As set out in section 1002.112(b)
and in comment 112(b)–1, a financial institution is presumed
to maintain procedures reasonably adapted to avoid errors with respect
to a given data field if the number of errors found in a random sample
of a financial institution’s data submission for a given data
field do not equal or exceed the threshold in column C of the following
table:
Table 1. Tolerance Thresholds for Bona Fide Errors
| Small business lending
application register count |
Random sample size1 |
Threshold (#) |
Threshold (%) |
| (A) |
(B) |
(C) |
(D) |
| 100–130 |
47 |
3 |
6.4 |
| 131–190 |
56 |
3 |
5.4 |
| 191–500 |
59 |
3 |
5.1 |
| 501–100,000 |
79 |
4 |
5.1 |
| 100,001+ |
159 |
4 |
2.5 |
1 For a financial institution with
fewer than 30 entries in its small business lending application register,
the full sample size is the financial institution’s total number
of entries. The threshold number for such financial institutions remains
three. Accordingly, the threshold percentage will be higher for financial
institutions with fewer than 30 entries in their registers.
The size of the random sample, under column B, shall
depend on the size of the financial institution’s small business
lending application register, as shown in column A of the Threshold
Table.
The thresholds in column C of the Threshold
Table reflect the number of unintentional errors a financial institution
may make within a particular data field (e.g., the credit product
data field within the credit type data point or the ethnicity data
field for a particular principal owner within the ethnicity, race,
and sex of principal owners data point) in a small business lending
application register that would be deemed bona fide errors for purposes
of section 1002.112(b).
For instance, a financial
institution that submitted a small business lending application register
containing 105 applications would be subject to a threshold of three
errors per data field. If the financial institution had made two errors
in reporting loan amount and two errors reporting gross annual income,
all of these errors would be covered by the bona fide error provision
of section 1002.112(b) and would not constitute a violation of the
Act or this part. If the same financial institution had made four
errors in reporting loan amount and two errors reporting gross annual
income, the bona fide error provision of section 1002.112(b) would
not apply to the four loan amount errors but would still apply to
the two gross annual income errors.
Even when the
number of errors in a particular data field do not equal or exceed
the threshold in column C, if either there is a reasonable basis to
believe that errors in that field were intentional or there is evidence
that the financial institution did not maintain procedures reasonably
adapted to avoid such errors, then the errors are not bona fide errors
under section 1002.112(b).
For purposes of determining
bona fide errors under section 1002.112(b), the term “data field”
generally refers to individual fields. Some data fields may allow
for more than one response. For example, with respect to information
on the ethnicity or race of an applicant’s principal owners,
a data field may identify more than one race or more than one ethnicity
for a given person. If one or more of the ethnicities or races identified
in a data field are erroneous, they count as one (and only one) error
for that data field.
