Each depositor insured to at least $250,000 per insured bank



Home > News & Events > Financial Institution Letters




Financial Institution Letters

INTERAGENCY FAIR LENDING EXAMINATION PROCEDURES
APPENDIX

TABLE OF CONTENTS

Compliance Management Analysis Checklist

Credit Scoring Analysis

Evaluating Responses to Evidence of Disparate Treatment Fair Lending Sample Size Tables

Marginal Transactions

Potential Scoping Information

Special Analyses

Streamlining the Examination

Compliance Management Analysis Checklist

This checklist is for use in conjunction with Part II of these procedures as a device for evaluating the quality of preventive and corrective measures, identifying worthwhile innovations and offering suggestions for improvement. The checklist is not, however, intended to be an absolute test of a lender's compliance management program. Lender programs containing all or most of the features described in the list may nonetheless be flawed for other reasons; conversely, a compliance program which encompasses only a portion of the factors listed below may nonetheless adequately support a strong program under appropriate circumstances. In short, the examiner must exercise his or her best judgment in utilizing this list and in assessing the overall quality of a lender's efforts to ensure fair lending compliance.

If the transactions within the proposed scope are covered by a listed self-compliance measure, check the box in the left column. Reduce the intensity (mainly the sample size) of the planned comparative file review to the degree that the self-compliance measures cover transactions within the proposed scope. Document your findings in sufficient detail to justify any resulting reduction in the intensity of the examination.

You are not required to learn whether self-compliance measures apply to specific products outside the proposed scope. However, if the information you have obtained shows that the self-compliance measure is a general practice of the lender, check the box in the second column in order to assist future examination planning.

A. Preventive Measures

Determine whether policies and procedures exists that tend to prevent illegal disparate treatment in the transactions you plan to examine. There is no legal or agency requirement for institutions to conduct these activities. The absence of any of these policies and practices is never, by itself, a violation.

1. Lending practices and standards:

--Within the proposed scope.

--Lender-wide

3. Correcting Discriminatory Conduct

a. Determine whether the lender has provisions to take appropriate corrective action and provide adequate relief to victims for any violations in the transactions you plan to review.

  • Who is to receive the self-evaluation results?
  • What decision process is supposed to follow delivery of the information?
  • Is feedback to be given to staff whose actions are reviewed?
  • What types of corrective action may occur?
  • Are customers to be:
a. Principal policy issues

Are underwriting practices clear and similar to industry standards?
Is pricing within reasonably confined ranges with guidance linking variations to risk and/or cost factors?
Does management monitor the nature and frequency of exceptions to its standards?
Are denial reasons accurately and promptly communicated to unsuccessful applicants?
NOTE: The items above are not compliance measures, but they are fundamental features of lending that tend to work against disparate treatment.
b. Do training, application-processing aids, and other guidance correctly and adequately describe:
Prohibited bases under ECOA, Regulation B, and the Fair Housing Act?
Other substantive credit access requirements of Regulation B (e.g., spousal signatures, improper inquiries, protected income
c. Is it specifically communicated to employees that they must not, on a prohibited basis:
Refuse to deal with individuals inquiring about credit?
Discourage inquiries or applicants by delays, discourtesy, or other means?
Provide different, incomplete, or misleading information about the availability of loans, application requirements, and processing and approval standards or procedures (including selectively informing applicants about certain loan products while failing to inform them of alternatives)?
Encourage or more vigorously assist only certain inquirers or applicants?
Refer credit seekers to other lenders?
Waive or grant exceptions to application procedures or credit standards?
State a willingness to negotiate?
Use different procedures or standards to evaluate applications?
Use different procedures to obtain and evaluate appraisals?
Provide certain applicants opportunities to correct or explain adverse or inadequate information, or to provide additional information?
Accept alternative proofs or creditworthiness?
Require co-signers?
Offer or authorize loan modifications?
Suggest or permit loan assumptions?
Impose late charges, reinstatement fees, etc.?
Initiate collection or foreclosure?
d. Has the institution taken specific initiatives to prevent forms of unintentional discrimination, including:
Basing credit decisions on assumptions derived from racial, gender, and other stereotypes, rather than facts?
Seeking customers from a particular racial, ethnic, or religious group, or of a particular gender, to the exclusion of other types of customers, on the basis of how "comfortable" the employee may feel in dealing with those different from him/her?
Because of their discomfort or unease in dealing with customers from certain racial, ethnic, or religious groups, or of a certain gender, limiting the exchange of credit-related information or their effort to qualify the applicant?
Is the institution's CRA assessment area drawn without unreasonably excluding minority areas?
e. Does the institution have procedures to ensure that it does not:
State racial or ethnic limitations in advertisements?
Employ code words in advertisements that convey racial or ethnic limitations?
Place advertisement that a reasonable person would regard as indicating minority customers are less desirable?
Advertise only in media serving non-minority areas of the market?
Conduct other forms of marketing only in non-minority areas of the market?
Market only through brokers known to serve only one racial or ethnic group in the market?
Use a prohibited basis in any pre-screened solicitation?


2. Compliance Audit Function: Does the Bank Attempt to DetectProhibited Disparate Treatment by Self-Evaluation?
NOTE: Do not request the results of self-evaluations. The following items are intended to obtain information about the bank's approach for self-evaluation, not its findings. Evaluating the voluntarily disclosed results of self-evaluations is described in Streamlining the Examination in the Appendix.

