from a depository institution – when, in fact, the

opposite may be true. A credit scoring system that relies on this false premise penalizes the

borrower who simply may not have had access to a mainstream lender but had abundant access

to fringe lenders.

Indeed, credit scoring mechanisms are a reflection of the lending and finance systems which

produce the data upon which the mechanisms are built. Oftentimes, credit scoring mechanisms

assess the riskiness of the lending environment, product type or loan features a consumer uses

rather than the risk profile of the consumer.

Let us use an analogy to illustrate this point. Suppose a test has been developed to determine

how safe or risky someone is as a car driver. In this test, the driver has to drive through a path

and navigate a series of cones and obstacles. However, the driver is placed in a car that is

essentially a lemon. The brakes do not work, there is no steering wheel fluid in the car so that it

does not turn well, and the transmission is malfunctioning along with other problems. The

driver completes the course and is given a low score having knocked over several cones or run

into some of the obstacles on the course. But then, this same driver is placed into a different car

and asked to drive the same course again. This time the car is not a lemon. It is in pristine

condition – with no problems. The second time through, the driver passes with flying colors

and receives a high score.

Did the driver change? No. But what did change is the vehicle in which the driver was placed.

So the test, while accurately measuring how well the driver navigated the course, was more a

reflection of the quality of the vehicle in which the driver was placed than the ability or

riskiness of the driver. Similarly, credit scoring mechanisms are often a reflection or

measurement of the lending environment or loan product type – and not so much the risk

profile of the borrower.

Consumers of color are ill-served by the financial mainstream and disproportionately access

credit in more volatile financial environments – these consumers disproportionately get the

lemons of the financial services world. As a result, current credit scoring mechanisms which do

not evaluate or calibrate scores based on the safety or soundness of the lending environment,

may actually cause harm to borrowers of color by misjudging them.

A. Limited Scope, Quality and Transparency of Credit Information

Discriminatory Effects of Credit Scoring / page 15 National Fair Housing Alliance

The information used to build credit scoring models can come from many different sources;

however, modelers have tended to rely heavily on credit reporting data from credit bureaus.

The quality or accuracy of the scoring model is intrinsically tied to the quality of data upon

which the model is based: the better the data quality, the better the scoring system. If modelers

are relying on limited data or inaccurate data, they will develop scoring models that are less

effective and have limited predictive power and market applicability. The less predictive a

scoring model, the greater the likelihood for miscalculating risk.

Companies can use data purchased from third party sources or their own privately held data to

develop their scoring systems. Larger companies that have abundant information about a large

number of consumers oftentimes use their own in-house data to either develop their own

unique scoring systems or to enhance systems that they might obtain from an outside source.

But, by and large, the data upon which scoring models are built are purchased from large credit

repositories, and these data are often flawed. A study conducted by the National Association of

State Public Interest Research Groups revealed the following:

 Four out of five credit reports contained errors;

 25 percent of credit reports contained significant errors that would result in the

denial of credit;

 54 percent had inaccurate personal information;

 30 percent listed closed accounts as open; and

 8 percent did not list major credit accounts.

Not only can the data that credit modelers use be flawed but it can be incomplete. Not all

creditors report consumer information to credit repositories. Indeed, some positive credit

information from fringe lenders is typically not reported while negative information is almost

always reported. Take the case of payday lenders, which, as illustrated above, are concentrated

in communities of color. According to the Community Financial Services Association of

America, “Payday advances are not reported to traditional credit bureaus.”

If a consumer

obtains a payday loan, the fact that the consumer has paid off the debt on time is not reported

to credit bureaus. However, unpaid payday loans are often reflected on the consumer’s credit

report. The Consumer Federation of America reports that unpaid payday loans can lead to

negative credit ratings as well as difficulty in opening bank accounts.

Creditors are not required to report consumer data to the credit repositories. Nor, if they do

report, are they required to report positive data as well as negative data. A creditor can decide

to forgo submitting any data, or to report only negative data to the credit repositories. Some

Cassady and Mierzwinski, Mistakes Do Happen: A Look at Errors in Consumer Credit Reports, National

Association of State PIRGs, June, 2004.

The Consumer Financial Services Association of America is a national organization for “small dollar,

short-term lending or payday advances.” http://cfsaa.com/about-cfsa.aspx. See especially

http://cfsaa.com/what-is-a-payday-advance/frequently-asked-questions.aspx.

See “How Payday Loans Work,” at http://www.paydayloaninfo.org/facts.

Discriminatory Effects of Credit Scoring / page 16 National Fair Housing Alliance

creditors may opt not to submit data because they do not wish to pay reporting costs. Others

may not want other companies to be able to identify – and poach – their best paying customers.

