Committee on Ways and Means, Full Committee, 5-8-97 Testimony

Statement of
Lynda D. Willis, Director
Tax Policy and Administration Issues
U.S. General Accounting Office

We are pleased to be here today to participate in the Committee's inquiry into noncompliance surrounding the Earned Income Credit (EIC)--a refundable tax credit available to low-income, working taxpayers. As used in connection with the EIC, "noncompliance" occurs when persons either claim credits to which they are not entitled or claim credits in excess of the amount to which they are entitled. This statement is based on our past work on the EIC,⁽¹⁾ a review of limited data on the results of IRS' study of EIC filers for tax year 1994,⁽²⁾ and a review of various Department of the Treasury proposals to reduce EIC errors.

-- EIC noncompliance has been a concern for a number of years and is a major factor underlying our designation of filing fraud as one of the federal program areas at high risk because of vulnerability to waste, fraud, abuse, and mismanagement.⁽³⁾ Through design changes and administrative actions, noncompliance (expressed as a percentage of total EIC dollars paid out) has been reduced since 1988 but, because of increases in the number of claimants and changes in credit amounts over the past few years, the amount of dollars erroneously paid out has increased dramatically. A root cause of EIC noncompliance is the self-determination of eligibility by taxpayers combined with IRS' limited ability to verify eligibility before the refund is issued.

-- IRS has undertaken, with some success, a variety of efforts to reduce EIC noncompliance in recent years. While the impact of IRS' efforts cannot be precisely quantified, it is reasonable to expect that recent declines in the noncompliance rate were in part the result of IRS' efforts. How much further it can be reduced with available resources is uncertain.

-- It will not be easy to significantly reduce EIC noncompliance because of the nature of the credit and the design of IRS' systems. Treasury has announced eight proposals, six of which would involve legislation, to reduce EIC noncompliance. Those proposals provide a starting point for deliberations on what can reasonably be done to address this difficult problem. Various questions need to be answered in assessing those proposals, the most significant being whether they get at the real causes of noncompliance.

Congress established the EIC in 1975 to (1) offset the impact of Social Security taxes on low-income families and (2) encourage low-income families to seek employment rather than welfare.

EIC eligibility depends on taxpayers' amount of earned income⁽⁴⁾ or, in some cases, adjusted gross income (AGI).⁽⁵⁾ Credit amounts depend on the number of qualifying children who meet age, relationship, and residency tests. The credit gradually increases with increasing income (the phase-in range), plateaus at a maximum amount (the plateau range), and then gradually decreases until it reaches zero (the phase-out range). Taxpayers with earned income or AGI exceeding the maximum qualifying income level are not eligible for the credit. Taxpayers with AGI falling in the credit's phase-out range receive the lesser amount resulting from using their earned income or AGI in calculating the credit.

EIC coverage and benefit rules have been modified several times since 1990. In the Omnibus Budget Reconciliation Act (OBRA) of 1990, Congress made two major changes to the EIC that took effect in tax year 1991. These changes (1) adjusted the credit structure to grant different credit amounts to taxpayers with one qualifying child and taxpayers with two or more qualifying children and (2) added two supplemental credits--one for taxpayers with a child under 1 year of age and another for taxpayers who paid health insurance premiums on policies covering their children. OBRA 1990 also allowed taxpayers with a filing status of single to claim the credit, as long as they had a qualifying child, and specified a general increase in credit rates that was to be phased-in over 4 years (the planned increase for 1994, however, was superseded by 1993 legislation).

OBRA 1993 made two changes in the credit's structure that went into effect in tax year 1994. First, to simplify EIC filing, the act repealed the supplemental young child and health insurance credits. Second, the act expanded EIC eligibility to include certain taxpayers without qualifying children or "childless adults."⁽⁶⁾ OBRA 1993 also increased, over a 3-year period beginning in tax year 1994, the maximum credit for families with children.

The maximum basic credit amount for EIC families with two or more children was $953 in tax year 1990, $1,511 in tax year 1993 (reflecting OBRA 1990), and $3,556 in tax year 1996 (reflecting OBRA 1993). The maximum credit for childless adults in tax year 1996 was $323.

The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 made three additional changes to the EIC. First, beginning with tax year 1996 returns, the act made taxpayers with certain investment income greater than $2,200 ineligible for the EIC, regardless of their earned income or AGI. Second, the act created a "modified AGI" to be used in calculating the credit. Modified AGI disregards certain losses from investments and businesses. And third, the act denied the EIC to filers without valid Social Security Numbers (SSN).⁽⁷⁾ Taxpayers were already required to provide valid SSNs for qualifying children.