Mark the box if the answer is "yes" for the transactions within the scope.

a. Are the transactions reviewed by an independent analyst who:

Is directed to report objective results?
Has an adequate level of expertise?
Produces written conclusions?
b. Does the bank's approach for self-evaluation call for:

Attempting to explain major patterns shown in the HMDA data?
Determining whether actual practices and standards differ from stated ones and basing the evaluation on the actual practices?
Evaluating whether the reasons cited for denial are supported by facts relied on by the decision maker at the time of the decision?
Comparing the treatment of prohibited basis group applicants to control group applicants?
Obtaining explanations from decision makers for any unfavorable treatment of the prohibited basis group that departed from policy or customary practice?
Covering significant decision points in the loan process where disparate treatment might occur, including:
      The approve/deny decision?
      Pricing?
      Other terms and conditions?
Covering at least as many transactions as examiners would independently, if using the "Fair Lending Sample Size Guide" for a product with the application volumes of the product to be evaluated?


c. In the bank's plan for comparing the treatment of prohibited basis group applicants with that of control group applicants:

Are control and prohibited basis groups based on a prohibited basis found in ECOA or the FH Act and defined clearly to isolate that prohibited basis for analysis?
Are appropriate data to be obtained to document treatment of applicants and the relative qualifications vis-a-vis the requirement in question?
Are the data to be obtained the data on which decisions were based, not later or irrelevant information?
Does the plan call for comparing the denied applicants' qualifications related to the stated reason for denial with the corresponding qualifications for approved applicants?
Are comparisons designed to identify instances in which prohibited basis group applicants were treated less favorably than control group applicants who were no better qualified?
Is the evaluation designed to determine whether control and prohibited basis group applicants were treated differently in the processes by which the bank helped applicants overcome obstacles and by which their qualifications were enhanced?
Are responses and explanations to be obtained for any apparent disparate treatment on a prohibited basis or other apparent violations of credit rights?
Are reasons cited by credit decision makers to justify or explain instances of apparent disparate treatment to be verified?
Offered credit if they were improperly denied?
Compensated for any damages, both out of pocket and compensatory?
Notified of their legal rights?

b. Other corrective action:
Are institutional policies or procedures that may have contributed to the discrimination to be corrected?
Are employees involved, to be trained and/or disciplined?
Is the need for community outreach programs and/or changes in marketing strategy or loan products to better serve minority segments of the lender's market to be considered?
Are audit and oversight systems to be improved in order to ensure there is not recurrence of any identified discrimination?

Credit Scoring Analysis

These procedures are intended to assist an examiner in arriving at supportable conclusions with respect to an institution's record of non-discrimination when the Focal Point involves a product for which the institution uses automated underwriting or when credit scoring risk factors make such a product the Focal Point

A. Structure and Organization of the Scoring System

Determine the utilization of credit scoring at the institution including:

  • 1. For each customized credit scoring model or scorecard for any product, or for any credit scoring model used in connection with a product held in portfolio, identify:
    1. The number and inter-relationship of each model or card applied to a particular product,
    2. The purposes for which each card is employed (e.g., approval decision, set credit limits, set pricing, determine processing requirements, etc.)
    3. The developer of each card used (e.g., in-house department, afFILiate, independent vendor name) and describe the development population utilized;
    4. The types of monitoring reports generated (including front-end, back-end, account management and any disparate impact analyses), the frequency of generation and recent copies of each;
    5. All policies applicable to the use of credit scoring;
    6. Training materials and programs on credit scoring for employees, agents and brokers involved in any aspect of retail lending;
    7. Any action taken to revalidate or re-calibrate any model or scorecard used during the exam period and the reason(s) why;
    8. The number of all high-side and low-side overrides for each type of override occurring during the exam period and any guidance given to employees on their ability to override;
    9. All cutoffs used for each scorecard throughout the examination period and the reasons for any change made during the exam period;
    10. All variables scored by each product's scorecard(s) and the values that each variable may take; and
    11. The method used to select for disclosure those adverse action reasons arising from application of the model or scorecard.
  • For each judgmental underwriting system that includes as an underwriting criteria a standard credit bureau or secondary market credit score identify:
    1. The vendor of each credit score and any vendor recommendation or guidance on the usage of the score relied upon by the institution.
    2. The institution's basis for using the particular bureau or secondary market score and the cutoff standards for each product's underwriting system and the reasons for any changes to the same during the exam period;
    3. The number of exceptions or overrides made to the credit score component of the underwriting criteria and the basis for those exceptions or overrides, including any guidance given to employees on their ability to depart from credit score underwriting standards, and;
    4. Types of monitoring reports generated on the judgmental system or its credit scoring component (including front-end, back-end, differential processing and disparate impact analysis), the frequency of generation and recent copies of each.

B. Adverse Action Disclosure Notices

Determine the methodology used to select the reasons why adverse action was taken on a credit application denied on the basis of the applicant's credit score. Compare the methodology used to the examples recited in the Commentary to Regulation B and decide acceptability against that standard. Identify any consumer requests for reconsideration of credit score denial reasons and review the action taken by management for consistency across applicant groups.

Where a credit score is used to differentiate application processing, and an applicant is denied for failure to attain a judgmental underwriting standard that would not be applied if the applicant had received a better credit score (thereby being considered in a different-presumably less stringent--application processing group), ensure that the adverse action notice also discloses the bases on which the applicant failed to attain the credit score required for consideration in the less stringent processing group.

C. Disparate Treatment in the Application of Credit Scoring Programs

  1. Determine what controls and policies management has implemented to ensure that the institution's credit scoring models or credit score criteria are not applied in a discriminatory manner; in particular:
    1. Examine institution guidance on using the credit scoring system, on handling overrides and on processing applicants and how well that guidance is understood and observed by the targeted employees and monitored for compliance by management
    2. Examine institution policies that permit overrides or that provide for different processing or underwriting requirements based on geographic identifiers or borrower score ranges to assure that they do not treat protected group applicants differently than other similarly situated applicants.
  2. Evaluate whether any of the bases for granting credit to control group applicants who are low-side overrides are applicable to any prohibited basis denials whose credit score was equal to or greater than the lowest score among the low-side overrides. If such cases are identified, obtain and evaluate management's reason for why such different treatment is not a fair lending violation.
  3. Evaluate whether any of the bases for denying credit to any prohibited basis applicants who are high side overrides are applicable to any control group approvals whose credit score was equal to or less than the highest score among the prohibited basis high-side overrides. If such cases are identified, obtain and evaluate management's reason for why such different treatment is not a fair lending violation.
  4. If credit scores are used to segment applicants into groups that receive different processing, or are required to meet additional underwriting requirements (e.g., "tiered risk underwriting"); perform a comparative FILe review, or confirm the results and adequacy of management's comparative FILe review, that evaluates whether all applicants within each group are treated equally.

D. Credit Scoring Systems that Include Age

Regulation B imposes certain requirements on credit scoring systems that include age as a variable in the determination of creditworthiness. This examination section applies only to credit scoring systems that consider age.