And while a creditor may not be able or willing to report positive data on a regular basis, the

creditor can report negative data to the credit repositories by having the matter referred to a

collections agency or by filing an action against the consumer to collect on the alleged debt.

This tilts the entire system against the consumer, especially those who access credit outside of

the financial mainstream.

Smaller creditors like community development financial institutions (CDFIs) that want to report

positive data may be prohibited from doing so because of their size. An informal survey

conducted by NFHA underscores the difficulty of collecting comprehensive information on

consumer credit habits. The major credit repositories are structured to collect data from larger

creditors with a large number of consumer files. Some repositories require creditors to have at

least 500 files when reporting data; others require 1,000 files. These numbers are often beyond

the reach of CDFIs and other community-based institutions.

In addition to posing accuracy and access challenges, credit scoring mechanisms lack

transparency. The formulas are proprietary and not disclosed to the public. In addition, there

are a number of individual factors that help determine the score, only some of which are public.

It is not clear exactly how the factors used in the credit scoring systems affect a consumer’s

score. There are potentially thousands of variables that can be included in a scoring system.

These variables can be comprised of individual components as well as combined components

and might include such elements as the number of: 30-day late payments; inquiries; inquiries

by subprime lenders; open trade lines; late mortgage loan payments; or installment loans. They

might also include length of employment or length of individual revolving loan accounts.

Each variable is purportedly tested to determine first if it is related to a particular outcome, such

as likelihood for a mortgage loan default or for filing an insurance auto claim. Then the

variables are tested to determine how they should be weighted within the credit scoring

formula. There is a level of subjectivity to the process and experts who develop the systems

make the final determination as to which variables are to be included in the formulas and how

much weight each is given.

It is important to note that credit scoring modelers are trying to determine whether a particular

variable has a correlation to a particular outcome. But the mere presence of a correlative

relationship between a particular variable and a certain outcome does not in and of itself

indicate a causal relationship. For example, variable testing may indicate that there is a

correlation between gas company credit cards and higher rates of mortgage loan defaults; but

this does not mean that having a gas company credit card will cause a consumer to default on a

mortgage.

It stands to reason that not all variables with a correlative relationship can or should be used in

a credit scoring system. For example, some analyses have shown that hair or eye color can

Discriminatory Effects of Credit Scoring / page 17 National Fair Housing Alliance

correlate to certain types of insurance claims. Other analyses have revealed links between

zodiac signs and frequency of auto claims.

If we were to follow these data, those born under

the sign of Taurus or Virgo would pay higher premiums than Cancers or Aquarians. It also

follows that neither race nor national origin nor any proxies that stand for them should be used

in a credit scoring system, not only because it flies in the face of our nation’s laws and policies,

but because it makes as little sense as using a zodiac sign to price car insurance.

B. Disparate Impact of Credit Scoring Factors

While it is no longer legal to evaluate risk using protected class characteristics, current credit

scoring systems still have a significant disparate impact on people of color and other

underserved consumers because some seemingly facially-neutral factors actually have

discriminatory effects.

Take, for example, the factors used by the FICO scoring system, which is widely-known and

often touted as the industry standard for use in mortgage lending. While many independent

variables and their weighting in the FICO scoring system are unknown and proprietary, several

broad categories that impact the score are public: payment history; amounts owed; length of

credit history; new credit; and types of credit used. The chart below from one of FICO’s

websites illustrates the value assigned to each of these categories.

Chart: www.myfico.com

All of these categories pose concerns about disparate impact and unintended discriminatory

outcomes, and affect access to sustainable, affordable, and fair credit. Below is a more detailed

description of the fair lending concerns related to each category of the FICO scoring system.

Payment History – 35% of FICO Score

“Allstate: Virgos have most crashes,” United Press International

http://www.upi.com/Odd_News/2011/01/28/Allstate-Virgos-have-most-crashes/UPI-28291296246090/.

While the Allstate press release announcing the findings was tongue-in-cheek, the data and analysis were

real. “Allstate zodiac joke bombs,” CNN Money, February 2, 2011.

http://money.cnn.com/2011/02/02/news/companies/allstate_zodiac/index.htm See also “Aquarius with

the fewest claims, Taurus lives more dangerously,“ Allianz Suisse, February 17, 2011.

https://www.allianz.com/en/press/news/studies/news_2011-02-17.html

Chart developed by Fair Isaac: http://www.myfico.com/CreditEducation/WhatsInYourScore.aspx.

