![]() |
|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
|
Utilization of Tax Incentives for Retirement Saving August 2003 |
![]() |
The tabulations presented in this report come from a sample of 1997 tax returns created by the Statistics of Income (SOI) Division of the Internal Revenue Service (IRS). That file was enhanced by attaching data from two different tax information returns--Form 5498 and Form W-2--to each record. Those forms do not undergo the same degree of "cleaning" by the SOI Division as the tax returns themselves do. Without that cleaning, superseded records remain in the file as do current records with amounts and codes that are not consistent with amounts reported on the same taxpayer's Form 1040. Therefore, the Congressional Budget Office (CBO) developed separate cleaning and reconciliation procedures for the two forms to delete superfluous records and impose consistency with the amounts reported on Forms 1040.
Forms 5498Forms 5498 are prepared annually by the administrators of individual retirement accounts (IRAs) and sent to the accounts' owners to report their contributions, rollovers, and end-of-year balances. Individuals receive one 5498 form for each IRA they own, even if the IRAs are administered by the same party. The original file of 5498s that CBO received from the IRS contained 179,791 forms, of which 1,933 were eliminated because they were for years other than 1997. Of the remainder, 907 were coded as amended or as "amended original" returns. CBO developed a set of rules to determine when amended returns should be used in place of original returns. In general, amended returns were retained, and original returns that had been amended were deleted. In fact, however, most amended returns were not associated with an original return, so CBO actually deleted only 123 records from the file. Because contributions reported on Form 5498 may be either deductible or nondeductible (the reporting institution has no way of knowing which), it is impossible to precisely reconcile an individual's 5498s with his or her Form 1040. If a deduction was claimed on the 1040 without an accompanying 5498, CBO's reconciliation procedure created a dummy 5498 with the deductible amount as the amount contributed. Otherwise, the procedure focused on classifying the amount as a simplified employee pension (SEP) contribution or as an IRA contribution that was deductible, nondeductible, or "unallowable." Some SEPs were easy to identify. SEPs are the only contributions reported on Form 5498 that can legitimately exceed $2,000, so all contributions in excess of that amount were instant candidates for SEP status. Not all such contributions, however, could be so counted. SEPs are reported on the Keogh line of Form 1040 instead of the IRA line; consequently, if the Keogh line on a given return was zero, none of the contributions could be counted as SEPs. Frequently, the Keogh line was not zero but was still less than the amount of the contribution. In such cases, CBO counted as a SEP the portion of the contribution up to the amount reported on the Keogh line. It counted amounts in excess of the Keogh line (or the maximum SEP contribution of $9,500) as unallowable. After reconciling the W-2 forms (discussed below), CBO's reconciliation procedure for the 5498s determined which taxpayers (or spouses) were covered by a qualified plan. For those that were so covered, it calculated the maximum allowable IRA deduction using the filing status and adjusted gross income (AGI) from the individual's Form 1040. It then compared actual IRA deductions claimed with the calculated maximum amount. Deductions within the limit were counted as "deductible/deducted." Deductions in excess of the limit were counted as "nondeductible/deducted." If the sum of all contributions exceeded the amount allowed as an IRA deduction, three tests were applied to determine if any 5498s should be deleted at this stage:
If the answer to any of those questions was yes, the 5498 was deleted. Contributions reported on Form 5498 but not deducted on Form 1040 were first tested to see whether they could be SEPs. Contributions not deducted as IRAs were counted as SEPs up to the amount reported on the Keogh line. Additional amounts up to the maximum allowable deduction were counted as "deductible/not deducted." Amounts above the maximum were counted as "nondeductible/not deducted." Those counting rules allowed the Keogh line on the 1040 form to be split between Keogh plans and SEPs. SEPs totaled $564 million, leaving $9,674 million from Keoghs. With regard to IRAs, tabulations using the above rules can be presented in two different ways: according to whether the contribution was deducted and according to whether the contribution was, in fact, deductible (see Table A-1). On the one hand, those tabulations showed that 5.3 percent of IRA deductions taken were not legitimately deductible. Of contributions that were legitimately deductible, on the other hand, 14.6 percent were not actually claimed as deductions. Overall, if taxpayers in 1997 had used all of the deductions to which they were entitled (and only those), IRA deductions would have been 10.9 percent higher than they were. An alternative source of data on nondeductible IRAs is Form 8606, which must be filed for every nondeductible IRA contribution and for every distribution from an account to which nondeductible contributions have been made. Nondeductible contributions derived from Forms 5498 contrast markedly with those reported on Forms 8606. According to the 5498s, nondeductible contributions totaled $5.9 billion, whereas the 8606s yielded a total of $3.3 billion. There are, however, more reasons to doubt the accuracy of the 8606 figures than the accuracy of the 5498 figures. Form 8606 is filed by the taxpayer, and failure to complete and file it has no immediate tax consequences. Form 5498, however, is filed by the financial institution administering the IRA. Failure to file the form would deny customers a documentable tax deduction, not to mention trigger the possibility of IRS sanctions. Clearly, the incentive of the financial institutions to provide accurate information to both IRA owners and the IRS is greater than the incentive of taxpayers to fill out one more form that does not affect their tax liability for the current year. For that reason, CBO judged the data from the 5498s to be more credible than the data from the 8606s.
