Fair Lending: Federal Oversight and Enforcement Improved but Some Challenges Remain

Fair Lending: Federal Oversight and Enforcement Improved but Some
Challenges Remain (Chapter Report, 08/96, GAO/GGD-96-145).

Pursuant to congressional requests, GAO reviewed federal oversight and
enforcement of fair lending laws, focusing on: (1) efforts to strengthen
law enforcement procedures; and (2) challenges bank regulators face in
their efforts to detect discrimination and ensure compliance.

GAO found that: (1) since 1992, bank regulators and other enforcement
agencies have increased enforcement of fair lending laws and encouraged
greater compliance by the lending industry; (2) bank regulators have
also overhauled their compliance policies and examination procedures
which strengthened their ability to detect discriminatory lending
practices; (3) several agencies have recommended a number of compliance
programs and activities that could help lenders ensure fair treatment of
all loan applicants; (4) referrals to the Department of Justice (DOJ)
have increased significantly and DOJ has settled several civil suits
that alleged fair lending violations; (5) challenges to fair lending
oversight and enforcement include nonuniform examination procedures
among regulatory agencies, inadequate provisions to detect
discrimination prior to submission of a formal loan application,
inadequate disclosure data, examiners' inexperience and lack of
training, and insufficient time allowances; (6) unresolved legal issues
associated with the interpretation and application of these laws impede
formulation of concise guidance and leave lending institutions confused
and reluctant to implement voluntary compliance programs; and (7)
resolution of these problems may require civil or administrative
judicial proceedings or legislative action.

--------------------------- Indexing Terms -----------------------------

REPORTNUM: GGD-96-145
TITLE: Fair Lending: Federal Oversight and Enforcement Improved
but Some Challenges Remain
DATE: 08/96
SUBJECT: Mortgage loans
Racial discrimination
Sex discrimination
Banking regulation
Bank examination
Lending institutions
Reporting requirements
Law enforcement
Interagency relations
Civil procedure
IDENTIFIER: HUD Fair Lending Best Practices Agreement
FRB Fair Lending Comparative-File Analysis Model
HUD Fair Housing Initiatives Program
HUD Fair Housing Assistance Program
FHEO Fair Lending Complaint Investigation Program
Cleveland Residential Housing and Mortgage Credit Project

******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO report. Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved. Major **
** divisions and subdivisions of the text, such as Chapters, **
** Sections, and Appendixes, are identified by double and **
** single lines. The numbers on the right end of these lines **
** indicate the position of each of the subsections in the **
** document outline. These numbers do NOT correspond with the **
** page numbers of the printed product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
** A printed copy of this report may be obtained from the GAO **
** Document Distribution Center. For further details, please **
** send an e-mail message to: **
** **
** **
** **
** with the message 'info' in the body. **
******************************************************************

Cover
================================================================ COVER

Report to Congressional Requesters

August 1996

FAIR LENDING - FEDERAL OVERSIGHT
AND ENFORCEMENT IMPROVED BUT SOME
CHALLENGES REMAIN

GAO/GGD-96-145

Fair Lending

(233463)

Abbreviations
=============================================================== ABBREV

CCPA - Consumer Credit Protection Act
CMP - Civil Money Penalty
CRA - Community Reinvestment Act
DOJ - Department of Justice
ECOA - Equal Credit Opportunity Act
FCA - Farm Credit Administration
FDIC - Federal Deposit Insurance Corporation
FDICIA - Federal Deposit Insurance Corporation Improvement Act
FFIEC - Federal Financial Institutions Examination Council
FHA - Fair Housing Act
FHAA - Fair Housing Amendments Act
FHAP - Fair Housing Assistance Program
FHEO - Office of Fair Housing and Equal Opportunity
FHFB - Federal Housing Finance Board
FHIP - Fair Housing Initiatives Program
FIRREA - Financial Institutions Reform, Recovery, and Enforcement
Act
FRB - Federal Reserve Board
FTC - Federal Trade Commission
GSE - Government-Sponsored Enterprise
HMDA - Home Mortgage Disclosure Act
HUD - Department of Housing and Urban Development
MBA - Mortgage Bankers Association
MOU - Memorandum of Understanding
MSAs - Metropolitan Statistical Areas
NCUA - National Credit Union Administration
NFHA - National Fair Housing Alliance
OCC - Office of the Comptroller of the Currency
OFHEO - Office of Federal Housing Enterprise Oversight
OTS - Office of Thrift Supervision
SEC - Securities and Exchange Commission

Letter
=============================================================== LETTER

B-272004

August 13, 1996

The Honorable Henry B. Gonzalez
Ranking Minority Member
Committee on Banking and
Financial Services
House of Representatives

The Honorable Joseph P. Kennedy II
House of Representatives

This report responds to requests concerning federal oversight and
enforcement of the nation's fair lending laws, principally the Equal
Credit Opportunity Act and the Fair Housing Act. The report reviews
federal efforts to strengthen enforcement of the fair lending laws,
discusses the challenges federal regulators face in their efforts to
detect discrimination and ensure compliance, and recommends actions
to meet some of those challenges.

We are sending copies of this report to the U.S. Attorney General,
the Secretary of Housing and Urban Development, the Chairman of the
Board of Governors of the Federal Reserve System, The Chairwoman of
the Federal Deposit Insurance Corporation, the Comptroller of the
Currency, the Acting Director of the Office of Thrift Supervision,
and the Chairman of the National Credit Union Administration. We are
also sending copies to Members of the Senate Banking Committee, other
interested Committees and Subcommittees, and other interested
parties.

This report was prepared under the direction of Mark J. Gillen,
Assistant Director. Other major contributors are listed in appendix
XIII. If you have any questions, please call me on (202) 512-8678.

James L. Bothwell
Director, Financial Institutions
and Markets Issues

EXECUTIVE SUMMARY
============================================================ Chapter 0

PURPOSE
---------------------------------------------------------- Chapter 0:1

Concerns about discrimination in the credit markets, particularly in
the market for home mortgages, have existed for some time but have
recently moved to the fore among civil rights issues. This
resurgence has been due in large part to repeated reports in the
media and elsewhere that members of various racial and ethnic groups
are more likely to be denied credit for a home mortgage than are
white applicants with comparable income. Although no single one of
these reports has offered conclusive evidence that discrimination is
pervasive, their collective weight has raised concern.

Partly because of these reports, Members of Congress and others began
to question the effectiveness of federal oversight and enforcement in
the fair lending area and the zeal with which the nation's principal
fair lending laws--the Equal Credit Opportunity Act (ECOA) and the
Fair Housing Act (FHA)--have been enforced. The former Chairmen of
the House Committee on Banking, Finance and Urban Affairs and the
Subcommittee on Consumer Credit and Insurance asked GAO to (1) review
federal efforts to oversee and enforce the fair lending laws and (2)
discuss the challenges federal regulators face in their efforts to
detect discrimination and ensure compliance. To accomplish this
work, GAO reviewed agency fair lending policies and examination
procedures, interviewed federal agency officials and staff, consulted
with legal and academic experts, industry groups, and fair housing
advocates, and attended conferences, seminars, and workshops on fair
lending issues.

BACKGROUND
---------------------------------------------------------- Chapter 0:2

ECOA and FHA, enacted in 1974 and 1968, respectively, comprise the
federal civil rights statutes applicable to extensions of credit by
banks and other lending institutions. Together, these statutes--
referred to in this report as the "fair lending laws"--prohibit
discrimination in all forms of credit transactions, including
consumer and business loans as well as mortgage loans. To support
the enforcement of the fair lending laws, the Home Mortgage
Disclosure Act (HMDA), as amended, provides for disclosure to the
regulatory agencies and the public, information about mortgage loan
applicants and borrowers at certain lending institutions. Such
information is intended to be useful for identifying possible
discriminatory lending patterns.\1

General rulemaking authority for implementing the fair lending laws
is split between the Federal Reserve Board (FRB), which has such
authority for ECOA and HMDA, and the Department of Housing and Urban
Development (HUD), which has similar authority under FHA. For
depository institutions, compliance with ECOA, FHA, and HMDA is
primarily assessed through regularly scheduled consumer compliance
examinations conducted by the primary federal banking regulators.\2
In contrast, other nonbank lending institutions, such as independent
mortgage companies, are generally not subject to regularly scheduled
compliance examinations but may be periodically investigated for
noncompliance by HUD, the Department of Justice (DOJ), the Federal
Trade Commission (FTC), or other responsible federal agencies.

Enforcement responsibilities under the three acts follow a similar
pattern. The federal banking regulatory agencies are responsible for
administrative enforcement of ECOA and HMDA with respect to financial
institutions within their jurisdictions.\3 HUD, on the other hand,
has administrative enforcement authority with respect to FHA
violations for all institutions and HMDA compliance responsibilities
for independent mortgage companies.\4 Both ECOA and FHA provide for
civil suits by the DOJ and by private parties. Current law also
dictates that whenever the banking regulatory agencies or HUD have
reason to believe that an institution has engaged in a "pattern or
practice" of illegal discrimination, the agencies must refer these
cases to DOJ for possible civil action. Such cases include repeated,
regular, or institutionalized discriminatory practices. Other types
of cases also may be referred.

In 1992, DOJ filed its first major lawsuit alleging discriminatory
lending against a financial institution, Decatur Federal Savings and
Loan of Atlanta. Since then, DOJ has filed nine additional lawsuits
against financial institutions for "pattern and practice" violations
of the fair lending laws. Although most DOJ-initiated fair lending
lawsuits against financial institutions to date have been settled by
consent decrees before trial, significant remedies, including
affirmative initiatives and substantial monetary payments, have been
prescribed in the settlement agreements.

--------------------
\1 The Community Reinvestment Act (CRA) is often included in
discussions of the fair lending laws. However, CRA is distinct from
the antidiscrimination laws and is thus discussed in this report only
as it relates to fair lending. CRA is, however, the subject of a
separate GAO report, Community Reinvestment Act: Challenges Remain
To Successfully Implement CRA (GAO/GGD-96-23, Nov. 28, 1995).

\2 The primary banking regulatory agencies are FRB, the Comptroller
of the Currency (OCC), the Federal Deposit Insurance Corporation
(FDIC), the Office of Thrift Supervision (OTS), and the National
Credit Union Administration (NCUA).

\3 ECOA specifically charges certain other agencies with
responsibility for enforcing the act with respect to persons and
matters under their jurisdictions. The Federal Trade Commission
(FTC) has enforcement responsibility in all other areas. 15 U.S.C.
� 1691(c).

\4 The federal banking regulators may also take enforcement actions
for FHA violations to the extent provided in their general
enforcement authorities, which are contained in section 8 of the
Federal Deposit Insurance Act, 12 U.S.C. 1818.

RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:3

Following years of little activity, the banking regulators, DOJ, HUD,
and other responsible federal agencies have begun to devote
additional effort toward improving compliance with the nation's fair
lending laws. Since 1992, these agencies have moved to step up
enforcement of the fair lending laws and to heighten the level of
awareness and sensitivity of the lending community to its
responsibilities under these laws. The banking regulatory agencies
have also moved to strengthen their ability to detect discrimination
through improved examination procedures. In addition, these and
other federal agencies have recommended a number of compliance
programs and activities that, if implemented, could help lenders
ensure that their loan applicants are treated fairly.

Yet, despite the overall improvement in federal fair lending
oversight and enforcement efforts, challenges still remain in some
areas. On the basis of its review of fair lending compliance
policies and examination procedures of the federal banking regulatory
agencies, GAO identified several areas related to the regulators'
existing examination procedures where it believes the agencies have
not taken full advantage of opportunities to strengthen their ability
to detect lending discrimination in all of its forms and improve the
consistency of oversight and enforcement. For example, fair lending
examination procedures of the federal banking regulatory agencies
currently lack adequate means to detect discrimination that could
occur prior to submission of a formal loan application.
Additionally, compliance examiners at the FDIC, OCC, FRB, and OTS
have indicated that detecting discrimination during fair lending
examinations was made more difficult by poor quality HMDA data,
examiner inexperience, and to some extent by insufficient time
allowances. In addition, GAO also learned that uncertainty persists
among some federal agency officials and examiners about what
constitutes a referable "pattern or practice" violation under ECOA
and FHA.

GAO also found that a number of other legal issues related to
interpretation and application of the fair lending laws remain
unresolved, creating uncertainty among both lenders and regulators.
This uncertainty has, in turn, somewhat impeded current attempts by
the federal banking regulatory agencies, DOJ, and HUD to provide
clearer and more concise guidance regarding fair lending
policies--leaving banks and other lending institutions confused about
what, precisely, is needed to ensure that they are in compliance.

Unresolved legal issues also pose potential barriers to achieving
wider adoption of some of the programs and activities recommended by
federal agencies to ensure compliance. Perhaps principal among these
issues are questions surrounding the establishment of proof under the
disparate impact theory of lending discrimination,\5 and the use by
regulators and third parties of data acquired or generated by lenders
through self-testing programs.

Compounding the problems created by uncertainty is that solutions to
many of the legal questions may be difficult to achieve by regulation
and thus may require resolution through civil or administrative
judicial proceedings or legislation. Consequently, it could take
some time before the contours of the fair lending laws become more
distinct. In the meantime, controversies surrounding the
interpretation and application of the fair lending laws are likely to
remain, as is the reluctance of some lending institutions to
implement additional compliance programs.

--------------------
\5 According to this theory, a lender commits discrimination if the
lender applies a seemingly innocuous policy or practice equally to
all credit applicants but with the result that the policy or practice
has a disproportionate adverse impact on applicants from a protected
group. Furthermore, it must be the case that the policy or practice
is not justified by a business necessity.

PRINCIPAL FINDINGS
---------------------------------------------------------- Chapter 0:4

FAIR LENDING OVERSIGHT AND
ENFORCEMENT HAVE IMPROVED
-------------------------------------------------------- Chapter 0:4.1

Prior to DOJ's filing of its first lending discrimination lawsuit
under ECOA and FHA in 1992, federal enforcement of the nations's fair
lending laws was limited. Before that time, only a few complaints
involving lending discrimination had been filed under either ECOA or
FHA by federal agencies. However, actions taken since then by
Congress, the administration, federal banking regulatory agencies,
DOJ, HUD, and others have increased federal oversight and, as a
consequence, have contributed to a regulatory environment that is
more likely to engender greater compliance by the lending industry.

More vigilant enforcement of the fair lending laws was made possible,
in part, by Congress, which over the years has passed a progression
of amendments to the fair lending statutes and related laws aimed at
strengthening their public and private enforcement mechanisms. Among
the more notable of these were amendments to HMDA, which provided for
the creation of a database on mortgage lending activity for use by
both regulators and the public, and the 1988 amendments to FHA, which
created an administrative enforcement system at HUD that did not
previously exist. These legislative initiatives, combined with the
attention focused on the issue by the media and concerned community
groups, spurred DOJ to initiate, in 1989, its first investigation of
lending discrimination under ECOA. DOJ's investigation culminated in
the 1992 complaint against Decatur Federal, which was significant in
that it signaled the beginning of a more aggressive federal
enforcement program and set forth DOJ's view of evidentiary
requirements sufficient for claiming a "pattern or practice" of
unlawful lending discrimination under ECOA. The investigation also
marked the beginning of a trend toward increased reliance on
statistical analysis to establish evidence of discriminatory lending
patterns.

The movement by regulatory agencies toward more energetic enforcement
of the fair lending laws received additional momentum in 1993 when
the incoming administration declared that the development of an
effective and aggressive fair lending enforcement program would be a
top priority and promptly formed an interagency task force charged
with clarifying federal regulatory policies in the fair lending area.
Marking the first time that the federal agencies primarily
responsible for fair lending enforcement had spoken with one voice on
the topic of lending discrimination, the task force released a
landmark policy statement in March 1994 outlining how the regulatory
agencies are to define and enforce ECOA and FHA.

Another significant effort by federal agencies, which has contributed
to an overall improvement of the regulatory environment, has been the
encouragement of preventative measures by the lending industry. In
May 1993, the bank and thrift regulators released a list of
recommended programs and activities for financial institutions to
consider when planning their fair lending compliance programs. Among
these recommendations were such things as second-review programs for
applications that may otherwise be denied, participation on
multilender mortgage review boards, affirmative marketing programs,
enhanced employee training on fair lending issues, and self-testing
programs. In 1994, HUD initiated its more formal "best practices"
program in which independent mortgage companies were urged to sign
individually negotiated, nonbinding agreements with HUD that
encourage initiatives similar to those recommended for the banking
industry. The regulators' intent in encouraging lenders to consider
these types of programs was to help financial institutions design
compliance programs that would ensure that unlawful discrimination
does not affect the fair and even-handed distribution of credit in
society.

Finally, over the last several years, the federal banking regulatory
agencies have undertaken a major overhaul of their policies and
procedures for ensuring compliance with the fair lending laws. The
new procedures broadened the regulators' search for discriminatory
lending practices by, among other things, emphasizing
"comparative-file analysis"--a methodology that seeks to detect
unequal or disparate treatment among similarly qualified applicants.

CHALLENGES REMAIN IN
OVERSIGHT AND ENFORCEMENT
-------------------------------------------------------- Chapter 0:4.2

Despite the progress federal regulators have made toward improving
oversight and enforcement of the fair lending laws, there remain
several issues that present significant and continuing challenges to
the efforts of federal regulators to detect discrimination and ensure
consistent oversight and enforcement.

First, current fair lending examination procedures are not uniform
across banking regulatory agencies--creating a situation in which
application and enforcement of the laws could vary by regulator.
Since a failed effort at developing uniform examination procedures in
1992, the federal banking regulatory agencies have each pursued
independent efforts to improve their own procedures. This has
resulted in a situation in which some depository institutions may be
subject to compliance examinations involving the use of newer
detection methodologies like regression analysis or testing, while
others may not. In light of the legal and economic consequences of
even being accused of discriminatory lending practices, as
illustrated in recent DOJ settlement agreements, the lack of uniform
examination procedures raises an important equity issue regarding the
evenhanded application of the law to all depository institutions.

Second, procedures to detect discrimination prior to submission of a
formal application are inadequate. At this point in the mortgage
application process, there are no documents or records for bank
examiners to review. Although current examination procedures call
for examiners to routinely interview those bank personnel who serve
as initial contact points for potential applicants, it is extremely
difficult, if not impossible, for an examiner to determine (on the
basis of information gathered through interviews) whether any
applicant was illegally discouraged from making a formal application
or was steered to a less advantageous product or institution.

Third, concerns about the accuracy of HMDA data--which is intended to
assist regulatory agencies and the public in identifying possible
discriminatory lending patterns--have been long standing and have
recently received credence from special examination efforts by FRB
and FDIC. Yet, not until fiscal year 1994 did a federal agency use
its full range of enforcement authorities in an effort to ensure
compliance with HMDA data reporting requirements. Typically,
enforcement of HMDA reporting requirements by the responsible federal
agencies has been limited to requiring institutions to correct and
resubmit the required data. In fiscal year 1994, HUD and FDIC began
imposing civil money penalties for inaccurate or untimely submission
of HMDA data. However, the other federal regulatory agencies with
HMDA oversight responsibilities have not yet taken similar action,
although OTS has adopted specific guidelines for assessing civil
money penalties (CMP) for future HMDA violations. Hence, even though
HMDA data are widely used by both federal regulatory agencies and the
public in overseeing compliance with the fair lending laws, the
quality of the data remains suspect.

Fourth, responses to GAO surveys of bank compliance officers and
agency examiners at FRB, OCC, FDIC, and OTS, and other sources, have
indicated that additional examiner training in the latest fair
lending examination and detection techniques would be beneficial.
Also, in some instances, examiners felt they were not allowed
sufficient time to develop evidence of substantive violations.

Finally, several officials and examiners at the federal agencies
responsible for fair lending oversight have expressed some
uncertainty regarding the meaning of statutory phrases like "reason
to believe" and "pattern or practice," which are pivotal to the
referral of suspected cases of discrimination to DOJ.

While issues such as those just mentioned can be directly confronted
and addressed by the responsible agencies, others, because of their
nature, are more problematic and defy immediate resolution. For
example, lending discrimination has often proven to be subtle,
leaving the victim or victims unaware that they have been
discriminated against. Also, in some cases, discriminatory lending
patterns may become apparent only with the use of complex statistical
analyses. This subtlety makes detection difficult even with newer,
more advanced techniques. Additionally, because the contours of the
fair lending laws are still evolving, there remain a number of
unresolved legal issues associated with the interpretation and
application of the fair lending laws. These unresolved issues
contribute to uncertainty within the lending industry and pose
potential barriers to achieving wider adoption of some of the
programs and activities recommended by federal agencies to ensure
compliance with the fair lending laws. Perhaps principal among these
issues are questions surrounding the establishment of proof under the
disparate impact theory of lending discrimination, and the use by
regulators and third parties of data acquired or generated by lenders
through self-testing programs.\6

As interpreted by the courts and the federal regulatory agencies,
there currently exists under the fair lending laws a disparate impact
(effects) test for discrimination. Under this test a lender commits
lending discrimination if the lender maintains a neutral policy or
practice that has a disproportionate, adverse effect on members of a
protected group and which cannot be justified by "business
necessity," or for which a less discriminatory alternative is shown
to exist.\7 The application of disparate impact analysis to some
common practices inherent to the financial services industry,
however, could prove to be problematic. For example, some legal
experts have questioned whether and how the disparate impact test
would affect the use of differential and tiered pricing systems based
on perceived credit risk. Also, others have indicated that the
disparate impact test could pose compliance problems for banks that
employ computerized underwriting systems. Some bankers are uncertain
whether such standardized systems using uniform criteria would pass a
disparate impact test, given the relative socioeconomic status of
some protected groups.

In addition to the questions raised about the application of the
disparate impact test, the banking industry has also expressed
concern regarding the recommended use of self-testing programs. The
data obtained through self-testing generally are not protected from
disclosure to federal agencies or private litigants. Hence,
institutions that undertake such programs in good faith run the risk
of exposing fair lending violations that could result in
administrative or civil sanctions.

Resolution of some of the legal questions related to the disparate
impact test and other issues may only be possible through civil or
administrative judicial proceedings, or by legislative action.
Hence, it could be some time before settlement of these issues is
forthcoming. Meanwhile, application and enforcement of the law will
probably continue to generate controversy--with some voluntary
compliance activities possibly discouraged as a consequence.

Despite these and other potential difficulties with some of the
recommended fair lending compliance activities, it has been reported
that many financial institutions have, nevertheless, moved to adopt a
number of the agencies' suggestions into their compliance programs.
Although GAO has no basis for evaluating these voluntary efforts, it
believes voluntary compliance is key to achieving the ultimate goal
of the fair lending laws--preventing unlawful discrimination before
it occurs.

--------------------
\6 Self-testing programs are voluntarily initiated internal auditing
procedures that may employ the use of "mystery shoppers" to check on
an institution's compliance with the fair lending laws.

\7 Even if a lender's policy or practice were justified by business
necessity, the policy or practice nonetheless may be discriminatory
if it is shown that a less discriminatory alternative would achieve
the same purposes or results and the challenged institution refused
to adopt it. See 42 U.S.C. � 2000e-2(k)1(A) (ii).

MATTER FOR CONGRESSIONAL
CONSIDERATION
---------------------------------------------------------- Chapter 0:5

Congress may wish to consider measures that would remove or diminish
the disincentives associated with self-testing by alleviating the
legal risks of self-testing when conducted by lenders who in good
faith are seeking to prevent discriminatory lending activity and who
move to correct such discriminatory practices when they are
identified.

RECOMMENDATIONS
---------------------------------------------------------- Chapter 0:6

Despite significant improvement in federal fair lending oversight and
enforcement, GAO believes that further efforts can still be made in
some areas, which would strengthen the ability of federal banking
regulators to detect lending discrimination in all of its forms and
help ensure greater consistency in oversight and enforcement. To
this end, GAO recommends that the heads of FRB, FDIC, OCC, OTS, and
NCUA:

-- work together to develop and adopt uniform fair lending
examination procedures and provide all compliance examination
staff with the necessary training to implement those procedures;

-- adopt as a component of their fair lending examination and
training programs, guidelines and procedures for the use of
testing methodologies for detection of discrimination at the
preapplication stage of the lending process; and

-- use their full range of enforcement authority, including the use
of civil money penalties, to ensure that the HMDA data is
submitted in a timely and accurate manner.

GAO also recommends that the U.S. Attorney General provide updated
guidance to the banking regulatory agencies and HUD on the
characteristics of referable "pattern or practice" cases under ECOA
and FHA.