Discriminatory Effects of Credit Scoring / page 18 National Fair Housing Alliance

The payment history component of the score includes information about whether borrowers

make timely debt payments, including some subprime loans. As mentioned above, subprime

loans carry much higher default and delinquency rates

– not necessarily because of the

borrower traits, but more often because of the aspects and features of the loans themselves.

Because African-Americans and Latinos are targeted for subprime loans, the data suggest that

these groups will undoubtedly experience higher rates of poor performance in payment history.

A unique study that compared two similar groups of low- and moderate-income borrowers

demonstrates this point.

The study compared two mortgage loan portfolios, one comprised of

loans made through a program that provided low-cost fixed rate loans, and the other a portfolio

of subprime loans.

Using propensity score match methodology, the researchers were able to

isolate borrowers with similar characteristics in the two groups. The divergent variables

between the two groups were the loan terms and conditions, and the channel borrowers used to

obtain the mortgages. While the traits of both groups of borrowers were similar, the loan

performance outcomes were not. The default rate for the subprime portfolio was four times

higher than that for the lending program portfolio for low- and moderate-income borrowers.

Moreover, the study found compelling evidence that loan characteristics and origination

channel had a significant impact on loan performance. Specifically, the existence of prepayment

penalties, adjustable interest rates, and elevated costs negatively impacted the loans’

performance – even after controlling for credit score. Additionally, loans originated through

broker channels resulted in higher default rates.

These data conflict with the underlying assumption behind scoring mechanisms. This study

and others suggest that a borrower may well end up with a damaged credit score not because

the borrower was more risky or negligent but rather because the borrower obtained a loan

through a broker or received loan terms that increase the likelihood of delinquency and default.

Existing credit scoring systems do not distinguish between risk caused by borrower behavior

and risk caused by loan terms and conditions. Thus, risky loans are likely to have a negative

According to Mortgage Bankers Association National Delinquency Survey Data released 5/19/2010, the

seasonally adjusted delinquency rate was 6.17% for prime fixed loans, 13.52% for prime ARM loans,

25.69% for subprime fixed loans, 29.09% for subprime ARM loans, 13.15% for FHA loans, and 7.96% for

VA loans. Foreclosure starts rate was .69% for prime fixed loans, 2.29% for prime ARM loans, 2.64% for

subprime loans, 4.32% for subprime ARM loans, 1.46% for FHA loans, and .89% for VA loans. These

trends have held steady. The same data released 8/29/2009 revealed the following: the seasonally

adjusted delinquency rate was 6.41% for prime loans, 25.35% for subprime loans, 14.42% for FHA loans,

and 8.06% for VA loans. The foreclosure inventory rate was 3% for prime loans, 15.05% for subprime

loans, 2.98% for FHA loans, and 2.07% for VA loans.

Lei Ding, Roberto G. Quercia, Janneke Ratcliff, and Wei Li, Risky Borrowers or Risky Mortgages:

Disaggregating Effects Using Propensity Score Models, Center for Community Capital, University of

North Carolina at Chapel Hill, September 13, 2008.

Discriminatory Effects of Credit Scoring / page 19 National Fair Housing Alliance

impact on the borrowers’ credit scores, even though those borrowers may have had a perfect

payment record had they been able to obtain a less risky loan.

Amounts Owed – 30% of FICO Score

The FICO score calculation of amounts owed is comprised of multiple factors and FICO does

not reveal in detail all of these factors and how they are weighted. However, the company does

report that this category takes into consideration the amount of credit available to a borrower

for certain types of revolving and installment loan accounts. To the extent that underserved

communities have restricted access to credit, and in particular, the type of credit that will likely

be reported in a positive fashion to credit repositories, this category can pose a disparate

discriminatory impact.

A study by the San Francisco Federal Reserve Board provide an analysis of individuals who do

not have a checking or savings account in the region. These “unbanked” tend to be low-income,

young, non-white adults who lack a college degree.

The analysis goes on to reveal that

approximately half of African-Americans and Latinos fall into this category and that the

unbanked are concentrated in lower-income census tracts without a checking or savings

account. This analysis also documents the preponderance of payday lenders and check cashers

in predominately African-American and Latino neighborhoods.

The lack of access to mainstream lenders may well impact the ability of underserved consumers

to obtain revolving or installment lines of credit from such lenders. And if these borrowers

experience undue difficulty in accessing quality credit, they may well suffer a lower credit score

from a system that considers how much “extra” credit they may have available in certain

revolving and installment accounts.

Here again, this component is not only measuring the ability of the borrower to effectively

manage credit accounts but is also measuring the extent to which a consumer actually has

access to certain types of credit accounts.

