Section 2. Safe Harbor — Internal Revenue Bulletin 2016-29

A. Populations .

The population of the first sample must consist of QI accounts, taking into account each of the strata described in section 2.C(a) of this Appendix. If QI acts as a QDD, the population of the second sample must consist of accounts for which QI is acting as a QDD, taking into account each of the strata described in section 2.C(b) of this Appendix. If QI assumes primary withholding responsibility for substitute interest payments, the population of the third sample must consist of accounts receiving such substitute interest payments, taking into account each of the strata described in section 2.C(c) of this Appendix.

(a) Sample of QI Accounts. The reviewer selects a random sample of accounts from a portion of the population of all QI accounts. The portion of the population will consist of (1) all accounts held by U.S. persons (or account holders presumed to be U.S. persons) that received a reportable payment and (2) all accounts held by non-U.S. persons (or account holders presumed to be non-U.S. persons) that received a reportable amount. If QI is acting as a QDD, accounts for which QI is acting as a QDD are not to be considered in determining the population for this sample. If QI assumes primary withholding responsibility for substitute interest payments, accounts receiving such substitute interest payments are also not are to be considered in determining the population for this sample.

(b) Sample of QDD Accounts. The reviewer selects a random sample of accounts from the population consisting of accounts for which QI is acting as a QDD.

(c) Sample of Accounts Receiving Substitute Interest Payments. The reviewer selects a random sample of accounts from the population consisting of accounts receiving substitute interest payments for which QI assumes primary withholding responsibility.

B. Sample Sizes .

(a) The sample sizes for each of the three samples are calculated independently. If PAI accounts have been added to the sample of QI accounts because the PAI did not perform its own periodic review, a separate sample size calculation should also be performed for the PAI accounts as if they were part of a separate sample. The sample size for each of the three samples, and for any additional PAI accounts, is the lesser of (i) the number of sample units determined using the sample formula in paragraph (c) of this section, or (ii) 25 percent of the total number of sample units in the population. However, in determining the sample size, the reviewer must adhere to the guidelines for minimum stratum sample sizes in sections 2.C and 2.D of this Appendix. This may result in a sample size greater than the maximum sample size resulting from using the formula in paragraphs (b) and (c) of this section. The minimum sample size of any sample shall not be less than 50.

July 18, 2016 128 Bulletin No. 2016–29

(b) Sample Size Adjustments. The variable P “error rate” should be set equal to (1) 5 percent for the sample of QI accounts, resulting in a maximum sample size of 321; (2) 4 percent for the sample of QDD accounts, resulting in a maximum sample size of 259; and (3) 1.5 percent for the sample of accounts receiving substitute interest payments, resulting in a maximum sample size of 100.

t ² PQ

d ²

Sample Size = ^{___________________________}

1 + 1 ( ^{t2 PQ _} 1) N d ²

where t� 1.645 (confidence coefficient at 95 percent one-sided) P� 5 percent (error rate) for the QI account sample, 4 percent for the QDD sample, and 1.5 percent for the substitute interest

sample. Q� 1-P d - 2 percent (precision level) N� total population

C. Strata .

(a) Sample of QI Accounts. The reviewer must segregate all of the QI accounts into the following strata. If QI makes payments to a single account that result in the account meeting the qualifications for more than one stratum, the account should be treated as multiple accounts, with each newly redefined account consisting of the payments received for each relevant stratum. Further substratification by dollar amounts may be used in accordance with section 2.H of this Appendix.

(1) A stratum of accounts held by recalcitrant account holders and non-participating FFI account holders. (2) A stratum of accounts not included in the previously defined stratum of nonwithholding foreign partnerships and nonwithholding foreign trusts to which the QI applied the joint account option or the agency option.

(3) A stratum of all accounts held by direct account holders that are not U.S. non-exempt recipients and are not included in any previously defined strata.

(4) A stratum of all accounts that are held by direct account holders that are U.S. non-exempt recipients that are not included in any of the previously defined strata.

(5) A stratum of all accounts held by indirect account holders not included in any previously defined strata. (b) Sample of QDD Accounts. The reviewer must segregate all of the accounts for which QI is acting as a QDD into the following strata: (1) A stratum of all accounts where the QDD makes or is contractually obligated to make a dividend equivalent payment to recalcitrant account holders and non-participating FFIs.

(2) A stratum of all accounts where the QDD makes or is contractually obligated to make a dividend equivalent payment to foreign recipients that are not included in the previously defined stratum.

(3) A stratum of all accounts where the QDD makes or is contractually obligated to make a payment of a qualifying dividend equivalent offsetting payment that are not included in any of the previously defined strata.

(4) A stratum of all accounts where the QDD makes reportable payments with respect to potential section 871(m) transactions that are not dividend equivalents to foreign recipients that are recalcitrant account holders and non-participating FFIs.