AGENCY COMMENTS
---------------------------------------------------------- Chapter 0:7

GAO requested and received comments on a draft of this report from
FRB, OCC, FDIC, OTS, NCUA, HUD, and DOJ. These written comments
appear along with GAO's responses in appendixes VI - XII. NCUA did
not comment on the conclusions and recommendations. Overall, the
agencies were in general agreement with the report findings,
conclusions, and recommendations. Agency remarks regarding
individual recommendations are summarized below.

All agencies expressed general agreement with GAO's recommendation
that the banking regulatory agencies adopt uniform fair lending
examination procedures and provide appropriate training with respect
to such procedures. OCC, however, maintained that some differences
in examination procedures were appropriate, given the supervisory
needs of the agencies and the varying sizes and risk profiles of the
financial institutions they regulate. FRB commented that they have
been working on the development of uniform procedures and anticipated
submitting a draft for consideration by other agencies in the very
near future. FRB noted, however, that their draft procedures did not
include a testing component or recommend the routine use of
regression analysis.

Finally, FDIC, OCC, and HUD expressed support for our suggestion that
Congress consider legislative initiatives that remove or diminish the
disincentives associated with self-testing by protecting institutions
from having to release results of self-testing reviews when they are
followed by actions to correct any discriminatory behavior that may
have been discovered. In their comments HUD stressed how important
it was that any protection be granted only in those cases in which
lenders promptly corrected the problems found through self-testing.

While there are some differences in the agencies' response to our
recommendations, these differences generally reflect the agencies'
desire to retain the discretion necessary to consider the specific
facts and circumstances of individual cases. GAO agrees that agency
discretion is necessary once due consideration has been given to the
full range of regulatory alternatives and analytical techniques
available to ensure effective fair lending oversight of financial
institutions.

INTRODUCTION
============================================================ Chapter 1

Concerns about discrimination in the credit markets, particularly in
the market for home mortgages, have existed for some time and have
recently moved to the fore among civil rights issues. This
resurgence of interest is due in large part to repeated reports in
the media and elsewhere suggesting that the problem of lending
discrimination is of such magnitude that it demands immediate and
increased public attention. Most prevalent among these reports are
ad hoc statistical analyses of data collected under the Home Mortgage
Disclosure Act (HMDA), which reveal that applicants who are members
of various racial and ethnic groups are more likely to be denied
credit for a home mortgage than are white applicants with comparable
incomes. In addition, a few widely-cited econometric studies of
mortgage lending decisions have also reported results that indicate
that lenders discriminate against minority applicants.

While none of these studies or analyses has offered conclusive
evidence that lending discrimination is pervasive or represents an
industrywide problem, their collective weight has raised concerns.
In light of these continuing reports, Congress and others have
questioned the effectiveness of supervision and oversight in the fair
lending area and the zeal with which the nation's antidiscrimination
laws governing the credit markets have been enforced. As a
consequence, responsible federal agencies have rededicated themselves
to upholding the fair lending laws, and have revamped outdated
supervisory and compliance policies as well as examination procedures
in the fair lending area. Even so, the fair lending laws still pose
many difficult policy and legal questions and offer a number of
operational challenges to federal oversight enforcement efforts.

This report responds to a request from the former Chairmen of the
House Committee on Banking, Finance and Urban Affairs, and its
Subcommittee on Consumer Credit and Insurance, asking us to review
federal efforts to oversee and enforce the nation's "fair lending
laws"--principally the Equal Credit Opportunity Act (ECOA) and the
Fair Housing Act (FHA)--and to discuss the challenges federal
regulators face in their efforts to detect discrimination and ensure
compliance. In conjunction with the Chairmen's request to review
enforcement of the fair lending laws, we were also asked to evaluate
compliance with the Community Reinvestment Act (CRA) and report on
the strengths and weaknesses of current reform efforts. Our response
to the request regarding CRA is contained in a separate report.\1

--------------------
\1 See Community Reinvestment Act: Challenges Remain to Successfully
Implement CRA, (GAO/GGD-96-23, Nov. 28, 1995).

BACKGROUND
---------------------------------------------------------- Chapter 1:1

THE FAIR LENDING LAWS
-------------------------------------------------------- Chapter 1:1.1

During the late 1960's and 1970's, Congress enacted a number of laws
for the purpose of ensuring fair and equitable access to credit for
both individuals and communities. These laws included the Fair
Housing Act (1968), the Equal Credit Opportunity Act (1974), the Home
Mortgage Disclosure Act (1975), and the Community Reinvestment Act
(1977). Two of these laws, ECOA and FHA, comprise the federal
antidiscrimination statutes applicable to lending practices and have
become commonly referred to as the "fair lending laws." Additionally,
HMDA, as amended in 1988, is intended to support enforcement of ECOA,
FHA, and CRA by requiring certain lending institutions to provide
federal regulators and the public with information on mortgage loan
applicants and borrowers. Such information can be useful in
identifying possible discriminatory lending patterns. Unlike ECOA
and FHA, HMDA does not prohibit any specific activity of lenders, but
only establishes a reporting obligation for particular institutions.
For more detail on the scope, applicability, and evolution of these
laws and their related implementing regulations, see appendix I.

OVERSIGHT RESPONSIBILITY
-------------------------------------------------------- Chapter 1:1.2

General rulemaking authority for implementing the fair lending laws
is divided between the Federal Reserve Board (FRB), which has such
authority for ECOA and HMDA, and the Department of Housing and Urban
Development (HUD), which has similar authority for FHA. Oversight
and enforcement responsibilities, however, are divided among at least
12 separate federal agencies, including but not limited to the 5
federal banking regulatory agencies\2 , the Department of Justice
(DOJ), HUD, the Federal Trade Commission (FTC), the Federal Housing
Finance Board (FHFB), the Office of Federal Housing Enterprise
Oversight (OFHEO), the Securities and Exchange Commission (SEC), and
the Farm Credit Administration (FCA).\3 These agencies use varying
approaches to discharge their oversight and enforcement
responsibilities.

For depository institutions, compliance with ECOA, FHA, and HMDA is
primarily assessed through regularly scheduled consumer compliance
examinations conducted by the federal bank regulators. These
agencies are also responsible for administrative enforcement of both
ECOA and HMDA, whereas HUD has such authority for FHA (and for HMDA
in the case of independent mortgage companies). In addition, federal
regulatory agencies are required to refer to DOJ any matter in which
an agency has reason to believe that a pattern or practice of lending
discrimination has occurred. Moreover, the agencies are also
required to notify HUD of apparent FHA violations.\4 In contrast,
mortgage companies and other nondepository lending institutions are
generally not subject to regularly scheduled compliance examinations,
but may be periodically investigated for noncompliance by their
primary regulator or HUD, DOJ, or FTC.

--------------------
\2 The primary banking regulatory agencies are FRB, the Comptroller
of the Currency (OCC), the Federal Deposit Insurance Corporation
(FDIC), the Office of Thrift Supervision (OTS), and the National
Credit Union Administration (NCUA).

\3 Other agencies that enforce ECOA include the Department of
Transportation, the Interstate Commerce Commission, the Small
Business Administration, and the Packers and Stockyards
Administration of the Department of Agriculture.

\4 Specifically, the agency must notify HUD when the agency has
reason to believe that an ECOA violation also violated FHA and the
matter was not referred to DOJ as a pattern or practice case. See 15
U.S.C. � 1691e(k). Executive branch agencies are also required to
notify HUD of FHA violations and complaints under Executive Order
12892. See Federal Register, Vol. 59, No. 13, Thursday, Jan. 20,
1994.

THE ROLE OF PRIVATE
LITIGATION
-------------------------------------------------------- Chapter 1:1.3

In addition to federal supervision, private litigation also serves as
a prominent enforcement method under the fair lending laws. Under
both ECOA and FHA, however, private civil suits must be brought
within prescribed periods--generally within 2 years of the date of
the occurrence or termination of the alleged violation. Individual
claimants under ECOA can recover actual damages and punitive damages
up to $10,000. In class action suits, punitive damages are limited
to the lesser of $500,000 or 1 percent of the institution's net
worth. Under FHA, plaintiffs can also recover both kinds of damages,
but the amount of punitive damages is not limited.

To assist private enforcement of FHA, amendments to FHA in 1988
created within HUD an administrative system within which a
complainant could pursue a fair lending complaint at little or no
cost. Within HUD's system, a case can be heard in a formal
administrative proceeding before an administrative law judge, or the
complainant or respondent can elect to move the case to federal
district court.

OBJECTIVES, SCOPE, AND
METHODOLOGY
---------------------------------------------------------- Chapter 1:2

The objectives of this study were to (1) review recent federal
efforts to oversee and enforce the fair lending laws and (2) discuss
the challenges federal regulators face in their efforts to detect
discrimination and ensure compliance. To achieve this, we began by
reviewing the legislative and case history of the laws to discern
their original intent and to see how the interpretation of that
intent has evolved over time. We also undertook an extensive review
of the literature on lending discrimination. We then interviewed
officials from the five federal banking regulatory agencies, HUD, and
DOJ. These interviews served to help us more fully understand the
existing regulatory system, including the practices and procedures
used to detect discrimination, and the policies in place to deal with
violations, especially violations of a serious or substantive nature.
Information and documentation regarding various aspects of the
agencies' fair lending oversight and enforcement activities were
collected from all of the federal banking regulators, HUD, FTC, and
DOJ. Also, the impressions and experiences of 40 bank compliance
officers and agency examiners, who were participants in compliance
examinations during 1993, were obtained through surveys and
interviews. These surveys and interviews were conducted as part of
our work related to our review of CRA. In that review, the case
study approach was used to review in detail compliance examinations
at 40 banks and thrifts. The institutions included in the review
were judgmentally chosen to represent a cross section of geographic
areas, federal banking regulatory agencies, and depository
institutions. Of the institutions studied, 6 were examined by FRB,
13 by FDIC, 9 by OCC, and 12 by OTS.

In addition to contacts with federal agency officials, we also
judgmentally selected and interviewed individuals from the banking
industry, industry trade groups, and consumer groups, as well as a
number of private consultants, academic and legal experts, and
officials at firms active in the secondary mortgage market. We also
attended workshops and conferences on fair lending sponsored by a
variety of industry and community groups, professional law
associations, and the federal banking regulatory agencies. Finally,
we reviewed letters submitted by bankers and other concerned parties
responding to public policy statements and proposed rule changes
related to the fair lending laws.

We did the work underlying this report between January 1994 and
December 1995 in accordance with generally accepted government
auditing standards. We requested comments on a draft of this report
from FRB, FDIC, OCC, OTS, NCUA, HUD, and DOJ. Their written
comments, along with our evaluation, are summarized at the end of
chapter 4 and are presented in appendixes VI through XII.

ASSESSING DISCRIMINATION IN THE
CREDIT MARKETS
============================================================ Chapter 2

The fair lending laws prohibit discrimination with regard to any
aspect of a consumer, commercial, or real estate credit transaction
based on a number of protected personal characteristics, including
but not limited to race, color, religion, gender, and national
origin. However, interpretation and applicability of the fair
lending laws is unclear in a number of areas and federal regulatory
agencies have had to translate and apply the laws to a wide variety
of lending practices, including some that are not specifically
mentioned in the statutes. Lenders, federal regulators, and
consumers have not always agreed on these interpretations. As a
result, there has been, and continues to be, some controversy and
confusion regarding the scope and applicability of the laws and what
it means in practice to discriminate against a credit applicant.
Moreover, this uncertainty has been compounded by numerous
difficulties encountered in measuring and assessing the nature of
lending discrimination.

DEFINING DISCRIMINATION
---------------------------------------------------------- Chapter 2:1

In fair lending cases, courts and the federal agencies have adopted
the analytical framework applicable to employment discrimination
cases under Title VII of the Civil Rights Act of 1964.\1 Under this
framework, two methods of analysis are used to evaluate
discrimination claims. The first method, known as disparate
treatment analysis, determines whether a borrower has been treated
less favorably than his or her peers due to race, sex, or other
characteristic that places the individual within a group protected by
ECOA or FHA. The second method, known as disparate impact analysis,
determines (a) whether a seemingly innocuous lending policy or
practice has a disproportionately adverse effect on a protected group
and, if so, whether the policy or practice can be justified by
business necessity; and (b) whether a less adverse alternative to
such a policy or practice exists.\2

Both of these tests have generally recognized that discriminatory
behavior can enter the credit search and lending processes in a
number of forms and at various stages. For example, the courts have
held that discrimination can be blatant or it can consist of
extremely subtle behavior that may leave the victim or victims
unaware that they have been treated unfairly. Furthermore, the
courts have ruled that in disparate impact cases, discrimination can
occur even without actual, subjective animus against an individual or
a protected group. In other words, when a protected group is
adversely affected, discrimination may be found even in the absence
of any discriminatory intent.

--------------------
\1 See 42 U.S.C. � 2000e - 2000e-17.

\2 For a discussion of the disparate treatment and disparate impact
models as they apply to employment law, see Richey (1994). For
relevant case law and a description of the legal requirements
necessary for establishing proof of discrimination under ECOA, see
Lieberman (1994).

THREE TYPES OF LENDING
DISCRIMINATION
-------------------------------------------------------- Chapter 2:1.1

In practice, all potential violations of the fair lending laws have
been categorized by federal regulators into three distinct types of
discriminatory behavior: (1) blatant or overt, (2) disparate
treatment, and (3) disparate impact (see table 2.1). To help lenders
and examiners better understand the various forms that lending
discrimination can take, the Interagency Task Force on Fair Lending
released a policy statement in March 1994 that provides examples of
the three types of discriminatory behavior and illustrates through
hypothetical case examples the types of lending-related activities
that might be considered to constitute illegal discrimination.\3

Table 2.1

Three Types of Lending Discrimination

Blatant or overt Disparate treatment Disparate/adverse impact
-------------------- ---------------------------- ----------------------------
Definitions
--------------------------------------------------------------------------------
Occurs when a lender Occurs when two or more Occurs when a seemingly
openly discriminates applicants, who are similar innocuous policy or practice
against an in most respects except for is applied equally to all
individual or group some protected credit applicants but with
based on a protected characteristic like race, the result that the policy
factor under ECOA or age, sex, etc., receive or practice has a
FHA (race, marital different treatment based on disproportionate adverse
status, sex, age, that characteristic. impact on applicants from a
etc.). Disparate treatment can protected group. It must be
range from blatant the case that the policy or
discrimination to more practice is not justified by
subtle disparities in a business necessity and
treatment and does not that a less discriminatory
require that such treatment alternative does not exist.
be motivated by prejudice or
a conscious intent.

Ease of Identification
--------------------------------------------------------------------------------
Blatant/overt Disparate treatment may more Of the three types,
discrimination is likely occur in the detection of disparate
usually easily treatment of applicants who impact is the most difficult
identified. are neither clearly well- and encompasses several
qualified nor clearly steps. First, the existence
unqualified. Hence, it is of disparate impact must be
generally more difficult to established. Frequently this
detect and necessarily is done by quantitative or
involves detailed statistical analysis. Once a
comparative-file analysis or disparate impact has been
some other means of identified, it must then be
detection capable of determined if the policy or
identifying differences in practice is justified by
treatment among similar business necessity. If so,
applicants. it must then be shown that a
less discriminatory
alternative to the policy or
practice existed and that
the lender refused to adopt
the alternative policy.

Examples
--------------------------------------------------------------------------------
It may include, A nonminority applicant A lender's policy is not to
e.g., lender's applied for a loan. The extend loans for single
outright refusal to lender found adverse family residences for less
assist applicant or information in the than $60,000. This minimum
refusal to accept applicant's credit report. loan amount policy is shown
application on a The lender discussed the to disproportionately
prohibited basis. report with the applicant exclude potential minority
and determined that the applicants from
information was incorrect. consideration because their
The nonminority applicant income levels or the value
was granted the loan. A of the houses in the areas
minority applicant applied where they live would
for a similar loan with the qualify them only for
same lender. Upon smaller loan amounts. The
discovering adverse lender would be required to
information in the minority justify the business
applicant's credit report, necessity for the policy.
the lender denied the loan
application on the basis of
the adverse information
without giving the minority
applicant an opportunity to
discuss the report.
--------------------------------------------------------------------------------
Source: Adopted from the Interagency Policy Statement on
Discrimination in Lending, Federal Register, Vol. 59, No. 73,
Friday, Apr. 15, 1994, pp. 18266-18274.

Under ECOA and FHA, the statement explained, lenders may not, because
of a prohibited factor,

-- fail to provide information or services or provide different
information or services regarding any aspect of the lending
process, including credit availability, application procedures,
or lending standards;

-- discourage or selectively encourage applicants with respect to
inquiries about or applications for credit;

-- refuse to extend credit or use different standards in
determining whether to extend credit;

-- vary the terms of credit offered, including the amount, interest
rate, duration, or type of loan;

-- use different standards to evaluate collateral;

-- treat a borrower differently in servicing a loan or invoking
default remedies;

-- use different standards for pooling or packaging a loan in the
secondary market;

-- express, orally or in writing, a preference for or against
protected applicants;

-- discriminate because of a person associated with a credit
application (for example, a co-applicant, spouse, business
partner, or live-in aide); or

-- discriminate because of the present or prospective occupants of
the area where property to be financed is located.

The policy statement is particularly notable in part because it
represents the first time that the federal regulatory agencies
comprising the Task Force have spoken with one voice on the subject
of fair lending. As such, the respective agencies believe it
represents a significant step in arriving at a uniform and effective
fair lending policy. However, individual lenders and banking trade
associations, while applauding the agencies' efforts, have expressed
serious concerns about certain aspects of the policy and its
implementation. Among other things, lenders have expressed concern
that the new fair lending policy will not be implemented in a manner
consistent with safety and soundness; i.e., they are concerned that
safety and soundness examiners will question the adequacy of relaxed
underwriting standards for loans made to applicants in protected
groups who may not otherwise be deemed creditworthy. Lenders have
also voiced concern that enforcement will not be administered
equitably across industry segments, since depository institutions are
examined regularly for fair lending compliance and other lenders,
like mortgage companies, are not. Lenders have further argued that,
in some instances, the statement goes beyond the strictures of
settled law, particularly with respect to disparate impact theory.

In addition, the agencies' interpretation of the fair lending laws
could, lenders contend, result in unintended consequences in a number
of areas. For example, bankers point out that the policy has the
potential to eliminate so-called "character lending" by forcing the
adoption of credit scoring systems in an effort to demonstrate that
all applicants are treated alike. They suggest that this could
actually restrict, rather than expand, credit to marginal applicants
as banks amend underwriting criteria and practices to remove elements
of judgment in the lending process. Hence, while the federal
regulatory agencies and the lending industry are both on record as
being committed to combating lending discrimination, the two do not
necessarily agree on the legal definitions and techniques to be used
in implementing fair lending policy.

--------------------
\3 The task force consists of representatives from HUD, DOJ, FRB,
OCC, FDIC, OTS, NCUA, FHFB, FTC, and OFHEO. The statement was
subsequently published as "Policy Statement on Discrimination in
Lending," Federal Register, Vol. 59, No. 73, Friday, Apr. 15,
1994, pp. 18266-18274.

EXTENT OF PROBLEM REMAINS
UNKNOWN
-------------------------------------------------------- Chapter 2:1.2

Much has been written on the topic of discrimination in housing and
housing-related practices over the last 20 years, and reviews of the
literature by a number of reputable researchers aptly assess and
critique the myriad of individual studies related to those
subjects.\4 Upon review, the literature on discrimination in housing
seems generally well developed, and there appears to be widespread
recognition that discrimination in the housing market exists and has
been and continues to be a serious problem. Evidence of racial and
ethnic steering and discriminatory real estate marketing and
advertising practices, for example, is particularly well documented.

In contrast, the literature on the subject of lending discrimination
is not nearly as extensive and a consensus has not yet been reached
regarding the nature and pervasiveness of discrimination in the
mortgage markets and elsewhere in the credit industry. Indeed, one
scholar well known in the discrimination literature may have said it
best when he observed that "given its social importance and media
attention, it is staggering that researchers in fact have little
definitive knowledge about the existence and severity of
discrimination in mortgage markets."\5

The absence of consensus on the nature and pervasiveness of lending
discrimination may be explained, at least in part, by the complexity
of the lending process itself. The home-buying process, for example,
is a multifaceted and perplexing process that involves many different
participants. The process and the roles of the various agents,
brokers, underwriters, insurers, lenders, and even the buyers
themselves are not well understood. Additionally, the
multifariousness of the process has presented a number of
methodological obstacles to researchers that have not yet been
resolved. Thus, our perspective is limited by the imprecise and
often contradictory nature of the work to date. A description of the
various press accounts, HMDA data analyses, statistical studies, and
other sources of information is provided in appendix II.

For example, statistical studies examining the lending discrimination
issue have not yielded consistent results. Although several studies
of loan application denial rates have reported finding evidence of
disparate treatment among mortgage lenders, recent studies of
redlining (the refusal of lenders to make loans in certain geographic
areas regardless of the creditworthiness of the individual loan
applicant) and mortgage default rates have generally not found such
evidence. Additionally, a number of reputable researchers have
argued that the findings from contemporary statistical studies are
severely limited by the data sources, the accuracy of model
specifications, and the current state of knowledge regarding the
mortgage search and underwriting processes.

Insights regarding the pervasiveness of discrimination from other
sources have also proven to be contradictory. Information derived
from the examination of depository financial institutions (together
with information from consumer complaints), for example, has not
indicated a major problem. In contrast, testing programs recently
conducted around the country by private fair housing groups have
exposed numerous instances of differential treatment during the
mortgage application process. If confirmed by further tests, the
findings from these programs would suggest a fair lending problem of
greater dimensions. Hence, despite all that has been said and done,
there remains much to be learned about the forms, occurrence, and
magnitude of discrimination in the credit markets, particularly in
markets other than the mortgage credit market.

--------------------
\4 See for example, Galster (1992), Goering (1990), Lake (1986),
Benston (1979), and Yinger et al (1979).

\5 Galster (1992).

FEDERAL OVERSIGHT AND ENFORCEMENT
HAVE IMPROVED
============================================================ Chapter 3

Prior to DOJ's initiation of its first major fair lending
investigation in 1989 against Decatur Federal Savings and Loan,
federal efforts to detect, deter, and punish instances of lending
discrimination were quite limited--confined for the most part to the
examination programs of the federal banking regulatory agencies.\1
Before that time, few, if any, federal lawsuits involving lending
discrimination had been filed under ECOA. Federal enforcement
efforts under FHA were similar. However, actions taken since 1988 by
Congress, the Executive Branch, federal banking regulatory agencies,
DOJ, HUD, and others have contributed to a more aggressive regulatory
environment designed to engender greater compliance by the lending
industry.

--------------------
\1 Prior to 1980, only FRB had in place an organization and cadre of
examiners trained and dedicated to perform examinations to assess
banks' compliance with consumer protection laws, including the fair
lending laws.

FAIR LENDING NAMED A TOP
PRIORITY
---------------------------------------------------------- Chapter 3:1

In the late 1980s, Congress, the administration, and the banking
regulatory agencies began taking actions to more effectively oversee
and enforce the fair lending laws. Congress passed several
amendments to the laws aimed at strengthening the statutes'
enforcement provisions and to provide regulators with additional
information on loan applicants and mortgage lending patterns. At
about the same time, DOJ began its first investigation into alleged
discriminatory lending practices, and the federal banking regulators
began to revise their fair lending policies and examination
procedures.

The movement toward more aggressive enforcement of the fair lending
laws received additional momentum in 1993 when the new administration
declared that the development of an effective and aggressive fair
lending enforcement program would be a top priority. In 1993, the
administration formed an interagency task force charged with
clarifying regulatory policies in the fair lending area. The
Interagency Task Force on Fair Lending subsequently released a major
policy statement on fair lending enforcement policies in March 1994
and has continued to meet regularly.\2 Numerous other federal
initiatives were also underway by this time. Figure 3.1 provides a
chronology of events and key fair lending initiatives undertaken at
the federal level over the last several years.

Figure 3.1: Chronology of Key
Fair Lending Enforcement
Initiatives, 1988-1995

(See figure in printed
edition.)

--------------------
\2 See Interagency Policy Statement on Discrimination in Lending,
Federal Register, Vol. 59, No. 73, Friday, Apr. 15, 1994.

FAIR LENDING INITIATIVES BY
BANKING REGULATORY AGENCIES
---------------------------------------------------------- Chapter 3:2

The administration also spurred the executive branch regulatory
agencies and others to reconsider their fair lending policies and
procedures. On June 10, 1993, four of the five federal banking
regulatory agencies (FRB, OCC, FDIC, and OTS) responded by announcing
initiatives that they would undertake to improve the effectiveness of
their examination and enforcement efforts. Specifically, the
agencies pledged to

-- develop new training programs for examiners in fair lending
detection techniques,

-- develop and sponsor regional fair lending seminars for senior
industry executives to foster increased sensitivity and
awareness among lenders for discrimination issues,

-- explore the use of statistically-based discrimination detection
models as tools in the examination process,

-- implement an internal process for making referrals to DOJ for
violations of ECOA, and

-- refine their consumer complaint systems to improve the agencies'
ability to detect and correct credit discrimination.