Length of Credit History- 15% of FICO Score

Presumably, the longer a borrower has had an account, and to the extent that the account is

reported to the credit repositories, the higher the borrower’s credit score. If this is indeed the

case, then borrowers with little access to credit that is reported to the credit repositories will be

negatively impacted by this component.

We provide a fairly detailed analysis above of how mainstream creditors historically

discriminated against communities of color. Moreover, as referenced above, borrowers of color

“Understanding the Unbanked Market in San Francisco”, Federal Reserve Bank of San Francisco.

Presentation available at http://www.frbsf.org/community/resources/banksfpresentation.pdf.

Discriminatory Effects of Credit Scoring / page 20 National Fair Housing Alliance

are much less likely than their white counterparts to have access to mainstream banks and

consequently are much more likely to access credit from fringe lenders who do not report

positive data to the credit repositories. This means that borrowers of color will be less likely to

have trade lines with a significant amount of history.

This factor also penalizes borrowers who deal on a cash basis, access credit outside of the

financial mainstream, have been shut out from accessing traditional credit, or obtain credit from

lenders who do not report positive data. Borrowers with these circumstances are

disproportionately persons of color.

New Credit- 10% of Credit Score

This component takes into consideration the number of newly opened accounts a consumer has.

FICO does not provide details on just how a consumer’s credit score will be affected if the

consumer establishes new credit. FICO advises consumers to avoid opening new lines of credit

as this might result in lowering the credit score.

Further, opening new accounts will lower the

average account age of credit lines and this will result in a lower credit score.

This component also considers the number of credit accounts a consumer pursues. So if a

consumer is shopping for a mortgage or applying for credit at different places, the consumer’s

credit score can be negatively impacted. To guard against any negative impact, FICO advises

consumers to shop for a mortgage loan within a short window of time.

There are two areas of concern with respect to disparate outcomes under this component. The

first is the higher likelihood that consumers of color will be among those who are accessing new

credit accounts. As discussed above, credit access is a major challenge for underserved groups

and these groups are much more likely to be unbanked. It stands to reason, therefore, that

underserved groups will be among those who are newly entering the credit markets and

therefore, establishing new accounts.

The second area of concern emanates from the higher mortgage loan declination rates for

borrowers of color. As described earlier, HMDA data reveal that borrowers of color are much

more likely than their white counterparts to be declined for a loan. These higher declination

rates suggest that borrowers of color may be more likely to apply to additional lenders for a

loan approval.

If mortgage loan inquiries or applications are undertaken in a short time frame, there may be no

negative impact on a consumer’s credit score. However, if a consumer applies for a mortgage

with one lender, is declined, and then applies for a mortgage with another lender, this process

may well negatively impact the consumer’s credit score due to the longer lapse in time between

“How to Repair Your Credit and Improve Your FICO Credit Score,” Available at:

http://www.myfico.com/crediteducation/improveyourscore.aspx

Discriminatory Effects of Credit Scoring / page 21 National Fair Housing Alliance

loan inquires. More analysis and research needs to be conducted to determine if borrowers of

color have a higher incidence of shopping for a mortgage with different lenders over longer

periods of time and ultimately how that might impact their credit scores.

Types of Credit Used – 10% of FICO Score

Again, FICO does not reveal exactly how it calculates the type of credit a borrower may use in

generating a credit score; however, there is evidence that certain types of credit, like credit

provided by finance companies, are treated less favorably than credit provided by mainstream

lenders, like depository banking institutions. According to the Federal Reserve Board, “Many

credit-scoring models consider the number and type of credit accounts you have. A mix of

installment loans and credit cards may improve your score. However, too many finance

company accounts or credit cards might hurt your score.”

If this is indeed the case, this

category also presents dangerous implications for borrowers of color.

FICO, in a guide developed to advise consumers on how to improve their credit scores,

suggests that consumers who have installment loans and credit cards that are reported to the

credit repositories will have a more favorable analysis in the FICO credit scoring system.

Here again, consumers who access credit outside of the financial mainstream will be penalized

by this type of an analysis.

This component may more largely assess the quality of the environment or type of loan product

a consumer accesses rather than the risk characteristics of the consumer.

C. Existing Credit Scoring Systems Do Not Adequately Predict Risk

The current crisis has revealed that credit scoring mechanisms are an insufficient measure for

predicting and managing performance. While FICO is designed to assess risk and predict a

borrower’s performance, recent analyses demonstrate the ineffectiveness of the scoring

mechanism. Default rates for all borrowers have increased precipitously, regardless of credit

score, and one study found that “higher FICO scores have been associated with bigger increases

in default rates over time.”

