DEPARTMENT OF THE TREASURY INTERNAL REVENUE SERVICE WASHINGTON, D.C. 20224

September 6, 2001 Number: 200142022 Release Date: 10/19/2001 U.I.L. #s: 6213.07-01, 408A.00-00, 72.20-00, 4973.00-00 CC:TEGE:EB:QP1:CAVOHS TL-N-1972-01 MEMORANDUM FOR: Associate Area Counsel, Brooklyn (Small Business/Self-Employed: Area 1) Attention: Patricia A. Riegger FROM: Chief, Qualified Plans (Employee Benefits) Office of the Division Counsel/Associate Chief Counsel (Tax Exempt and Government Entities Division) SUBJECT: Ineligible Roth Conversions - Significant Service Center Advice This Significant Service Center Advice responds to your memorandum dated April 18, 2001, regarding an inquiry you received from the Brookhaven Service Center. ISSUES Issue 1: Whether the additional tax (including the 10 percent additional tax imposed under section 72(t) of the Internal Revenue Code (IRC) for early distributions) due for the taxable year in which taxpayers took a distribution from a traditional IRA to convert the funds to a Roth IRA may be assessed pursuant to the mathematical error procedures, if the reason for the adjustment was that the taxpayer’s modified adjusted gross income (AGI) was above the threshold amount permitted for making such a conversion. Issue 2: Whether penalties under IRC § 4973 ensuing as the result of a failed Roth IRA conversion may be assessed under mathematical error procedures. BRIEF ANSWER Issue 1: No, statutory deficiency procedures must be used. Issue 2: No, statutory deficiency procedures must be used.

FACTS Taxpayers are permitted to convert traditional individual retirement accounts (IRAs) to Roth IRAs. If a taxpayer converts an amount from a traditional IRA to a Roth IRA, the amount distributed or transferred from the traditional IRA is treated as a

2 1 Under Treas. Reg. § 1.408A-5 A-6(b), the election to recharacterize an IRA contribution and the trustee-to-trustee transfer of such contribution (including allocable net income) must occur on or before the due date (including extensions) for filing the individual ’ s Federal income tax return for the taxable year for which the recharacterized contribution was made. However, under the provisions of Treas. Reg. § 301.9100-2(b), an individual has an automatic 6 month extension from the return due date (excluding extensions) to recharacterize an IRA contribution. In addition, under Treas. Reg. § 301.9100-3, the Commissioner may grant an additional extension of time to recharacterize an IRA contribution. Pursuant to § 6.04 of Revenue Procedure 2001-4, 2001-1 I.R.B. 121, a request for an extension of time to recharacterize an IRA contribution under Treas. Reg. § 301.9100-3 is considered a letter ruling request. 2 Excess contributions are defined, in general, as the excess amounts contributed to IRAs over the amount allowable as a deduction under I.R.C. § 219. See I.R.C. 4973(b). distribution and is fully taxable in the year of the conversion (i.e., distribution or transfer), unless the conversion was made in 1998. If the conversion was made in 1998, the taxpayer may report the income from the distribution over four years. In order to be eligible to convert an IRA to a Roth IRA, the taxpayer ’ s modified AGI must be $100,000 or less in the year of the distribution or transfer. Married taxpayers are permitted to convert only if their modified AGI on a joint return is $100,000 or less in the year of the distribution or transfer. In 1998, many married taxpayers filing joint returns, attempted to convert their IRAs to Roth IRAs even though they were not eligible to do so because their joint income for the 1998 taxable year was more than $100,000. However, these ineligible taxpayers were allowed to make corrections to their failed conversions until December 31, 1999. APPLICABLE LAW The regulations implementing IRC § 408 provide that if there is a failed conversion, a taxpayer may recharacterize amounts which he attempted to contribute to the Roth IRA as contributions to the traditional IRA. 1 Treas. Reg. § 1.408A-4, Q & A 3. If the contribution is not recharacterized, it will be treated as a Roth contribution and thus an excess contribution will be subject to the excise tax under IRC § 4973 to the extent it exceeds the individual ’ s regular contribution limit. Id.