By year-end 1994, the agencies had made progress in several of these
areas. For example, since the announcement, each of the federal
banking regulatory agencies had developed and initiated new or
enhanced fair lending training programs for their examiners and had
substantially increased the number of examiners who received training
in fair lending detection techniques in a given year. For example,
in 1994, FRB implemented a new 2-week school for its examiners
devoted solely to fair lending issues. Seventy-one FRB examiners
attended the 2-week program in its first year. NCUA, meanwhile,
retained a private consulting firm to instruct its examiners on
compliance with the fair lending laws. The training was completed in
February 1996.

The agencies have also followed through on their pledge to sponsor
fair lending seminars for top-level industry executives. The first
of three regional seminars was held in Washington, D.C., on July 18,
1994. Other seminars were held later in the year in Chicago and San
Francisco. In total, the seminars drew more than 900 executives from
bank and thrift institutions around the country. In responses to
questionnaires circulated by the Federal Financial Institutions
Examination Council (FFIEC)\3 , numerous participants indicated that
the seminars provided a useful forum for them to exchange information
and open channels of communication between top bank management and
federal regulators and enforcement authorities.

The banking regulatory agencies have also made a commitment to
reforming their examination procedures to reflect a new emphasis on
the more subtle forms of discrimination like disparate treatment.
These reforms have included, among other things, the development and
use of sophisticated statistical methods for detecting discriminatory
lending patterns. For example, FRB has already adopted statistical
techniques as a means to more efficiently identify loan files for
comparative analysis-- eliminating the need for manual sampling at
many institutions. Several of the agencies have also supported the
examination function by adding fair lending specialists to their
staffs, not only to support the examination function but to provide
technical assistance to financial institutions on fair lending
matters as well.

Finally, the agencies have also made some progress in refining their
consumer complaint systems. At the time of the agencies' June 1993
announcement that each agency would undertake an evaluation of the
effectiveness of its consumer complaint system in detecting credit
discrimination, several of the agencies had already begun the
process. The FRB initiative in the area, for example, had begun in
November 1992, and new complaint procedures were adopted in June
1995. By the end of fiscal year 1995, OTS had joined FRB in
completing revisions to its consumer complaint procedures. In its
response, FDIC advised us that it planned to complete revisions to
its procedures by June 1996.

--------------------
\3 Members of the council include FRB, FDIC, OCC, OTS, and NCUA.

AGENCIES RECOMMEND VOLUNTARY
FAIR LENDING COMPLIANCE
ACTIVITIES
-------------------------------------------------------- Chapter 3:2.1

Another major effort by the responsible federal agencies to ensure
compliance with the fair lending laws has been the encouragement of
preventative measures in the lending industry. In May 1993, the
federal banking regulators released the following list of recommended
activities or practices for financial institutions to consider when
planning their fair lending compliance programs:\4

-- Use of an internal second review system for consumer, mortgage,
and small business loan applications that would otherwise be
denied.

-- Enhanced employee training that engenders greater sensitivity by
financial institution management and employees to racial and
cultural differences in our society.

-- Training of loan application processors to assure that any
assistance provided to applicants in how to best qualify for
credit is provided consistently to all applicants.

-- Efforts to ensure that all persons inquiring about credit are
provided equivalent information and encouragement.

-- Use of flexible underwriting and appraisal standards that
preserve safety and soundness criteria while responding to
special factors in low- and moderate-income and minority
communities.

-- Efforts to encourage equal employment opportunity at all levels
throughout the institution, including lending, credit review,
and other key positions related to credit applications and
decisions.

-- Affirmative marketing and call programs designed to assure
minority consumers, realtors, and business owners that credit is
available on an equal basis. Marketing may involve sustained
advertising programs covering publications and electronic media
that are targeted to minority audiences.

-- Ongoing outreach programs that provide the institution with
useful information about the minority community, its resources,
credit needs, and business opportunities.

-- Participation on multilender Mortgage Review Boards that provide
second reviews of applications rejected by participating
lenders.

-- Participation in public or private subsidy or guarantee programs
that would provide financing on an affordable basis in targeted
neighborhoods and communities.

-- Use of commissions or other monetary or nonmonetary incentives
for loan officers to seek and make safe and sound consumer and
small business loans in minority communities.

Variations of the list of recommended activities have also been
published by other agencies, lending institutions, compliance
consultants, and a number of financial industry trade groups. These
recommendations reflect the input of a variety of sources, including
mortgage lenders, bankers' associations, credit counseling agencies,
fair housing organizations, and social research groups. Among the
recommended activities are such things as enhanced employee training
programs, participation in multilender review boards, self-assessment
activities, and affirmative marketing programs. HUD formalized its
push for "best practices" by asking mortgage lenders to sign
voluntary Fair Lending Best Practices Agreements that incorporate
many of the suggested activities from the above list. Despite some
delays due to resource and staffing constraints and extended
one-on-one negotiations with interested mortgage lenders, HUD
reported that as of November 30, 1995, 70 lenders had either signed
agreements or had agreed in principle on agreements and were expected
to sign soon.

Many banking institutions have already incorporated at least some of
the recommended activities into their lending operations. For
example, in a recent survey of bank chief executives conducted by
KPMG Peat Marwick, of the 660 chief executive officers who responded,
85 percent reported that they had reviewed fair lending policies and
procedures at their institutions; 43 percent reported having done a
quantitative analysis of application and loan files; and 23 percent
acknowledged having tested their banks using mystery shoppers.

All recommendations, however, have not been enthusiastically endorsed
by the entire financial services industry. Some bankers see certain
aspects of the Interagency Policy Statement on Discrimination in
Lending, as well as some of the recommended activities, as
problematic. In public letters to the regulatory agencies and in the
press, individual bankers and industry trade groups have cited the
high costs associated with some activities, and the legal liabilities
associated with others, as potential barriers to acceptance. Other
bankers expressed concern that special-purpose lending and credit
assistance programs developed by banks to increase lending to certain
groups may run afoul of the fair lending laws by excluding
nontargeted groups and individuals in a discriminatory manner. While
ECOA makes exceptions for such programs, FHA does not directly
address the issue. Although failure to adopt any of the recommended
practices is to have no express bearing on an institution's
compliance rating, DOJ has suggested that by undertaking such
activities a lender could possibly avoid severe sanctions if the
government were to discover discriminatory activity at the
institution at a later date.

--------------------
\4 Joint News Release from Federal Banking Regulatory Agencies, May
27, 1993.

FAIR LENDING OVERSIGHT AND
ENFORCEMENT
---------------------------------------------------------- Chapter 3:3

Federal efforts to ensure compliance with the fair lending laws have
not always been so energetic. In the period immediately following
the passage of FHA, federal enforcement efforts were almost
exclusively limited to matters related to housing discrimination
rather than lending discrimination. For example, HUD and DOJ had
initiated programs to identify discriminatory behavior by landlords
and other agents in the housing search process. Similarly, no
significant enforcement actions were taken under ECOA for more than a
decade after its enactment. In the late 1980s, however, Congress,
DOJ, and other responsible federal agencies began to respond to the
growing number of media reports suggesting that lending
discrimination was widespread.

In 1988 and 1989, respectively, Congress amended FHA and HMDA to both
expand the scope and breadth of the laws and to strengthen FHA's
enforcement provisions. Soon thereafter, DOJ began laying the
groundwork for an intensive fair lending enforcement campaign. Other
federal agencies with oversight responsibilities in the fair lending
area also began, at about this time, to respond to the increasing
calls for stepped up fair lending enforcement efforts. Similarly,
the banking regulatory agencies recognized shortcomings in their
existing fair lending compliance policies and examination procedures
and began extensive revisions.

OVERSIGHT BY THE FEDERAL
BANKING REGULATORY AGENCIES
-------------------------------------------------------- Chapter 3:3.1

For depository institutions, compliance with the fair lending laws is
primarily assessed through regularly scheduled consumer compliance
examinations conducted by the federal banking regulatory agencies.
Fundamentally, the objectives of fair lending examinations are to
determine (1) whether a lender's written policies and standards for
creditworthiness are nondiscriminatory and (2) whether those
standards are applied uniformly and without discrimination. While
the purpose of the examination process is to reveal any unlawful
practices that affect large numbers of people, as well as more
isolated cases of discrimination, the subtle nature of nonovert forms
of discrimination might defy detection by conventional means. This
was especially true prior to 1993 when fair lending compliance
examination procedures were, for the most part, designed to identify
cases of blatant discrimination--i.e., cases in which clearly
qualified minority loan applicants were unjustifiably denied credit.
For example, under the pre-1993 examination procedures, examiners
were directed to select a small judgmental sample of rejected
minority applications and review them for consistency with a lender's
written underwriting standards. With this focus, examiners rarely
detected instances of discriminatory lending activity.

IMPROVEMENTS MADE TO FAIR
LENDING EXAMINATION
PROCEDURES
------------------------------------------------------ Chapter 3:3.1.1

By the early 1990s, the federal banking regulatory agencies had
recognized the inability of their examination process to detect the
more subtle forms of discriminatory behavior and moved to revamp
their examination procedures. Since announcing in June 1993 their
intention to work on improving fair lending examination procedures,
in general, and detection capabilities, in particular, the agencies
have made considerable progress. By 1994, FRB, OCC, FDIC, OTS, and
NCUA had each adopted revised or interim procedures that abandoned
the past process of only comparing rejected applications with
underwriting standards and emphasized a "comparative-file" approach.
The comparative-file approach seeks primarily to detect disparate
treatment by comparing the outcomes of the lending process for
similarly qualified, but racially or ethnically different,
applicants.\5

--------------------
\5 FDIC, OCC, and NCUA adopted interim fair lending examination
procedures emphasizing the comparative-file approach in mid-1993.
FRB formally updated its fair lending examination procedures in May
1994, but said it had been using the comparative-file technique since
it adopted a compliance examination program in 1979.

REGULATORS INCREASE USE
OF COMPARATIVE-FILE
ANALYSIS
------------------------------------------------------ Chapter 3:3.1.2

The interim procedures for fair lending oversight vary by regulatory
agency but essentially share the same focus--a hands- on search for
evidence of differential treatment of applicants based on prohibited
factors like race or gender. In theory, a comparative-file review
for disparate treatment seeks to answer two questions: (1) are the
outcomes of the lending process equivalent for racially or ethnically
different applicants with equivalent qualifications and (2) did the
lender give equivalent levels of assistance during the application
and underwriting processes to applicants from different racial or
ethnic groups?

When conducting a comparative-file analysis, an examiner can employ a
variety of techniques for selecting files to compare. For example,
files may be selected by choosing a particular period (time-period
approach), or be chosen based on reasons for denial
(questionable-transaction approach), or by matching comparable files
(matched-pair analysis), or through scientific sampling, or some
other method. Regardless of the method chosen, the comparative-file
approach attempts to find individual cases in which disparate
treatment may have occurred or when an institution's credit standards
were not applied consistently.

The effectiveness of the comparative-file technique was recently
illustrated in DOJ's fair lending investigation of the Northern Trust
Company and several of its affiliates. By examining numerous
applications from potential borrowers who were deemed by DOJ to be
only marginally creditworthy, the department was able to find what it
considered to be substantial evidence of differential or disparate
treatment of minority loan applicants. The differences in treatment
described included disparities in the level of assistance and advice,
dissimilarities in the way financial information was analyzed, and
variations in how and when "offsetting" qualifications were
considered as compensation for credit deficiencies.\6

Despite its obvious benefits, however, comparative-file analysis is
time-consuming and is often limited in its application because of the
difficulties that can be encountered in finding a sufficient number
of comparable files to review. Because of these difficulties, the
agencies have continued to seek ways to improve their practical
ability to detect discrimination.

--------------------
\6 DOJ indicated that it is not necessary to obtain truly
"comparable" files when using the comparative-file technique to
document a type of disparate treatment they referred to as
"processing discrimination." Processing discrimination could include,
for example, instances in which minority loan applicants were not
given the same level of assistance and advice as were white
applicants.

NEW STATISTICAL MODELS
BEING TESTED AS
EXAMINATION TOOLS
------------------------------------------------------ Chapter 3:3.1.3

In 1994, FRB began to use, on a regular basis, a computerized
statistical model in bank examinations. OCC and other agencies are
also experimenting with statistical models. Essentially, FRB's model
automates the comparative-file analysis of the fair lending
examination (see app. III).

The use of statistically-based methods to detect discrimination can
offer a more systematic means of examination than that of older
methods. In combination with HMDA data, census data, and geographic
information, computerized statistical analysis can allow examiners to
more quickly identify institutions that may require a more intensive
review of their mortgage lending decisions. Use of statistical
models also essentially automates the approach of on-site fair
lending examinations and allows examiners to quickly sort through
vast quantities of data, focus on data for specific lending markets,
select a sample of files, and draw comparisons.

However, the use of statistics and statistical models as examination
tools is not a panacea. Economists and statisticians have repeatedly
pointed out that statistical approaches have limitations and can be
misleading. For example, the application and effectiveness of such
models as examination tools can be limited by an institution's size
and volume of lending activity. Moreover, statistical models of the
loan underwriting process are difficult to construct and data errors
or omitted variables can cause such models to give unreliable
results. These and other limitations associated with the use of
statistics in overseeing and enforcing the fair lending laws are
further discussed in appendix IV.

REFERRALS TO DOJ HAVE
INCREASED
------------------------------------------------------ Chapter 3:3.1.4

Adoption of the new statistical-based examination techniques,
combined with certain statutory reforms, has led to an increase in
the number of cases being reviewed by DOJ. ECOA, as amended by the
Federal Deposit Insurance Corporation Improvement Act (FDICIA) in
1991, requires the federal banking regulatory agencies to refer
pattern and practice violations of ECOA to DOJ, and to notify HUD
when it appears that ECOA violations not referred to DOJ would also
violate FHA. To implement this requirement, each of the banking
regulatory agencies has put in place an internal process under which
apparent fair lending violations are to be referred to the
responsible department. As a result, agency referrals to DOJ have
risen noticeably since 1990 (see figure 3.2).

Figure 3.2: Number of
Referrals by Banking Regulatory
Agency to DOJ for Violations of
the Fair Lending Laws,
1990-1995

(See figure in printed
edition.)

Source: U.S. Department of Justice.

Upon receiving a referral, DOJ is to assess the nature of the case to
determine if the violation warrants a full investigation. If not,
the case is to be returned to the appropriate agency to be handled
administratively. In practice, this process can be informal, with
DOJ and the banking regulatory agencies engaged in two-way
discussions regarding the merits of individual cases so as to
ascertain how best to pursue enforcement. Table 3.1 provides a
breakdown of the number of referrals by banking regulatory agencies
to DOJ by year. Table 3.2 briefly describes the nature of the
referrals and any subsequent action taken by DOJ.

Although the number of referrals to DOJ has increased since passage
of the 1992 amendments to ECOA, there remains a degree of uncertainty
on the part of some agency officials and examiners as to the
characteristics of possible "pattern or practice" cases of lending
discrimination. Such uncertainty arises, in part, from two terms
within the statutory language, "reason to believe" and "pattern or
practice." Because the meaning of these terms is often defined on a
case-by-case basis, the standard for referrals to DOJ is unclear.
Such uncertainty could possibly result in inconsistent application of
the referral mandate and ultimately to inconsistent enforcement of
the fair lending laws across agencies and financial institutions.
This issue is discussed in greater detail in chapter 4.

Table 3.1

Number of Referrals by Bank Regulatory
Agencies and HUD to DOJ for Violations
of the Fair Lending Laws, by Agency,
1990-1995

Year FRB FDIC OCC OTS NCUA HUD Total
--------------------- ----- ----- ----- ----- ----- ----- =====
1990 0 0 1 0 0 0 1
1991 0 0 0 0 0 0 0
1992 1 3 0 0 0 0 4
1993 0 7 4 1 0 1 13
1994 1 12 7 5 0 0 25
1995 5 0 5 0 0 0 10
======================================================================
Total 7 22 17 6 0 1 53
----------------------------------------------------------------------
Source: U.S. Department of Justice.

Table 3.2

Number of Referrals by Banking
Regulatory Agencies and HUD to DOJ for
Violations of the Fair Lending Laws, by
Agency, 1990-1995

Agen Action
cy Year No. taken Details of referral
---- ---- ---- ----------- ---------------------------------------
FRB 1992 1 Legal DOJ/FTC lawsuit filed against Shawmut
action Mortgage Company charging racial
discrimination; settled by consent
agreement.

1994 1 Legal DOJ complaint filed against Security
action State Bank alleging discrimination in
loan pricing based on national origin;
settled by consent agreement.

1995 4 Returned to Alleged discrimination based on marital
agency status\a and spousal signature
violations; to be handled
administratively.

1995 1 Legal DOJ complaint filed against Fleet
action Financial Group for alleged
discrimination in the pricing of home
mortgage loans based on race and
national origin.

OCC 1990 1 Returned to None.
agency

1993 3 Returned to Alleged discrimination based on age,
agency sex, and marital status; violations to
be handled administratively.

1993 1 Legal DOJ lawsuit filed against First
action National Bank of Vicksburg charging
racial discrimination; settled by
consent agreement.

1994 1 Returned to Alleged racial discrimination;
agency administrative remedy achieved through
HUD.

1994 1 Legal DOJ lawsuit filed against Huntington
action Mortgage Company alleging price
discrimination based on race; settled
by consent agreement.

1994 5 Returned to Marital status violation;\a to be
agency handled administratively.

1995 2 Returned to Marital status violation; to be handled
agency administratively.

1995 2 Returned to Alleged age discrimination in use of
agency credit scoring models; to be handled
administratively.

1995 1 Legal DOJ complaint filed against First
action National Bank of Gordon for alleged
price discrimination against Native
Americans.

FDIC 1992 3 Returned to CRA violations by small lenders; to be
agency handled administratively.

1993 7 Returned to Insufficient information.
agency

1994 1 DOJ intends Alleged racial discrimination case;
to close referred to and being handled by HUD.
and return

1994 1 Returned to Alleged violation of FHA based on
agency appraisal rules; isolated incident with
administrative remedy achieved.

1994 10 Returned to Marital status violation; to be handled
agency administratively.

OTS 1993 1 Returned to Alleged racial discrimination in
agency marketing; no cause found.

1994 1 Under Alleged discrimination based on race,
investigati national origin, sex, and age.
on by DOJ

1994 2 Returned to Alleged discrimination based on
agency national origin; failure to serve
entire community; to be handled
administratively.

1994 1 Returned to Alleged discrimination based on age; to
agency be handled administratively.

1994 1 To be Alleged racial discrimination in
returned marketing; insufficient documentation;
to be handled administratively.

HUD 1993 1 Returned to Racial discrimination case; nonserious
agency violation to be handled
administratively.
----------------------------------------------------------------------
\a Failure to combine income/assets and credit histories of joint
applicants who are not married, while routinely doing so for married
joint applicants.

Source: U.S. Department of Justice.

ENFORCEMENT BY THE
DEPARTMENT OF JUSTICE
-------------------------------------------------------- Chapter 3:3.2

DOJ has independent authority under both ECOA and FHA to conduct
investigations and bring a civil suit against financial institutions
when it appears that a pattern or practice of lending discrimination
has occurred.\7 However, like other federal agencies, DOJ did not use
its authority to aggressively enforce these laws prior to 1988.
According to a senior DOJ official, DOJ's lack of involvement prior
to that time was primarily due to a lack of sufficient information
and because evidentiary standards for lending discrimination had not
yet been developed. The implementation of expanded HMDA reporting
requirements in 1990, however, provided DOJ with sufficient
information to begin investigations of mortgage lending patterns at
financial institutions. In 1992, DOJ filed its first major complaint
alleging discriminatory lending practices against Decatur Federal
Savings and Loan (N.D. GA. 1992). That action resulted in a
consent decree in which Decatur Federal, while denying the
allegations of discrimination, agreed to set aside $1 million to
compensate the alleged victims and to adopt a detailed business plan
designed to make home mortgage loans more available to
African-Americans (see app. V). Since then, DOJ has filed nine
additional lawsuits against lending institutions alleging illegal
discriminatory lending practices under ECOA and FHA. They are (1)
U.S. vs. Shawmut Mortgage Company (D. Conn. 1993); (2) U.S. vs.
First National Bank of Vicksburg (S.D. Miss. 1994); (3) U.S. vs.
Blackpipe State Bank (D. S.D. 1994); (4) U.S. vs. Chevy Chase
Savings and Loan (D. D.C. 1994); (5) U.S. vs. Northern Trust
Company (N.D. Ill. 1995); (6) U.S. vs. Security State Bank of
Pecos (W.D. Tex. 1995); (7) U.S. vs. Huntington Mortgage Company
(N.D. Ohio 1995); (8) U.S. vs. First National Bank of Gordon (D.
S.D. 1996); and (9) U.S. vs. Fleet Mortgage Corporation (E.D.
N.Y. 1996). None of these cases has gone to court. Rather, in
every case but one, DOJ obtained a consent agreement with the accused
institution. Only the tiny First National Bank of Gordon, Nebraska,
refused to agree to a settlement at the time the complaint was filed.
Disposition of that case was still pending at the time of our review
(see table 3.3 and figure 3.3 for details on all these cases).

Table 3.3

Description of Complaints Made by DOJ
Against Financial Institutions for
Alleged Violations of the Fair Lending
Laws, 1992-1995

Financial
institution Date Description of DOJ complaint
------------------ -------- ----------------------------------------
Decatur Federal 9/17/92 Complaint alleged that Decatur Federal
Savings and Loan engaged in policies and practices that
discriminated against potential and
actual loan applicants on the basis of
race by, among other things, conducting
its home mortgage loan marketing in a
manner that excluded potential black
borrowers and that it discriminated
against those blacks who did apply for
home mortgage loans.

Shawmut Mortgage 12/13/ Complaint alleged that the mortgage
Co. of Boston 93 company engaged in policies and
practices that discriminated on the
basis of race and national origin in its
home mortgage lending business by, among
other things, requiring a higher level
of documentation of black and Hispanic
applicants' information, failing to make
the same effort to assist minority
applicants obtain qualifying information
as it did for other applicant groups,
and by applying more stringent
underwriting standards to black and
Hispanic applicants.

National Bank of 1/21/94 Complaint alleged that the bank engaged
Vicksburg in a pattern or practice of racial
discrimination by charging blacks higher
interest rates than those charged whites
on unsecured home improvement loans.

Blackpipe State 1/21/94 Complaint alleged that the bank
Bank discriminated on the basis of race,
color, and/or national origin by, among
other things, refusing to make secured
loans to Native Americans when the
collateral was located on a reservation
and by applying different underwriting
standards to Native Americans than those
applied to whites.

Chevy Chase 8/22/94 Complaint alleged that the bank and its
Federal Savings affiliated mortgage company
Bank and B.F. Saul discriminated on the basis of race by
Mortgage Co. adhering to policies and practices that
denied an equal opportunity to residents
of African-American neighborhoods to
obtain mortgage financing and other
types of credit transactions--practices
commonly referred to as redlining.

Northern Trust 6/1/95 Complaint alleged that in processing
Company and named home mortgage loan applications, the
affiliates banks engaged in lending practices that
constituted unlawful discrimination on
the basis of race and national origin;
i.e., that African-American and Hispanic
loan applicants were treated differently
(and less favorably) than white
applicants.

Security State 10/18/ Complaint alleged that the bank
Bank of Pecos 95 discriminated on the basis of national
origin by charging Hispanic borrowers
higher annual percentage rates for
general consumer, nonmortgage related
installment and single payment loans
than for similarly situated non-
Hispanic borrowers.

Huntington 10/18/ Complaint alleged that the mortgage
Mortgage Company 95 company engaged in a pattern of racial
discrimination in the pricing of home
mortgage loans by charging African-
Americans higher up-front fees or
"overages" for home mortgage loans than
similarly situated white borrowers.

First National 04/15/ Complaint alleged that the bank
Bank of Gordon 96 discriminated on the basis of race,
color, and/or national origin by
charging Native American borrowers
higher interest rates for consumer loans
than similarly situated white borrowers.

Fleet Mortgage 05/07/ Complaint alleged that the mortgage
Corporation 96 company engaged in discriminatory
treatment of African American and
Hispanic borrowers at two of its branch
offices by charging them higher prices
in the form of greater "overages" on
home mortgage loans than it charged
white borrowers.
----------------------------------------------------------------------
Source: Compiled by GAO from DOJ case records.