As shown in figure 1, both the number of taxpayers claiming the credit and EIC program costs (in 1996 dollars) increased steadily from tax years 1990 through 1994. In large part, this growth reflects the impact of either eligibility or benefit expansions implemented in tax years 1991 and 1994, as discussed earlier.

Despite efforts over time to change design and administration of the EIC, it is still a major source of noncompliance. That continuing noncompliance is one reason why refund fraud remains on our list of high-risk federal program areas.

IRS' study of tax year 1994 EIC filers showed that of $17.2 billion in EIC claimed, 25.8 percent ($4.4 billion) was overclaimed. While that percentage of noncompliance is an improvement over the level identified in the 1988 Taxpayer Compliance Measurement Program (TCMP), the dollars involved have increased significantly.⁽⁸⁾ The 1988 TCMP showed that about $1.9 billion, or 34 percent of the total EIC paid out, was awarded erroneously.

The lower rate of noncompliance since 1988 may be due, at least in part, to legislative changes since the 1988 TCMP. In that regard, IRS data indicated that taxpayers who claimed the wrong filing status were the most frequent source of EIC error in the 1988 TCMP. Legislative action in 1990 simplified the rules for qualifying for the credit by eliminating different eligibility rules for different filing statuses. Even with that change, eligibility-related compliance issues remain. For claimants of the credit for families with children, eligibility for the credit is still self-determined by the taxpayer using a three-part test based on the relationship of the child to the taxpayer, the length of time the child lived with the taxpayer, and the child's age.

Congress has long been concerned about the high level of EIC noncompliance. However, reducing it to more acceptable levels will be difficult. Many of the noncompliance problems identified by IRS are the result of a process whereby taxpayers self-determine their eligibility for the credit and/or the amount of credit they are due. These erroneous claims frequently related to qualifying children or choosing the wrong filing status.⁽⁹⁾ In both instances, IRS faces difficulty in verifying the information on the return without using field resources to determine taxpayer eligibility in a fashion similar to that used by organizations administering welfare programs.

The reported level of EIC noncompliance is much higher than the reported level of noncompliance in some other federal outlay programs but those other programs also have much higher administrative costs. For example, according to the Committee on Ways and Means 1996 Green Book, about 6.1 percent of the dollars paid out under the Aid to Families With Dependent Children (AFDC) program was overpaid in fiscal year 1993 and 7.3 percent of the dollars paid out in the Food Stamp program was overpaid in fiscal year 1995. As we noted in June 1995 testimony before the Senate Finance Committee, those programs not only have lower noncompliance rates than the EIC program but also have administrative costs that likely are many times higher than those of the EIC program.⁽¹⁰⁾ Data available at that time showed AFDC and Food Stamp administrative costs of about 12 percent of total program expenditures in 1993. In comparison, we estimated EIC administrative costs to be about 1 percent of EIC program costs.⁽¹¹⁾

Before deciding on how to reduce EIC noncompliance, it is important to know the major causes of that noncompliance. Data IRS made available on the results of its study shed little light on that question. According to an analysis of IRS' study by Treasury's Office of Tax Analysis, however, the three most common causes were (1) taxpayers claiming qualifying children who did not reside with them for over half the year, (2) taxpayers claiming the wrong filing status, and (3) complicated living arrangements involving more than one custodial caregiver.

It seems clear that a major share of the problem can be traced back to the nature of the credit, as explained by an IRS consultant in 1993. According to the consultant, the EIC, and other tax credits, have historically caused problems for IRS because IRS' systems were designed and, for the most part, are operated with the overriding objective of enabling anyone who wants to pay their taxes to do it. The distribution of EIC is philosophically different from the issuance of traditional refunds, where the government returns the taxpayer's own money after excess withholding. The consultant noted that payment of the EIC has much more in common with government distribution of welfare benefits through other agencies but that the standards of proof required to prove eligibility for EIC are not comparable to the standards of proof required for receipt of welfare benefits. As he pointed out, establishing eligibility for benefits delivered through other agencies normally requires the claimant to deal with government employees face to face; to produce proof of identification; and to prove the existence of, and relationship with, any relevant dependents. These types of controls are foreign to traditional IRS modes of operation.

A Treasury Task Force on Tax Refund Fraud made similar observations in 1994. According to the Task Force, (1) every refundable credit provides some incentive for the filing of problematic returns, and the incentive rises as the amount of the refundable credit rises and (2) the incentive to file problematic returns is likely to increase as IRS' capability to verify information on the return decreases. With the EIC, there are various important pieces of information, such as filing status and the existence of qualifying children, that IRS cannot easily verify. The relationship between verifiable information and compliance is not unique to the EIC. Throughout the tax system, noncompliance tends to be higher whenever there is an absence of easily verifiable data.