IRC § 4973(a) provides that, in the case of excess contributions to an individual ’ s retirement account, there is imposed for each taxable year a tax in an amount equal to 6% of the amount of the excess contributions 2 to such individual ’ s accounts or annuities.

3 3 Subsection (M), the 13 th applicable situation, was added on June 7, 2001, when President Bush signed the Economic Growth and Tax Relief Reconciliation Act of 2001 ( “ the Act ” ), Pub. L. No. 107-16, 115 Stat. 38. IRC § 72(t) provides, in general, that if any taxpayer receives a distribution from a qualified retirement plan prior to attaining the age of 59 ½ , there shall be added to the tax an amount equal to 10% of such distribution. The penalty under this section is only applicable in certain specific circumstances, for example, if the taxpayer is under age 59 ½ . For recharacterizations that were untimely or were not made, additional tax is due. IRC § 408(d)(1). The question is whether the deficiency procedures or the mathematical error procedures should be used. In general, IRC § 6212(a) prohibits the assessment and collection of a deficiency until a notice of deficiency has been mailed to the taxpayer and until certain statutory steps have been taken. IRC § 6213 imposes certain restrictions on the assessment of deficiencies. Section 6213(g)(2) defines “ mathematical or clerical error ” and enumerates 13 exceptions to the general rule where deficiency procedures need not be used before assessment of the tax. 3 Two of the exceptions may be applicable in answering the above questions. Under IRC § 6213(g)(2)(C), a mathematical error includes “ an entry on a return of an item which is inconsistent with another entry of the same or another item on such return. ” According to its legislative history, subsection (g)(2)(C) is intended to encompass those cases where it is apparent which of the inconsistent entries is correct and which is incorrect. Summary assessment procedures cannot be used where it is not clear which of the inconsistent entries is the correct one. They are also not to be used where the Service is merely resolving an uncertainty against the taxpayer. See H.R. Rep. No. 658, 94 th Cong., 1 st Sess. 291-292, 1976-3 C.B. (Vol.2) 695, 983. Under IRC § 6213(g)(2)(E), a math error may also be “ an entry on a return of a deduction or credit in an amount which exceeds a statutory limit imposed by subtitle A or B, or chapter 41, 42, 43, or 44, if such limit is expressed - (i) as a specified monetary amount, or (ii) as a percentage, ratio, or fraction, and if the items entering into the application of such limit appear on such return. ” An example of a math error under IRC § 6213(g)(2)(E) would be an entry claiming the medical expense deduction under IRC § 213 without meeting the 7.5%-of-adjusted-gross-income floor. ANALYSIS

4 4 G.C.M. 39131, dated February 16, 1984, used the analysis in G.C.M. 39019 in reaching its conclusion that the math error procedures could not be used in assessing the amount of tax withheld on nonresident aliens under section 1441. G.C.M. 39738, dated June 1, 1988, used the analysis in G.C.M. 39019 in reaching the conclusion that the mathematical error procedures were applicable because the failure to attach a Form 8396, Mortgage Interest Credit, where a mortgage interest credit amount was claimed on the Form 1040 is the precise situation in which I.R.C. § 6213(g)(2)(D) was intended to apply. 5 There is no overall monetary amount statutory maximum that applies to contributions to IRAs and SEPs in general. The maximum deduction that could be claimed based on section 219(b) for contributions to individual IRAs would be $4,000. However, theoretically, there is no limit on the number of employers who may contribute In order to analyze the issues, it is helpful to look at specific examples of the exceptions. In GCM 39019 4 , dated August 3, 1983, the conclusion was reached that “ the exception to the normal assessment procedures for mathematical or clerical errors in IRC § 6213(b) may not be used to correct deductions for contributions to individual retirement accounts . . . claimed on the Form 1040 and to summarily assess additional tax . . . because the types of errors involved are not within the scope of the definition of mathematical or clerical errors in IRC § 6213(g)(2). ” Some of the examples of errors not eligible for the mathematical error procedures were: (1) a deduction was claimed on line 25, Form 1040 for contributions to IRAs but no code was entered to identify the type and number of IRAs; (2) a code was entered but the deduction claimed exceeded the monetary amount of the statutory limit for the type and number of IRAs indicated by the code; (3) an excess contribution was made to the IRA based on the deduction claimed, but no excise tax was reported on line 57-Form 1040 and Form 5329 was not attached to the return. In example (1), the GCM relied on the legislative history and concluded there was no math error because the error did not meet the definition of IRC § 6213 (g)(2)(D). Subsection (g)(2)(D) defines a mathematical or clerical error as “ an omission of information which is required to be supplied on the return to substantiate an entry on the return. ” The GCM reasoned that mathematical error procedures could not be used in example 1 because “ omission of information ” was construed to mean the omission of an entire schedule; the omission of a code is not the omission of an entire schedule. In example (2), the GCM reasoned that there was no math error because the code that was entered did not cover all the possible variations on the type and manner of IRA an individual might have. The GCM concluded that there was no math error under IRC 6213 § (g)(2)(E) because the monetary amount of the statutory limit varies according to which code is entered. 5 Although it could be argued that the