In the years before the economic crisis, it was common for lenders to put aside more thorough

and comprehensive underwriting criteria which allowed unique and compensating factors to be

evaluated, and instead to substitute them with flimsy underwriting standards. If a borrower

had a higher credit score, the lender could truncate the underwriting process by foregoing a

fully documented underwriting review. In order to maximize short-term profits, lenders took

See “5 Tips: Improving Your Credit Score,” available at:

http://www.federalreserve.gov/consumerinfo/fivetips_creditscore.htm.

Ibid.

Demyanyk, Yuliya, “Did Credit Scores Predict the Subprime Crisis?” The Regional Economist, Federal

Reserve Bank of St. Louis, October, 2008.

Discriminatory Effects of Credit Scoring / page 22 National Fair Housing Alliance

great strides to increase volume. One way to increase volume was to shorten the time it took to

approve a loan.

Sound underwriting criteria such as verifying savings and other deposits, income and

employment or documenting timely rental payments were largely disregarded. Lenders gave

substantially more weight to the credit score factor. In that environment, the FICO score

became a proxy for sound underwriting. Whereas the credit score might have been an

important tool to add to the underwriting toolbox, instead it was over-valued in the

underwriting process. Even FICO admits that lenders were too reliant on the model.

A study published by the Federal Reserve Bank of St. Louis looked at credit scores and

borrowers who received subprime mortgages.

The study revealed that, for borrowers with the

lowest FICO scores (500 – 600), the rate of seriously delinquent loans was twice as large in 2007

as it was in 2005. Comparatively, for borrowers with the highest FICO scores (above 700), the

rate of seriously delinquent loans was almost four times as large in 2007 as it was in 2005.

Borrowers with lower FICO scores saw a 100 percent increase in seriously delinquent loans

while borrowers with higher FICO scores saw a 300 percent increase in seriously delinquent

loans. The study’s author concludes that “the credit score has not acted as a predictor of either

true risk of default of subprime mortgage loans or of the subprime mortgage crisis.” The heavy

reliance on FICO during the most recent housing boom has contributed to the system’s

ineffectiveness. Even industry analysts have recognized the flaws in FICO.

In a document

written to clients, an analyst at CIBC World Markets called FICO scores "virtually

meaningless."

Borrowers with higher FICO scores are in many cases acting the way borrowers with very low

scores are predicted to act. Some analysts in reviewing private loan portfolios have found that

in some cases loan characteristics were more predictive of loan performance than the borrower’s

FICO score. Indeed, both FICO

and TransUnion have released reports that indicate that

borrowers with higher FICO scores are performing in uncharacteristic ways. These borrowers,

in a trend never before seen, are more likely to pay their credit card debt than their mortgage

loan debt. This offers additional proof that a credit score alone cannot predict long-term

mortgage performance.

Many lenders that either do not rely on credit scoring mechanisms at all or minimally rely on

them experience default rates that are lower than the industry average. For example, Golden

West Financial, a lender that did not rely on the FICO score because of its non-predictive nature,

Sullivan, Bob. “Credit Scores 102: A Crisis, and Some Changes,” MSNBC, The Red Tape Chronicles,

March 18, 2008.

Ibid footnote 49.

Gandel, Stephen, “Lenders Look Beyond Credit Scores to Gauge Who’s a Risk”, Time, January 9, 2009.

Foust, Pressman, “Credit Scores: Not-So-Magic Numbers,” Bloomberg Businessweek, February 7, 2008.

Tedeschi, Bob, “Even High Score Borrowers at Risk of Mortgage Default,” New York Times, March 10,

2010.

Discriminatory Effects of Credit Scoring / page 23 National Fair Housing Alliance

experienced a default rate of 0.75 percent while the industry average for the same class of loans

was 1.04 percent.

Golden West relied on careful underwriting, including income and asset

verification and employed a different mechanism for compensating appraisers. Instead of

compensating an appraiser based on the number of appraisals completed, Golden West

compensated appraisers on the accuracy of the appraisal over the life of the loan. Underscoring

the tenuous reliability of the FICO score, a Golden West representative reported that some of

Golden West’s best clients had very low FICO scores and some of their worst clients had high

FICO scores. In addition, the North Carolina State Employees’ Credit Union indicated that for

their borrowers who would be classified as subprime, the default rate is 1.25 percent, well

below the industry average. NCSE attributes the higher default rates among subprime loans to

higher interest rates and poor underwriting practices.

D. Risky Loan Products and Unsafe Lending Environments – Not Borrowers – Were

Clearly the Culprit

When looking at which loans failed and which were successful over the past ten years, the

picture becomes clear. Loan terms and conditions were the largest part of the problem, not the

borrowers. Failed underwriting processes and unsuitable loan products were higher

contributors to poor loan performance than were the credit characteristics of the borrower.