Figure 3.3: Checklist of
Specific Provisions Agreed to
in Consent Decrees in Order to
Settle Claims of Lending
Discrimination

(See figure in printed
edition.)

Source: Compiled by GAO from DOJ case records.

In initiating these legal actions, DOJ has adhered to a broadened
philosophy regarding what constitutes discrimination-- scrutinizing
not only an institution's lending policies and loan files for
evidence of discrimination but also its geographic branching patterns
and marketing efforts as well. For example, in the Decatur and Chevy
Chase cases particularly, DOJ alleged that given the racial
characteristics of the thrifts' lending areas, the institutions'
credit standards and practices and branching and marketing strategies
discriminated against protected minority groups. Initiatives to
address these problems were an integral part of the consent decrees
agreed to by these institutions.

In addition to the ten lawsuits brought against financial
institutions to date, DOJ officials have acknowledged that several
other investigations are underway both within and outside of the
banking industry. For example, DOJ has already entered into an
agreement with HUD to investigate independent mortgage companies and
is also investigating, in cooperation with FTC, the finance units of
the "Big Three" U.S. automobile companies\8 for possible lending
bias in automobile loans. DOJ officials believe that joint
investigations such as these can be particularly effective given the
mix of knowledge, experience, and resources that could be brought to
bear in these efforts. However, Decatur-like investigations, i.e.,
those involving in-depth statistical analyses of loan files, are time
consuming and expensive. According to DOJ, a full and thorough
investigation typically takes 6 to 9 months to complete and can cost
as much as $500,000.\9 Because of the increased workload stemming
from lending discrimination investigations and from FHA "election"
cases; i.e., those cases in which the complainant elects to have
their FHA-related complaint heard in federal court rather than by
HUD's administrative law judge, Attorney General Reno has added 34
new positions to DOJ's Housing Section since 1992 (the authorized
staffing level for fiscal year 1994 was 76 full-time positions).
Additionally, she has also asked that U.S. District Attorneys make
some of their staff members available for these investigations.
However, although DOJ has increased its enforcement efforts,
officials suggest that limited resources and the high cost of
Decatur-like investigations will necessarily limit the number of
cases that can be pursued.

DOJ investigations and legal actions appear to have had a significant
impact both on lenders' compliance efforts and on the enforcement
efforts of other federal agencies. By illustrating their willingness
to single out individual financial institutions for prosecution and
by extracting consent agreements with substantial monetary penalties,
DOJ has reawakened the lending industry to its responsibilities under
the fair lending laws. Although we cannot confirm the extent to
which the financial industry has reacted by increasing its efforts to
ensure compliance with the fair lending laws, indications are that
the industry has been generally responsive. For example, it has been
reported in the press that the industry has been voluntarily
implementing many of the preventative practices, like second review
programs, affirmative marketing programs, and self-testing, as
prescribed in the settlement agreements with Decatur Federal and
others, and as recommended by federal banking regulators.

In addition to apparently spurring increased industry compliance
activity, DOJ's enforcement efforts may also have had an impact on
the enforcement efforts of other federal agencies--prompting
stepped-up enforcement by all of the federal banking regulatory
agencies, HUD, and FTC. Moreover, the findings from DOJ
investigations have aided the banking regulators in their efforts to
improve their own fair lending policies and procedures, and in
developing and implementing more sophisticated examination
techniques.

--------------------
\7 In addition, FHA authorizes an independent DOJ action when it
appears that any group of persons has been denied rights to fair
housing as provided in the act, including equal access to
housing-related credit, and the denial "raises an issue of general
public importance." 42 U.S.C. � 3614(a).

\8 General Motors, Ford, and Chrysler.

\9 DOJ's first fair lending investigation, that of Decatur Federal
Savings and Loan, took nearly 3 years and cost more than $1 million.

OVERSIGHT AND ENFORCEMENT BY
HUD
-------------------------------------------------------- Chapter 3:3.3

The Secretary of Housing and Urban Development (HUD) is authorized to
administer FHA. However, under the original act, HUD's enforcement
authority was limited to complaint investigation, referral, and
attempts at voluntary conciliation. It was not until 1988, with the
passage of the Fair Housing Amendments Act, that a full
administrative enforcement system was put in place to administer and
enforce the law.\10 Also, through the establishment of the Fair
Housing Initiatives Program (FHIP) in 1987, Congress made additional
funding available to HUD for the purpose of conducting
investigations--through grants with private organizations--of
possible FHA violations and to take such enforcement actions as
necessary to remedy violations detected by those organizations.\11

Several units within HUD share fair lending oversight and enforcement
responsibilities. HUD's Office of Fair Housing and Equal Opportunity
(FHEO) has overall responsibility for enforcing FHA and other civil
rights laws and makes referrals to DOJ when appropriate. The Federal
Housing Administration is responsible for ensuring that HUD-approved
mortgagees (lenders) comply with all fair housing and fair lending
laws and promulgates rules implementing fair lending requirements.
In this capacity, the Federal Housing Administration has outlined
requirements covering such fair lending-related topics as minimum
loan amounts, tiered pricing, overages, and second review
programs.\12 HUD-approved lenders that violate the fair lending laws
are subject to administrative sanctions and civil money penalties by
HUD's Mortgagee Review Board. In 1992, HUD was also given authority
to oversee the two major housing government-sponsored
enterprises--Fannie Mae and Freddie Mac (the GSEs) to, among other
things, ensure that they were not discriminating in their mortgage
purchases based on factors protected under FHA.\13

--------------------
\10 The Fair Housing Amendments Act of 1988 created an administrative
enforcement system subject to judicial review and allowed the use of
court enforcement by private litigants and DOJ (Pub. Law No.
100-430).

\11 See Housing and Community Development Act of 1992, Pub. Law No.
102-550, Title IX, � 905(b) (1992).

\12 See for example, HUD Mortgagee Letter 94-22 (May 4, 1994) and
94-43 (Sept. 29, 1994), Office of the Assistant Secretary for
Housing-Federal Housing Commissioner.

\13 See 12 U.S.C. 4501-4641, the Federal Housing Enterprises
Financial Safety and Soundness Act of 1992, enacted as Title XIII of
the Housing and Community Development Act of 1992 (Pub. Law
102-550).

RESPONSIBILITIES OF FHEO
------------------------------------------------------ Chapter 3:3.3.1

As just mentioned, FHEO has been delegated primary responsibility for
enforcing FHA's provisions. FHEO's enforcement program has several
components: (1) complaint investigations, (2) Fair Housing
Initiatives Program grants, and (3) voluntary "best practices"
agreements.

Complaint Investigations

The largest component of FHEO's fair lending enforcement program is
its complaint investigation program. As described by HUD, it
consists of two parts: consumer complaints, and Secretary- initiated
complaints.\14 Consumer complaints can be narrow or broad in scope
depending on the number and type of allegations. In contrast,
Secretary-initiated complaints are normally broader in scope, require
more resources, and take longer to investigate than do consumer
complaints. All Secretary- initiated complaints are processed by HUD
staff in FHEO. Consumer complaints are primarily processed by HUD,
but about one- third of them are processed by state and local
government agencies under the Fair Housing Assistance Program (FHAP).
The FHA requires that HUD allow state and local government agencies,
which have been determined to have fair housing enforcement programs
substantially equivalent to those under FHA, to process complaints
filed in those agencies' jurisdictions.\15 Under FHAP, HUD is to
provide financial support and technical assistance to these agencies
for complaint processing.

In the past 7 years since the Fair Housing Amendments Act of 1988 was
implemented (in 1989), HUD and the FHAP agencies have processed 2,356
fair lending complaints. Of this number, HUD processed 1,598 (68
percent), and the FHAP agencies processed 758 (32 percent). In
fiscal year 1995, HUD closed 456 complaints alleging discrimination
in housing finance (6.5 percent of all FHA-related complaints closed
by HUD that year).

HUD's Secretary-initiated complaint program began focusing on
mortgage lending issues in fiscal year 1993. FHEO and HUD's Federal
Housing Administration reviewed 16 FHA-approved lenders to determine
whether a formal investigation was warranted. The review resulted in
two Secretary-initiated investigations. In all, HUD has initiated
four investigations of mortgage lenders, all of which were nearing
completion at the time of our review.

Although HUD is required to process all consumer complaints, FHA
requires that HUD attempt to conciliate complaints and try to avoid a
time-consuming and costly investigation or enforcement action.
Conciliation is voluntary and both the respondent and complainant
must agree for it to occur. Through this process, a large percentage
(38 percent) of the fair lending complaints that HUD processes are
conciliated (See table 3.4).

Table 3.4

Number and Percentage of Fair Lending
Complaints Closed by HUD During FYs
1990-1995, by Closure Type

Number of
Closure type complaints Percentage
------------------------------ ------------------ ------------------
Administrative\a 720 30.6%
Conciliated 896 38.0
No violation 717 30.4
Violation 14 0.6
Referred to DOJ 9 0.4
======================================================================
Total 2,356 100.0%
----------------------------------------------------------------------
\a HUD may close complaints administratively with or without an
investigation. This can occur, for example, when the complainant
refuses to cooperate with HUD during an investigation, or when HUD is
unable to contact the complainant after the complaint is filed.

Source: HUD, Office of Fair Housing and Equal Opportunity.

Conciliation agreements may provide compensatory and equitable relief
for the complainant(s) and relief in the public interest. They may
also enjoin the respondents from further discriminatory behavior.
According to HUD, in fiscal year 1995, it conciliated 105 mortgage
lending complaints that resulted in some form of monetary
compensation to the complainant. As a result of these conciliations,
complainants received $1.3 million in monetary compensation--an
average of over $12,000 per complainant. However, this is only a
fraction of the amount that lenders agreed to expend in the public
interest. For example, HUD stated that it had recently negotiated a
settlement of $175,000 for a Hispanic female employee who was
discharged by a lender for making loans to minority persons and for
questioning why one of her clients, a Hispanic male, was denied a
loan. As part of the settlement, the lender agreed to advertise the
availability of loans in minority media, contribute $2,500 to a
nonprofit housing organization for the purpose of promoting fair
housing, and to make $250,000 available to a lending program for low-
to moderate-income borrowers.

Another case described by HUD involved a black developer who was
denied a commitment to finance homes to be constructed in a minority
area. The lender agreed to settle the case by paying the complainant
$68,000 and establishing a $500,000 fund to counsel credit applicants
under a program established pursuant to the Community Reinvestment
Act (CRA). The lender also agreed to provide mortgages for qualified
applicants presented by the complainant for up to 250 newly
constructed homes at an average of $70,000. According to HUD, this
represents a commitment of over $17 million in predominantly minority
areas.

Under FHA, the Secretary is to investigate complaints of
discriminatory lending practices. However, according to HUD,
investigations of complaints are generally only conducted if the
complainant and/or respondent refuse to participate in the
conciliation process or when conciliation is attempted but fails. As
a result, HUD actually investigates only about one-third of all fair
lending complaints. Of the 740 fair lending complaints that were not
administratively closed or successfully conciliated by HUD during
fiscal years 1990-1995, and which required an investigation, only 23
(or 3 percent) were found to substantiate a violation of FHA. If a
conciliation agreement is not reached and HUD believes that
discrimination has occurred, the Secretary must file a charge on
behalf of the aggrieved person. Once the charge is filed, the
aggrieved person on whose behalf the charge was filed may elect to
have the matter administratively adjudicated within HUD or have HUD
commence a civil action on behalf of that person.\16

If during the investigation of a consumer complaint, or when
considering matters for Secretary-initiated complaints, HUD has
reason to believe that a pattern or practice of discriminatory
behavior is evident, the Department must transmit the pertinent
information to DOJ. In addition to its referral obligations, HUD
officials have told us that they have agreed to cooperate with DOJ in
investigations of independent mortgage companies. In the past, these
investigations have taken place either sequentially or concurrently,
but not jointly.

As mentioned earlier, the federal banking regulatory agencies must
notify HUD of cases not referred to DOJ when the agencies believe
that an institution's apparent violation of ECOA also would violate
FHA. Moreover, under an agreement with member agencies of the FFIEC
and by executive order, FHA-related complaints of lending
discrimination received by the agencies must be referred to HUD for
investigation. (Likewise, if the complaint is received by HUD and it
involves a federally regulated financial institution, HUD is to
inform the appropriate regulatory agency of the complaint and any
pending investigation.) During the past 3 fiscal years, HUD has
received 560 complaint referrals from the regulators, and, during the
same period, HUD notified the banking regulators of 139 complaints it
received against lending institutions regulated by them.

The Fair Housing Initiatives Program (FHIP)

The Fair Housing Initiatives Program, which was established by the
Housing and Community Development Act of 1987,\17 provides grants to
private organizations and State and local government agencies to
provide education and outreach and to conduct enforcement-related
activities. As currently administered, FHIP provides funding for
activities in four program areas: (1) administrative enforcement,
(2) education and outreach, (3) private enforcement, and (4) fair
housing organizations.\18 FHIP funds are awarded on a competitive
basis. In fiscal year 1993, HUD awarded $9.6 million in grants under
FHIP, with almost $5 million of that targeted to projects related to
insurance redlining and mortgage lending discrimination. In fiscal
year 1994, congressional appropriations for FHIP were increased to
$20.5 million, with approximately $12 million targeted for
enforcement projects in those same two areas. Through fiscal year
1994, FHIP competitions have awarded more than $21.5 million in funds
to support fair housing enforcement efforts. Included among those
awards was a fiscal year 1992 grant for $1 million to support a
large-scale national testing program to assess mortgage lending
discrimination. Information obtained from FHIP-funded projects can
be used by either public or private nonprofit organizations, or HUD,
as the basis for a formal complaint against individuals or lending
institutions.

Following passage of the Housing and Community Development Act of
1992, HUD lifted a requirement that FHIP-funded testing activities be
restricted to bona fide complaints and expanded FHIP to permit
contract recipients to carry out testing programs whenever there is a
reasonable basis for doing so. According to HUD, a reasonable basis
may be obtained from complaints, allegations, statistical
disparities, or other types of substantive information. Testing
programs involve the use of "testers" posing as renters, purchasers,
or borrowers in order to ascertain if a similarly situated member of
a protected group has been subject to discrimination. Beginning with
the fiscal year 1990 and 1991 FHIP competitions, funding started to
become available for testing programs focusing on mortgage lending
discrimination. Those first testing projects were largely
experimental but have served as the prototype for other FHIP- funded
testing programs now under way in a variety of locations. Several
FHIP-funded projects involving testing of mortgage lenders and
insurance companies were completed in 1995 and, as a result,
complaints have been filed with HUD against three of the largest home
insurance companies and five of the largest independent mortgage
companies in the country. Additional large-scale investigations are
still under way.

HUD "Best Practices" Agreements

Section 809 of FHA charges the Secretary of HUD to endeavor to work
out programs of voluntary compliance and enforcement with the housing
industry and other interested parties. In 1994, as part of these
efforts, HUD and the Mortgage Bankers Association (MBA) signed a best
practices agreement designed to encourage individual mortgage banking
firms to use fair lending "best practices"; i.e., practices or
activities that firms could undertake to deter discriminatory
activity and further lending to underserved groups and communities.
Under this agreement, MBA agreed to urge its members, which are not
regulated like banks, to sign individually negotiated, nonbinding
agreements with HUD to increase credit and homeownership
opportunities for historically underserved borrowers. Generally,
these agreements encourage initiatives similar to those recommended
for the banking industry, and include initiatives like second review
programs, self-testing, and minority targeted advertising and
recruitment. In addition, HUD has also included tailor-made lending
targets within the agreements for lending to underserved communities.
According to HUD, as of November 30, 1995, 70 mortgage lenders had
either signed, or agreed in principle to sign, best practices
agreements. In January 1996, HUD also reported that it had signed
its first voluntary fair lending agreements with commercial banking
institutions.

It should be noted that while HUD, MBA, and consumer groups have
generally hailed the agreements, the banking industry has publicly
decried the use of formalized "best practices" agreements--citing
them as credit allocation and an attempt by the unregulated mortgage
banking industry to ensure that laws like CRA are not amended to
apply to mortgage bankers.

--------------------
\14 Section 810 of FHA sets forth the requirements for HUD
investigations of potential discriminatory housing practices, which
include discriminatory mortgage lending and related activities. The
section requires that HUD investigate all consumer complaints and
authorizes the agency to self-initiate investigations and file
complaints (Secretary-initiated complaints). 42 U.S.C. � 3610.

\15 See 42 U.S.C. � 3610(f) (1988).

\16 See 42 U.S.C. � 3612.

\17 Pub. L. No. 100-242, Title V, � 561, 101 Stat. 1942 (1987).
The law was passed in 1987 but became effective Feb. 5, 1988. FHIP
was subsequently amended in 1990 and 1992 (codified, as amended, at
42 U.S.C.A. � 3616a (West Supp. 1993).

\18 Section 905 of the Housing and Community Development Act of 1992
allows for money appropriated under FHIP to be used to conduct
investigations, through contracts with private nonprofit
organizations of violations of rights granted under FHA. See 42
U.S.C. � 3616 note.

HUD'S OVERSIGHT OF THE
SECONDARY MORTGAGE MARKET
------------------------------------------------------ Chapter 3:3.3.2

Partly in response to public concerns regarding the potential for
discrimination (disparate impact) by primary lenders who follow the
underwriting standards created by secondary mortgage market
institutions, Congress passed the Federal Housing Enterprises
Financial and Safety and Soundness Act (the GSE Act) in October
1992.\19 In the GSE Act, Congress reaffirmed the Secretary of HUD's
role as the programmatic regulator for Fannie Mae and Freddie
Mac--the two largest housing GSEs.\20 Among the provisions of the GSE
Act were requirements for the Secretary of HUD to

-- set levels of congressionally-mandated housing goals that
require the GSEs to purchase mortgages for very low- and
moderate-income families and families living in areas
underserved by the mortgage markets;

-- establish fair lending requirements for the GSEs;

-- monitor the GSEs' performance in meeting housing goals; and

-- create a public-use database, making available information on
the GSEs' activities.

More specifically, the GSE Act required HUD to issue regulations to
prohibit the housing GSEs from discriminating in their mortgage
purchases and to carry out a number of other fair lending
obligations. Under the GSE Act, the Secretary of HUD must

-- prohibit discrimination by the GSEs in their mortgage purchases
because of race, color, religion, sex, handicap, familial
status, age, or national origin, including any consideration of
the age or location of a dwelling or age of the neighborhood or
census tract where the dwelling is located in a manner that has
a discriminatory effect;

-- require that the GSEs submit information to the Secretary to
assist enforcement of FHA and ECOA;

-- advise the GSEs of violations of FHA, ECOA, and state and local
fair lending laws by lenders;

-- periodically review the underwriting and appraisal guidelines of
the GSEs to ensure compliance with FHA and the Federal Housing
Enterprises Financial Safety and Soundness Act;

-- review the annual assessment by the GSEs, in their statutorily
required Annual Housing Activities Report, of their underwriting
standards, business practices, repurchase requirements, pricing
fees and procedures that affect the purchase of mortgages of
low- and moderate-income families, or that may yield disparate
results based on the race, status, age, or national origin of
the borrower;

-- direct the GSEs to take action, following adjudication, against
lenders for violations of FHA and ECOA; and

-- refer potential violations of the fair lending provisions of the
Federal Housing Enterprises Financial Safety and Soundness Act
to the Director of the Office of Federal Housing Enterprise
Oversight (OFHEO) for enforcement action.

HUD published a proposed rule implementing the Secretary's regulatory
authorities on February 16, 1995, with a 75-day comment period.
After reviewing comments on the proposed rule, including extensive
comments from the GSEs, HUD met with each of the GSEs, industry trade
groups, public interest groups, the Treasury Department, the U.S.
Department of Agriculture, the federal banking regulatory agencies,
and various state and local officials to discuss issues related to
the rule. HUD issued its final rule implementing the GSE Act on
December 1, 1995.\21

--------------------
\19 12 U.S.C. � 4501-4641.

\20 The GSEs were created by Congress and, in return for their
publicly provided benefits--such as exemption from state and local
taxes--the GSEs are required to extend the benefits of the secondary
market to a broad range of Americans, including low- income, working
class families, first-time homebuyers, and residents of communities
that may be underserved by mortgage credit.

\21 See 24 CFR Part 81, Federal Register, Vol. 60, No. 231, Friday,
Dec. 1, 1995.

OVERSIGHT AND ENFORCEMENT BY
FTC
-------------------------------------------------------- Chapter 3:3.4

The Federal Trade Commission has the authority to investigate certain
lenders suspected of lending discrimination. The FTC's authority is
provided in ECOA and extends to lenders subject to ECOA whose
activities are not regulated by the federal agencies specified in the
act.\22 FTC, through DOJ or on its own, may file suit in federal
district court against lenders suspected of violating the law. In so
doing, FTC can seek injunctions to prohibit future illegal conduct,
civil money penalties of up to $10,000 for each violation, and
redress for consumers unfairly denied loans. In addition, FTC can
impose recordkeeping and reporting requirements on defendants to
assist FTC in monitoring compliance.

Since fiscal year 1991, FTC has obtained consent decrees against
eight small-loan and mortgage finance companies for alleged
violations of ECOA and Regulation B, which implements the act. In
December 1993, FTC and DOJ jointly entered into a settlement
agreement with Shawmut Mortgage Company. In signing the agreement,
Shawmut agreed to pay almost $1 million into a redress fund to
compensate minority applicants who allegedly were unfairly denied
mortgage loans during the period from 1990 through late 1992. The
agreement with Shawmut is of particular note in that it organized a
framework for recordkeeping and reporting that could be followed in
future cases to identify and compensate applicants that have been the
victims of lending discrimination.\23

--------------------
\22 As described in Appendix A to FRB's Regulation B (12 C.F.R. Part
202), which implements ECOA, FTC's authority under the act typically
extends to retailers, finance companies, and other creditors not
subject to oversight by the agencies specified in ECOA. See 15
U.S.C. � 1691c(c).

\23 As indicated in table 3.3, FRB referred the Shawmut case to DOJ
on the basis of its examination of the bank holding company.
However, pursuant to ECOA, FTC has jurisdiction over mortgage
companies that are subsidiaries of a bank holding company. Because
of these overlapping authorities, both FTC and DOJ filed the
complaint against the institution.

AGENCY EDUCATIONAL AND
OUTREACH PROGRAMS EXPANDED
-------------------------------------------------------- Chapter 3:3.5

In addition to their oversight and enforcement initiatives, the
federal banking regulatory agencies, HUD, DOJ, and FTC have also
increased their education and outreach efforts to assist institutions
in creating effective antidiscrimination programs. By 1993, FRB,
FDIC, OCC, and OTS had created or substantially expanded their
community affairs programs, especially in the areas of fair lending
and community reinvestment. Included among these educational and
outreach efforts are training workshops, seminars, major conferences,
and the development of new publications and educational materials to
assist lenders, community organizations, and others to better
understand and respond to fair lending concerns. The Boston Federal
Reserve Bank, for example, developed and distributed more than 90,000
copies of Closing the Gap: A Guide to Equal Opportunity Lending, a
publication highlighting various techniques that banks can use to
help combat possible discrimination in lending and to ensure
equitable treatment for loan applicants. It is the most widely
circulated publication ever developed by FRB. FDIC also achieved
great success with its publication on self-testing, Side-by-Side: A
Guide to Fair Lending. By late 1995, FDIC reported that more than
35,000 copies of the guide had been distributed.

In addition, the federal banking regulatory agencies, HUD, DOJ, and
FTC have supported or participated in a number of public/ private
working groups that provide forums for the development of a public
consensus on actions to ensure equal access to mortgage and other
types of credit. One example of this is the Cleveland Residential
Housing and Mortgage Credit Project, in which nearly 100 housing,
real estate, and lending organizations throughout the Cleveland area
formed a group to identify potential discriminatory lending practices
in the home-buying process and to recommend ways of eliminating
them.\24 The project, begun in 1993, identified 18 points in the
mortgage lending process where discrimination could occur. Smaller
task forces were then formed to study what the larger forum thought
were the four most critical points. These areas were explored in
depth, problems were identified, and possible solutions were
recommended to the forum. The project is still active and the
different industries involved in the project are currently
implementing some of the task force's recommendations.

--------------------
\24 The Project was initiated and co-sponsored by the Federal Reserve
Bank of Cleveland. Other co-sponsors include the Cuyahoga County
Department of Development, the Greater Cleveland Roundtable, and the
Ohio Civil Rights Commission.