5 to SEPs on behalf of an individual and thus there is no limit on the amount of money an individual may deduct for employer contributions to his SEP. Consequently, because contributions to SEPs are included on line 25, there is no specific statutory limit on the deduction claimed on line 25 that may be determined without regard to the code entered. discrepancy could fall under an “ inconsistent entry ” as defined in (g)(2)(C), the math error procedures still could not be used because the taxpayer may have entered the wrong code rather than claimed the wrong deduction. See S. Rep. No. 94-938, 94 th Cong., 2d Sess. 377 (1976). In example (3), the GCM further concluded that math error procedures did not apply because the Service could not determine that the deduction exceeds the amount of compensation without determining what was compensation. Thus it would not be possible to determine whether an excess contribution was made. Even if it were possible to determine that an excess contribution had been made, the GCM concluded that the excise taxes imposed by IRC §§ 4972 and 4973 could not be summarily assessed as math errors based on the Form 1040. Moreover the GCM concluded that there was no substantiation omission, within the meaning of IRC § 6213(g)(2)(D) where a taxpayer does not attach Form 5329 to report excess contribution excise taxes. Excess contributions and the appropriate excise tax are reported on Form 5329, which must be attached to Form 1040. However, Form 5329 calculates a tax not otherwise reported on Form 1040 and Form 1040 does not provide for reporting the excise tax. Therefore, the GCM concluded that math error procedures may not be used when Form 5329 is not attached, even if it were possible to determine from other information on the return that an excess contribution had been made. CONCLUSION Consistent with the legislative history of IRC § 6213, the math error assessment procedures may only be employed if it is apparent from the face of the return which entry is incorrect. In the situation you describe, it is not apparent from the face of the return which return entry is incorrect, e.g., whether the taxpayer ’ s AGI or conversion amount is the incorrect entry, nor can you determine the taxpayer ’ s modified AGI. The examples above show that even in cases where it would appear that the procedure could be used because the deduction claimed was inconsistent with the other information on the return, the math error procedures still would not apply if there was some uncertainty about what the taxpayer intended. The math error exception may only be utilized if it is absolutely clear what the taxpayer intended. We do not believe that in the Roth conversion situation you describe the taxpayer ’ s intention is absolutely clear. Since Form 5329 is not attached to the return, you cannot determine whether the amount distributed from the IRA is subject to the 10

6 percent additional tax on early distributions imposed under IRC § 72(t) because the taxpayer may qualify for an exception to the additional 10 percent tax. In addition, you cannot determine from the face of the return whether the taxpayer made other Roth IRA contributions for the taxable year, so you cannot determine whether the failed Roth IRA conversion is subject to the excise tax imposed on excess Roth IRA contributions under IRC § 4973, nor is Form 5329 attached. Accordingly, since the Internal Revenue Service is not permitted to make determinations about a taxpayer ’ s intention when making assessments pursuant to the math error procedures, assessments made as a result of a failed Roth IRA conversion (i.e., the additional tax under IRC § 72(t) and the excise tax on excess contributions under IRC § 4973) must be made pursuant to deficiency procedures.