CA 17-2: Revised Interagency Home Mortgage Disclosure Act Sampling, Verification, and Resubmission Procedures — FRB — US

FFIEC HMDA Examiner Transaction Testing

Guidelines1

The Federal Financial Institutions Examination Council (FFIEC) member agencies (Agencies) promote compliance with federal consumer protection laws and regulations through supervisory and outreach programs.2 Among these laws and regulations are the Home Mortgage Disclosure Act (HMDA), 12 U.S.C. 2801 et seq., implemented by Regulation C, 12 C.F.R. Part 1003. HMDA requires certain financial institutions to collect, record, and report information about their mortgage lending activity. Pursuant to the FFIEC’s efforts to prescribe principles and standards for the federal examination of financial institutions and to make recommendations providing consistency and coordination in the supervision of financial institutions,3 the Agencies are issuing HMDA Examiner Transaction Testing Guidelines (Guidelines). To support the evaluation of financial institutions’ compliance with HMDA’s requirements, the Agencies’ examiners will use the Guidelines in assessing the accuracy of the HMDA data that financial institutions record and report. The Agencies use HMDA data to support a variety of activities. For example, some Agencies use HMDA data as part of their fair lending examination process4 and other Agencies use HMDA data in conducting Community Reinvestment Act examinations.5 Moreover, HMDA disclosures provide the public with information on the home mortgage lending activities of particular reporting entities and on activity in their communities. These disclosures are used by local, state, and federal officials to evaluate housing trends and issues and by community organizations to monitor institution lending patterns. Because HMDA data serve numerous important purposes, validating the accuracy of HMDA data is a key element of the Agencies’ supervisory activities. Used in conjunction with HMDA examination procedures, the Guidelines describe how to validate the accuracy of HMDA data collected beginning on January 1, 2018 and the circumstances in which examiners may direct institutions to correct and resubmit HMDA data. Testing Procedures
1) To conduct HMDA transaction testing, examiners select a random sample of entries
from the financial institution’s HMDA Loan Application Register (LAR) (Total Sample) and ask the financial institution to provide the loan or application files (loan files) that correspond to the HMDA LAR sample entries. The size of the Total Sample will depend on the size of the financial institution’s HMDA LAR, as shown in Column A of the HMDA Transaction Testing Sample Sizes and Thresholds Table (HMDA Table) below. 1 These guidelines apply to HMDA data collected by financial institutions in or after 2018. 2 The FFIEC member agencies are the Board of Governors of the Federal Reserve System, the Consumer Financial

Protection Bureau, the Federal Deposit Insurance Corporation, the National Credit Union Administration, the Office of

the Comptroller of the Currency, and the State Liaison Committee.
3 12 U.S.C. 3301 et seq.
4 42 U.S.C. 3605; 15 U.S.C. 1691 et seq.; 12 C.F.R. Part 1002.
5 12 U.S.C. 2901 et seq.; 12 C.F.R. Parts 25, 195, 228, and 345.

2) If a financial institution’s HMDA data are collected through multiple data collection and
reporting systems, examiners may test a single sample from the financial institution’s entire HMDA LAR, test separate samples from each system, or test samples from selected systems chosen based on risk. If examiners do not take a single sample from the entire HMDA LAR, they should document in their work papers from which system(s) they chose the sample(s) and why.
3) Once examiners receive the loan files from the financial institution, they should verify
the accuracy of the data in the entries in the HMDA LAR sample(s) against the corresponding loan files. Examiners should document in their work papers any differences between the data in the HMDA LAR and information in files, and determine whether the differences may be explained by any additional information that the financial institution may provide. Differences that are not adequately explained should be identified as errors.
4) All data fields within the sample may be reviewed, or the supervisory agency may prioritize
designated fields for review.
5) HMDA transaction testing can be divided into two stages. Both stages test for errors only
in individual data fields that are selected for review as provided above in Paragraph 4. In Stage 1, examiners review only a subset of the sample (Initial Sample). The size of the Initial Sample will depend on the size of the financial institution’s HMDA LAR, as shown in Column B of the HMDA Table. If the number of errors identified in the Initial Sample falls below the Initial Sample Threshold in Column C of the HMDA Table for each and every data field reviewed, no further sample review is required and the examiners may conclude the transaction testing. If the number of errors in any data field reviewed equals or exceeds the Initial Sample Threshold in Column C of the HMDA Table, examiners should proceed to Stage 2 and review the remainder of the Total Sample. In Stage 2, examiners must review all data fields that had one or more errors in the Initial Sample and may review any or all Initial Sample data fields reviewed and found to have no errors in Stage 1.
6) If, after reviewing the remainder of the Total Sample in Stage 2, the total number of errors
in any data field equals or exceeds the Resubmission Threshold in Column D of the HMDA
Table, examiners should direct the financial institution to correct any such data field in its
full HMDA LAR and resubmit its HMDA LAR with the corrected data field(s).
7) A financial institution may also be directed to correct one or more individual data fields and
resubmit its HMDA LAR, even if errors in that field or fields do not meet the Resubmission Threshold in Column D of the HMDA Table, if examiners have a reasonable basis to believe that errors in that field or fields will likely make analysis of the HMDA data unreliable. To illustrate, assume examiners discover that a financial institution has incorrectly coded withdrawn applications as denials to such an extent that it likely prevents reliable analysis of underwriting disparities in a fair lending examination. Examiners may direct a financial institution to correct the Action Taken data field and resubmit the HMDA LAR even if the number of Action Taken errors found in the Total Sample does not equal or exceed the Resubmission Threshold in Column D in the HMDA Table. 2