Age is considered in a credit scoring system either (a) when it is explicitly included as a variable in the items scored by the system, or (b) when the population is segmented into more than two groups that are separated by the age of the members and each group is then separately scored.

  1. Age as an Explicit Variable.
    1. When age is an explicit variable, the examiner must first determine whether 202.6(b)(2)(ii) is met by the system. A credit scoring system meets this regulatory requirement when it assigns applicants age 62 or older the same or higher number of points for their age as are assigned to the most favored age (or age group) among all other applicants less than 62 years old. If the system fails this requirement, then, even if it is empirically derived and statistically sound, it still violates Regulation B. For purposes of assessing the degree of damage caused by the violation, the examiner should obtain a count of the number of applicants age 62 or older who, but for the difference in the score received for age as compared to the age (or age group) most favored by the system, (i) would have received credit, (ii) received credit on terms other than those for which they applied, or (iii) received credit on terms less favorable than which they would otherwise have qualified.
    2. If the system examined assigns applicants age 62 or older a value for "age" that is equal to or greater than the value assigned to any other age or age group and the values assigned to the variable "age" for all applicants under age 62 is the same, then the system is permissible without further analysis. This conclusion results from the fact that even if the system was not empirically derived and demonstrably and statistically sound, age is only being used to favor the elderly and accordingly is permissible under 202.6(b)(2)(iv). Therefore, it is immaterial to the examiner whether the system is an empirically derived, demonstrably and statistically sound credit scoring system or is a judgmental system.
    3. If the system assigns applicants age 62 or older a value for "age" that is equal to or greater than the value assigned to any age or age group less than 62, but among applicants less than 62 the values for "age" are not equal, then the system must be empirically derived, demonstrably and statistically sound. To examine for these properties, the examiner should consult the agency's most recent guidance on that issue.
  2. Age-split Scorecards

    The commentary to Regulation B declares that a creditor may segment the population of applicants into scorecards based on the age of an applicant. When a system uses a card covering a wide age range that encompasses elderly applicants, the credit scoring system does not score age. But if a system segments the population by age into multiple scorecards, and includes elderly applicants in a narrower age range, the credit scoring system does score age. For those systems falling into the first category of being deemed as not scoring age, there is no obligation to conduct a review for empirical derivation and statistical soundness.

    If age-split scorecards are deemed to score age, they must be empirically derived, demonstrably and statistically sound, and must treat persons 62 or older at least as favorably as any other age group.

E. Examination for Empirical Derivation and Statistical Soundness

Regulation B requires credit scoring systems that use age to be empirically derived, and demonstrably and statistically sound. This means that they must fulFILl the requirements of 202.2(p)(1)(i) - (iv). Obtain documentation provided by the developer of the system and consult the agency's most recent guidance for making that determination.

Evaluating Responses to Evidence of Disparate Treatment

A. Responses to Comparative Evidence of Disparate Treatment

The following are responses that a lender may offer -- separately or in combination -- to attempt to explain that the appearance of illegal disparate treatment is misleading, and that no violation has in fact occurred. The responses, if true, rebut the appearance of disparate treatment. The examiners must evaluate the validity and credibility of the responses.

1. The lender's personnel were unaware of the prohibited basis identity of the applicant(s).

If the lender claims to have been unaware of the prohibited basis identity (race, etc.) of an applicant or neighborhood, ask it to show that the application in question was processed in such a way that the institution's staff who made the decisions could not have learned the prohibited basis identity of the applicant.

If the product is one for which the institution maintains prohibited basis monitoring information, assume that all employees could have taken those facts into account. Assume the same when there was face-to-face contact between any employee and the customer.

If there are other facts about the application from which an ordinary person would have recognized the applicant's prohibited basis identity (for example, the surname is an easily recognizable Hispanic one), assume that the institution's staff drew the same conclusions. If the racial character of a community is in question, ask the institution to provide persuasive evidence why its staff would not know the racial character of any community in its service area.

2. The difference in treatment was justified by differences in the applicants (applicants not "similarly situated")

Ask the lender to account for the difference in treatment by pointing out a specific difference between the applicants' qualifications, or some factor not captured in the application but that legitimately makes one applicant more or less attractive to the lender, or some non-prohibited factor related to the processing of their applications. The difference identified by the lender must be one that is important enough to justify the difference in treatment in question, not a meaningless difference.

The factors commonly cited to show that applicants are not similarly situated fall into two groups: those that can be evaluated by how consistently they are handled in other transactions, and those that cannot be evaluated in that way.

a. Verifying "not similarly situated" explanations by consistency

The appearance of disparate treatment remains if a factor cited by the lender to justify favorable treatment for a control group applicant also exists for an otherwise similar prohibited basis applicant who was treated unfavorably. Similarly, the appearance of disparate treatment remains if a factor cited by the lender to justify unfavorable treatment for a prohibited basis applicant also exists for a control group applicants that got favorable treatment. If this is not so, ask the lender to document that the factor cited in its explanation was used consistently for control group and prohibited basis applicants.

Among the responses that should be evaluated this way are:

  • Customer relationship. Ask the lender to document that a customer relationship was also sometimes considered to the benefit of prohibited basis applicants and/or that its absence worked against control group customers.
  • "Loan not saleable or insurable." If FILe review is still in progress, be alert for loans approved despite the claimed fatal problem. At a minimum, ask the lender to be able to produce the text of the secondary market or insurer's requirement in question.
  • Difference in standards or procedures between branches or underwriters. Ask the lender to provide transactions documenting that each of the two branches or underwriters applied its standards or procedures consistently to both prohibited basis and control group applications it processed, and that each served similar proportions of the prohibited basis group.
  • Difference in applying the same standard (difference in "strictness") between underwriter, branches, etc. Ask the lender to provide transactions documenting that the stricter employee, branch, etc., was strict for both prohibited basis and control group applicants and that the other was lenient for both, and that each served similar proportions of the prohibited basis group. The best evidence of this would be prohibited basis applicants who received favorable treatment from the lenient branch and control group applicants who received less favorable treatment from the "strict" branch.
  • Standards or procedures changed during period reviewed. Ask the lender to provide transactions documenting that during each period the standards were applied consistently to both prohibited basis and control group applicants.
  • Employee misunderstood standard or procedure. Ask the lender to provide transactions documenting that the misunderstanding influenced both prohibited basis and control group applications. If that is not available, find no violation if the misunderstanding is a reasonable mistake.

b. Evaluating "not similarly situated" explanations by other means.