IRS took several steps in the past few years to combat EIC noncompliance, with some success. Although the impact of IRS' efforts cannot be precisely quantified, it is reasonable to assume that those efforts contributed to the recent decline in the rate of EIC noncompliance.

Improved verification of SSNs was a key objective of IRS' recent efforts. For electronic returns, IRS increased the number of automated filters that are designed to identify and reject submissions that involve missing, invalid, or duplicate SSNs. Through those filters, IRS identified 4.1 million SSN problems on tax year 1994 returns, 1.3 million of which involved the EIC. This year, as of April 24, the filters identified about 3.2 million SSN problems, of which 1.1 million involved the EIC.

IRS also emphasized SSN verification on paper returns. For tax year 1994, it identified 3.3 million paper returns with missing or invalid SSNs (how many involved EIC returns is unknown) and followed up on 1 million of those cases (it did not have sufficient resources to follow up on all 3.3 million). IRS continued that effort, but at a reduced level, for tax year 1995 paper returns. According to IRS, these verification efforts resulted in recommended changes to taxpayers' refunds or tax liabilities of about $900 million in fiscal year 1996. Starting with tax year 1996 returns filed this year, IRS was authorized to treat missing or invalid SSNs on filed returns as math errors. As such, IRS can automatically reduce or deny the taxpayer's EIC claim, if there is any.

Also for tax year 1994, IRS (1) improved the Questionable Refund Program, (2) strengthened the process for checking the suitability of persons applying to participate in the electronic filing program as return preparers or transmitters, and (3) eliminated the direct deposit indicator.⁽¹²⁾

Despite the various changes discussed above, IRS' study of tax year 1994 EIC filers indicates that much more needs to be done. For example, even with the many electronic filing filters, which are intended to keep erroneous returns from being submitted electronically, the percent of tax year 1994 electronic returns with EIC overclaims, according to IRS' data, was almost as high as the percent of paper returns (25.3 compared with 26.1). Also, IRS determined that even if its study results were adjusted to reflect the impact of its enforcement efforts in 1995 and the new math-error procedure being used this year, the overall noncompliance rate would still be about 21 percent.

It is also interesting to note, as shown in table 1, that the number of fraudulent returns detected by the Questionable Refund Program has declined since 1994, with a dramatic decrease in 1996.

Calendar year	Number of fraudulent returns detected	Amount of fraudulent dollars detected (in millions)	Percent of fraudulent returns that involved the EIC
1993	77,840	$136.8	98
1994	77,781	160.5	91
1995	62,309	131.7	73
1996	24,919	82.5	72

According to program officials, a major reason for the decline in fraudulent returns and dollars detected in 1996 was a staffing reduction from 553 full-time equivalent staff in 1995 to 379 full-time equivalent staff in 1996.

It will not be easy to significantly reduce EIC noncompliance without somehow addressing the basic underlying problem--the self-determination of eligibility by taxpayers and IRS' limited ability to verify that eligibility before issuing the refunds. On April 23, 1997, Treasury announced eight proposals, six of which would require legislation, that it believes will help reduce noncompliance.

The six proposals requiring legislation would (1) deny future EIC claims from persons who are found to have claimed the EIC fraudulently or through reckless or intentional disregard of the rules and regulations; (2) require taxpayers who have had an EIC claim denied after an audit to prove their eligibility to IRS before being allowed future credits; (3) allow IRS to place liens and execute levies on a portion of unemployment compensation, welfare benefits, and other types of assistance in order to recapture EIC claims that were found to be erroneous after IRS had paid them; (4) penalize preparers who did not meet certain due diligence requirements; (5) clarify the definition of a foster child; and (6) conduct state tests of new ways to provide the EIC and to verify eligibility. The other two proposals would (1) increase IRS' enforcement efforts and (2) expand access to volunteer return preparation services.

There are not enough details in IRS' study and Treasury's announcement to identify the major causes of EIC noncompliance or to assess the costs, benefits, and administrability of Treasury's proposals. However, based on available information as well as our past work on the EIC specifically and tax administration in general, we have identified several issues that Congress needs to consider in deliberating on those proposals.

The most important issue is whether the various proposals get at the real causes of noncompliance. According to IRS' data, for example, a disproportionate segment of the noncompliant returns involved incorrect filing status. However, IRS' report provides no information that helps explain what it is about those claims that made them noncompliant, and none of Treasury's proposals directly addresses that issue.