8) A financial institution may be directed to resubmit its HMDA LAR in order to include
reportable applications or loans that examiners determine were previously omitted from the HMDA LAR. Tolerances
9) For the sole purpose of determining whether the number of errors equals or exceeds the
Initial Sample Threshold in Column C or the Resubmission Threshold in Column D of the HMDA Table, examiners should not count the following differences between data in the HMDA LAR and in the loan files as errors:

Three calendar days or less in the date the application was received or the date shown
on the application form reported pursuant to 12 CFR 1003.4(a)(1)(ii);
One thousand dollars or less in the amount of the covered loan or the amount applied
for, as applicable, reported pursuant to 12 CFR 1003.4(a)(7);
Three calendar days or less in the date of the action taken by the financial institution
reported pursuant to 12 CFR 1003.4(a)(8)(ii), provided that such differences do not result in reporting data for the wrong calendar year; and
Rounding errors in reporting the dollar amount, rounded to the nearest thousand, of the
gross annual income relied on in making the credit decision or, if a credit decision was not made, the gross annual income relied on in processing the application, reported pursuant to 12 CFR 1003.4(a)(10)(iii). To illustrate, if a loan file indicates June 4th as the application date, a LAR application date of June 1st or June 7th would not be counted as an error because it is within three calendar days of June 4th, but a LAR application date of May 31st or June 8th would be counted as an error because it is more than three calendar days from June 4th. Ethnicity or Race Data Errors
10) For purposes of these guidelines, the term “data field” generally refers to individual HMDA
Filing Instructions Guide (FIG) fields, each identified by a distinct Data Field Number and Data Field Name. However, with respect to information on the ethnicity or race of an applicant or borrower, or co-applicant or co-borrower, a data field consists of a group of FIG fields as follows:
The Ethnicity of Applicant or Borrower data field group—comprised of six FIG fields
with information on an applicant’s or borrower’s ethnicity (FIG Data Field Numbers 19 24);
The Ethnicity of Co-Applicant or Co-borrower data field group—comprised of six FIG
fields with information on a co-applicant’s or co-borrower’s ethnicity (FIG Data Field Numbers 25-30);
The Race of Applicant or Borrower data field group—comprised of eight FIG fields
with information on an applicant’s or borrower’s race (FIG Data Field Numbers 33-40); and 3
The Race of Co-Applicant or Co-borrower data field group—comprised of eight FIG
fields with information on a co-applicant’s or co-borrower’s race (FIG Data Field Numbers 41-48).6 To illustrate, for an applicant who indicates “Hispanic or Latino” and “Mexican” in response to the question of ethnicity, a financial institution reports the information in two FIG fields, for example, Ethnicity of Applicant or Borrower: 1 (1: Hispanic or Latino) and Ethnicity of Applicant or Borrower: 2 (11: Mexican). If one or more of the six Ethnicity of Applicant or Borrower FIG fields have errors, they would count as one (and only one) error for that data field group. If the Ethnicity of Applicant or Borrower data field group has errors in the Total Sample that meet or exceed the Resubmission Threshold in Column D of the HMDA Table, examiners should direct the institution to correct the six Ethnicity of Applicant or Borrower FIG fields and resubmit its HMDA LAR with those FIG fields corrected. See Example 4 below. Prospective Changes

Examiners may direct the financial institution to make any appropriate changes in its
policies, procedures, audit processes, or other aspects of its compliance management system needed to prevent the reoccurrence of errors identified within the sample that are—absent such changes—capable of repetition, even if the number of errors does not equal or exceed either the Initial Sample Threshold in Column C or the Resubmission Threshold in Column D of the HMDA Table, or even if the errors fall within the tolerances provided in paragraph 9. HMDA Transaction Testing Sample Sizes and Thresholds LAR Count Total Sample Size (A) Initial Sample Size (B) Initial Sample Threshold (C) Resubmission Threshold (D)