If consistency cannot be evaluated, consider an explanation favorably even without examples of its consistent use if:

  • the factor is documented to exist in (or be absent from) the transactions, as claimed by the institution;
  • the factor is one a prudent lender would consider;
  • FILe review found no evidence that the factor is applied selectively on a prohibited basis (in other words, the lender's explanation is "not inconsistent with available information"); and
  • the lender's description of the transaction is generally consistent and reasonable.

Some factors that may be impossible to compare for consistency are:

  • Unusual underwriting standard. Ask the lender to show that the standard is prudent. If the standard is prudent and not inconsistent with other information, accept this explanation even though there is no documentation that it is used consistently.
  • "Close calls." The lender may claim that underwriters' opposite decisions on similar applicants reflects legitimate discretion that the examiners should not second-guess. That is not an acceptable explanation for identical applicants with different results, but is acceptable when the applicants have differing strengths and weaknesses that different underwriters might reasonably weigh differently. However, do not accept the explanation if other FILes reveal that these "strengths" or "weaknesses" are counted or ignored selectively on a prohibited basis.
  • "Character loan." Expect the lender to identify a specific history or specific facts that make the applicant treated favorably a better risk than those treated less favorably.
  • "Accommodation loan." There are many legitimate reasons that may make a transaction appealing to a lender apart from the familiar qualifications demanded by the secondary market and insurers. For example, a customer may be related to or referred by an important customer, be a political or entertainment figure that would bring prestige to the institution, be an employee of an important business customer, etc. It is not illegal discrimination to make a loan to an otherwise unqualified control group applicant who has such attributes while denying a loan to an otherwise similar prohibited basis applicant without them. However, be skeptical when the lender cites reasons for "accommodations" that an ordinary prudent lender would not value.
  • "Gut feeling." Be skeptical when lenders justify an approval or denial by a general perception or reaction to the customer. Such a perception or reaction may be linked to a racial or other stereotype that legally must not influence credit decisions. Ask whether any specific event or fact generated the reaction. Often, the lender can cite something specific that made him or her confident or uncomfortable about the customer. There is no discrimination if it is credible that the lender indeed considered such a factor and did not apply it selectively on a prohibited basis.

    c. Follow up customer contacts

    If the lender's explanation of the handling of a particular transaction is based on customer traits, actions, or desires not evident from the file, consider obtaining agency authorization to contact the customer to verify the lender's description. Such contacts need not be limited to possible victims of discrimination, but can include control group applicants or other witnesses.

    3. The different results stemmed from an inadvertent error.

    If the lender claims an identified error such as miscalculation or misunderstanding caused the favorable or unfavorable result in question, evaluate whether the facts support the assertion that such an event occurred.

    If the lender claims an "unidentified error" caused the favorable or unfavorable result in question, expect the lender to provide evidence that discrimination is inconsistent with its demonstrated conduct, and therefore that discrimination is the less logical interpretation of the situation. Consider the context (as described below).

    4. The apparent disparate treatment on a prohibited basis is a misleading portion of a larger pattern of random inconsistencies.

    Ask the institution to provide evidence that the unfavorable treatment is not limited to the prohibited basis group and that the favorable treatment is not limited to the control group. Without such examples, do not accept a lender's unsupported claim that otherwise inexplicable differences in treatment are distributed randomly.

    If the lender can document that similarly situated prohibited basis applicants received the favorable treatment in question approximately as frequently and in comparable degree as the control group applicants, conclude there is no violation.

    NOTE: Transactions are relevant to "random inconsistency" only if they are "similarly situated" to those apparently treated unequally.

    5. Loan terms and conditions.

    The same analyses described in the preceding sections with regard to decisions to approve or deny loans also apply to pricing differences. Risks and costs are legitimate considerations in setting prices and other terms and conditions of loan products. However, generalized reference by the lender to "cost factors" is insufficient to explain pricing differences.

    If the lender claims that specific borrowers received different terms or conditions because of cost or risk considerations, ask the lender to be able to identify specific risk or cost differences between them.

    If the lender claims that specific borrowers received different terms or conditions because they were not similarly situated as negotiators, consider whether application records might provide relevant evidence. If the records are not helpful, consider seeking authorization to contact customers to learn whether the lender in fact behaved comparably toward prohibited basis and control group customers. The contacts would be to learn such information as the lender's opening quote of terms to the customer and the progress of the negotiations.

    If the institution responds that an average price difference between the control and prohibited basis groups is based on cost or risk factors, ask it to identify specific risk or cost differences between individual control group applicants with the lowest rates and prohibited basis group applicants with the highest that are significant enough to justify the pricing differences between them. If the distinguishing factors cited by the institution are legitimate and verifiable as described in the sections above, remove those applications from the average price calculation. If the average prices for the remaining control group and prohibited basis group members still differ more than minimally, consult within the agency about obtaining an analysis of whether the difference is statistically significant. Find a violation only if (1) there is evidence of disparate treatment of similarly situated borrowers or (2) there is a particular risk factor that meets all the criteria for a disproportionate adverse impact violation.

    B. Responses to Overt Evidence of Disparate Treatment

    1. Descriptive references vs. lending considerations

    A reference to race, gender, etc., does not constitute a violation if it is merely descriptive -- for example, "the applicant was young." In contrast, when the reference reveals that the prohibited factor influenced the lender's decisions and/or customer behavior, treat the situation as an apparent violation to which the lender must respond.

    2. Personal opinions vs. lending considerations

    If an employee involved with credit availability states unfavorable views regarding a racial group, gender, etc., but does not explicitly relate those views to credit decisions, review that employee's credit decisions for possible disparate treatment of the prohibited basis group described unfavorably. If there are no instances of apparent disparate treatment, treat the employee's views as permissible private opinions. Inform the lender that such views create a risk of future violations.

    3. Stereotypes related to credit decisions

    There is an apparent violation when a prohibited factor influences a credit decision through a stereotype related to creditworthiness, even if the action based on the stereotype seems well-intended -- for example, a loan denial because "a single woman could not maintain a large house." If the stereotyped beliefs are offered as "explanations" for unfavorable treatment, regard such unfavorable treatment as apparent illegal disparate treatment. If the stereotype is only a general observation unrelated to particular transactions, review that employee's credit decisions for possible disparate treatment of the prohibited basis group in question. Inform the lender that such views create a risk of future violations.