CONCLUSIONS
---------------------------------------------------------- Chapter 3:4

In recent years, the federal banking regulatory agencies, DOJ, HUD,
and most other responsible federal agencies have devoted considerable
effort toward improving compliance with the nation's fair lending
laws. Beginning in the late 1980s, these agencies have stepped-up
enforcement of the fair lending laws and have tried to heighten the
level of awareness and sensitivity of the lending community. The
federal banking regulatory agencies have also moved to strengthen
their ability to detect discrimination through improved examination
procedures and techniques. A number of these agencies have also put
forward a list of recommended compliance activities and programs for
use by lenders who seek to ensure that all loan applicants are
treated fairly. Most of these agencies have also intensified their
efforts over the last several years to develop and deliver
educational and informational programs designed to help lenders
ensure equal access to credit. All of these efforts are ongoing.

We believe the totality of the actions taken by the responsible
federal agencies over the last several years has served to increase
the level of awareness and sensitivity to the issue of fair lending
throughout most of the lending industry. If maintained, we believe
these efforts are also likely to lead to increased lender compliance
and improved enforcement of the fair lending laws in the future.

CHALLENGES REMAIN IN OVERSIGHT AND
ENFORCEMENT
============================================================ Chapter 4

Even though responsible federal agencies have made substantial
progress in the area of fair lending oversight and enforcement, there
remain a number of issues that present significant and continuing
challenges to the efforts of federal regulators to combat lending
discrimination. For instance, during the course of our review, we
identified several areas related to the banking regulatory agencies'
existing fair lending examination policies and procedures where we
believe the agencies have not taken full advantage of opportunities
to (1) strengthen their ability to detect discrimination in all of
its forms and (2) improve the consistency of oversight and
enforcement. For example, because fair lending examination
procedures are not uniform across agencies, the likelihood of finding
evidence of discrimination may vary by regulator due to differences
in examination techniques. Additionally, the ability of the agencies
to detect discrimination during the early stages of the lending
process is constrained because they have not incorporated
pre-application testing as an examination tool. Finally, most
agencies have not used the full range of their enforcement authority
to ensure that HMDA reporting requirements are adhered to in a timely
and accurate manner.

Although issues such as those mentioned above can be directly
confronted and addressed by the agencies, others, by their nature,
are more problematic and defy immediate resolution. For example, the
subtle and sometimes statistical nature of some types of lending
discrimination makes detection difficult even with newer, more
advanced techniques. Additionally, some key legal issues associated
with the interpretation and application of the fair lending laws
remain unresolved. Nevertheless, it remains important to understand
that these issues exist, and that they pose substantial challenges to
oversight and enforcement of the fair lending laws.

FURTHER IMPROVEMENTS POSSIBLE
IN FAIR LENDING EXAMINATION
POLICIES AND PROCEDURES
---------------------------------------------------------- Chapter 4:1

In June 1993, the banking regulatory agencies indicated that they had
begun several efforts to promote compliance with the fair lending
laws. These efforts included revisions to their fair lending
examination procedures aimed at strengthening their ability to detect
discrimination. We began our review of the agencies' fair lending
examination procedures by conducting in-depth interviews with bank
management, compliance officers, and federal bank examiners who were
involved in recently completed compliance examinations at 40
financial institutions around the country. On the basis of this
work, and on the insights and knowledge we gained from other aspects
of our review of the lending discrimination issue, we identified
several areas where we believe the agencies have not taken full
advantage of opportunities to strengthen their fair lending
examination procedures. Although our case studies of 40 compliance
examinations were specific to examination practices in effect during
1993, the issues are still relevant to the revised and interim
examination procedures now in use.

On the basis of our review, we found that past and current fair
lending examination policies and procedures of the federal banking
regulators: (1) lack uniformity across agencies--a situation that
could result in inconsistent application and enforcement of the laws;
(2) have inadequate methods for detecting discrimination prior to a
prospective borrower's submission of a formal application; and (3)
have not resulted in vigorous enforcement of HMDA data reporting
requirements by all agencies. Also, responses to our surveys of bank
compliance officers and agency examiners indicated that additional
examiner training in the latest fair lending examination and
detection techniques would be beneficial and that, in some instances,
examiners felt they were not allowed sufficient time to develop
evidence of substantive violations. Finally, some officials and
examiners at several of the federal agencies responsible for fair
lending oversight have expressed some uncertainty about the
identification of "pattern or practice" cases.

FAIR LENDING EXAMINATION
PROCEDURES NOT UNIFORM
ACROSS AGENCIES
-------------------------------------------------------- Chapter 4:1.1

In June 1993 the federal banking regulatory agencies announced their
intention to work on improving fair lending examination procedures
and to improve their detection capabilities. Initially, FRB, OCC,
OTS, FDIC, and NCUA had sought to develop uniform fair lending
examination procedures through the FFIEC-- the interagency body
charged with bringing uniformity to all examination procedures and
processes. As part of this effort the FFIEC awarded, in late 1992, a
$75,000 contract to Arthur Andersen & Co. to review the agencies'
fair lending examination procedures and training programs and to
recommend improvements. However, agency officials told us that the
contract yielded little of value. Partly as a result, the drive
toward development of uniform examination procedures stalled. Each
of the banking regulatory agencies then began independent efforts to
improve their own fair lending examination procedures. These efforts
included experimentation with such things as alternative ways of
analyzing HMDA data, new sampling paradigms for comparative-file
analysis, and the use of regression models for detecting
discriminatory lending patterns.

By the end of 1994, each of the federal banking regulatory agencies
had made significant revisions to, or had replaced, older examination
procedures with procedures that, while similar in some respects,
lacked uniformity in a number of areas. For example, while FRB's
revised procedures have formalized a statistically-based approach to
aid in the detection of discriminatory lending patterns at large
institutions, most other bank regulatory agencies have not adopted
similar procedures. Also, in 1994, OCC initiated a pilot testing
program involving the use of mystery shoppers to search for disparate
treatment of similarly situated loan applicants and has subsequently
adopted testing as an optional component of its examination
procedures.

If continued, we believe the agencies' independent efforts to revise
their examination procedures could result in a situation in which the
same degree of oversight is not necessarily applied to all lenders;
e.g., some depository institutions may be subject to compliance
examinations involving the use of advanced detection methodologies
such as regression analysis or testing, while others may not.
Consequently, it may be that the likelihood of finding evidence of
lending discrimination, and/or being referred to DOJ, will vary by
regulator due to differences in examination techniques.\1 In light of
the economic and legal consequences that could arise from allegations
of discriminatory lending practices, as illustrated in recent DOJ
settlement agreements, the lack of uniform examination procedures
raises an important issue regarding the evenhanded application of the
law. Furthermore, adoption of uniform examination procedures would
also help to avoid confusion within the banking industry--a situation
that could possibly inhibit voluntary compliance efforts.

--------------------
\1 Differences across agencies in examiner experience and training,
and in examination time and resource constraints, could also affect
the likelihood of detecting violations of the fair lending laws.

It should also be noted that uneven oversight and enforcement are
probably more pronounced across different segments of the financial
industry. For example, while depository institutions are regularly
examined for compliance with the fair lending laws, most finance
companies and other nondepository lending institutions are seldom so
closely scrutinized.

PROCEDURES TO DETECT
DISCRIMINATION PRIOR TO
SUBMISSION OF A FORMAL
APPLICATION ARE INADEQUATE
-------------------------------------------------------- Chapter 4:1.2

Both examiners and their respective agencies agree that the ability
of examiners to detect illegal credit discrimination in the
preapplication stage is limited. At this point in the mortgage
application process there is no paper trail for the bank examiner to
review. Although examination procedures call for examiners to
routinely interview those bank personnel who serve as initial contact
points for potential applicants, it is extremely difficult, if not
impossible, for an examiner to determine whether any applicant was
illegally discouraged from making a formal application or steered to
a less advantageous product or institution--unless a complaint had
been filed. Furthermore, it is not possible for examiners to know
how many prospective applicants have even approached an institution
to inquire about credit. Inadequate procedures for detecting
discrimination at the preapplication stage represent a serious
omission, especially since pre-application testing programs conducted
by private groups have uncovered evidence of disparate treatment
among prospective loan applicants.\2

One technique that has the potential to be a useful means of
detecting and preventing illegal activities, especially in the
preapplication stage, is testing. Generally, the testing methodology
involves having matched pairs of "testers" pose as prospective loan
applicants. After discussing loan possibilities on an individual
basis, the testers document their treatment and the completeness of
the information given to them by the institution's personnel.
Although they do not actually complete an application, the testers do
experience the important preapplication phase of the loan process.\3

Although the use of testers as an examination tool has been formally
discussed by FFIEC and individual banking regulatory agencies, the
regular use of a testing procedure has never been widely supported
except by HUD and DOJ. For example, in a 1991 feasibility study on
the application of the testing methodology to the detection of
lending discrimination, FRB expressed reservations regarding the use
of testers because of ethical concerns involving entrapment or
self-incrimination.\4 FRB also expressed concern about the ability to
measure treatment accurately, objectively, and in quantifiable terms,
and about the high costs associated with a reliable testing program.
It should be noted, however, that FRB's concerns about costs were
primarily associated with using testing methodologies to obtain
statistically valid results regarding the extent and nature of
possible mortgage discrimination in a given marketplace (for example,
a selected metropolitan area). A testing project with this objective
would necessarily involve hundreds of tests and entail significant
labor and overhead expenses. In contrast, the FRB acknowledged that
an alternative approach that they considered--referred to as an
"enforcement design," whose objective was only to determine if
systemic differences in treatment based on race were present at
individually selected institutions--could yield useful information
for enforcement purposes from a relatively small number of tests per
institution, but only if differential treatment of testers was
unambiguous and very commonplace.\5 Indeed, this approach has been
employed by a number of private fair housing groups for detecting
disparate treatment by mortgage lenders (See app. II).

The renewed emphasis on lending discrimination and the impotence of
current examination procedures in the preapplication stage prompted
OCC to undertake a pilot testing program based on the "enforcement
design" in 1994 and early 1995. Of the eight institutions tested for
illegal preapplication discrimination under OCC's pilot program, no
evidence of discrimination was found in six of them. In the
remaining two institutions, questionable treatment was reported by
the testing contractor, but after reviewing the raw data, OCC
concluded that discrimination did not occur. Even though the pilot
testing program proved to be labor-intensive and failed to uncover
evidence of illegal lending discrimination, OCC concluded that
testing can be a valuable tool in its overall fair lending program.
Consequently, in April 1996, OCC formerly adopted a policy to use
matched-pair testing as an examination technique on a case-by-case
basis when information received from examiners, consumers, or the
media indicated that an institution might be engaged in illegal
discrimination, particularly when such information indicated a
problem at the institution's preapplication stage. FDIC also
recently considered a testing program but postponed any action on the
proposal indefinitely-- opting instead to wait and see if efforts to
encourage voluntary testing programs bear fruit.

--------------------
\2 See discussion of findings from preapplication testing programs in
app. II.

\3 The testing methodology could also be used to investigate
individual complaints. In this case, a single tester could assume
financial, demographic, or other characteristics similar to a
complainant's and attempt to obtain information about the same loan
product. Observing the treatment received by the tester can help
investigators ascertain if a complaint has cause.

\4 See A Feasibility Study on the Application of the Testing
Methodology to the Detection of Discrimination in Mortgage Lending,
prepared for the Federal Reserve's Consumer Advisory Council by Staff
of the Board of Governors of the Federal Reserve System, Washington,
D.C., (1991), 80 p.

\5 The FRB study, however, goes on to question whether the issues of
cost and effectiveness it raised regarding marketplace testing can be
overcome under the "enforcement design" if the discrimination is
subtle. In such circumstances, they argue, discrimination may be
more difficult to identify and, therefore, more likely to require a
larger number of tests in order to obtain a reliable outcome that
could be used to take action against a bank. This could possibly
undermine the agency's ability to conduct the tests without being
identified, and negate the cost advantage obtained by conducting
fewer tests.

AGENCIES HAVE NOT USED FULL
ENFORCEMENT AUTHORITY TO
ENSURE COMPLIANCE WITH HMDA
DATA REPORTING REQUIREMENTS
-------------------------------------------------------- Chapter 4:1.3

Amendments to HMDA in 1989, requiring the collection and reporting of
data on race, gender, and income characteristics of mortgage
applicants, were intended to provide data to assist in identifying
discriminatory lending practices. Although the banking agencies and
HUD are supposed to ensure that the lenders they supervise provide
complete and accurate HMDA information, concerns have arisen
regarding the accuracy of reported data. These concerns were echoed
repeatedly in our interviews with bank examiners and FRB staff
involved in the processing of HMDA data. Findings from recently
completed internal agency audits at FRB and FDIC have also lent some
credence to these longstanding concerns about HMDA data accuracy.
For example, based on the results of its 1994 review, FRB required
one out of every five banks it examined to resubmit their reported
HMDA data for the year 1992 (See figure 4.1).

We believe that large and frequent errors in HMDA data, especially in
critical variables like applicant income, could impair fair lending
examination and enforcement efforts of the federal agencies. For
example, despite its limitations, HMDA data is still widely used by
regulators and the public to target institutions for further scrutiny
and to perform mechanical analyses of institutional lending patterns.
If based on inaccurate data, these analyses could be misleading and
result in a misappropriation of time and resources by regulatory
agencies, the public, and lending institutions. Until recently,
enforcement of HMDA reporting requirements by the federal regulatory
agencies has been limited to verbal warnings or requiring
institutions to correct and resubmit the required data. Not until
fiscal year 1994 did a federal regulatory agency levy a civil money
penalty against a financial institution for untimely and/or
inaccurate HMDA reporting.\6

Figure 4.1: Federal Reserve
Board Survey of HMDA Reporters

(See figure in printed
edition.)

Source: FRB.

In December 1993, HUD imposed a civil money penalty of $500 on an
independent mortgage company it supervised for violating HMDA
reporting requirements. Since then, it has assessed penalties of up
to $2,000 on an additional 16 lenders for violating HMDA.\7 In June
1994, FDIC announced that it had fined six institutions for late
submissions of 1992 and 1993 HMDA data. By the end of fiscal year
1995, FDIC had levied civil money penalties totaling $79,000 against
31 institutions. The penalties ranged from $1,000 to $5,000.

The willingness on the part of HUD and FDIC to impose monetary
penalties on HMDA violators is an important step toward ensuring more
accurate HMDA data reporting. However, adherence to a consistent
enforcement policy across all federal agencies for late and
inaccurate HMDA data submissions is still needed to help ensure the
integrity of the entire HMDA dataset. Otherwise, enforcement of HMDA
will not be perceived to be a priority of the regulators--thereby
making HMDA data quality less likely to become a priority of
financial institution management. In its response, OTS indicated
that it had recently adopted a new policy and issued specific
guidelines regarding the use of civil money penalties (CMP) against
HMDA reporters who submit late or inaccurate reports. According to
OTS, after this policy was announced, institutions that had been
significantly tardy in their 1993 and 1994 reporting filed timely
reports covering 1995.

--------------------
\6 All federal banking regulatory agencies have similar enforcement
authorities. See section 8(i) of the Federal Deposit Insurance
Corporation Act, 12 U.S.C. � 2804, as amended; see also 12 C.F.R.
part 203 (FRB Regulation C). Also, HUD's Mortgage Review Board is
authorized to levy fines under 24 C.F.R. 25.9(j) when a Federal
Housing Administration-approved lender is in "[v]iolation of the
requirements set forth in any statute, regulation, handbook,
mortgagee letter, or other written rule or instruction[.]"

\7 The data submitted by HUD indicated that other lenders may also
have been assessed civil money penalties for HMDA violations, but
because these lenders had multiple violations, which included HMDA
violations, it was not possible to determine the specific purpose of
the civil penalty.

EXAMINER TRAINING AND
EXAMINATION TIME ALLOWANCES
CITED BY EXAMINERS AS BEING
INSUFFICIENT
-------------------------------------------------------- Chapter 4:1.4

Agency examiners have cited as hindrances to their examination
efforts a lack of sufficient training in HMDA data analysis and in
discrimination detection techniques, and an insufficient time
allowance in which to uncover discriminatory conduct. The lack of
examiner training in some fair lending procedures was mentioned by a
number of consumer compliance examiners in our case studies of
institutions examined by FRB, OCC, FDIC, and OTS in 1993. Of the 39
examiners responding to our survey questions about training, 17
reported not having received HMDA-related training as part of their
own agencies' general compliance training program and 7 reported that
as of December 30, 1993, they had no HMDA-related training
whatsoever.

Inexperience in HMDA data analysis and fair lending detection
techniques was also cited by the examination staff at FDIC. During a
year-long, agency-initiated review of FDIC-supervised institutions
with large disparities in minority/white mortgage application denial
rates, which was completed in 1995, examination staffs cited the lack
of formal training, unclear guidance, and the need for improved
resource tools as causing difficulties in conducting fair lending
examinations.

Another problem cited by some examiners in our case studies was an
insufficient time allowance during examinations to develop evidence
of substantive violations. During interviews and in survey
responses, a number of examiners told us that they had insufficient
time during consumer compliance examinations to conduct useful HMDA
analyses or comparative-file analyses. The HMDA data, they said, are
simply too voluminous and too complex to analyze in the time
allotted. Examiners often criticized the FFIEC-standardized HMDA
output for being too narrow in scope--it only includes mortgage
lending--or too institution-specific to give a full picture of
lending patterns in a community. It would be more helpful, they
suggested, if HMDA analyses were done by well-trained specialists
prior to each examination so that examiners could focus more quickly
on suspicious patterns. Also, some examiners felt that a graphical
representation of HMDA and census data combined could be more quickly
and easily interpreted than the summary tables currently being
provided.

If not addressed, the time constraints and lack of training cited by
examiners may prove to be major obstacles to using the new
comparative-file techniques designed to detect disparate treatment.
The comparative-file approach, as currently practiced, is relatively
more labor intensive and time consuming than the older procedures.
Not only do the newer comparative-file procedures require examiners
to review many more loan files than was previously dictated, but they
also may require them to search for "matched-pairs" to compare for
equal treatment.

While our survey results and interviews highlighted the needs of some
examiners for training in fair lending issues and examination
techniques, we acknowledge the fact that the federal banking
regulatory agencies and HUD have already instituted advanced HMDA
training programs and have upgraded their automated systems to
improve their ability to analyze HMDA data and to integrate it with
other databases. It will take some time, however, before all agency
examination staff receive such training and become familiarized with
the new software programs for analyzing HMDA data. Furthermore,
adoption of uniform fair lending examination procedures, as
recommended later in this report, would necessitate an additional
training effort focusing on those examination techniques and
procedures not previously common to all the banking regulatory
agencies.

SOME UNCERTAINTY REMAINS
REGARDING THE REFERRAL
REQUIREMENT
-------------------------------------------------------- Chapter 4:1.5

Since 1992, ECOA has required the regulatory agencies to refer
certain violations of ECOA to DOJ. Specifically, section 706(g) of
ECOA states that an agency charged with enforcing the act "shall
refer the matter to the Attorney General whenever the agency has
reason to believe that one or more creditors has engaged in a pattern
or practice of discouraging or denying applications for credit in
violation of [the Act]".\8 A similar provision in FHA also charges
HUD with the referral mandate.\9 However, officials at HUD and
several of the banking regulatory agencies have expressed some
uncertainty about how to identify a "pattern or practice" in a
particular case.

Such uncertainty arises, in part, from two terms within the statutory
language--"reason to believe" and "pattern or practice." Since the
law does not precisely define the meanings of these terms, they must
be discerned on a case-by-case basis. Hence, the standard for
referrals to DOJ is sometimes unclear. For instance, if a "pattern
or practice" decision is not necessarily a mathematical process (the
Interagency Policy Statement on Discrimination in Lending says it is
not), then whatever decisionmaking process should be used merits
clarification. Although both DOJ and the Fair Lending Task Force
have discussed some of these issues informally and in the Interagency
Policy Statement on Discrimination in Lending, agency officials and
examiners indicated during discussions with us that further
clarification would be helpful.

Additional guidance on the characteristics of a referable case should
be possible. Since the agencies have frequently sought such guidance
from DOJ on a case-by-case basis over the last several years, it
should now be possible for DOJ to share its accumulated experience
regarding "pattern or practice" inquiries.

--------------------
\8 15 U.S.C. 1691e(g).

\9 See 42 USC 3610(e)(2).

DESPITE IMPROVEMENTS, DETECTION
OF DISCRIMINATION WILL REMAIN
DIFFICULT
---------------------------------------------------------- Chapter 4:2

At the core of many of the difficulties encountered in attempts to
improve enforcement of the fair lending laws is the fact that
detection of lending discrimination has been, and continues to be, a
difficult and time-consuming task. One that, in the end, often
requires examiners to use their professional judgment in determining
whether violations have occurred. For example, research studies have
suggested that the problem of disparate treatment is highly
concentrated among marginal loan applications-- those that have one
or more deficiencies. This observation has been independently
supported by both agency and DOJ investigations. Yet, to uncover
disparate treatment at the margin is extremely difficult and time
consuming, even with improved examination techniques. Automated
statistics-based detection methods may aid examiners in their search
for possible discriminatory lending patterns or practices, but they
are as yet not deemed accurate or reliable enough by economists and
agency officials to replace examiner judgment--nor is that the
intention of the agencies. Even though several of the agencies have
adopted revised fair lending examination procedures that emphasize
comparative-file analysis, it seems unlikely that all instances of
discriminatory treatment could be discovered. It is, therefore,
critical that the agencies continue to research and develop better
detection methodologies in order to increase the likelihood of
detecting illegal practices. Moreover, we encourage the agencies'
efforts to broaden their knowledge and understanding of the credit
search and lending processes in general. Such knowledge is
prerequisite to both improved detection and prevention of
discriminatory lending practices.

UNRESOLVED LEGAL ISSUES POSE
BARRIERS TO VOLUNTARY
COMPLIANCE EFFORTS
---------------------------------------------------------- Chapter 4:3

To date, all but one of the cases brought by DOJ under ECOA and FHA
have been settled by consent agreements wherein the defendants have
not admitted any wrongdoing. While these agreements serve a public
purpose by speedily bringing cases to closure, remedying the alleged
wrongs, and highlighting the government's commitment to enforce the
fair lending laws, they leave some complex legal issues unanswered.
For example: what liabilities are associated with self-assessment
programs? How will the disparate effects test be applied across
widely divergent geographic markets with different demand and supply
pressures? Is credit scoring and accompanying differential pricing
illegal? As long as these issues remain unresolved, they could
contribute to the reluctance of some financial institutions to
initiate self-testing and other voluntary compliance programs
suggested by the bank regulatory agencies to improve compliance with
the fair lending laws. Some of the more pressing areas in need of
clarification are discussed below.

CONCERNS ABOUT LIABILITY FOR
RESULTS OF SELF-ASSESSMENT
PROGRAMS
-------------------------------------------------------- Chapter 4:3.1

Among the banking regulatory agencies' list of suggested activities
for banks and other lending institutions to employ to ensure
compliance with the fair lending laws is a continuing program of
self-assessment. As described by FDIC, self- assessment is a way of
measuring, in a controlled manner, differences in treatment of
customers and potential customers. It can consist of a variety of
programs, including preapplication testing and comparative-file
analysis. The goal of these programs is to help find potential
problems so that corrective actions can be taken and to help ensure
that an institution's lending practices and decisions are not
discriminatory. In addition, an institution can also gain insight
into how its lending personnel and practices are perceived by
prospective loan applicants, a valuable insight not readily available
through other audit methods.

However, some self-assessment activities, like self-testing, pose a
dilemma for lending institutions in that under current law the
results of self-testing programs may not be privileged or protected
from disclosure to federal regulatory agencies or private litigants.
Hence, despite the obvious preventative benefits to be gained from
having lenders adopt continuous self- testing programs, many
institutions are reluctant to undertake such programs out of fear
that the findings could be used as evidence against them, especially
by third-party litigants.

A few states have recognized this dilemma faced by financial
institutions and have moved to eliminate, or at least partially
mitigate the fears institutions have regarding self-incrimination.
For example, Maryland has passed legislation that partially protects
from discovery by third parties information obtained through the use
of self-testing programs. Additionally, the Attorney General of
Massachusetts has entered into an agreement with the Massachusetts
Banker's Association to seek the enactment of legislation to ensure
that institutions engaging in self- testing and comparative-file
review evaluation by senior level management, as it relates to
residential financing, will not be (a) forced to disclose to private
litigants in civil actions the results of the use of such methods and
(b) subject to legal action by nonregulatory government agencies
based on the results of such self-testing and comparative-file
reviews. The legislation was under consideration by the
Massachusetts State legislature at the time of our review. Such
legal protection is intended to engender voluntary self-testing
programs and place greater emphasis on prevention rather than ex-post
government enforcement through regulation and legal action.