Even borrowers with good credit who paid their bills on time, quickly found themselves in

trouble after getting a predatory or subprime loan or accessing credit in an unsafe environment.

We saw similar outcomes among corporations like Lehman Brothers and Bear Stearns that

turned more and more to risky investment products and tenuous financial deals. Just as the

creation and sale of unregulated complex derivative investment products was a bad idea, and

led otherwise sound companies into ruin, so was the creation and sale of unwise mortgage loan

products with highly risky features, like pre-payment penalties, and negative amortization

which led otherwise good consumers into default.

Some lenders might improve overall loan performance by improving the quality of the

underwriting process. In a presentation on the impact of the Qualified Residential Mortgage

requirements, a number of organizations, including NFHA, the National Association of

Realtors® and the Mortgage Bankers Association, highlighted a number of factors that are most

important in decreasing default risk. Those factors included full loan documentation and

verification processes. These critical underwriting components were identified as key elements

in improving loan portfolio performance and management.

The organizations also cite risky loan features including:

Ibid.

The presentation deck is available by contacting the National Fair Housing Alliance or any of the other

sponsoring organizations.

Discriminatory Effects of Credit Scoring / page 24 National Fair Housing Alliance

 negative amortization loans;

 interest-only loans;

 loans with balloon payments;

 loans exceeding 30 years in maturity;

 prepayment penalties;

 unverified income, employment, assets and other debts i.e., no-doc or low-doc

loans;

 underwriting for ARMs based on an introductory rate rather than the fully-

indexed interest rate;

 total points and fees exceeding three percent of loan amount;

 unstable or undocumented payment history;

 ARM reset caps above two percentage points per year;

 investor loans;

 yield spread premiums; and

 piggyback seconds.

These same risky loan features have been identified in proposed regulations for the Qualified

Mortgage and Qualified Residential Mortgage requirements.

Perhaps instead of concentrating so much of the risk analysis on the borrower, more attention

should be paid to evaluating the products themselves, the environment in which the credit is

provided and the underwriting process used by the mortgage lender.

III. Why the Federal Government and Lenders Have an Obligation to Change the System

All federal agencies and their grantees associated in any way with housing and community

development have a special obligation to further the purposes of federal Fair Housing Act. The

law covers policies and practices that have a disparate impact on protected classes. To the

extent that credit scoring has a disparate impact, the federal government and its grantees must

take action.

The federal Fair Housing Act – passed in 1968 – has the dual mission of eliminating housing

discrimination and promoting residential integration. The Fair Housing Act requires that

government agencies spend funds dedicated to housing and community development in a

manner that “affirmatively furthers fair housing.” This obligation is not limited to the

Department of Housing and Urban Development; rather it applies to a wide range of

government agencies, including those with regulatory or supervisory authority over financial

institutions, as stated in Section 808(d) of the Fair Housing Act:

All executive departments and agencies shall administer their programs and

activities relating to housing and urban development (including any Federal

agency having regulatory or supervisory authority over financial institutions) in

a manner affirmatively to further the purposes of this subchapter and shall cooperate

Discriminatory Effects of Credit Scoring / page 25 National Fair Housing Alliance

with the Secretary [of Housing and Urban Development] to further such

purposes.

(emphasis added)

Executive Orders and other provisions of the Fair Housing Act related to affirmatively

furthering fair housing provide additional guidance on this obligation.

The Obama

Administration has also affirmed its commitment to fair housing and fair lending.

This affirmative obligation has been interpreted to apply to efforts to eliminate segregation.

This is important to the well-being of our nation because where we live determines our access to

opportunities, wealth, and resources.

In this context, equal access to credit, financial services

and products cannot be overstated. The largest federal housing program ever, the Troubled

Asset Relief Program (TARP) provided funding for major banks and insurance companies. As

recipients of these funds, these entities are also required to affirmatively further fair housing

with TARP and any other government funds.

Credit scoring systems, which are clearly

related to housing and community development, are also covered by this provision of our

nation’s fair housing laws.

IV. Policy and Enforcement Solutions to Improve Credit Scoring Systems

Because of the significance that credit scoring has for a wide range of access issues, such as

credit access, employment opportunity, and insurance availability, credit scoring mechanisms

need major improvements if not a complete overhaul. Intrinsic and persistent discrimination in

the lending markets and America’s dual and unequal credit market have contributed to serious

credit access problems for borrowers and communities of color. Below are some

recommendations on how to improve credit scoring mechanisms and suggestions on how to

monitor and evaluate these systems.