    4. Indirect reference to a prohibited factor

    If negative views related to creditworthiness are described in non-prohibited terms, consider whether the terms would commonly be understood as surrogates for prohibited terms. If so, treat the situation as if explicit prohibited basis terms were used. For example, a lender's statement that "It's too risky to lend north of 110th Street" might be reasonably interpreted as a refusal to lend because of race if that portion of the lender's lending area north of 110th Street were predominantly black and the area south white.

    5. Lawful use of a prohibited factor

    a. Special Purpose Credit Program (SPCP)

    If a lender claims that its use of a prohibited factor is lawful because it is operating an SPCP, ask the lender to document that its program conforms to the requirements of Regulation B. An SPCP must be defined in a written plan that existed before the lender made any decisions on loan applications under the program. The written plan must:

    • demonstrate that the program will benefit persons who would otherwise be denied credit or receive credit on less favorable terms; and
    • state the time period the program will be in effect or when it will be re-evaluated.

    No provision of an SPCP should deprive people who are not part of the target group of rights or opportunities they otherwise would have. Qualified programs operating on an otherwise-prohibited basis will not be cited as a violation.

    NOTE: Advise the lender that an agency finding that a program is a lawful SPCP is not absolute security against legal challenge by private parties. Suggest that an institution concerned about legal challenge from other quarters use exclusions or limitations that are not prohibited by ECOA or the FH Act, such as "first-time home buyer."

    b. Second review program

    Such programs are permissible if they do no more than ensure that lending standards are applied fairly and uniformly to all applicants. For example, it is permissible to review the proposed denial of applicants who are members of a prohibited basis groups by comparing their applications to the approved applications of similarly qualified individuals who are in the control group to determine if the applications were evaluated consistently.

    Ask the lender to demonstrate that the program is a safety net that merely attempts to prevent discrimination, and does not involve underwriting terms or practices that are preferential on a prohibited basis.

    Statements indicating that the mission of the program is to apply different standards or efforts on behalf of a particular racial or other group constitute overt evidence of disparate treatment. Similarly, there is an apparent violation if comparative analysis of applicants who are processed through the second review and those who are not discloses dual standards related to the prohibited basis.

    c. Affirmative marketing/advertising program:

    Affirmative advertising and marketing efforts that do not involve application of different lending standards are permissible under both the ECOA and the FH Act. For example, special outreach to a minority community would be permissible.

Fair Lending Sample Size Tables

Table A
Underwriting (Accept/Deny) Comparisons
Sample 1
Prohibited Basis Denials
Sample 2
Control Group Approvals
Number of Denials or Approvals

5 - 50

51 - 150

> 150

20 - 50

51 – 250

> 250

Minimum to review:

All

51

75

20

51

100

Maximum to review:

50

100

150

5x prohibited basis sample (up to 50)

5xprohibited basis sample (up to 125)

5x prohibited basis sample (up to 300)

Table B
Terms and Conditions Comparisons
Sample 1
Prohibited Basis Denials
Sample 2
Control Group Approvals
Number of Approvals

5-25

26 - 100

> 100

20 -50

51 – 250

> 250

Minimum to review:

All

26

50

20

40

60

Maximum to review:

25

50

75

5x prohibited basis sample (up to 50)

5x prohibited basis sample (up to 75)

5x prohibited basis sample (up to 100)

Explanatory Notes to Sample Size Tables

  1. When performing both underwriting and terms and conditions comparisons, use the same control group approval sample for both tasks.
  2. If there are fewer than 5 prohibited basis denials or 20 control group approvals, refer to "Sample Size" instructions in the procedures.
  3. "Minimum" and "maximum" sample sizes: select a sample size between the minimum and maximum based on the outcome of the Compliance Management Review conducted in Part II of these procedures. Once the sample size has been determined, select individual transactions judgmentally. Refer to procedures.
  4. If two prohibited basis groups (e.g., black and Hispanic) are being compared against one control group, select a control group that is 5 times greater than the larger prohibited basis group sample, up to the maximum.
  5. Where the institution's discrimination risk proFILe identifies significant discrepancies in withdrawal/incomplete activity between control and prohibited basis groups, or where the number of marginal prohibited basis group FILes available for sampling is small, an examiner may consider supplementing samples by applying the following rules:
  6. If prohibited basis group withdrawals/incompletes occur after the applicant has received an offer of credit that includes pricing terms, this is a reporting error under Regulation C (the lender should have reported the application as approved but not accepted) and therefore these applications should be included as prohibited basis group approvals in a terms and conditions comparative FILe analysis.
  7. If prohibited basis group incompletes occur due to lack of an applicant response with respect to an item that would give rise to a denial reason, then include them as denials for that reason when conducting an underwriting comparative FILe analysis.

Marginal Transactions

A. Marginal Denials

Denied applications with any or all the following characteristics are "marginal." Such denials are compared to marginal approved applications. Marginal applications include those that:

  • Were close to satisfying the requirement that the adverse action notice said was the reason for denial;
  • Were denied by the lender's rigid interpretation of inconsequential processing requirements;
  • Were denied quickly for a reason that normally would take a longer time for an underwriter to evaluate;
  • Involved an unfavorable subjective evaluation of facts that another person might reasonably have interpreted more favorably (for example, whether late payments actually showed a "pattern," or whether an explanation for a break in employment was "credible");
  • Resulted from the lender's failure to take reasonable steps to obtain necessary information.
  • Received unfavorable treatment as the result of a departure from customary practices or stated policies. For example, if it is the lender's stated policy to request an explanation of derogatory credit information, a failure to do so for a prohibited basis applicant would be a departure from customary practices or stated policies even if the derogatory information seems to be egregious;
  • Were similar to an approved control group applicant who received unusual consideration or service, buy were not provided such consideration or service;
  • Received unfavorable treatment (for example, were denied or given various conditions or more processing obstacles) but appeared fully to meet the lender's stated requirements for favorable treatment (for example, approval on the terms sought);
  • Received unfavorable treatment related to a policy or practice that was vague, and/or the FILe lacked documentation on the applicant's qualifications related to the reason for denial or other factor.
  • Met common secondary market or industry standards even though failing to meet the lender's more rigid standards;
  • Had a strength that a prudent lender might believe outweighed the weaknesses cited as the basis for denial;
  • Had a history of previously meeting a monthly housing obligation equivalent to or higher than the proposed debt; and/or
  • Were denied for an apparently "serious" deficiency that might easily have been overcome. For example, an applicant's total debt ratio of 50 percent might appear grossly to exceed the lenders guideline of 36 percent, but this may in fact be easily corrected if the application lists assets to pay off sufficient non-housing debts to reduce the ratio to the guideline, or if the lender were to count excluded part-time earnings described in the application.