Forms W-2Each year, employers send every employee a Form W-2 that reports wages subject to tax, certain forms of deferred compensation, pension coverage, and taxes paid. Individuals receive one W-2 from each employer they had during the year. The original file of W-2s that CBO received from the IRS contained 169,695 forms, of which 3,024 were eliminated because they were for years other than 1997. Of the remainder, 230 were coded as amended returns. CBO developed a set of rules, similar to those it used for the 5498 forms, to determine when to use amended returns in place of original returns. Because IRS cleaning procedures for W-2s are more thorough than for other forms, CBO deleted only 72 records as a result of those rules. The purpose of analyzing the W-2s was to determine taxpayers' eligibility for IRA deductions and to estimate contributions to 401(k), 403(b), and 457 plans and to the federal Thrift Savings Plan.(1) The form contains a box for "Deferred Compensation," the amount in which was assigned to a type of plan according to the following criteria:
To fully reconcile the W-2s with the 1040s, CBO compared the taxable amounts (including tips and excess deferrals) from the W-2s with the amount reported as salaries and wages on the 1040. If the amounts on the two types of forms were within $10 of one another, they were deemed a match. A frequent cause of mismatches was the failure to report on the 1040 tips and "excess deferrals" that appeared on a W-2. (Excess deferrals are amounts deferred in excess of the statutory maximum for a specific type of plan.) CBO used several approaches to reconcile the W-2s with the 1040s. The first method was to temporarily exclude all amounts not included in Box 1 of the W-2 (wages, tips, and so forth). If the amount in Box 1 matched the amount on the 1040 and there were no excess deferrals, the excess was attributed to tips, and the amount shown as "Allocated Tips" in Box 8 was replaced with zero. If there were excess deferrals of less than $2,000 attributable to a 457 plan, CBO converted the 457 contributions into 401(k)-type contributions, which have a higher statutory maximum, by imputing a check in the pension coverage box. If excess deferrals exceeded $2,000 or were attributable to a 401(k)-type plan, CBO reduced deferred compensation to eliminate the excess. The remaining cases fell into two categories: returns in which wages and salaries on a 1040 exceeded the corresponding amounts on all associated W-2s, and returns in which wages and salaries on a 1040 fell short of the corresponding amounts on the W-2s. In the former case, dummy W-2s that contained the missing wages were created. CBO assumed that all such returns had no deferred compensation and no pension coverage. In the latter case, specific W-2s were deleted, or--if no combination of W-2s matched the 1040--the wages on all W-2s were scaled to hit the proper total. CBO scaled deferred compensation amounts on those W-2s by the same factor that it used for wages. Finally, CBO compared IRA deductions from the 1040 with the maximum allowable deduction based on AGI and the status of the pension coverage box. If the deduction was less than $2,000 but exceeded the maximum allowable amount, and no amount was reported in the deferred compensation box, then CBO unchecked the pension coverage box. As a result of those procedures, CBO deleted 3,549 W-2s and created 20,098 from scratch. Of the resulting 183,148 records, 1 percent were reclassified from "covered by a pension plan" to "not covered by a pension plan," or vice versa; nearly 2½ percent had their compensation modified in some way; and less than ½ percent had other fields modified (see Table A-2). Once the W-2s were reconciled to the amounts on the 1040s, it was possible to tabulate deferred compensation by type (see Table A-3). That tabulation showed that of the nearly $100 billion in deferred compensation, 92 percent was in the form of 401(k)-type plans, and 7 percent was in the form of 457 plans.
|