42 U.S.C. Sec. 3601 et seq.

Section 805 of the Fair Housing Act lays the groundwork for this mandate by detailing discrimination

in residential real estate-related transactions; Section 808 of the Act spells out the responsibility of the

Secretary of Housing and Urban Development (HUD) to administer the Act, and the Act’s application to

other federal agencies; and Executive Order 11063,

signed on November 20, 1962, and Executive Order

12892,

signed on January 17, 1994, together state the responsibilities of all federal agencies to administer

their programs in a manner that affirmatively furthers fair housing and clarify what is meant by

programs and activities relating to housing and urban development.

See Remarks by Shaun Donovan at the National Fair Housing Alliance Conference available at:

http://www.hud.gov/news/speeches/2009-06-08.cfm See also HUD Statement No. 09-206 available at:

http://portal.hud.gov/portal/page/portal/HUD/press/press_releases_media_advisories/2009/HUDNo.09-

206.

Carr and Kutty, p. 2.

Swesnik, Deidre; Clark, Benjamin; Goldberg, Deborah. How Tarp Funds Could (and Should) Be Used to

Improve Our Neighborhoods, National Fair Housing Alliance, November 2009.

Discriminatory Effects of Credit Scoring / page 26 National Fair Housing Alliance

Broaden the scope of financial data utilized by underwriting and credit scoring models

One way to improve credit scoring models is to broaden the scope and quality of data upon

which the systems are based. Currently the primary source of data is major credit repositories.

Credit repositories should make it easier for smaller financial institutions to report positive

data. Moreover, credit repositories must be proactive and ensure that positive data from non-

traditional sources can be submitted. Data should also be included from state housing finance

agencies, community development financial institutions, micro-lending organizations, credit

unions, and affiliation or community groups such as churches, faith-based institutions and

benevolent organizations.

Broadening the scope of credit information will create a more robust data pool with additional

information about and from consumers who access credit in safe, but non-traditional

environments. It will also enable the credit scoring systems to more accurately assess a broader

range of consumers. This will in turn lessen the likelihood that a consumer will be incorrectly

characterized or categorized in various credit scoring systems.

Finally, credit repositories must create mechanisms to correct the current system’s slant toward

the reporting of only negative data. For example, a mechanism that allows consumers to report

and submit verifiable and documented information about their credit payment histories could

be designed. Consumers are paying debt obligations on time that do not get reflected in the

credit repository data and this has a huge negative impact on communities of color.

Improve the quality of data

Credit bureaus must make it easier for consumers to correct erroneous information on their

credit reports. Incorrect information can lead to low credit scores and credit denials and limit

access to quality, affordable credit.

Improving data quality will also contribute to better scoring models that more accurately assess

consumer risk. Improving the performance of scoring models should be the goal of everyone

involved with providing credit to consumers. It should also be the goal of regulators that

oversee financial institutions and credit reporting agencies. Ensuring that consumers have

access to quality credit will expand opportunities for consumers, promote healthy financial

practices, and contribute to the growth of consumer net worth.

Make the system more transparent

It has taken years to get agencies to reveal the limited information that they currently do about

how various factors impact a consumer’s credit score. Yet, there is much we don’t know. This

can lead housing professionals and credit and housing counselors to give inaccurate

information to consumers about how effectively to manage their credit. Moreover, since

Discriminatory Effects of Credit Scoring / page 27 National Fair Housing Alliance

different scoring mechanisms are used for different reasons, it may well be the case that when a

consumer does something to improve their insurance score, for example, the consumer’s credit

score will be negatively impacted.

Consumers and consumer counselors are generally uninformed about what should be done to

impact positively the consumer’s score. Making the scoring systems more transparent will help

consumers better manage their financial affairs. Making the system and the data more

transparent will also help advocates, financial institutions, federal regulators, and legislators.

Adequately assess the impact of credit scoring mechanisms on underserved groups

The Consumer Financial Protection Bureau, other federal banking regulators, and federal

enforcement agencies including DOJ and HUD should examine the impact of credit scoring

mechanisms on underserved groups and the population as a whole. The regulators should also

conduct disparate impact analyses of credit scoring systems. It is imperative that the data that

regulatory and enforcement agencies use to undertake these analyses come from a broad range

of sources. For example, regulators cannot rely predominantly on industry-developed data

upon which to conduct these evaluations.

Credit score developers should also conduct similar analyses of their own systems to identify

any fair lending concerns and to implement less discriminatory alternatives.