B. Marginal Approvals

Approved applications with any or all of the following characteristics are "marginal." Such approvals are compared to marginal denied approved applications. Marginal approvals include those:

  • Whose qualifications satisfied the lender's stated standard, but very narrowly;
  • That bypassed stated processing requirements (such as verifications or deadlines);
  • For which stated creditworthiness requirements were relaxed or waived;
  • That, if the lender's own standards are not clear, fell short of common secondary market or industry lending standards;
  • That a prudent conservative lender might have denied;
  • Whose qualifications were raised to a qualifying level by assistance, proposals, counteroffers, favorable characterizations or questionable qualifications, etc.; and/or
  • That in any way received unusual service or consideration that facilitated obtaining the credit.

Potential Scoping Information

This Appendix offers a full range of documentation and other information that might conceivably be brought to bear in an examination. In that sense, it is a "menu" of resources to be considered and selected from, depending on the nature and scope of the examination being conducted. Any decision to select one or more particular items from this Appendix for inclusion in a particular examination should, of course, include consideration of any burdens to the agency and lender in assembling and providing the selected item(s).

A. Internal Agency Documents and Records.

1. Previous examination reports and related workpapers for the most recent Compliance / CRA and Safety and Soundness Examinations.

2. Demographic data for the institution's community.

Comment: The examiner should obtain the most recent agency demographic data, for information on the characteristics of the institution’s assessment/market areas.

B. Information from the institution.

Comment: Prior to beginning a compliance examination, the examiner should request the institution to provide the information outlined below. This request should be made far enough in advance of the on-site phase of the examination to facilitate compliance by the institution. In some institutions, the examiner may not be able to review certain of this information until the on-site examination.

1. Institution's Compliance Program. (For examinations that will include analysis of the lender’s compliance program.)

a. Organization charts identifying those individuals who have lending responsibilities or compliance, HMDA or CRA responsibilities, together with job descriptions for each such position.

b. Lists of any pending litigation or administrative proceedings concerning fair lending matters.

c. To the degree that your agency’s policy permits you to solicit and utilize the results of self-evaluations, copies of audit or compliance reviews of the institution's program for compliance with fair lending laws and regulations, including both internal and independent audits.

NOTE: The request should advise the lender that it is not required to disclose whether it has engaged in self-testing programs of the type protected under amendments to ECOA and the FH Act nor the results of such programs.

d. Complaint FILe.

e. Any written or printed statements describing the lender’s fair lending policies and/or procedures.

f. Training materials related to fair lending issues including records of attendance.

2. Lending Policies / Loan Volume.

a. Internal underwriting guidelines and lending policies for all consumer and commercial loan products.

Comment: If guidelines or policies differ by branch or other geographic location, request copies of each variation.

b. A description of any credit scoring system(s) in use now or during the exam period.

Comment: Inquire as to whether a vendor or in-house system is used; the date of the last verification; the factors relied on to construct any in-house system and, if applicable, any judgmental criteria used in conjunction with the scoring system.

c. Pricing policies for each loan product, and for both direct and indirect loans.

Comment: The lender should be specifically asked whether its pricing policies for any loan products include the use of "overages". The request should also ask whether the lender offers any "sub-prime" loan products for "B", "C" or "D" risk level customers or otherwise uses any form of risk-based pricing. A similar inquiry should be made regarding the use of any cost-based pricing. If any of these three forms are or have been in use since the last exam, the lender should provide pricing policy and practice details for each affected product, including the lender’s criteria for differentiating between each risk or cost level. Regarding indirect lending, the lender should be asked to provide any forms of agreement (including compensation) with brokers/dealers, together with a description of the roles that both the lender and the dealer/broker play in each stage of the lending process.

 

d. A description of each form of compensation plan for all lending personnel and managers.

e. Advertising copy for all loan products.

f. The most recent HMDA / LAR, including unreported data if available. Information should be provided on diskette if possible.

Comment: The integrity of the institution's HMDA / LAR data should be verified prior to the pre-examination analysis. Verification should take place approximately two to three months prior to the on-site phase of the examination.

g. Any existing loan registers for each non-HMDA loan product.

Comment: Loan registers for the 3 month period preceding the date of the examination, together with any available lists of declined loan applicants for the same period should be requested. Registers / lists should contain, to the extent available, the complete name and address of loan applicants and applicable loan terms, including loan amount, interest rate, fees, repayment schedule and collateral codes.

h. A description of any databases maintained for each loan product, including a description of all data fields within the database.

i. Forms used in the application and credit evaluation process for each loan product.

Comment: At a minimum, this request should include all types of credit applications, forms requesting financial information, underwriter worksheets, any form used for the collection of monitoring information, and any quality control or second review forms or worksheets.

j. Lists of service providers.

Comment: Service providers may include: realtors, real estate developers, appraisers, home improvement contractors and private mortgage insurance companies. Request the full name and address and geographic area served by each provider. Also request documentation as to any fair lending requirements imposed on, or commitments required of, any of the lender’s service providers.

k. Addresses of any Internet Site(s)

Comment: Internet "Home Pages" or similar sites that a lender may install on the Internet may provide information concerning the availability of credit, or means for obtaining it. All such information would have to comply with the nondiscrimination requirements of the fair lending laws. Moreover, future enhancements to the Internet may include the capacity to conduct partial or complete credit transactions via that medium. Accordingly, it is important for examiners to review a lender’s Internet sites to ensure that all of the information or procedures set forth therein are in compliance with any applicable provisions of the fair lending statutes and regulations.