CONCERNS ABOUT THIRD-PARTY
LIABILITY
-------------------------------------------------------- Chapter 4:3.2

Some legal experts have suggested that a lending institution's
relationship with appraisers, loan brokers, and other financial
institutions could pose potential compliance problems if the
third-party were found to be in violation of FHA or ECOA. For
example, if loan brokers were found to be charging, in an unlawfully
discriminatory manner, "overage" or rates above that rate at which
the underwriting bank was willing to lend, then the funding
institution could possibly be held liable as well. This argument is
based on the premise that the funding institution would be in a
position to know the magnitude and distribution of the price
differentials charged by its affiliate brokers. In light of this
potential liability, institutions may be reluctant to monitor
third-party lending practices.

POTENTIAL PROBLEMS
ASSOCIATED WITH THE
DISPARATE IMPACT TEST
-------------------------------------------------------- Chapter 4:3.3

As interpreted by the federal regulators and others, there exists
under the fair lending laws a disparate impact test for
discrimination.\10 Under this test, unlawful discrimination is
presumed to occur if a lender maintains a neutral lending policy or
practice that has a disproportionate adverse effect on members of a
protected group and for which there is no business necessity and no
less discriminatory alternative. Yet, because no case involving the
disparate impact test within the context of lending discrimination
has been decided by a court, considerable controversy exists as to
how the test should or will be applied in certain lending scenarios.
Although DOJ and the federal banking regulatory agencies have tried
to provide some guidance as to what constitutes a discriminatory
impact, the applicability of the disparate impact test to certain
lending practices remains hotly debated. Additionally, legal
scholars, bankers, and others have pointed out some common practices
in the financial services industry which may prove to be
problematical under the disparate impact test as defined in the
agencies' fair lending policy statement. Because of the lack of case
law, many significant legal questions also remain regarding the
nature of the evidence required to prove a disparate impact claim
under ECOA and FHA. Some of the areas that may prove to be
problematic are described below.

Standard of Proof: Business Necessity. Some controversy has arisen
regarding the threshold showing a lender must make in order to prove
that a practice having a disparate impact is not discriminatory.
Under the Civil Rights Act of 1991, a defendant in an employment
discrimination case can rebut a presumption that a disparate impact
was discriminatory by demonstrating that the challenged practice is
"job related ... and consistent with business necessity."\11 The
federal agencies charged with administering ECOA and FHA have adopted
this standard with adjustments to practices relevant to fair lending.
However, disagreement exists as to what constitutes a "business
necessity." Some legal experts have argued that the test can be
satisfied when the lender shows that the challenged practice serves a
"legitimate business purpose." Others have contended that the
necessity of the practice in question must be more closely
scrutinized. Resolution of this issue could prove to be significant
given the far-reaching effects such a change could have on the
development of fair lending law and policy and the burden of proof
that financial institutions and the industry would have to meet in
order to avoid liability.

Differential and Tiered Pricing Systems. Half of the fair lending
lawsuits filed against financial institutions to date by DOJ have
alleged disparate treatment in the pricing of mortgage and consumer
loans. Legal experts have indicated that some common loan pricing
strategies also have the potential to cause compliance problems for
many lending institutions under the disparate impact test. For
example, experts questioned whether differential and tiered pricing
strategies based on cost and risk factors could be said to constitute
a necessary business practice under the disparate impact test.

With differential pricing, the interest rate charged on a loan (its
price) typically will vary based on the credit risk of the borrower
or some other factor or factors that affect the lender's perception
of the borrower's ability to repay the loan. Under the disparate
impact test, however, the lender's use of such a pricing scheme (and
uniform pricing schemes as well) could prove to be problematic,
especially if a protected group were to be disproportionately
adversely affected. If that were to occur, the experts contended, a
lender could conceivably be required to prove the business necessity
of each price differential and to show that no less discriminatory
pricing alternative was available--a potentially costly proposition.

Similar problems were envisioned by legal experts for tiered pricing
systems. Under a tiered pricing system, interest rates for loans or
mortgages vary by size, with higher rates charged on loans of lesser
amounts. For example, for mortgages under a certain amount, say
$25,000, the policy might be to increase the interest rate charged by
0.5 percentage points for each $5,000 increment below the floor
amount. Fair housing advocates have argued, however, that this
pricing scheme has a disparate impact on minority or female
applicants because they tend to borrow smaller amounts. Lenders
counter, however, that higher rates on smaller denomination loans are
justified because of the business need to cover origination and
servicing expenses.

Questions have also arisen regarding profit margins under tiered
pricing systems. For example, some in the banking industry query
whether profit margins would need to be similar for each tier for an
institution to be in compliance. How much of a problem the disparate
effects test proves to be for banks will depend on how broadly the
courts, DOJ, and the regulators apply disparate impact theory.

Computerized Underwriting Programs. Some bankers argue that rigorous
interpretation and enforcement of the fair lending laws will result
in more and more institutions adopting computerized underwriting
systems to alleviate any chance of unequal or disparate treatment.
Although it is generally agreed that overall loan processing costs
are expected to decline as a result of automated underwriting,
lenders are uncertain whether such standardized systems using uniform
criteria will pass a disparate impact test, given the relative
socioeconomic status of protected groups. While some experts believe
that the use of computerized underwriting programs would be within
the bounds of the law, they point out that defending the selection
and weighting of underwriting factors could raise costs and further
limit the availability of credit to marginal applicants.

As can be seen, there remain a number of practical issues unique to
the business of lending that raise significant legal questions under
the disparate effects test for discrimination. The issues are
substantive and regulatory interpretations are likely to be
controversial. Due to the complexities of these questions, it
appears unlikely that more concrete judicial or administrative
guidance on these issues will be forthcoming in the near future.

--------------------
\10 The disparate impact test has been borrowed from employment law
as established from cases brought under Title VII of the Equal
Employment Opportunity Act.

The Senate report accompanying the 1976 amendments to ECOA expresses
the view that disparate impact applies to ECOA. Senate Report No.
94-589, 94th Congress, 2nd Session, at 4. See also the related House
Report, H.R. Rep. No. 94-210, 94th Congress, 2nd Session at 5.
See also Betsey v. Turtle Creek, 736 F.2d 983 (4th Cir. 1984) and
Mountain Side Mobile Estates v. HUD, 1995 U.S. App. Lexis 13043
(10th Cir. 1995).

\11 See 42 U.S.C. � 2000e-2(k)(l)(A)(i).

CONCLUSIONS
---------------------------------------------------------- Chapter 4:4

Since 1988, the federal regulatory agencies have made significant
strides in strengthening their oversight and enforcement of the fair
lending laws. However, during the course of our review, we
identified several areas related to the banking regulatory agencies'
existing fair lending examination policies and procedures where we
believe that the agencies have failed to take full advantage of
opportunities to ensure thorough and consistent supervision.
Additionally, unresolved legal issues involving, among other things,
interpretations of statutory language and practical applications of
the laws appear to present barriers to an immediate formulation of a
more effective policy and greater acceptance of voluntary compliance
efforts by financial institutions.

Finally, it should be remembered that the detection of discrimination
in its more subtle forms can be a difficult and time-consuming task.
Even with improved detection methodologies and clearer legal
interpretations, detection of lending discrimination in all its forms
will continue to pose a significant challenge to regulators.

MATTER FOR CONGRESSIONAL
CONSIDERATION
---------------------------------------------------------- Chapter 4:5

RECOMMENDATIONS
---------------------------------------------------------- Chapter 4:6

Despite significant improvement in federal fair lending oversight and
enforcement, we believe that further efforts can still be made in
some areas, which would strengthen the ability of federal banking
regulators to detect lending discrimination in all of its forms and
help ensure greater consistency in oversight and enforcement. To
this end, we recommend that the heads of FRB, FDIC, OCC, OTS, and
NCUA:

-- work together to develop and adopt uniform fair lending
examination procedures and provide all compliance examination
staff with the necessary training to implement those procedures;

-- adopt, as a component of their fair lending examination and
training programs, guidelines and procedures for the use of
testing methodologies for detection of discrimination at the
preapplication stage of the lending process; and

-- use their full range of enforcement authority, including the use
of civil money penalties, to ensure that the HMDA data is
submitted in a timely and accurate manner.

We also recommend that the U.S. Attorney General provide updated
guidance to the banking regulatory agencies and HUD on the
characteristics of referable "pattern or practice" cases under ECOA
and FHA.

AGENCY COMMENTS AND OUR
EVALUATION
---------------------------------------------------------- Chapter 4:7

GAO requested and received comments on a draft of this report from
FRB, OCC, FDIC, OTS, NCUA, HUD, and DOJ. These written comments
appear along with GAO's responses in appendices VI-XII. NCUA did not
comment on the conclusions and recommendations. Overall, the
agencies were in general agreement with the report findings,
conclusions, and recommendations. Agency remarks regarding
individual recommendations and our evaluations are summarized below.

All agencies expressed general agreement with GAO's recommendation
that the banking regulatory agencies adopt uniform fair lending
examination procedures and provide appropriate training with respect
to such procedures. FRB commented that they have been working on the
development of uniform procedures and anticipated submitting a draft
for consideration by other agencies in the very near future. OCC,
however, maintained that some differences in examination procedures
were appropriate, given the supervisory needs of the agencies and the
varying sizes and risk profiles of the financial institutions they
regulate. We agree that the initiatives by OCC and the other banking
regulatory agencies to establish their own fair lending procedures in
the absence of an interagency agreement on uniform procedures seems
prudent given the pressing need for improvements in the interim. We
reiterate, however, that the adoption of uniform fair lending
examination procedures would increase the likelihood that each of the
banking agencies would use the most advanced and proven techniques to
detect discrimination while applying the same degree of oversight to
all depository institutions.

Among the regulatory agencies, only HUD and OCC fully supported GAO's
recommendation that the agencies adopt guidelines and procedures for
the use of testing methodologies for detection of discrimination at
the preapplication stage of the lending process. In general, the
banking regulatory agencies had concerns about the routine use of
testing because of its associated costs and time requirements.
Despite these concerns, OCC and FRB have already authorized its
limited use in individual cases when compelling evidence exists that
an institution may be discriminating. OTS commented that it would
consider the future use of testing only after careful study, while
FDIC preferred to promote voluntary self-testing by the financial
institutions themselves. As we discuss in chapter 4, we believe that
the ability of examiners to detect illegal discrimination at the
preapplication stage of the lending process is limited. Thus, we
believe that our recommendation to adopt testing as a tool for
examinations is well grounded. However, our recommendation does not
necessarily suggest that testing be used routinely, only that the
technique be employable when a situation may warrant its use--such as
when compelling evidence from other sources suggests that
discriminatory behavior may be occurring prior to submission of a
formal written loan application.

FRB, FDIC, OCC, OTS, and HUD agreed with GAO's recommendation that
the regulatory agencies use their full range of enforcement
authorities to ensure timely and accurate HMDA data. Although FRB
and OCC expressed a willingness to consider civil money penalties in
certain cases, current policy and preference of these agencies
regarding violations of HMDA has been to require institutions they
supervise to resubmit HMDA data if it contains errors that compromise
its integrity. Although we agree that the cost associated with
correcting and resubmitting HMDA data can be significant, we believe
that the use of CMPs represents a more formal and public deterrent to
future inaccurate HMDA data submissions for both the violating
institution and others in the lending industry.

DOJ accepted our recommendation to provide updated guidance to the
bank regulatory agencies and HUD on the characteristics of referable
"pattern or practice" cases. FRB, OTS, and HUD said they would
welcome DOJ's insights regarding these cases. FRB pointed out,
however, that the ultimate responsibility to make determinations of
the meaning of statutory phrases in the absence of court opinions
rested with the agency. We are in agreement with FRB that the
absence of clear statutory language regarding the characteristics of
referable pattern and practice cases under ECOA and FHA necessitates
independent interpretations by the various agencies responsible for
fair lending enforcement. It is for precisely this reason we
recommend that DOJ provide additional guidance to the agencies
regarding the referral mandate. It is hoped that such guidance would
be of assistance to the agencies in making their own determinations
by providing a "case history" of prior referral decisions made in
agreement with DOJ.

Finally, FDIC, OCC, and HUD expressed support for our suggestion that
Congress consider legislative initiatives that remove or diminish the
disincentives associated with self-testing by protecting institutions
from having to release results of self- testing reviews when they are
followed by actions to correct any discriminatory behavior that may
have been discovered. In their comments HUD stressed how important
it was that any protection be granted only in those cases in which
lenders promptly corrected the problems found through self-testing.
We agree in principle that the implementation of corrective measures
should be a prerequisite for gaining protection for self-testing
activities and have used language to that effect in the Matter for
Congressional Consideration.

Although there are some differences in the agencies' response to our
recommendations, these differences generally reflect the agencies'
desire to retain the discretion necessary to consider the specific
facts and circumstances of individual cases. GAO agrees that agency
discretion is necessary once due consideration has been given to the
full range of regulatory alternatives and analytical techniques
available to ensure effective fair lending oversight of financial
institutions.

THE FAIR LENDING LAWS
=========================================================== Appendix I

THE FAIR HOUSING ACT (FHA)
------------------------------------------------------- Appendix I:0.1

FHA, the first of the fair lending laws, was passed by Congress in
1968 as Title VIII of the Civil Rights Act of 1968 and amended in
1974 and 1988.\1 FHA prohibits discrimination in residential real
estate-related transactions against any person because of race,
color, religion, handicap, gender, familial status, or national
origin. Under FHA it is unlawful for anyone who engages in the
business of making or purchasing residential real estate loans, or in
the selling, brokering, or appraising of residential real property,
to discriminate on the basis of any of the aforementioned factors.
Prohibited activities include, for example, denying a loan or
discriminating in fixing the amount, interest rate, duration,
application procedures, or other terms and conditions on a prohibited
basis. Any financial institution that extends housing loans is
subject to FHA.

Congress amended FHA in 1988 to, among other things, extend the
provisions of the act to certain secondary mortgage market
transactions and other purchase and sales transactions involving
residential loans and residential-related securities, and to add
"handicap" and "familial status" (having one or more children under
the age of 18) to the group of protected categories under FHA. The
1988 amendments also provided an administrative enforcement system
for HUD that did not previously exist, and removed barriers to the
use of court enforcement by private litigants and DOJ.\2

FHA is implemented by HUD Regulation (24 C.F.R. Parts 100-121).
Under the act, HUD has administrative enforcement responsibilities.
The act also provides for enforcement through civil actions initiated
by the aggrieved party or, in certain cases, HUD or DOJ. By law, the
U.S. Attorney General has authority to file suit in Federal District
Court whenever it has reason to believe that a pattern or practice of
violations exists, or when it appears that rights granted under the
act have been denied to any group, and the matter raises an issue of
general public importance. Additionally, the banking regulatory
agencies must notify HUD of apparent FHA violations in cases that are
not referred by them to DOJ.

--------------------
\1 Table I.1 provides a brief description of major amendments to the
various fair lending laws from the time of their passage through
December 31, 1994.

\2 For a detailed discussion of the 1988 amendments to FHA, see
Kushner (1992) and U.S. Commission on Civil Rights (1994).

THE EQUAL CREDIT OPPORTUNITY
ACT (ECOA)
------------------------------------------------------- Appendix I:0.2

ECOA, originally enacted in 1974 and amended in 1976 and 1991, is
broader than FHA in that it prohibits discrimination with regard to
any aspect of a consumer, commercial, or real estate credit
transaction based on race, color, religion, gender, national origin,
marital status, age, receipt of public assistance, or the exercise,
in good faith, of rights granted by the Consumer Credit Protection
Act.\3 The law applies to all persons who are creditors, including
but not limited to, banks, thrifts, credit unions, federal land
banks, investment companies, and finance companies.

ECOA is implemented by Federal Reserve Board (FRB) Regulation B (12
C.F.R. Part 202), which requires, among other things, that creditors
notify credit applicants of the action taken on their applications
and the reasons for any adverse credit decisions. Regulation B also
requires that creditors collect certain monitoring information about
the applicants. However, such data are currently only permitted to
be collected on mortgages.

The federal agencies that regulate financial institutions have
authority to enforce Regulation B administratively. However, the
1991 amendments to ECOA require the banking regulatory agencies to
refer certain cases to the U.S. Attorney General when an agency has
a "reason to believe that one or more creditors has engaged in a
pattern or practice of discouraging or denying applications for
credit" on a prohibited basis. Specifically, section 223 of FDICIA
amended ECOA to prescribe the following courses of action by the
banking regulatory agencies for specific types of apparent violations
of ECOA or FHA:

-- Pattern or practice of ECOA violations, with or without related
FHA violations: mandatory referral to DOJ;

-- Isolated ECOA violation with or without related FHA violation:
optional referral to DOJ for civil action when compliance is not
obtained administratively;

-- FHA violation that is also a violation of ECOA, but not related
to violations referred to DOJ: mandatory notice to HUD;

-- Other violations: to be handled administratively.

--------------------
\3 Note that while the protected categories under FHA and ECOA are,
for the most part, the same, there are important differences. For
example, the categories "handicap" and "familial status" are
protected under FHA but not under ECOA. Conversely, factors
protected under ECOA but not under FHA include "marital status,"
"age," "receipt of public assistance," and "the exercise of certain
consumer protection rights."

HOME MORTGAGE DISCLOSURE ACT
(HMDA)
------------------------------------------------------- Appendix I:0.3

HMDA was enacted by Congress in 1975 to provide public officials and
U.S. citizens with information to enable them to determine whether
financial institutions were serving the housing needs of their
communities. HMDA is a disclosure law. It does not prohibit any
specific activity of lenders, rather, it merely establishes a
recordkeeping and reporting obligation for certain institutions.

FRB Regulation C (12 C.F.R. Part 203), which implements HMDA,
requires most mortgage lenders located in metropolitan statistical
areas (MSA) to report annually to their respective federal
supervisory agency, and disclose to the public information about
their home mortgage and home improvement lending activity. The
reports and disclosures cover loan originations, applications that
don't result in originations, e.g., applications that are denied or
withdrawn), and purchases of loans. FRB processes the data and
prepares disclosure statements on behalf of HUD and member agencies
of the Federal Financial Institutions Examination Council (FFIEC).
Members of the council include FRB, FDIC, OCC, OTS, and NCUA. The
FFIEC also prepares aggregate reports that contain data for all
lenders in a given MSA. The reports are to be publicly available at
a central depository within each MSA. Additionally, each institution
subject to HMDA must also make its individual mortgage loan
application registers available to the public.

As originally enacted, HMDA required only the collection of
information regarding the number and total dollar amount of loans
originated or purchased by a covered institution during the fiscal
year, itemized by the geographic location of the property within a
MSA. In 1989, the Financial Institutions Reform, Recovery, and
Enforcement Act (FIRREA) amended HMDA to require, among other things,
the collection and reporting of additional data about the racial,
gender, and income characteristics of mortgage applicants and
borrowers.\4 These amendments were intended to provide data to assist
in identifying discriminatory lending practices and enforcing the
fair lending statutes. Amendments to HMDA in 1988 and 1991 extended
HMDA reporting requirements to most mortgage banking subsidiaries of
bank and thrift holding companies and independent mortgage companies
not affiliated with depository institutions. Finally, the Housing
and Community Development Act of 1992 further amended HMDA by
requiring financial institutions to make available to the public,
upon request, their loan application registers, which contain data on
loans and applications covered by HMDA.

Table I.1

Major Amendments to FHA, ECOA, and HMDA
Through December 31, 1994

Statute Year Comment
------------------ -------- ----------------------------------------
Fair 1968 Passed as Title VIII of the Civil Rights
Housing Act Act of 1968.
1974 Amended to include sex as a prohibited
factor.
1988 FHAA of 1988 extended the provisions of
FHA to secondary mortgage market
activity and other purchase and sales
transactions involving residential
loans and residential-related
securities, and added two protected
categories, handicap and familial
status. The amendments also enhanced
the enforcement responsibilities and
authority for HUD and the U.S. Attorney
General.
Equal Credit 1974 Passed as Title V of the Depository
Opportunity Act Institution Amendments Act of 1974 as
new Title VII to the Consumer Credit
Protection Act (CCPA).
1976 The Equal Credit Opportunity Act
Amendments of 1976 added prohibitions
for discrimination on the basis of
race, color, religion, national origin,
age, participation in public assistance
programs, and exercise of consumer
protection rights under the CCPA.
1991 Amended by FDICIA to require the federal
banking regulatory agencies to refer
cases to the U.S. Attorney General when
they have "reason to believe that one
or more creditors has engaged in a
pattern or practice of discouraging or
denying applications for credit" on a
discriminatory basis.
Home Mortgage 1975 Passed.
Disclosure Act
1988 Amended to extend the term "depository
institution" to include any mortgage
banking subsidiary of a bank holding
company or savings and loan holding
company, or savings and loan service
corporation, that originates or
purchases mortgage loans.
1989 Amended by FIRREA to, among other
things, require collection and
reporting of data on the disposition of
applications, and race, gender, and
income characteristics of mortgage
applicants.
1991 Amended by FDICIA to expand HMDA
reporting requirements to most
independent mortgage companies not
affiliated with depository
institutions.
1992 Amended by the Housing and Community
Development Act of 1992 to require,
among other things, financial
institutions to make available to the
public, upon request, their loan
application registers.
----------------------------------------------------------------------
Source: Compiled by GAO from FHA, ECOA, and HMDA legislative
histories.

--------------------
\4 The effect of the changes in the 1988 amendments was not actually
realized until late 1991, when the HMDA data was reported for
calendar year 1990.

LENDING DISCRIMINATION: A REVIEW
OF WHAT IS KNOWN
========================================================== Appendix II

Some insight into the nature and occurrence of lending discrimination
can be obtained by reviewing evidence from a variety of sources,
including reports in the press, analyses of HMDA data, statistical
studies, regulatory examinations, and private testing programs.
Although the evidence is often contradictory and inconclusive, such a
review can at least serve to put the current controversy surrounding
the issue of lending discrimination in perspective.

PRESS ACCOUNTS
-------------------------------------------------------- Appendix II:1

Public interest in the issue of lending discrimination was heightened
in the mid-1980s by a growing number of reports on the topic in the
press. Of particular note was the Pulitzer Prize winning series
entitled, "The Color of Money," published in May 1988 by the Atlanta
Journal-Constitution. In this influential article, the newspaper
reported that race was a decisive factor in determining the lending
patterns of metropolitan Atlanta's largest financial institutions.
This conclusion was based on the observation that among stable
neighborhoods with similar income levels, African-Americans were less
likely than whites to receive conventional mortgage credit from
financial institutions in the Atlanta area. For example, using
1981-1986 HMDA data matched with 1980 census data, the article showed
that white neighborhoods in the Atlanta area always received the
highest number of bank loans per 1,000 single-family homes, whereas
black neighborhoods always received the fewest. Because, at the
time, HMDA did not require collection of race and income data for
individual mortgage applications, the newspaper's analysis stopped
short of a conclusion that deliberate racial discrimination was
behind the difference.

Since the "Color of Money" series, similar articles have also
appeared in The Detroit Free Press, The San Francisco Chronicle, The
Washington Post, The Wall Street Journal, and elsewhere. Nearly all
of these articles have relied on analyses of HMDA data to support
their claims of racial bias in lending patterns. While none of these
reports provides conclusive evidence of discrimination, their
recurring avowal that racial bias pervades the lending practices of
financial institutions has led some to question the effectiveness of
bank oversight and oversight in the fair lending area.

EXPANDED HMDA DATA
-------------------------------------------------------- Appendix II:2

Congress responded to the concerns raised by the "Color of Money" and
similar articles by amending HMDA in 1988, 1989, and 1991. These
amendments extended HMDA coverage to additional mortgage lenders and
required covered institutions to report additional information about
loan applicants, in particular their race, gender, and income level.
Also, institutions were required to report whether each application
was approved or denied. This additional information is intended to
assist regulatory agencies in identifying possible discriminatory
lending patterns and enforcing related fair lending laws. Analyses
of HMDA data became instrumental in focusing public attention on
mortgage credit availability, especially in low- and moderate-income
neighborhoods, and on the possible existence of discrimination in
mortgage lending decisions.

The expanded HMDA data, first released in 1991 for the year 1990, and
annually thereafter, has consistently shown that while the majority
of applications for home purchase loans are approved (about 75
percent in 1994), some minority applicants experienced a
substantially higher denial rate than did white applicants (see Table
II.1).\1 For 1994, the latest year for which data were available,
HMDA data showed that for conventional home purchase loans,
approximately 33 percent of black applicants, 25 percent of Hispanic
applicants, 16 percent of white applicants, and 12 percent of Asian
applicants were denied credit. As can be seen from the table, the
difference in denial rates between blacks and whites has been both
large and persistent. However, it is widely recognized that because
of limitations with the HMDA data, conclusions regarding the nature
or pervasiveness of discrimination cannot be drawn on the basis of
HMDA data alone.