Reduce the over-reliance on credit scoring mechanisms

Credit-scoring mechanisms are an insufficient measure for predicting and managing

performance as the current crisis has revealed. Borrowers are not behaving as their credit

scores would indicate. Lenders, investors, regulators and legislators must caution against using

credit scores as a replacement for underwriting or the only assessment of risk. There are many

factors that affect loan risk including the presence of pre-payment penalties, inefficient

appraisals, poor documentation practices, and other abusive loan features. The credit score

may be the least significant factor when it comes to risk analysis. Therefore, lenders, investors,

regulators and legislators must adopt approaches that objectively consider other elements that

impact risk.

Evaluate product risk

In addition to reducing the reliance on credit scoring systems, federal regulators and legislators

should push for the evaluation of credit and financial services products. Additionally,

underwriting systems and practices should be evaluated for their level of risk. This information

should be readily available so consumers will know which products and which underwriting

practices pose the most risk and will therefore likely contribute to a negative credit score.

Providing objective information to consumers about threats associated with the products or

Discriminatory Effects of Credit Scoring / page 28 National Fair Housing Alliance

services they are considering will enable consumers to make informed and sound financial

decisions.

As discussed above, multiple studies reveal that unsafe products and unsavory underwriting

practices have a significant impact on loan performance and credit risk. It therefore is quite

practical to consider these functions in the risk analysis. Focusing analyses on borrower

characteristics will not improve the quality of the assessment of risk; rather, objectively

considering all factors that contribute to credit risk will result in a better analysis of risk

exposure.

Fix credit scores for victims of discrimination

Repairing credit scores damaged by discrimination or any other practice should be included in

complaint settlements and remedies. For example, the recent Justice Department settlement

with Countrywide demonstrated the discrimination against African Americans and Latinos in

steering and fees. Thousands of families who should have received prime loans were steered to

subprime loans. It is reasonable to assume that their credit scores were negatively impacted by

the mere fact that they received a more expensive subprime loan. Those borrowers should be

made whole and their remedies should include restoring their damaged credit scores.

Regulators, enforcement agencies, and the courts should fix credit scores as a matter of course

as part of remedies and settlements. In fact, this has already been done in some settlements

between banks and fair housing organizations and consumer groups. When predatory lending

was especially rampant in the early 2000s, fair housing organizations were sometimes

successful in complaint settlements in getting a borrower’s credit history amended. In

consultation with the credit reporting agency, the bank would have the trade line that applied

to the predatory loan deleted from the borrower’s credit report. This, in turn, erased the loan

from the borrower’s history as if it had never been made. In recent years, however, some

lenders have not agreed to delete the trade line entirely and instead have agreed only to report

the loan as “satisfied.” This means that the credit report shows that there is no debt remaining

on the loan but any history of late payments and other blemishes remains on the credit report

for the time allotted by the credit scoring agency. Unfortunately, because of the opacity of

credit scoring mechanisms, it is hard to tell which approach might be best for a specific

consumer at any given time.

Discriminatory Effects of Credit Scoring / page 29 National Fair Housing Alliance

Conclusion

By 2042, the majority of people in this country will be people of color. Given these changing

demographics, it is past time to figure out how to make our nation’s credit system work equally

for everyone. When civil rights groups called for a foreclosure moratorium on subprime loans

more than five years ago, predicting that the nation was headed for a financial and foreclosure

crisis and referencing the disproportionate damage these loans were causing in communities of

color,

Federal Reserve Chair Ben Bernanke told the groups that the problem of foreclosure

would be contained and restricted to the subprime market. The Mortgage Bankers Association

responded that, “Each loan is an individual transaction and situation, one which needs to be

addressed individually between the lender and the borrower.”

We all know now that these responses to the burgeoning crisis did not make sense and that

regulators and the industry failed to see the breadth of the ensuing crisis, despite the warnings

made by civil rights and consumer protection groups. The foreclosure problems not only went

beyond the U.S. subprime market but turned into an international economic crisis of

proportions not seen since the Depression.

Credit scoring mechanisms are negatively impacting the largest growing segments of our

country and economy. America cannot be successful if increasing numbers of our residents are

isolated from the financial mainstream and subjected to abusive and harmful lending practices.

Credit scores affect more and more of our daily activities and determine everything from

whether we can get a job and provide for our families to whether we will be able to successfully

own a home and build wealth for future generations. The current credit scoring systems work

against the goal of moving qualified consumers into the financial mainstream because they are

too much a reflection of our broken dual credit market. This paradigm must change.

We believe that the recommendations presented here are important steps towards broadening

access to good credit for all qualified borrowers.

“Civil rights groups urge freeze on foreclosures,” Los Angeles Times, April 6, 2007.

“MBA Chairman Robbins Responds to Call for Moratorium on Foreclosures,” Mortgage Bankers

Association news release, April 4, 2007.