%

25 – 50 30* 15 2 3 10.0*
51 – 100 30 20 2 3 10.0
101 – 130 47 29 2 3 6.4
131 – 190 56 29 2 3 5.4
191 – 500 59 30 2 3 5.1
501 – 100,000 79 35 2 4 5.1
100,001+ 159 61 2 4 2.5
*For institutions with fewer than 30 LAR lines, the full sample size is the institution’s total number of LAR lines. The Resubmission Threshold number remains at 3. Accordingly, the Resubmission Threshold percentage will be higher for institutions with fewer than 30 LAR lines. 6 Data fields indicating whether ethnicity or race information was collected on the basis of visual observation or surname (FIG Data Field Numbers 31, 32, 49, and 50) are not included in any data group enumerated in Paragraph 10 and are treated as individual data fields for purposes of these guidelines. 4

Examples

Financial Institution A’s HMDA LAR contains 35 entries. Examiners select a Total Sample
of 30 loans as shown in Column A of the HMDA Table.

Examiners test the Initial Sample of 15 as shown in Column B of the HMDA Table
and find two errors in the Action Taken data field, which equals the Initial Sample Threshold in Column C of the HMDA Table.
Accordingly, the examiners proceed to review the remaining 15 entries in the Total
Sample and find one additional error in the Action Taken data field for a total of three errors in that field, which equals the Resubmission Threshold in Column D of the HMDA Table. In the review of the remaining entries in the Total Sample, examiners also find two errors in the Rate Spread data field, which is below the Resubmission Threshold in Column D of the HMDA Table.
Therefore, Financial Institution A is directed to correct the Action Taken data field and
resubmit its HMDA LAR with that field corrected.

Financial Institution B’s HMDA LAR contains 125 entries. Examiners select a Total
Sample of 47 loans as shown in Column A of the HMDA Table.

Examiners test the Initial Sample of 29 loans as shown in Column B of the HMDA
Table and find one error in the Action Taken data field, which is less than the Initial
Sample Threshold in Column C of the HMDA Table; one error in the Loan Type data field, which is less than the Initial Sample Threshold; and no other errors.
Therefore, examiners end the HMDA transaction testing for Financial Institution B and
do not proceed to Stage 2 testing of the 18 remaining entries in the Total Sample because no Stage 1 errors in any single data field equaled or exceeded the Initial Sample Threshold.

Financial Institution C’s HMDA LAR contains 500,000 entries. Examiners select a Total
Sample of 159 loans as shown in Column A of the HMDA Table.

Examiners test the Initial Sample of 61 loans as shown in Column B of the HMDA
Table and find two errors in the Action Taken data field, which equals the Initial
Sample Threshold in Column C of the HMDA Table; and five errors in the Loan Amount data field, which also exceeds the Initial Sample Threshold in Column C of the HMDA Table.
Accordingly, examiners proceed to test the remaining 98 entries in the Total Sample
and find two additional errors in the Action Taken data field, for a total of four errors in that field, which equals the Resubmission Threshold in Column D of the HMDA
Table; five additional errors in the Loan Amount data field, for a total of ten errors in
that field, which exceeds the Resubmission Threshold in Column D of the HMDA
Table; and four errors in the Census Tract data field, which equals the Resubmission
Threshold in Column D of the HMDA Table.
Therefore, Financial Institution C is directed to correct the Action Taken data field, the
5

Loan Amount data field, and the Census Tract data field and resubmit its HMDA LAR with those fields corrected.
4. Financial Institution D’s HMDA LAR contains 1,000 entries. Examiners select a Total
Sample of 79 loans as shown in Column A of the HMDA Table.

Examiners test the Initial Sample of 35 loans as shown in Column B of the HMDA
Table and find one loan with an error in the FIG Applicant or Borrower Race: 1 field,
and a different loan with an error in the FIG Applicant or Borrower Race: 2 field, for a total of two errors in the Race of Applicant or Borrower data field group, which equals the Initial Sample Threshold in Column C of the HMDA Table.
Accordingly, the examiners proceed to test the remaining 44 entries in the Total Sample
and find one loan with an error in the FIG Applicant or Borrower Race: 2 field, and one loan with errors in both the FIG Applicant or Borrower Race: 1 field and the FIG Applicant or Borrower Race: 2 field, for a total of four loans with at least one error in one of the eight Race of Applicant or Borrower FIG fields, which equals the Resubmission Threshold in Column D of the HMDA Table.
Therefore, Financial Institution D is directed to correct all eight FIG fields in the Race
of Applicant or Borrower data field group and resubmit its HMDA LAR with those FIG fields corrected.
The following table summarizes how the errors in this example are counted toward the
Resubmission Threshold in Column D of the HMDA table:
FIG Applicant or
Borrower Race: 1 field
FIG Applicant or
Borrower Race: 2 field
Race of Applicant or Borrower data field group Loan #1 Error (Initial Sample) 1 Loan #2 Error (Initial Sample) 1 Loan #3 Error (Remaining Sample) 1 Loan #4 Error (Remaining Sample) Error (Remaining Sample) 1 Total errors 4 6

%

Lineage: No known changes