Principal among the limitations associated with HMDA data is that it
suffers from a missing variables problem. For example, HMDA data
does not include such basic indicators of an applicants'
creditworthiness as net worth, housing expense-to-income ratios, the
down-payment amount and the amount of other funds needed to close the
loan, payment history on other mortgage and consumer loans, and
employment stability of the prospective borrower. Also not included
are details about the property to be purchased, its appraised value,
credit terms of the loan, and local demand conditions. As a result,
HMDA numbers, in and of themselves, do not provide a sufficient basis
for determining whether the mortgage lending industry or any
individual lender is discriminating unlawfully. Although aware of
these limitations, HMDA data are still widely used by both federal
regulators and private groups for the purpose of targeting and
investigating financial institutions for possible fair lending
violations.

Table II.1

Percentage of Home Purchase Loan
Applications Denied, by Racial or Ethnic
Identity of Applicant, 1990-1994

Racial or ethnic
identity of Government- Government- Government- Government- Government-
applicant backed Conventional backed Conventional backed Conventional backed Conventional backed Conventional
----------------- ---------------- ---------------- ---------------- ---------------- -------------- ------------- ------------- ------------- ------------- -------------
Native American 22.5% 22.4% 22.1% 27.3% 17.5% 26.6% 17.5% 27.8% 15.3% 31.6%
\a
Asian/Pacific 12.8 12.9 12.5 15.0 13.5 15.3 11.7 14.6 10.5 12.0
Islander
Black 26.3 33.9 26.4 37.6 23.8 35.9 22.2 34.0 18.8 33.4
Hispanic 18.4 21.4 18.9 26.6 18.5 27.3 14.6 25.1 12.7 24.6
White 12.1 14.4 16.3 17.3 12.8 15.9 11.8 15.3 10.4 16.4
Other 18.4 19.0 16.3 19.9 16.0 21.0 17.8 23.1 15.0 23.8
Joint (all) 14.1% 14.9% 15.9% 17.5% 14.8% 17.6% 14.7% 17.3% 11.7% 17.2%
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Note: Government-backed mortgage loans are those guaranteed by the
Federal Housing Administration, the Department of Veterans Affairs,
or the Farmers Home Administration. These loans are generally
targeted at lower-income borrowers and tend to have less stringent
underwriting standards (e.g., lower downpayment requirements) than
conventional mortgage loans (i.e., mortgages backed by private
mortgage insurance).

\a American Indian/Alaskan Native.

Source: Board of Governors of the Federal Reserve System.

--------------------
\1 For an in-depth analysis and discussion of HMDA statistics, see
Canner and Smith (1991, 1992); Canner, Passmore, and Smith (1994);
and Canner and Passmore (1995).

STATISTICAL STUDIES
-------------------------------------------------------- Appendix II:3

A growing number of statistical studies have recently begun to
address the issue of lending discrimination in the mortgage and
credit markets. In general, these studies can be broadly categorized
into three groups: (1) redlining studies--which focus on geographic
lending patterns, (2) studies of denial rates--in which the emphasis
is on modeling the outcome of the lending decision process, and (3)
studies of default risk--which attempt to measure loan performance.

REDLINING STUDIES
------------------------------------------------------ Appendix II:3.1

Among the various types of lending discrimination, redlining--the
refusal of lenders to make mortgage loans in certain areas,
regardless of the creditworthiness of the individual loan
applicant--has probably received the most attention in the
literature. Generally, early studies of redlining often claimed to
verify the existence of this practice. However, in reviews of the
literature, Benston (1979) and Galster (1992) concluded that most
early studies were flawed, primarily because they failed to
adequately control for differences in the demand for credit across
neighborhoods and for variations in risk associated with different
geographic areas. More recently, a review of the literature by
Schill and Wachter (1993) also cited inconsistent findings with
respect to the existence of redlining and demonstrated how, in
regression models that test for redlining, the inclusion of proxies
for neighborhood risk can reverse results where redlining was
previously found to exist.

Several more recent studies also cast doubt on the existence of
redlining by lending institutions. First, a study by Benston and
Horsky (1992) failed to find evidence of redlining. Their finding
was based on interviews with several hundred home sellers and buyers
(including potential buyers and sellers who were not successful) in
allegedly redlined neighborhoods in three midwestern cities. In two
separate studies, Schill and Wachter (1993, 1994) tested for sources
of geographic disparities in lending decisions in several major
metropolitan areas, using accept/reject methodologies while
controlling for neighborhood risk. In both studies, these authors
found no support for the allegation that financial institutions
discriminate against borrowers on the basis of the racial and ethnic
compositions of their neighborhoods. And more recently, Holmes and
Horvitz (1994), using data and techniques that, they argued,
addressed the aforementioned methodological problems associated with
earlier redlining research, were able to show that racial disparities
in the flow of mortgage credit can be explained by differences in
risk and demand.

STUDIES OF DENIAL RATES
------------------------------------------------------ Appendix II:3.2

In contrast to recent redlining studies, several studies that have
attempted to model the lending decision process have found
statistically measurable disparities in loan denial rates along
racial and ethnic lines. Most notable among these are the work of
Siskin and Cupingood (1993) and Munnell, et. al. (1992). In both
of those studies, the researchers used a logit regression model to
estimate the probability that a mortgage application with specific
characteristics would be denied.\2

As part of DOJ's fair lending investigation of Decatur Federal
Savings and Loan of Atlanta, Siskin and Cupingood (1993) analyzed the
institution's mortgage lending patterns during the period 1988
through 1989 for possible racial discrimination. On the basis of
in-depth interviews with the thrift's loan officers and underwriters,
the researchers constructed a statistical model that assigned weights
to all the relevant factors in the loan decisionmaking process based
on their ability to predict the observed outcome of the event, i.e.,
the acceptance or rejection of a loan. The model was then augmented
with a race variable whose statistical significance indicated that
black applicants had a lower likelihood of obtaining conventional
fixed and adjustable-rate mortgage loans than did similarly situated
white applicants.

Similar results were also found in the widely-cited Boston Federal
Reserve Bank study by Munnell, et. al. (1992) [the Boston Fed
Study]. In that study, the researchers supplemented 1990 HMDA data
for 131 banks and thrifts in the Boston metropolitan area with
additional demographic and economic information about the applicants,
which the lenders said was relevant to their credit decisions. This
additional data was voluntarily provided by the lenders from their
original loan files. Then, using the logit regression approach, the
researchers found that, for a minority individual with the average
economic characteristics of a white applicant, the probability of
denial increased 55 percent, from 11 percent to 17 percent. Unable
to explain this disparity in denial probabilities, the economists
concluded that black and Hispanic mortgage applicants in the Boston
area were more likely to be denied credit because of their race or
ethnic background.

At the time of their release, the DOJ-sponsored study and the Boston
Fed study represented a turning point in the debate regarding the
existence of discrimination in the mortgage credit markets. Their
econometric (statistically-based) approach to investigating the issue
of lending discrimination overcame some of the reservations many had
regarding interpretations of HMDA data, and added additional weight
to the argument that race played a significant role in credit
decisions. Indeed, the findings from these studies were viewed by
many as providing proof of widespread, pervasive lending
discrimination.

Since its release, however, the Boston Fed study has been the center
of much controversy. Although its results were later duplicated by
other researchers using the same data base,\3 critics have argued
that the research suffers from problems with data accuracy and model
specification and that its results were driven, for the most part, by
a subsample of the data.\4 Many researchers, for example, contend
that studies like the Boston Fed study do not accurately capture the
standards of creditworthiness as legitimately applied by lenders, nor
do they adequately account for factors related to housing and
mortgage credit demand. Indeed, Horne (1994b) demonstrated that the
estimates of the race effect are quite sensitive to the selection of
variables and that relatively minor modifications to the Boston Fed's
statistical model are sufficient to eliminate the race effect
altogether. In a rebuttal to these criticisms, Browne and Tootell
(1995) maintained that an applicant's race still affected the
probability of getting a mortgage in the Boston metropolitan area in
1990, even when the concerns of some of the Boston Fed study's
strongest critics were incorporated into the original model.\5

Subsequent research has also questioned the appropriateness of the
single-equation regression approach to testing for discrimination
used in both the aforementioned studies. Some economists have argued
that the models suffer from debilitating flaws related to selection
or simultaneous-equations bias;\6 i.e., that the outcome of the
lending process reflects the decisions not only by the lender but by
both lenders and applicants.\7 This situation, they argue, would
necessitate not a single-equation but a complete system of equations
to explain the varied and numerous decisions required to be made by
both parties. Under such circumstances, parameter estimates from a
single-equation model would be unreliable and possibly misleading.
Other research has also shown that current model specifications have
a tendency to give "false positives," i.e., the models can
incorrectly indicate a discriminatory bias when in fact none exists
or when a thorough review of the loan files provides justification
for seemingly improbable outcomes.\8 As a result, most federal
banking regulators have not yet fully embraced the use of these types
of statistical models in their examination processes.

--------------------
\2 Logit regressions are particularly suited to modeling discrete
outcomes, i.e., in which the dependent variable can take on one of
only two values.

\3 See for example, Carr and Megbolugbe (1993); Glennon and Stengel
(1994); and Hunter (1995).

\4 See Day and Liebowitz (1993), Zandi (1993), Horne (1994, 1995).

\5 See also Munnell, et. al. (1996).

\6 Simultaneous equation bias can occur when feedback effects exist
between independent (explanatory) variables and the dependent
variable to be estimated. For example, if mortgage applicants choose
a downpayment amount (explanatory variable) based on the knowledge
that it will influence the probability of having their loan approved
(dependent variable), then the assumption of independence that
underlies regression analysis is violated and unbiased estimates of
the effects of the explanatory variables are no longer possible.

\7 See, for example, Rachlis and Yezer (1993) and Yezer, Phillips,
and Trost (1994).

\8 See Stengel and Glennon (1995) and Bauer and Cromwell (1994).

DEFAULT STUDIES
------------------------------------------------------ Appendix II:3.3

According to the economic theory of discrimination, if lenders
discriminate against minority groups, they would do so by imposing
stricter standards on loans to them than they would to whites with
truly comparable credit backgrounds.\9 In other words, lenders would
be willing to finance only the most profitable minority applications.
Theoretically then, mortgage loans approved for minority applicants
should be more profitable than loans to whites, not less profitable
or even equally profitable, and should exhibit lower default rates,
holding other factors constant.\10

Two recent studies are notable from among those that have attempted
to uncover discrimination based on the implications of discrimination
theory. Van Order, Westin, and Zorn (1993) examined the performance,
as of year-end 1991, of loans purchased by the Federal Home Loan
Mortgage Corporation (Freddie Mac) in 1983 that were originated
between 1975 and 1983. Also, Berkovec, Canner, Gabriel, and Hannan
(1994) analyzed default risk and loan losses associated with
mortgages insured by the Federal Housing Administration. In general,
both of these studies found that default rates in minority
concentrated census tracts are the same as or higher than in other
neighborhoods, not lower as predicted by economic theory. The study
by Berkovec, et. al., also concluded as one of its main findings
that black borrowers defaulted more often than did other similarly
situated borrowers.\11 While not conclusive, these findings are not
consistent with the hypothesis that lenders engage in prejudicial
discrimination against minority applicants.

--------------------
\9 See Becker (1971).

\10 A conceptual challenge to this thesis is enumerated in Galster
(1993).

\11 Berkovec, et. al., are careful to note, however, that because a
number of potentially important explanatory variables were not
accounted for in the analysis, their statistical results could
possibly suffer from bias if the omitted variables were related to
race or ethnicity. Also, in testing for discrimination, the authors
focused on only one type of lending bias--the application of
different standards of creditworthiness to different groups.
However, other types of discrimination, if present, could also lead
to similar loan default performance. For example, if prejudicial
discrimination led lenders to foreclose more quickly on black
borrowers than on other borrowers, this could result in higher
default rates for black borrowers--the very result observed in the
Berkovec study.

RESEARCH EFFORTS HAVE
GENERATED INCONCLUSIVE
RESULTS
------------------------------------------------------ Appendix II:3.4

In summary, it appears that recent empirical efforts to detect and
measure discriminatory lending activity using statistically- based
methodologies have met with mixed results. To date, researchers have
reported contradictory findings regarding the existence of lending
discrimination and are still debating the merits of various
statistical techniques and models used to study the issues. For
example, some studies of loan application denial rates have reported
finding evidence of discriminatory lending activity, but other
studies have disputed those claims and highlighted shortcomings in
the methodologies and data used in those studies. Elsewhere, recent
statistical studies of redlining and mortgage default rates have
generally not supported a finding of prejudicial discriminatory bias
on behalf of mortgage lenders, but the research cannot be considered
conclusive at this time due to ongoing methodological and data
limitations.

OTHER SOURCES OF INFORMATION
-------------------------------------------------------- Appendix II:4

In addition to news articles, HMDA data, and statistical studies of
mortgage lending patterns and practices, information regarding the
pervasiveness of lending discrimination can also be gleaned by
reviewing the results of fair lending examinations of depository
financial institutions, from agency consumer-complaint files, and
from private-sector testing programs. Unfortunately, the information
obtained from a review of these sources is as ambiguous as is that
obtained elsewhere.

The findings of routine compliance examinations, for instance, have
historically found little evidence of racial discrimination in
lending. In 1994, for example, the federal bank regulators examined
nearly 5,000 banks and thrifts for compliance with ECOA, yet
reportedly found less than two dozen violations of Regulation B
involving discrimination on the basis of race and/or sex. Of that
number, the agencies considered eight to be serious enough to merit a
referral to DOJ for further investigation. There were, however, 20
referrals to DOJ in 1994 for suspected discriminatory practices
involving protected factors others than race or sex.

Like the fair lending examination results, FRB investigations of
consumer complaints also suggest that lending discrimination may not
be widespread. In 1994, FRB reported receiving 98 complaints against
state-member banks alleging discrimination during loan- related
transactions (roughly 9 percent of the total received). After
investigation, FRB determined that in only four cases had a state
member bank possibly violated a law or regulation. In each of these
cases the bank took corrective measures voluntarily or as indicated
by FRB.\12 The resolution of an additional 16 cases, however, was
still pending at year-end 1994.

Although these numbers do not appear to be indicative of an endemic
problem, it should be noted that past and current examination
procedures, with their focus on reviews of mortgage applications
actually filed, are unlikely to detect discrimination at the
preapplication stage of the loan process (prior to the creation of
written records). For example, fair lending examinations conducted
by the banking regulators failed to uncover evidence of
discriminatory behavior in the Decatur, Chevy Chase, and Northern
Trust cases. Furthermore, some have suggested that the low number of
fair lending-related consumer complaints received by FRB and other
agencies could be the result of prospective applicants either being
unaware that they may have been treated unfairly or choosing not to
(or not knowing how to) file a formal complaint with federal
regulators.

In contrast, testing programs conducted by fair housing groups, like
the National Fair Housing Alliance (NFHA), with financial support
from HUD-administered programs, have exposed instances of
differential treatment based on race during the mortgage application
process. Moreover, findings from these programs appear to be
suggestive of a lending discrimination problem with greater
dimensions than that revealed by conventional fair lending
examinations, consumer complaints, and some other indicators. For
example, in 1991 the Chicago Fair Housing Alliance conducted
experimental tests at ten financial institutions in Chicago: two
banks, three thrifts, and five independent mortgage companies. Even
though the testing program was experimental and conducted primarily
to test the methodology, some evidence of differential treatment was
reported by the testing teams at seven of the ten institutions.
Similar findings were reported by the Philadelphia Commission on
Human Relations, which investigated mortgage lenders in and around
Philadelphia, Pennsylvania, in 1992. Of 96 completed matched-pair
tests, involving 68 institutions, the testing teams generated 11
complaints involving racial steering, discriminatory policies, and
disparate treatment. All complaints were filed with HUD and
negotiated settlements were reported to have been obtained in at
least five of the eleven cases.

Also, in the largest testing program undertaken to date, NFHA
reported that in March 1995 it had completed a FHIP-funded fair
lending testing program in which 81 lenders in 8 large metropolitan
areas across the country were tested. During this program, a total
of 760 individual tests were performed. The tests were of two types:
(1) neighborhood-based tests that attempted to detect discriminatory
behavior related to geography or housing location and (2)
applicant-based tests that were "person sensitive" so as to detect
disparate treatment of similarly qualified applicants. According to
NFHA, on average, 45 percent of the tests they conducted revealed
disparate treatment of minorities or some other type of
discriminatory lending activity. Among the prohibited activities
cited by NFHA testers were instances of minorities being charged
higher appraisal fees for houses in the same neighborhood, cases in
which minorities were told that they didn't qualify for refinancing
even though white testers with similar qualifications did, and
instances where minority testers were not given "good faith
estimates" of loan costs, or were not given the option of locking-in
interest rates, when their white counterparts were furnished these
preferences.

On the basis of these test results, in June and July of 1995, fair
lending complaints were filed by NFHA with HUD against five large
financial institutions, including two of the largest independent
mortgage companies in the nation. All were undergoing HUD's
complaint process at the time of our review and were to be further
investigated by HUD if no conciliation agreement is reached between
the respective parties. Other testing programs similar to that
conducted by NFHA were also being conducted elsewhere around the
country and results were expected in the near future.

--------------------
\12 In addition to these four cases, a complaint filed in another
case had resulted in litigation. However, the litigation was not the
result of FRB investigative findings. Rather, in this particular
case, litigation was initiated by the complainant prior to or during
FRB's investigation.

SUMMARY
-------------------------------------------------------- Appendix II:5

Findings from the expanded HMDA data, statistical studies, and other
sources justifiably raise concerns regarding the presence of
discriminatory activity in the lending industry. Results from
private-sector testing programs, for example, can be reasonably seen
to provide evidence of the existence of discrimination in lending by
some institutions in some parts of the country. Although there is
little to indicate that discrimination in the credit markets is
pervasive or widespread, the available evidence is often
contradictory and inconclusive.

For example, statistical studies examining the lending discrimination
issue have not yielded consistent results. While several studies of
loan application denial rates have reported finding evidence of
disparate treatment among mortgage lenders, recent studies of
redlining and default rates have not. Additionally, a number of
reputable researchers have argued that the findings from contemporary
statistical studies are severely limited by the data sources, the
accuracy of model specifications, and the current state of knowledge
regarding the mortgage search and underwriting processes.

As a result, despite all that has been said and done, there remains
much to be learned about the forms, occurrence, and magnitude of
discrimination in the credit markets, particularly in markets outside
the housing and mortgage credit markets. Obtaining such knowledge is
critical to development of efficient and effective methods to detect
discrimination and for proper allocation of limited public resources
toward enforcement of the fair lending laws.

DESCRIPTION OF FRB'S STATISTICAL
MODEL AND ITS USE DURING FAIR
LENDING COMPLIANCE EXAMINATIONS\1
========================================================= Appendix III

As described by FRB, its statistical model employs a two-step
process. First, HMDA data is to be used to identify institutions
that may require a more intensive review of their mortgage lending
decisions. This is to be done by conducting a statistical analysis
to determine whether applicants that are similar with respect to
income and loan amount, applying for the same loan product within the
same geographic area, are treated differently on the basis of race.
When the results of this analysis show measurable differences among
racial groups that cannot be explained by other HMDA variables, the
program automatically selects a sample of applications to be reviewed
more extensively.

For loans in the selected sample, examiners are to gather additional
data from the loan application files, including data on the
creditworthiness of the applicant and the related property. These
data are to then be analyzed in a more sophisticated regression model
to determine if the differences observed in denial rates across race,
ethnicity, or gender persist even after controlling for these other
influences. If so, the program is to pair a given applicant with one
or several other applicants (of different races or ethnic groups, for
instance) who have similar financial characteristics but who
experienced different outcomes on their loan requests. If these
matched-pair comparisons reveal any evidence of discriminatory
activity, further examination of the files is in order and an
explanation can be requested from bank management regarding any
inconsistencies. If satisfactory answers cannot be found,
enforcement actions, like cease and desist orders, may be taken, or a
referral to DOJ or HUD may be made if deemed appropriate.

--------------------
\1 As described in Canner, Smith, and Passmore (1994), and elsewhere.
For an evaluation of FRB's statistical model and its use in fair
lending examinations, see the report by FRB's Office of Inspector
General entitled, Report on the Audit of the Board's Consumer
Compliance Examination Process, Apr. 1996.

LIMITATIONS ASSOCIATED WITH THE
USE OF STATISTICAL MODELS TO
DETECT DISCRIMINATION
========================================================== Appendix IV

In the last several years, the federal regulatory agencies charged
with oversight and enforcement responsibilities under the fair
lending laws have increasingly relied on statistical analysis to
detect discriminatory lending behavior. Initially, simple
statistical measures based on HMDA data were used to screen or
identify individual financial institutions that appeared to have
unusual lending patterns. Often, these institutions became the focal
point of more in-depth regulatory review. More recently, however,
statistics have begun to play a more critical role in consideration
of lending discrimination cases. For example, DOJ now relies to a
great extent on sophisticated statistical analyses to provide
evidence of discrimination, especially the more difficult to detect
forms, like disparate treatment. In addition, FRB now employs on a
regular basis a statistical model as an examination tool for larger
institutions.

The use of such statistics and statistical models, however, is not a
panacea. Economists and statisticians have repeatedly pointed out
that statistical approaches have limitations and can, if relied on
too heavily, lead to erroneous conclusions or have unintended
consequences. For example, data errors, which will never be totally
eliminated, can lead analysts to draw incorrect conclusions. Also,
many loan decisions can be influenced by factors that are difficult
to capture in a statistical model. In general, statistical findings
will be unreliable if relevant information about loan applicants is
omitted from a model or input in an inappropriate form, especially if
the missing information is correlated with race or other important
explanatory variables. Other limitations and shortcomings associated
with the increasing use of statistics in the enforcement of the fair
lending laws are discussed below:

OTHER LIMITATIONS AND
SHORTCOMINGS
-------------------------------------------------------- Appendix IV:1

(1) Smaller institutions offer an insufficient loan pool from which
to sample. As described by FRB staff, FRB's newly employed
statistical model generally requires a database of at least 125
applications--a minimum of 25 from minority applicants and 100 from
white applicants. Under normal circumstances, however, examiners try
to obtain about 100 applications from each group. This is a
significant increase over the number of files reviewed under the
judgmental sample procedures. This large database requirement
effectively limits the use of the statistical program to the larger
institutions. For example, of the roughly 1,000 commercial banks
supervised by the Federal Reserve, only about 100 have a sufficient
volume of loan applications to be evaluated under the new statistical
technique. Institutions with insufficient loan files will still need
to be examined using current manual sampling procedures.
Furthermore, the statistical model currently employed by FRB relies,
in part, on HMDA data. This will further limit its application by
excluding its use in examinations of non-HMDA reporting institutions.
It will also make it susceptible to problems associated with poor
data quality.

(2) Prescreening or other forms of discrimination that may occur
prior to the formal submission of an application cannot be detected
with a statistical model. Comparative-file analysis in general,
whether done manually or using the new statistical modeling
techniques, looks only at the end result of the application process.
Thus, it cannot be used to detect discrimination prior to the
submission of a formal application. It does not, therefore, address
one of the fundamental weaknesses of existent examination procedures.

(3) It is not clear at this time whether the use of the new
statistical models will result in a reduction in the length of time
currently required for comparative-file analysis. Some examiners
interviewed during our case studies indicated that they were given
insufficient time to conduct comprehensive file comparisons. Yet,
the new statistical models will require examiners to construct an
expanded data set in order to competently test for disparate
treatment. And, as mentioned previously, there is no substitute for
hands-on examination of any suspicious files highlighted by the
model's matching program. As a result, it may be that the use of the
models will actually exacerbate, rather than alleviate, the time
crunch some examiners reported during their fair lending
examinations. Based on preliminary feedback from FRB examiners on
the use of the computer model, this appears to be a legitimate
concern.

For example, the modeling process requires FRB examiners not only to
pull a larger sample of loan files for review but to collect
information on an additional 12 to 15 non-HMDA variables. This has
been estimated to take from 30 to 45 minutes per loan file. For a
sample of 200 files, the number normally recommended by FRB
supervisory staff, use of the statistical modeling procedure could
increase examination time by as much as 100 to 150 hours. In light
of this, FRB has brought professional economists into the process of
conducting statistical analysis to mitigate the additional time
requirement.

(4) The expanded use of sophisticated statistical techniques to
conduct comparative-file reviews will necessitate a major training
effort on behalf of the regulatory agencies. Examiners will need to
be thoroughly familiarized with the use of the modeling techniques
and in the interpretation of the results. This will likely delay the
widespread use of the statistical models in the near term and require
a significant commitment of resources on behalf of the agencies for
years to come.