On the other hand, three Treasury proposals address issues that do not seem, on their face, to be major causes of noncompliance. Those proposals call for (1) clarifying the definition of a foster child; (2) testing alternate ways to provide advance EIC payments and, at the same time, to verify the eligibility of persons receiving the advance payments; and (3) increasing the availability of volunteer return preparation assistance. Although each of those actions probably has some merit, IRS' report and Treasury's proposals provide no evidence that the benefits, in terms of reducing EIC noncompliance, would be of any consequence.

-- IRS' report provides no information on the extent to which noncompliance can be traced back to confusion over the definition of a foster child or even how many EIC claims involve foster children,

-- only about 1 percent of all EIC recipients have historically used the advance payment option, and

-- IRS has provided no data comparing the noncompliance rate for returns done without the help of a preparer and those done by preparers, much less volunteer preparers.

A second issue relates to administrability. Much more information is needed on the proposals and how they will be implemented in order to determine if they can be easily administered. The proposals that most need attention in this regard, because they would apparently involve major operating changes, are the ones that would (1) automatically deny the EIC for several years to anyone who is found to have claimed the credit fraudulently or due to reckless or intentional disregard of the rules and regulations, (2) require taxpayers who had an EIC claim denied after an audit to prove their eligibility for future credits, and (3) penalize preparers who did not meet certain due diligence requirements.

To assess the administrability of the preparer penalty proposal, for example, we would want to know what preparers would have to do to demonstrate due diligence and what IRS would have to do enforce those requirements. Our interest in this proposal is heightened by the results of our past work on preparer penalties. In a 1991 report, we said that IRS' examiners and supervisors were reluctant to pursue return preparer penalties because of the low dollar amounts of the penalties.⁽¹³⁾ We wonder whether the new penalties proposed by Treasury would be viewed similarly.

In assessing administrability, it is also important that Congress consider whether the action being proposed is the most appropriate given the facts. In explaining its proposal that certain EIC claims be automatically denied, for example, Treasury says that "existing civil penalties do not appear to be an effective deterrent against ineligible taxpayers repeatedly claiming the [EIC]." We saw nothing from IRS' study related to the effectiveness of penalties. Nor have we seen any information on the extent to which IRS, in the past, has asserted available civil and criminal sanctions for fraudulent or reckless EIC claims and the extent to which penalized parties have submitted fraudulent claims afterwards. Such information would help determine if the answer lies in better administering existing procedures or establishing new procedures.

A third issue surrounding Treasury's proposals involves the need to ensure that any legislative or administrative change include adequate controls to protect taxpayer rights. For example, Treasury is proposing that IRS be allowed to place liens and execute levies on a portion of unemployment compensation, welfare benefits, and other types of assistance in order to recapture EIC claims that were found to be erroneous after IRS had paid them. Those types of income are now exempt from levy. It is for Congress to decide whether there is now sufficient reason to revise those exemptions. But, it is important in doing so that any such change include adequate controls to protect taxpayers' rights.

A final issue involves tradeoffs. It is important that concern about EIC noncompliance not become so encompassing that other areas of noncompliance are neglected. In that regard, one of Treasury's proposals calls for aggressive IRS action to prevent the payment of erroneous EIC claims. While we support the first part of that proposal--the development of new profiles to help IRS better target its enforcement efforts--we cannot support the second part--earmarking substantial resources for an intensified EIC compliance effort--without knowing the tradeoffs. IRS, like most of the federal government, is facing stagnant or declining resources through 2002. Consequently, knowing where the resources will come from is critical. Equity and financial concerns have already been raised about IRS reducing its audit coverage of high income individuals in order to target EIC returns.

To assess tradeoffs, it is important to know what the return on investment is for EIC-related enforcement efforts compared with other programs. In fiscal year 1995, for example, all examinations done by tax auditors (those IRS staff who would normally do EIC-related audits) resulted in average additional recommended taxes of about $3,500 per return, compared to the average EIC refund that year of less than $1,500.

Finally, as IRS becomes more and more involved in determining whether taxpayers are eligible to receive the EIC, the benefits of delivering income assistance to low income taxpayers through the tax code may erode. Specifically, administrative costs will increase and participation rates may fall. In addition, IRS employees will be faced with new job responsibilities traditionally more related to welfare programs than tax administration.