3. Community Information.

a. Demographic information prepared or used by the institution.

b. Any fair lending complaints received and lender responses thereto.

Special Analyses

  • Disproportionate Adverse Impact
  • Pre-Application Screening
  • Marketing

A. Disproportionate Adverse Impact Violations

When all five conditions below exist, consult within your agency whether to present the situation to the lender and solicit an explanation of the lender’s business justification for the policy or criterion that appears to cause the disproportionate adverse impact. Note that condition 5 can be satisfied by either of two alternatives.

The contacts between examiners and lenders described in this section are information-gathering contacts within the context of the examination and are not intended to serve as the formal notices and opportunities for response that an agency’s enforcement process might provide. Also, the five conditions are not intended as authoritative statements of the legal elements of a disproportionate adverse impact proof of discrimination; they are paraphrases intended to give examiners practical guidance on situations that call for more scrutiny and on what additional information is relevant.

NOTE: Even if it appears likely that a policy or criterion causes a disproportionate adverse impact on a prohibited basis (condition 3), do not proceed with this analysis if the policy or criterion is obviously related to predicting creditworthiness or to some other basic aspect of prudent lending, and there appears to be no equally effective alternative for it. Examples are reliance on credit reports or use of debt-to-income ratio.

Conditions

1. A specific policy or criterion is involved.

The policy or criterion suspected of producing a disproportionate adverse impact on a prohibited basis must be clear enough that the nature of action to correct the situation can be determined.

NOTE: Gross HMDA denial or approval rate disparities are not appropriate for disproportionate adverse impact analysis because they typically cannot be attributed to a specific policy or criterion. Similarly, a lender’s policies of allowing employees to exercise discretion and to negotiate terms or conditions of credit can better be described as the absence of policies or criteria than as a situation in which a policy or criterion generates a disproportionate adverse impact. Broad discretion and vague standards raise concerns about discrimination, but examiners should focus on possible disparate treatment.

2. The policy or criterion on its stated terms is neutral for prohibited bases.

3. The disparity on a prohibited basis is significant.

The difference between the rate at which prohibited basis group members are harmed or excluded by the policy or criterion and the rate for control group members must be large enough that it is unlikely that it could have occurred by chance. If there is reason to suspect a significant disproportionate adverse impact may exist, consult the supervisory office, compliance manager, district counsel, and/or compliance management department, as appropriate.

4. There is a causal relationship between the policy or criterion and the adverse result.

The link between the policy or criterion and the harmful or exclusionary effect must not be speculative. It must be clear that changing or terminating the policy or criterion would reduce the disproportion in the adverse result.

5. Either a or b:

a. The policy or criterion has no clear rationale, or appears to exist merely for convenience or to avoid a minimal expense, or is far removed from common sense or standard industry underwriting considerations or lending practices.

The legal doctrine of disproportionate adverse impact says that the policy or criterion that causes the impact must be justified by "business necessity" if the lender is to avoid a violation. There is very little authoritative legal interpretation of that term with regard to lending, but that should not stop examiners from making the preliminary inquiries called for in these procedures. For example, the rationale is not clear for basing credit decisions on factors such as location of residence, income level (per se rather than relative to debt), and accounts with a finance company. If black applicants were denied loans significantly more frequently than white applicants because they failed a lender’s minimum income requirement, it would appear that the first four conditions plus 5a existed; therefore, the examiners should consult within their agency about obtaining the lender’s response, as described in the next section below.

b. Alternatively, even if there is a sound justification for the policy, it appears that there may be an equally effective alternative for accomplishing the same objective with a smaller disproportionate adverse impact.

The law does not require a lender to abandon a policy or criterion that is clearly the most effective method of accomplishing a business objective. However, if an alternative that is approximately equally effective is available that would cause a less-severe impact, the policy or criterion in question will be a violation.

At any stage of the analysis of possible disproportionate adverse impact, if there appears to be such an alternative, and the first four conditions exist, consult within the agency how to evaluate whether the alternative would be equally effective and would cause a less-severe impact. If the conclusion is that it would, solicit a response from the lender, as described in the next section below.

Obtaining the lender’s response

If the first four conditions plus either 5a or 5b appear to exist, consult within your agency about whether and how to inform the lender of the situation and solicit the lender’s business justification. The communication with the lender should explain:

  • The specific neutral policy or criterion that appears to cause a disproportionate adverse impact.
  • How the examiners learned about the policy.
  • How widely the examiners understand it to be implemented.
  • How strictly they understand it to be applied.
  • The prohibited basis on which the impact occurs.
  • The magnitude of the impact.
  • The nature of the injury to individuals
  • The data from which the impact was computed.

The communication should state that no violation exists if the policy or criterion is used because of business necessity and there is no alternative that would accomplish the lender’s objective with a smaller disproportionate adverse impact. It should inform the lender that cost and profitability are factors the agency will consider in evaluating the lender’s business necessity. It should ask the lender to describe any alternatives it considered before adopting the policy or criterion at issue.

Evaluating and following up on the response

The analyses of "business necessity" and "less discriminatory alternative" tend to converge because of the close relationship of the questions of what purpose the policy or criterion serves and whether it is the most effective means to accomplish that purpose.

Evaluate whether the lender’s response persuasively contradicts the existence of the significant disparity or establishes a business justification. Consult the supervisory office, compliance manager, district counsel, and/or compliance management department, as appropriate.

Discriminatory pre-application screening.

Obtain an explanation for any:

  • Withdrawals by applicants in prohibited basis groups without documentation of customer intent to withdraw;
  • Denials of applicants in prohibited basis groups without any documentation whether qualified; or
  • On a prohibited basis, selectively quoting strongly unfavorable terms (for example, high fees or down payment requirements) to prospective applicants, or quoting strongly unfavorable terms to all prospective applicants but waiving such terms for control group applicants. (Evidence of this might be found in withdrawn or incomplete FILes.)

If the lender cannot explain the situations, examiners should consider obtaining authorization to contact the customers to verify the lender's description of the transactions. Information from the customer may help determine whether a violation occurred.

In some instances, such as possible "prescreening" of applicants by lender personnel, the results of the procedures discussed so far, including interviews with customers may be inconclusive in determining whether a violation has occurred. In those cases, examiners should, if authorized by their agency, consult with management regarding the possible use of "testers" who would pose as apparently similarly situated applicants, differing only as to race or other applicable prohibited basis characteristic, to determine and compare how the lender treats them in the application process.