(5) Reliance on Statistics as an Enforcement Tool Could Result in
Unintended Consequences. Statistics have played a major role in
consideration of the mortgage discrimination issue to date. And,
their role as an enforcement tool appears to be increasing
dramatically. In addition to FRB's use of a statistical model as an
examination tool, DOJ now relies to a great extent on a complex
statistical model to develop evidence of the disparate treatment form
of lending discrimination. However, economists and others have
cautioned that too heavy a reliance on statistics in fair lending
enforcement could also lead, in some cases, to unintended
consequences. For example, enforcement actions based on
statistically-derived evidence could ultimately lead to a complete
replacement of a bank officer's subjective and reasoned judgment
regarding loan approvals with a computerized underwriting system.
Such a system would, without doubt or bias, eliminate from
consideration for credit all marginal applicants who do not meet the
minimum criterion of the computerized system. Yet, this result is
counterproductive to efforts designed to increase the flexibility of
underwriting standards so that minorities and other disadvantaged
groups can have greater access to credit. Marginal applicants could
be similarly affected if a lender's costs associated with ensuring
that a computerized underwriting system did not result in disparate
impact, or some other violation of the fair lending laws, outweighed
the savings achieved from automating the application process. For
profit-minded institutions, higher marginal costs of making loans
would, according to economic theory, necessarily eliminate those
applicants at the margin of creditworthiness--the very ones that many
of the flexible underwriting programs are attempting to help.

THE CASE AGAINST DECATUR FEDERAL
SAVINGS AND LOAN
=========================================================== Appendix V

DOJ's first major effort to enforce the fair lending laws was
prompted in part by the "Color of Money" article that appeared in the
Atlanta Journal-Constitution in 1988. On the basis of that story,
DOJ began an investigation into mortgage lending discrimination in
the Atlanta area. Its investigation eventually focused on a single
institution, the Decatur Federal Savings and Loan, one of the largest
mortgage originators in the Atlanta area. Based on its findings, DOJ
ultimately filed a lawsuit in September 1992, alleging that Decatur
Federal engaged in unlawful racial discrimination in its mortgage
lending program. The case was quickly settled through a consent
agreement that, among other things, assessed $1 million in damages
for the victims of the alleged discrimination.

As part of the settlement, Decatur also agreed to make significant
changes to its operating procedures so as to avoid future
discriminatory activities. Among other things, the institution
agreed to (1) redraw its CRA boundary to include previously excluded
black neighborhoods, (2) open a branch or a regional loan office in
the black residential area, (3) revamp its advertising program to
target black residential areas and seek out real estate agents and
builders serving those areas as potential aids in producing
mortgages, (4) alter its account executive commission pay structure
to provide increased incentives to market mortgage loans in black
residential areas, (5) recruit more black applicants for job
openings, and (6) adopt a program of testing to help ensure that
potential applicants would be treated without regard to race when
they visited the institution.

DOJ's investigation of Decatur Federal was also significant in that
it relied heavily on a statistical analysis of the institution's
mortgage loan files, and in so doing, signaled DOJ's view of the
evidentiary standards sufficient to establish a "pattern or practice"
of unlawful discrimination. This analysis included an examination of
thousands of loan files over a multiyear period and the creation of a
data base of more than 70 variables relevant to the institution's
underwriting process. Ultimately, DOJ's statistical analysis
revealed that the race of an applicant was a significant factor in
determining whether or not Decatur would grant a loan, even after
controlling for other economic factors thought to underlie the
institution's mortgage lending decisions.

(See figure in printed edition.)Appendix VI
COMMENTS FROM THE FEDERAL RESERVE
BOARD
=========================================================== Appendix V

See comment 1.

(See figure in printed edition.)

See comment 2.

See comment 3.

See comment 4.

(See figure in printed edition.)

See comment 5.

The following are GAO's comments on the Federal Reserve Board's
letter dated May 20, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:1

1. Change made to footnote 5 on page 37 to reflect FRB's use of the
comparative-file technique since 1979.

Although it is true that FRB has used the comparative-file technique
since it adopted a compliance examination program in 1979,
examination procedures at that time directed FRB examiners to
judgmentally sample only a small number of loan applications. We
believe such a sampling procedure may have limited the technique's
effectiveness during past examinations. Revisions to FRB's fair
lending examination procedures, however, and to those of the other
federal banking regulatory agencies as well, have both greatly
increased the number of loan applications sampled and adopted more
systematic means of selecting those files. These changes, combined
with an overall greater emphasis on the use of comparative-file
analysis, have significantly enhanced the ability of the banking
regulatory agencies to detect possible discriminatory lending
behavior.

2. We agree that the use of regression analysis is still
experimental and that continued testing by FRB is warranted.
However, should regression analysis or some other technique prove to
be an effective and efficient means for detecting possible
discriminatory lending behavior, then we maintain that all agencies
have the responsibility to adopt such a technique as part of their
examination procedures and properly train their examination staffs.
Adoption of uniform examination procedures, as recommended by GAO,
will ensure that each of the banking regulatory agencies is using the
most advanced and proven techniques to detect discrimination while
still applying the same degree of oversight to all depository
institutions.

3. As we discuss in chapter 4, we believe that the ability of the
examiners to detect illegal discrimination in the preapplication
stage of the lending process is limited. During our review, none of
the agency examiners, officials, or other experts we spoke with
disagreed with this conclusion. Thus, we believe that our
recommendation to adopt a testing methodology for detecting
discrimination at this stage of the lending process is well grounded.
Our recommendation does not necessarily suggest that testing be used
as a routine part of an agencie's fair lending examination program,
only that the technique be employable when a situation may warrant
its use--such as when compelling evidence from other sources suggests
that discriminatory behavior may be occurring prior to submission of
a formal written loan application.

We acknowledge that current FRB policy authorizes limited testing in
individual cases, although FRB officials confirmed that they have not
yet employed the technique. We are also encouraged that FRB intends
to review the results of the OCC's pilot testing program and consider
its potential for enhancing its own detection capabilities.

4. We agree that the cost associated with correcting and
resubmitting HMDA data can be significant. However, we believe that
the use of civil money penalties represents a more formal and public
deterrent to future inaccurate HMDA data submissions for both the
violating institution and others in the lending industry. Use of
CMPs may be especially effective when used in combination with the
resubmission requirement.

5. We are in agreement with FRB that the absence of clear statutory
language regarding the characteristics of referable pattern and
practice cases under ECOA and FHA necessitates independent
interpretations by the various agencies responsible for fair lending
enforcement. It is for precisely this reason we recommend that DOJ
provide additional guidance to the agencies regarding the referral
mandate. It is hoped that such guidance would be of assistance to
the agencies in making their own determinations by providing a "case
history" of prior referral decisions made in agreement with DOJ.
Currently, DOJ is the only agency privy to this collective insight.

(See figure in printed edition.)Appendix VII
COMMENTS FROM THE FEDERAL DEPOSIT
INSURANCE CORPORATION
=========================================================== Appendix V

See comment 1.

(See figure in printed edition.)

See comment 2.

(See figure in printed edition.)

Now tables 3.1
and 3.2

See comment 3.

(See figure in printed edition.)

See comment 4.

See comment 5.

See comment 6.

The following are GAO's comments on the Federal Deposit Insurance
Corporation's letter dated May 22, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:2

1. Additional language has been added to our discussion of consumer
complaint processes on page 33.

2. We acknowledge and commend the myriad on-going efforts of FDIC
and the other banking agencies to improve their educational and
outreach programs related to fair lending. Our discussion on pages
55 and 56 makes reference to FDIC's publication, Side-by-Side, but is
not intended to be a complete accounting of all agency publications
or other educational initiatives.

3. DOJ has verified the number of referrals it received from FDIC in
1994 and 1995. The numbers reported by GAO in tables 3.1 and 3.2 are
those provided by DOJ.

4. We discuss the interagency initiatives in pursuit of consistency
in fair lending examination procedures on page 58. We are encouraged
by FDIC's commitment to continue to pursue interagency cooperation
and coordination.

5. We agree with FDIC that one of the most effective means of
preventing and detecting discrimination is to encourage
self-evaluation by financial institutions. We also agree that some
degree of protection from the legal liabilities associated with
self-testing may be necessary to encourage greater voluntary efforts
and have suggested that Congress consider the issue.

6. We discuss FDIC's commitment to ensuring accurate and timely HMDA
data submissions and their use of CMPs to deter noncompliance with
the law on page 63. Our chapter 4 discussion also goes on to
describe why it is important for the regulatory agencies to use their
full range of regulatory authority to ensure timely and accurate HMDA
data submissions.

(See figure in printed edition.)Appendix VIII
COMMENTS FROM THE OFFICE OF THE
COMPTROLLER OF THE CURRENCY
=========================================================== Appendix V

See comment 1.

(See figure in printed edition.)

See comment 2.

See comment 3.

The following are GAO's comments on the Office of the Comptroller of
the Currency's letter dated June 26, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:3

1. OCC's initiative to establish its own fair lending examination
procedures in the absence of an interagency agreement on uniform
procedures seems prudent considering the pressing need for
improvements in the interim and the difficulties the agencies have
encountered in past attempts to develop uniform fair lending
examination procedures (see discussion on page 58). OCC's expressed
willingness to modify their training programs to accommodate uniform
interagency procedures in the future is also to be commended.

2. We believe that the use of civil money penalties represents a
more formal and public deterrent to future inaccurate HMDA data
submissions for both the violating institution and others in the
lending industry. Use of CMPs may be especially effective when used
in combination with the resubmission requirement.

3. OCC's informal suggestions and technical comments on the draft
report were incorporated into the final report as appropriate.

(See figure in printed edition.)Appendix IX
COMMENTS FROM THE OFFICE OF THRIFT
SUPERVISION
=========================================================== Appendix V

See comment 1.

See comment 2.

(See figure in printed edition.)

See comment 3.

See comment 4.

The following are GAO's comments on the Office of Thrift Supervision
letter dated May 17, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:4

We address OTS' substantive comments below.

1. We discuss in chapter 4 the points OTS raises about how
unresolved legal issues associated with the interpretation and
application of the fair lending laws pose substantial challenges to
oversight and enforcement efforts of the federal agencies and
discourage some financial institutions from undertaking some
voluntary compliance programs like self-testing. We are encouraged
by OTS' intention to work with other agencies as we recommended.

2. We added language to our discussion on page 64 regarding OTS'
adoption in 1995 of a policy on the use of CMPs for HMDA violations,
and are encouraged by the positive results observed in thrift
reporting for 1996.

3. We concur with OTS that the use of testing methodologies for
detecting discrimination at the pre-application stage of the lending
process is still experimental and that further careful study remains
to be done. However, pilot programs like those conducted at OCC and
HUD have already shown testing to be a viable examination tool in
certain situations when properly employed. We, therefore, encourage
all agencies to collaborate in further developing the technique and
to incorporate it into their respective examination programs to be
used when deemed appropriate.

4. OTS' appended suggestions and technical comments on the draft
report were incorporated into the final report as appropriate.

In their appended comments, OTS suggested that we more clearly
pinpoint the salient equity issue regarding what we referred to as
"the evenhanded application of the law," as variation in
interindustry oversight rather than what OTS says is the "relatively
minimal variation among banking agencies." We agree that the
differences in oversight of banking institutions and others in the
financial services industry are much greater than that between
segments of the banking industry. In fact, we highlight this fact on
page 4 of the executive summary. In contrast, our discussion of the
equity issue that appears on pages 9 and 59 is in the context of our
review of the fair lending examination procedures of the banking
regulatory agencies, not on interindustry differences in oversight.

(See figure in printed edition.)Appendix X
COMMENTS FROM THE NATIONAL CREDIT
UNION ADMINISTRATION
=========================================================== Appendix V

Now on p. 32.

See comment 1.

The following are GAO's comments on the National Credit Union
Administration letter dated June 3, 1996.

GAO COMMENT
--------------------------------------------------------- Appendix V:5

1. Change made on page 32.

(See figure in printed edition.)Appendix XI
COMMENTS FROM THE DEPARTMENT OF
HOUSING AND URBAN DEVELOPMENT
=========================================================== Appendix V

(See figure in printed edition.)

See comment 1.

(See figure in printed edition.)

See comment 2.

See comment 3.

(See figure in printed edition.)

See comment 4.

The following are GAO's comments on the Department of Housing and
Urban Development's letter dated May 22, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:6

1. Much of the material referred to by HUD is included in appendix
II, which is itself a fairly extensive and objective review of the
literature related to lending discrimination and what can be
concluded regarding its nature and occurrence in the financial
services industry.

2. We discuss in chapter 3 HUD's oversight and enforcement
responsibilities under the fair lending laws and initiatives the
Department has undertaken to fulfill those responsibilities,
including those related to the GSEs.

3. We agree in principle that the implementation of corrective
measures should be a prerequisite for gaining protection for
self-testing activities and have adopted language to that effect in
the Matter for Congressional Consideration on pages 10 and 72, which
addresses the protection for self-testing issue.

4. HUD's appended comments on the draft report were incorporated
into the final report as appropriate.

(See figure in printed edition.)Appendix XII
COMMENTS FROM THE DEPARTMENT OF
JUSTICE
=========================================================== Appendix V

See comment 1.

(See figure in printed edition.)

The following are GAO's comments on the Department of Justice's
letter dated May 30, 1996.

GAO COMMENTS
--------------------------------------------------------- Appendix V:7

1. Our discussion on disparate impact theory in chapter 4 discusses
how some common practices in the financial services industry may
prove to be problematical under the disparate impact test. While
DOJ's description of its approach to assessing fair lending
performance explains that it first looks for disparate treatment, DOJ
also describes how it will proceed when a disparate impact is
encountered. This explanation may be helpful for lenders to better
understand DOJ's approach to enforcement, but it does not dispel the
uncertainty in the industry regarding how some common lending
practices will fare under the disparate impact test.

MAJOR CONTRIBUTORS TO THIS REPORT
======================================================== Appendix XIII

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C.
------------------------------------------------------ Appendix XIII:1

Mark Gillen, Assistant Director, Financial Institutions and
Markets Issues
Kenneth D. Jones, Senior Economist and Project Manager

OFFICE OF GENERAL COUNSEL,
WASHINGTON, D.C.
------------------------------------------------------ Appendix XIII:2

Paul G. Thompson, Senior Attorney

BIBLIOGRAPHY
=========================================================== Appendix 1

Bauer, Paul W., and Brian A. Cromwell. "A Monte Carlo Examination
of Bias Tests in Mortgage Lending." Economic Review, Federal Reserve
Bank of Cleveland, Vol. 30, No. 3, (1994) pp. 27-40.

Becker, Gary S. The Economics of Discrimination. Chicago: The
University of Chicago Press, 1971.

Benston, George J. "Mortgage Redlining Research: A Review and
Critical Analysis," Journal of Bank Research, Vol. 11, (1981) pp.
8-23.

Benston, George S., and D. Horsky. "The Relationship Between the
Demand and Supply of Home Financing and Neighborhood Characteristics:
An Empirical Study of Mortgage Redlining," Journal of Financial
Services Research, Vol. 5, (1992) pp. 235-260.

Berkovec, James, Glenn Canner, Stuart Gabriel, and Timothy Hannan.
"Race, Redlining, and Residential Mortgage Loan Performance," Journal
of Real Estate Finance and Economics, Vol. 9, (1994), pp. 263-294.

Browne, Lynn E. and Geoffrey M.B. Tootell. "Mortgage Lending in
Boston--A Response to the Critics," New England Economic Review,
Federal Reserve Bank of Boston, Sept./Oct. 1995, pp. 53-78.

Canner, Glenn B., and Dolores Smith. "Home Mortgage Disclosure Act:
Expanded Data on Residential Lending." Federal Reserve Bulletin, Vol.
77, Nov. 1991, pp. 859-881.

Canner, Glenn B., and Dolores Smith. "Expanded HMDA Data on
Residential Lending: One Year Later." Federal Reserve Bulletin, Vol.
78, Nov. 1992, pp. 801-824.

Canner, Glenn B., Dolores Smith, and Wayne Passmore. "Residential
Lending to Low-Income and Minority Families: Evidence from the 1992
HMDA Data." Federal Reserve Bulletin, Vol. 80, Feb. 1994, pp.
79-108.

Canner, Glenn B. and Wayne Passmore. "Home Purchase Lending in
Low-Income Neighborhoods and to Low-Income Borrowers." Federal
Reserve Bulletin, Vol. 81, Feb. 1995, pp. 71-103.

Carr, James H., and Isaac F. Megbolugbe. "The Federal Reserve Bank
of Boston Study on Mortgage Lending Revisited." Journal of Housing
Research, Vol. 4, Issue 2, 1993.

Day, Ted and S.J. Liebowitz. "Analyzing Mortgage Discrimination in
the Twilight Zone: Making Sense of the Boston Fed Study." Working
Paper, School of Management, University of Texas at Dallas, Oct. 5,
1993.

Federal Reserve Board of Governors, Office of Inspector General.
Report on the Audit of the Board's Consumer Compliance Examination
Process, Apr. 1996, 51 p.

Galster, George C. "Research on Discrimination in Housing and
Mortgage Markets: Assessment and Future Directions." Housing Policy
Debate, Vol. 3, Issue 2, (1992) pp. 639-83.

Galster, George C. "The Facts of Lending Discrimination Cannot be
Argued Away by Examining Default Rates," Housing Policy Debate, Vol.
4, Issue 1, (1993) pp. 141-146.

Glennon, Dennis and Mitchell Stengel. "An Evaluation of the Federal
Reserve Bank of Boston's Study of Racial Discrimination in Mortgage
Lending." Comptroller of the Currency, Economic and Policy Analysis
Working Paper 94-2, (Apr. 1994).

Holmes, Andrew and Paul Horvitz. "Mortgage Redlining: Race, Risk,
and Demand." The Journal of Finance, Vol. 49, No. 1 (Mar. 1994)
pp. 81-99.

Horne, David K. "Evaluating the Role of Race in Mortgage Lending."
FDIC Banking Review, Vol. 7, No. 1, Spring/Summer 1994, pp. 1-15.

Horne, David K. "Mortgage Lending, Race, and Model Specification."
Unpublished manuscript, 1995.

Hunter, William C., and Mary Beth Walker. "The Cultural Affinity
Hypothesis and Mortgage Lending Decisions," Working Paper Series,
Federal Reserve Bank of Chicago (WP-95-8), July 1995.

Kushner, James B. "Federal Enforcement and Judicial Review of the
Fair Housing Amendments Act of 1988," in Housing Policy Debate:
Discrimination in the Housing and Mortgage Markets, Proceedings from
the Fannie Mae Annual Housing Conference, Washington, D.C., May 19,
1992, Vol. 3, Issue 2, pp. 537-99.

Lake, R. "Postscript: Unresolved Themes in the Evolution of Fair
Housing," In Housing Desegregation and Federal Policy," ed. J.
Goering, Chapel Hill, NC: University of North Carolina Press, 1986,
pp. 313-26.

Lieberman, Carolyn B. "Equal Credit Opportunity and Fair Lending,"
in Fair Lending Compliance: The New Standards and Guidelines for
Financial Institutions, Proceedings from a Conference held in
Washington, D.C., Feb. 3-4, 1994. Prentice Hall Law & Business,
Englewood Cliffs, NJ, pp. 269-275.

Miles, Barbara L. "Discrimination in Mortgage Lending: What Do We
Know?," Congressional Research Service, CRS Report for Congress, No.
94-708E, Sept. 1, 1994.

Munnell, Alicia H., Lynn E. Browne, James McEneaney, and Geoffrey M.
B. Tootell. "Mortgage Lending in Boston: Interpreting HMDA Data."
Federal Reserve Bank of Boston, Working Paper No. 92-7, Oct. 1992.

Munnell, Alicia H., Geoffrey M. B. Tootell, Lynn E. Browne, and
James McEneaney. "Mortgage Lending in Boston: Interpreting HMDA
Data." American Economic Review, Vol. 86, No. 1, Mar. 1996, pp.
25-53.

Phillips-Patrick, Fred J., and David Senft. "Thrift Mortgage Lending
in the Washington, D.C. Area." Office of Thrift Supervision,
unpublished manuscript, Nov. 1993.

Rachlis, Mitchell B., and Anthony M. Yezer. "Serious Flaws in
Statistical Tests for Discrimination in Mortgage Markets." Journal of
Housing Research, Vol. 4, Issue 2, 1993.

Richey, Charles R., Manual on Employment Discrimination Law and Civil
Rights in the Federal Courts, Clark Boardman Callaghan, New York, NY
(1994).

Schill, Michael H., and Susan M. Wachter. "A Tale of Two Cities:
Racial and Ethnic Geographic Disparities in Home Mortgage Lending in
Boston and Philadelphia." Journal of Housing Research, Vol. 4, No.
2, 1993.

Schill, Michael H., and Susan M. Wachter. "Borrower and
Neighborhood Racial Characteristics and Financial Institution
Mortgage Application Screening," The Journal of Real Estate Finance
and Economics, Vol. 9, No. 3, Nov. 1994.

Siskin, Bernard R., and Leonard A. Cupingood. "Use of Statistical
Models to Provide Evidence of Discrimination in the Treatment of
Mortgage Loan Applicants: A Study of One Lending Institution." U.S.
Department of Housing and Urban Development, Home Mortgage Lending
and Discrimination: Research and Enforcement Conference, May 18-19,
1993.

Stengel, Mitchell and Dennis Glennon. "Evaluating Statistical Models
of Mortgage-Lending Discrimination: A Bank-Specific Analysis," Paper
presented at the Allied Social Sciences Association Annual
Conference, Washington, D.C., Jan. 6, 1995.

U.S. Commission on Civil Rights, The Fair Housing Act Amendments of
1988: The Enforcement Report, Washington, D.C., Sept. 1994.

U.S. Congress. Senate. Committee on Banking, Housing, and Urban
Affairs. Subcommittee on Consumer and Regulatory Affairs.
Discrimination in Home Mortgage Lending. Hearing, 101st Congress,
1st session. Oct. 24, 1989. Washington, D.C., G.P.O. 1990. 305
p.

U.S. Congress. House. Committee on Banking, Finance and Urban
Affairs. Subcommittee on Housing and Community Development and the
Subcommittee on Consumer Affairs and Coinage. Hearing, 103rd
Congress, 1st session. May 14, 1992. Washington, D.C., G.P.O.
1993. 370 p.

U.S. Congress. Senate. Committee on Banking, Housing, and Urban
Affairs. Problems in Community Development Banking, Mortgage Lending
Discrimination, Reverse Redlining, and Home Equity Lending.
Hearings, 103rd Congress, 1st session. Feb. 3-24, 1993.
Washington, D.C., G.P.O. 1993. 831 p.

U.S. Congress. House. Committee on Banking, Finance and Urban
Affairs. Subcommittee on Consumer Credit and Insurance. President
Clinton's Community Reinvestment Act Reform Initiative and
Enforcement of the Federal Fair Lending Laws. Hearing, 103rd
Congress, 1st session. Oct. 21, 1993. Washington, D.C., G.P.O.
1994. 370 p.

U.S. Congress. Senate. Committee on Banking, Housing, and Urban
Affairs. Fair Lending Enforcement and the Data on the 1992 Home
Mortgage Disclosure Act [HMDA]. Hearing, 103rd Congress, 1st
session. Nov. 4, 1993. Washington, D.C., G.P.O. 1994. 152 p.

Van Order, Robert., Margrett Weston, and Peter Zorn. "Racial
Discrimination in the Mortgage Credit Market: Evidence from Default
Behavior." Manuscript, Cornell University (1993).

Wienk, Ronald E., "Discrimination in Urban Credit Markets: What We
Don't Know and Why We Don't Know It," Housing Policy Debate, Vol. 3,
Issue 2 (1992), pp. 217-240.

Yezer, Anthony, Robert Phillips, and Robert Trost. "Bias in
Estimates of Discrimination and Default Mortgage Lending: The
Effects of Simultaneity and Self-Selection," Journal of Real Estate
Finance and Economics, Vol. 9, No. 3, Nov. 1994, pp. 197-216.

Yezer, Anthony M., ed., Fair Lending Analysis: A Compendium of
Essays on the Use of Statistics, American Bankers Association,
Washington, D.C., 1995.

Yinger, John. "Discrimination in Mortgage Redlining: A Literature
Review and Recommendations for Future Research," Paper presented at
Home Mortgage Lending and Discrimination: Research and Enforcement
Conference, May 18-19, 1993, Washington, D.C.

Zandi, Mark. "Boston Fed's Study Was Deeply Flawed," American
Banker, Aug. 19, 1993.

*** End of document. ***