4. Furnisher and Data
Screening
The NCRAs employ a number of methods to screen furnishers and incoming information for inaccuracies
and anomalies. This section examines the vetting and approving of furnishers and various quality screens
performed on data files received from furnishers. These methods focus on identifying formatting errors,
logical errors, internal inconsistencies, and anomalies.
4.1 New Furnisher Screening
The NCRAs’ data quality processes start with their screening of new furnishers.
The NCRAs report that a prospective furnisher can initiate a relationship with them by sending a letter of
intent to furnish.
Due to the resource and economic costs associated with adding a furnisher, the NCRAs
will generally require prospective furnishers to report a minimum of 100 to 200 active accounts per month.43
Each NCRA reportedly puts prospective furnishers through an initial security screening. Screening
generally includes an inspection of features of each business such as its physical headquarters, phone
number, website, and business license, as well as company records such as annual reports. Individual
NCRAs also may hire third-party investigation services to screen for illegal or unethical business history.
Sole proprietorships and new businesses (e.g., in business less than a year) may receive more specialized
screening. An NCRA may require the furnishers to submit test files which it will examine to make sure they
are Metro 2 compatible. Approved furnishers are trained on Metro
2.
After these initial inspections, NCRA policies may trigger reinspections after risk events such as consumer
complaints, suspicious trade lines, variations in data submissions, odd anomalies, and changes in company
ownership. At least one NCRA has policies to reinspect new furnishers six months after they start
submitting data to assess for data quality and fraud risk.
The NCRAs report that they continue to monitor for data quality and fraud once a furnisher starts
contributing live trade line data. One example of furnisher fraud is when a supposed credit repair
organization represents itself as a furnisher and attempts to boost the credit scores of consumers with bad
credit by reporting fictitious trade lines that the consumers purportedly used and paid back on time.
Overall, the objective of furnisher screening is to reduce the risk of fraud or of poor data quality by
screening out furnishers whose systems are not able to report accurate data on customers or report it in the
Metro 2 format.
4.2 Checking Furnished Data
Having passed this initial screening, furnishers can start providing data. Furnishers generally provide
monthly trade line updates through data file transfers that conform to the Metro or Metro 2 format and
contain trade line updates on all of the furnishers’ active accounts. All new furnishers are being added under
the Metro 2 format, which was first introduced in 1997. Data submitted by a furnisher to an NCRA
generally goes through a multi-stage process to identify data irregularities.
Typical data quality checks will identify issues such as blank fields or logical inconsistencies in the data –
both at the level of the data file and at the individual consumer’s trade line. If a furnished account is
reported as closed, and then in a subsequent data feed the furnisher reports a new account balance, the
NCRA might flag that inconsistency.
Other inconsistencies might be account balances higher than the
maximum credit line, duplicate instances of information on the same account being furnished, or data
patterns inconsistent with the furnisher’s historical pattern of transactions. It is not uncommon for
furnishers’ bulk files to be initially rejected by the NCRAs.44 The NCRAs report that furnishers tend to
correct most of the problems causing the file rejection, leaving only a small percentage of files permanently
rejected.
Some data rejections might not result from an error in the data but from format incompatibility
when the furnisher uses the wrong codes to update accounts, or the furnished data shows unfamiliar
formats because of system changes at the furnisher.
Within file submissions, individual consumer base records and tradeline updates are similarly screened for
formatting errors, logical errors, internal inconsistencies, and anomalies. The rejection rates for incoming
trade line data from furnishers appear to vary across multiple dimensions (e.g., by individual furnisher, by
furnishing industry, by the NCRA receiving the data). For example, submissions from collections agencies
tend to have a higher rejection rate than rejections for credit card trade lines.45
While the NCRAs’ data screens do find errors by identifying anomalies and inconsistencies, these checks
rely on underlying furnisher data to be valid.
The NCRAs do not conduct independent checks or audits to
determine if the data is accurate, such as contacting a consumer to ask if she is properly associated with an
account or if the balance reported on an account is true, or checking the record-keeping practices of a
furnisher.
The NCRAs generally rely on furnishers to report information on consumers that is complete
and accurate.
4.3 The Furnisher Rule
In 2009, the Federal Trade Commission, the Federal Reserve Board, the Federal Deposit Insurance
Corporation, the National Credit Union Administration, the Office of the Comptroller of the Currency, and
the Office of Thrift Supervision issued a joint rule (“Furnisher Rule”) implementing the accuracy and
integrity and direct dispute provisions for furnishers mandated by the Fair and Accurate Credit Transactions
Act (FACTA).46 The CFPB has since restated this rule.47
As a result of FACTA and the Furnisher Rule, furnishers have enhanced obligations to supply accurate data.
Each furnisher is required to “establish and implement reasonable written policies and procedures
concerning the accuracy and integrity of the information it furnishes to consumer reporting agencies.”48
The procedures should address “deleting, updating, and correcting information in the furnisher’s records, as
appropriate, to avoid furnishing inaccurate information.”