C. Possible discriminatory marketing

  1. Obtain full documentation of the nature and extent, together with management’s explanation, of any:
    • Prohibited basis limitations stated in advertisements;
    • Code words in advertisements that convey prohibited limitations; or
    • Advertising patterns or practices that a reasonable person would believe indicate prohibited basis customers are less desirable.
  2. Obtain full documentation as to the nature and extent, together with management’s explanation, for any situation in which the lender, despite the availability of other options in the market:
    • Advertises only in media serving non-minority areas of the market;
    • Markets through brokers or other agents that the lender knows, or could reasonably be expected to know, to serve only one racial or ethnic group in the market; or
    • Utilizes mailing or other distribution lists or other marketing techniques for pre-screened or other offerings of residential loan products* that:
    • Explicitly exclude groups of prospective borrowers on a prohibited basis; or
    • Exclude geographies (e.g., census tracts, ZIP codes, etc.) within the institution’s marketing area that have demonstrably higher percentages of minority group residents than does the remainder of the marketing area, but which have income and other credit-related characteristics similar to the geographies that were targeted for marketing.

    * NOTE: Pre-screened solicitation of potential applicants on a prohibited basis does not violate ECOA. Such solicitations are, however, covered by the FH Act. Consequently, analyses of this form of potential marketing discrimination should be limited to residential loan products subject to coverage under the FH Act.

  3. Evaluate management’s response particularly with regard to the credibility of any nondiscriminatory reasons offered as explanations for any of the foregoing practices. Refer to Evaluating Responses to Evidence of Disparate Treatment elsewhere in the Appendix for guidance.

    Streamlining the Examination

    Institutions may find it advantageous to conduct self-evaluations and, provided the examiners confirm the reliability and appropriateness of the self-evaluation (or even parts of it), they need not repeat those tasks. If the institution has performed a self-evaluation of any of the product(s) selected for examination, unless agency policy prohibits examiners from requesting the results, obtain a copy thereof and proceed through the remaining steps of this section on Streamlining the Examination. If the institution has conducted a self-evaluation of a product not selected in the scope of the examination, consider whether the product evaluated by the institution is appropriate under the scoping guidelines to substitute for another product that was selected. If such a substitution is considered appropriate, obtain the results of the self-evaluation for the substituted product and proceed through the remaining steps of this section.

    Determine whether the research and analysis of the planned examination would duplicate the institution’s own efforts. If the answers to Questions A and B below are both Yes, each successive Yes answer to Questions C through L indicates that the institution’s work up to that point can serve as a basis for eliminating steps for the examiners.

    If the answer to either Question A or B is No, the self-evaluation cannot serve as a basis for eliminating examination steps. However, you should still evaluate the self-evaluation to the degree possible in light of the remaining questions and communicate the findings to the lender so that it can improve its self-evaluation process.

    1. Did the transactions covered by the self-evaluation occur not longer ago than two years prior to the examination? If the self-evaluation covered more than two years prior to the examination, incorporate only results from transactions in the most recent two years.
    2. Did it cover the same product, prohibited basis, decision center, and stage of the lending process (for example, underwriting, setting of loan terms) as the planned examination?
    3. Did the self-evaluation include comparative FILe review? NOTE: One type of "comparative FILe review" is statistical modeling to determine whether similar control group and prohibited basis group applicants were treated similarly. If a lender offers self-evaluation results based on a statistical model, consult appropriately within your agency.
    4. Were control and prohibited basis groups defined accurately and consistently with ECOA and/or the FH Act?
    5. Were the transactions selected for the self-evaluation chosen so as to focus on marginal applicants or, in the alternative, selected randomly?
    6. F.Were the data abstracted from FILes accurate? Were those data actually relied on by the credit decision-makers at the time of the decisions?

      To answer these two questions and Question G below, for the institution’s control group sample and each of its prohibited basis group samples, request to review 10% (but not more than 50 for each group) of the transactions covered by the self-evaluation. For example, if the institution’s self-evaluation reviewed 250 white and 75 black transactions, plan to verify the data for 25 white and seven black transactions.

    7. Did the 10% sample reviewed for Question F also show that customer

      assistance and lender judgment that assisted or enabled applicants to qualify were recorded systematically and accurately and were compared for differences on any prohibited bases?

    8. Were prohibited basis group applicants’ qualifications related to the underwriting factor in question compared to corresponding qualifications of control group approvals? Specifically, for self-evaluations of approve/deny decisions, were the denied applicants’ qualifications related to the stated reason for denial compared to the corresponding qualifications for approved applicants?
    9. Did the self-evaluation sample cover at least as many transactions at the initial stage of review as examiners would initially have reviewed using the sampling guidance in these procedures?

      If the lender’s samples are significantly smaller than those in the sampling guidance but its methodology otherwise is sound, review additional transactions until the numbers of reviewed control group and prohibited basis group transactions equal the minimums for the initial stage of review in the sampling guidance.

    10. Did the self-evaluation identify instances in which prohibited basis group applicants were treated less favorably than control group applicants who were no better qualified?
    11. Were explanations solicited for such instances from the persons responsible for the decisions?
    12. Were the reasons cited by credit decision-makers to justify or explain instances of apparent disparate treatment supported by legitimate, persuasive facts or reasoning?

      If the questions above are answered Yes, incorporate the findings of the self-evaluation (whether supporting compliance or violations) into the examination findings. Indicate that those findings are based on verified data from the institution’s self-evaluation. In addition, consult appropriately within the agency regarding whether or not to conduct corroborative FILe analyses in addition to those performed by the lender.

      If not all of the questions in the section above are answered Yes, resume the examination procedures at the point where the lender’s reliable work would not be duplicated by the examiners. In other words, use the reliable portion of the self-evaluation and correspondingly reduce independent comparative FILe review by examiners. For example, if the institution conducted a comparative FILe review that compared applicants’ qualifications without taking account of the reasons they were denied, the examiners could use the qualification data abstracted by the institution (if accurate) but would have to construct independent comparisons structured around the reasons for denial.

Last Updated 07/17/1999 communications@fdic.gov