(5) A stratum of all accounts where the QDD makes reportable payments to foreign recipients with respect to potential section 871(m) transactions that are not dividend equivalents that are not included in any of the previously defined strata. (6) A stratum of all accounts where the QDD makes payments to U.S. persons with respect to potential section 871(m) transactions that are not qualifying dividend equivalent offsetting payments and that are not included not included in any of the previously defined strata.

(7) A stratum of all accounts not included in any of the previously defined strata. (c) Sample of Accounts Receiving Payments of Substitute Interest. The reviewer must segregate all of the accounts receiving payments of substitute interest for which QI has assumed primary withholding responsibility into the following strata:

(1) A stratum of accounts held by recalcitrant account holders and non-participating FFI account holders. (2) A stratum of all accounts held by account holders that are not U.S. non-exempt recipients and are not included in the previously defined stratum.

(3) A stratum of all accounts held by account holders that are U.S. non-exempt recipients and are not included in any of the previously defined strata.

D. Allocation of Sample Size to Each Stratum .

The reviewer must allocate the number of sample units for each sample, and for any PAI accounts added to the sample of QI accounts, independently of the other samples. For example, the reviewer must allocate the number of sample units in the sample

Bulletin No. 2016–29 129 July 18, 2016

determined under section 2.B of this Appendix to each stratum described in section 2.C(a) by multiplying the number of sample units in the sample, as determined under section 2.B, by a fraction, the numerator of which is the total number of sample units in the stratum and the denominator of which is the total number of sample units in the population. The same allocation should also be made for each stratum described in sections 2.C(b) and (c) of this Appendix, respectively. The minimum allocation to each stratum is the lesser of (1) 50 sample units or (2) the total number of sample units in the stratum. If there are fewer than 50 sample units in any stratum, all sample units in that stratum must be examined, and the difference between 50 and the number of sample units in the stratum must be reallocated to the remaining strata on a pro rata basis. If there are 50 or more sample units in the stratum, but the allocation, as determined under the fraction above, is less than 50 sample units, the number of sample units to be used in the sample from that stratum is 50.

E. Number Generator . The reviewer must select for review sample units from each stratum identified in section 2.C for each of the (a) sample of QI accounts; (b) sample of accounts for which QI is acting as a QDD; and (c) sample of accounts receiving substitute interest payments for which QI assumes primary withholding responsibility by using a random number generator. Random numbers should be drawn separately for each sample including the use of separate seeds. Information regarding the random number generator used must be included in the records required in section 3.E of this Appendix. This information must be sufficient to allow the IRS to replicate the random numbers. This information must include the name and version of the random number generator, the seed numbers used or generated, specification of any options selected, and the computer equipment on which it was run.

G. Selection of Accounts for Review . For purposes of reviewing and testing accounts in accordance with section 10.05 of the QI Agreement, the reviewer must review accounts from every stratum in all three applicable samples that meet the requirements for the review, taking into account the applicable presumption rules where documentation is missing, invalid, or cannot be relied upon (and any reclassification after applying the presumption rules).

To the extent the number of sample units listed above from the sample (or in the population, if the reviewer has not used statistical sampling) in any stratum is less than 20, the reviewer must also select for review (in the order selected by the random number generator under section 2.E of this Appendix or, if the reviewer has not used statistical sampling, in the order used by the QI for its record keeping) an additional number of sample units drawn from that stratum that equal the difference between 20 and the number of sample units from the sample in that stratum.

H. Optional Further Stratification by Dollar Amounts . For any of the three samples, the reviewer may further stratify by dollar amounts for that sample without submitting a request for approval when the reviewer is otherwise selecting the sample in accordance with this Appendix. For the QI account sample, reportable amounts for foreign recipients and reportable payments for U.S. recipients are to be considered in the substratification. If the reviewer chooses to substratify under this section, the reviewer must comply with the following rules:

(a) The strata consisting of sample units that have received payments of the highest dollar amounts during the audit year shall not

consist of more than 25 accounts. All items in these strata shall be reviewed. (b) The remaining strata shall be randomly selected to contain approximately equal amounts in each substratum. (c) The minimum strata size shall not be less than 25 sample units.

I. Determining Rate of Withholding for Partnerships and Trusts for Which the QI has Utilized Joint Account Treatment

When reviewing documentation of partners, beneficiaries, or owners to determine the rate of withholding QI should have applied to a partnership or trust, the reviewer may limit the review to the number of partners, beneficiaries, or owners by referring to the table below.

Number of partners, beneficiaries, or owners Number to be reviewed

0 - 10 all

11 - 14 10

15 - 19 13

20 - 24 16

25 - 29 18

30 - 34 20

35 - 39 21

40 - 49 22

50 - 74 24

75 - 99 26

100 - 199 27

200 - 499 29

500 - 4,999 31

4,999 32

July 18, 2016 130 Bulletin No. 2016–29