49 The procedures must be appropriate to the
“nature, size, complexity, and scope of each furnisher’s activities.”50 Appropriate procedures include using
standard data reporting formats, maintaining records for a reasonable period of time, providing appropriate
oversight of service providers (e.g., companies that provide core processing systems or software used for
recordkeeping and account management), furnishing information in a way that prevents re-aging,g
duplicative reporting, association of information with the wrong consumer, and providing sufficient
identifying information about consumers.51
5. Compiling Credit Files:
“Matching”
Once the NCRAs have received trade line information from a furnisher they must assign it to a specific
consumer’s identity. Each of the NCRAs has over 200 million active files on individual consumers which
are non-duplicative within the particular NCRA.52 The average credit file contains 13 past and current credit
obligations, including nine bank and retail cards and four installment loans (e.g., auto loans, mortgage loans,
student loans).53 In a typical month, an NCRA receives updates on over 1.3 billion trade lines.54 With this
much information included in and added to their databases, the NCRAs face technical and operational
challenges in attributing information to the proper consumer’s file.
5.1 Identifying the Correct Consumer
To locate and identify a consumer, NCRAs will use various combinations of personal identifying
information such as name, address, phone number, date of birth, address, and SSN. A given trade line
reported by a furnisher may not contain all of this identifying information. Typically, the furnisher reports
the personally identifying information that was provided by the consumer in the consumer’s original
application for credit or through updates (such as for current address or married name) that a consumer may
provide in the course of his or her relationship with the furnisher.
The fact that many consumers have the same or similar personal identifiers presents further challenges when
a credit bureau tries to match an incoming trade line with the correct consumer’s file. In the United States,
according to 2000 census figures (the most recent to have last name statistics available), there are more than
2.3 million Americans with the last name of Smith,
1.8 million Americans with the last name of Johnson, 1
million Americans with the last name of Davis, 850 thousand Americans with the last name of Garcia, and
600 thousand Americans with the last name of Lee.55 As one example, consider the matching challenges
posed by relatives with same first and last name, but different middle names, who reside at the same address,
and who do not regularly use their middle name when applying for credit.
Adding to the complexity, millions of individuals change how they identify themselves over time or between
furnishers.
Each year, a sizeable number of Americans change their name through marriage and divorce.
Separately, consumers do not necessarily refer to themselves consistently in credit applications. For
example, a woman named Elizabeth may use her full name on one application and then refer to herself with
a nickname “Betty,” “Beth,” “Liz,” or “Eliza” on another credit transaction. Finally, creditor practices may
vary as to the personally identifying information they require in their loan or credit applications, with the
result that the criteria one creditor uses to identify a consumer in a trade line update may vary from how
another creditor identifies him or her.
5.2 Posting and Organizing Account
Information in Consumer Files
Once a trade line has passed the NCRAs’ initial vetting and screening, the NCRAs assign or post that trade
line to the credit file of a specific consumer if they believe there is a match.
As discussed below,
inaccuracies may result from this process.
The manner in which each NCRA posts incoming data to a consumer’s file, and the way its files are
organized, depends on the particular structure of its database, or its unique “data architecture.” The NCRAs
take two different approaches to organizing personal data in their data networks: (1) flat file system and (2)
“PINning” technology.
5.2.1 Flat File Systems
At least one NCRA organizes its database like a traditional flat filing system so that each consumer is linked
to one file.56 Consumers’ files are distinguished through matching logic using a consumer’s personal
identifiers such as name, address, SSN, and date of birth. Multiple or fragmented files can occur for a single
person when information is reported with different identifying information such as a different last name.
Fragmented files on the same consumer will remain distinct until the NCRA receives new information about
the fragments (e.g., a unifying name, address, phone number) that indicates they should be combined.
In
some cases, matching algorithms will assign the trade line to a file that, according to the algorithm,
represents the best match even when all of the identifiers do not match up perfectly, or when only a limited
number of identifiers are contained with the trade line.
5.2.2 PINning Technology
Another method uses a unique personal identification number (PIN) to organize consumer files.57 Instead
of having a single file for each consumer, it uses the consumer’s assigned PIN to link information on the
consumer from multiple databases including inquiry, trade line, employment, public record, and address
databases. Each furnished trade line data element, inquiry, or public record is entered into the network with
an associated PIN in a relational database.
PINs are assigned to trade lines based on algorithms that find
the consumer that best matches the personal (header) information accompanying the trade line. When a
consumer report is requested by a creditor or a consumer requests a credit report, the NCRA assembles the
consumer report in real-time using the PIN as the central link to the different databases.
In this system, matching algorithms are used to assign a new incoming trade line or public record to the PIN
that represents the best possible fit based on the personally identifying information associated with the trade
line.
The CFPB has no data on the relative accuracy of flat-file vs. PIN-based architectures.