In conclusion, Mr. Chairman, many questions remain about EIC noncompliance and the most effective ways to reduce it. IRS efforts have reduced the level of noncompliance but it still remains above 20 percent, with several billion dollars in overclaimed credits annually. How much further it can be reduced with available resources is uncertain.

As I noted earlier, it will not be easy to significantly reduce EIC noncompliance without somehow addressing the basic underlying problem--the self-determination of eligibility by taxpayers and IRS' limited ability to verify that eligibility.

Treasury's proposals provide a starting point for deliberations on what can reasonably be done to address this difficult problem. Much more information is needed on the proposals and IRS' study findings before any overall judgment can be made on how much the proposals will contribute to reducing noncompliance.

It is important, in further deliberations on Treasury's proposals or those made by others, that Congress and the administration consider whether (1) the actions being proposed are feasible; (2) potential benefits justify expected costs, especially in light of tighter budgets and other compliance problems facing IRS; and (3) adequate procedures are built in to protect taxpayers' rights.

Earned Income Credit: IRS' 1995 Controls Stopped Some Noncompliance, But Not Without Problems (GAO/GGD-96-172, Sept. 18, 1996)

Earned Income Credit: Profile of Tax Year 1994 Credit Recipients (GAO/GGD-96-122BR, June 13, 1996).

Earned Income Credit: Noncompliance And Potential Eligibility Revisions (GAO/T-GGD-95-179, June 8, 1995)

Earned Income Credit: Targeting to the Working Poor (GAO/T-GGD-95-136, Apr. 4, 1995).

Earned Income Credit: Targeting to the Working Poor (GAO/GGD-95-122BR, Mar. 31, 1995)

Tax Administration: Earned Income Credit--Data on Noncompliance and Illegal Alien Recipients (GAO/GGD-95-27, Oct. 25, 1994).

Tax Policy: Earned Income Tax Credit: Design and Administration Could Be Improved (GAO/GGD-93-145, Sept. 24, 1993).

Earned Income Tax Credit: Advance Payment Option Is Not Widely Known or Understood by the Public (GAO/GGD-92-26, Feb. 19, 1992).

The New Earned Income Credit Form Is Complex and May Not Be Needed (GAO/T-GGD-91-68, Sept. 17, 1991).

2. We did not assess IRS' study methodology or the reliability of its reported results.

4. Earned income for calculating the EIC includes both taxable and nontaxable earned income. For the EIC, taxable earned income includes (1) wages, salaries, and tips; (2) union strike benefits; (3) long-term disability benefits received prior to minimum retirement age; and (4) net earnings from self-employment. Nontaxable earned income includes (1) voluntary salary deferral such as 401(k) plans or the federal thrift savings plan, (2) pay earned in a combat zone, (3) basic quarter and subsistence allowances from the U.S. military, (4) housing allowances or rental value of a parsonage for the clergy, and (5) excludable dependent care benefits.

5. In addition to taxpayers' taxable earned income, AGI includes their taxable income from other sources such as investments, alimony, and unemployment compensation. Beginning in tax year 1996, taxpayers are to use a newly defined "modified AGI" that excludes certain losses to determine EIC eligibility and to calculate the credit.

6. Although referred to as "childless adults," these taxpayers may be noncustodial parents or may live with a child who, for some reason, cannot be claimed as an EIC qualifying child.

8. Before IRS' study of tax year 1994 EIC filers, the 1988 TCMP provided the most current comprehensive data on EIC noncompliance. IRS did a study of tax year 1993 EIC filers, but that study only covered returns filed electronically during the last 2 weeks of January 1994.

9. A change in filing status, per se, will not necessarily disqualify a taxpayer from claiming the EIC. Only taxpayers who use the married-filing-separately status are ineligible for the credit. However, reporting an incorrect filing status has implications for correctly reporting income. For example, taxpayers who file as a head of household when they should have filed as married may underreport income, by excluding their spouse's income, and thus overclaim, in whole or in part, the EIC.

10. Earned Income Credit: Noncompliance and Potential Eligibility Revisions (GAO/T-GGD-95-179, June 8, 1995).

11. This estimate is for return and refund processing costs only and does not include the cost of IRS enforcement efforts related to EIC noncompliance.

12. The direct deposit indicator gave return preparers a quick signal from IRS that a taxpayer was going to receive a refund check and was relied on by providers of Refund Anticipation Loans. IRS' objective in eliminating the indicator was to give providers of Refund Anticipation Loans greater incentive to check the eligibility of EIC claimants before approving the loans.

13. Tax Administration: Effectiveness of IRS' Return Preparer Penalty Program Is Questionable (GAO/GGD-91-12, Jan. 7, 1991).