Earned Income Credit: Qualifying Child Certification Test Appears
Justified, but Evaluation Plan Is Incomplete (30-SEP-03,	 
GAO-03-794).							 
                                                                 
The Earned Income Credit (EIC), a tax credit available to the	 
working poor, has experienced high rates of noncompliance. Unlike
many benefit programs, EIC recipients generally receive payments 
without advance, formal determinations of eligibility; the	 
Internal Revenue Service (IRS) checks some taxpayers' eligibility
later. IRS estimated that tax year 1999 EIC overclaim rates, the 
most recent data available, to be between 27 and 32 percent of	 
dollars claimed or between $8.5 billion and $9.9 billion. To	 
address overclaims, IRS plans to test a new certification	 
program. Because IRS's plans have garnered much attention,	 
Congress asked us to (1) describe the design and basis for the	 
EIC qualifying child certification program, (2) describe the	 
current status of the program, including significant changes, and
(3) assess whether the program is adequately developed to prevent
unreasonable burden on EIC taxpayers and improve compliance so	 
that the test should proceed.					 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-03-794 					        
    ACCNO:   A08609						        
  TITLE:     Earned Income Credit: Qualifying Child Certification Test
Appears Justified, but Evaluation Plan Is Incomplete		 
     DATE:   09/30/2003 
  SUBJECT:   Federal taxes					 
	     Noncompliance					 
	     Strategic planning 				 
	     Tax administration 				 
	     Tax administration systems 			 
	     Tax credit 					 
	     Tax violations					 
	     Taxpayers						 
	     Eligibility determinations 			 
	     Program evaluation 				 
	     IRS Earned Income Credit Program			 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-03-794

United States General Accounting Office

GAO

                       Report to Congressional Requesters

September 2003

EARNED INCOME CREDIT

 Qualifying Child Certification Test Appears Justified, but Evaluation Plan Is
                                   Incomplete

                                       a

GAO-03-794

Highlights of GAO-03-794, a report to congressional requesters

The Earned Income Credit (EIC), a tax credit available to the working
poor, has experienced high rates of noncompliance. Unlike many benefit
programs, EIC recipients generally receive payments without advance,
formal determinations of eligibility; the Internal Revenue Service (IRS)
checks some taxpayers' eligibility later. IRS estimated that tax year 1999
EIC overclaim rates, the most recent data available, to be between 27 and
32 percent of dollars claimed or between $8.5 billion and $9.9 billion. To
address overclaims, IRS plans to test a new certification program.

Because IRS's plans have garnered much attention, you asked us to (1)
describe the design and basis for the EIC qualifying child certification
program, (2) describe the current status of the program, including
significant changes, and (3) assess whether the program is adequately
developed to prevent unreasonable burden on EIC taxpayers and improve
compliance so that the test should proceed.

September 2003

EARNED INCOME CREDIT

Qualifying Child Certification Test Appears Justified, but Evaluation Plan Is
Incomplete

The Assistant Treasury Secretary and IRS Commissioner convened a task
force to identify ways of reducing EIC overclaims while minimizing
taxpayer burden and maintaining the EIC's relatively high participation
rate. In August 2002, the Secretary approved a recommendation to certify
taxpayers' eligibility to claim EIC qualifying children. The proposal is
based on analyses of the leading sources of EIC errors, thus focusing
attention and burden on the subset of taxpayers most likely to make those
errors.

Since August 2002, IRS has made key changes to the certification program,
including concentrating on residency certification and postponing
relationship certification, delaying program implementation until later
this year, and reducing the test sample from 45,000 to 25,000. Despite the
changes, the process for selecting taxpayers, what taxpayers will receive
from IRS, what taxpayers are required to provide, and the program's goals
remain fundamentally the same as originally planned. In addition, IRS has
emphasized that program expansions, if any, will depend on the results of
this year's test. The process would involve three key stages as shown
below.

The EIC Certification Process as Envisioned

GAO recommends that the Commissioner of Internal Revenue accelerate the
development of IRS's plan to evaluate the certification test. The plan
should demonstrate how the program's objectives would be evaluated,
including milestones for conducting the evaluation. The Commissioner said
that IRS would further develop its evaluation plan as we recommended.

www.gao.gov/cgi-bin/getrpt?GAO-03-794.

To view the full report, including the scope and methodology, click on the
link above. For more information, contact Michael Brostek at (202)
512-9110 or [email protected].

These changes, including the most recent, help achieve a better balance
between preventing unreasonable taxpayer burden and addressing the EIC's
high overclaim rate and support IRS's plans to test the certification
program. However, IRS's plan for evaluating the test is incomplete,
presenting only some information on how IRS would evaluate whether
certification would reduce the EIC overclaim rate, minimize burden, and
maintain a relatively high participation rate. The plan proposes potential
options for identifying how and when certain critical data will be
obtained, but does not provide further details on when decisions will be
made or on the specific data that will be collected. Officials have
developed preliminary drafts identifying data to be obtained and have
begun considering how to use contractors to gather the data. Because the
data relate to taxpayers' actions that will occur next spring, IRS appears
to have some time to finalize its evaluation plan.

Contents

  Letter

Results in Brief

Background

Task Force Considered Much Information Related to EIC Compliance before
Recommending Qualifying Child Certification Program

IRS Has Made Key Changes to Its Initial Qualifying Child Certification
Program and More May Occur

Qualifying Child Certification Program Developed to Improve Compliance
While Considering Taxpayers' Burden, but Plan for Evaluating Test Is
Incomplete

Conclusions

Recommendations for Executive Action

Agency Comments and Our Evaluation 1 2 5

9

18

31 38 39 40

Appendixes

Appendix I:

Appendix II:

Appendix III:

Appendix IV:

                                       Appendix V: Appendix VI: Appendix VII:

Significant Noncompliance Rates Other Than for the EIC

Overclaim Rates, Administrative Costs, and Eligibility Verification
Processes of Benefit Programs

Objectives, Scope, and Methodology

Objectives
Scope and Methodology

Key Milestones for the Qualifying Child Certification Program

Documents Related to the Precertification Program

Comments from the Internal Revenue Service

GAO Contacts and Staff Acknowledgments

GAO Contacts Acknowledgments 41

44

50 50 51

54

55

65

71 71 71

Tables           Table 1: EIC Requirements for Tax Year 2002             6 
                   Table 2: EIC Overclaim Rates for 1997 and 1999          10 
                         Table 3: Recent Statutory Changes                 12 
           Table 4: Overclaim Rates, Administrative Costs, and Eligibility 
                         Verification Processes for EIC and Other Programs 45 
          Table 5: How Overclaim Rates Are Calculated for Selected Benefit 
                                      Programs                             49 

Contents

                      Table 6: Questions We Were Asked 50

Figures Figure 1: Estimated Percentage of EIC Taxpayers (with Children) 
                     Who Made Residency Errors and Filed Single or Head of 
                                  Household in Tax Year 1999               17 
           Figure 2: IRS's Process to Identify the Population of Taxpayers 
                      from Which the 25,000 Person Test Sample Was Drawn   22 
           Figure 3:      The EIC Certification Process as Envisioned      24 
           Figure 4:     Key Milestones for the Certification Program      54 

Abbreviations

EIC Earned Income Credit
IRS Internal Revenue Service
AGI Adjusted Gross Income
FCR Federal Case Registry
SB/SE Small Business/Self Employed
NRP National Research Program
TCMP Taxpayer Compliance Measurement Program

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

A

United States General Accounting Office Washington, D.C. 20548

September 30, 2003

The Honorable Amo Houghton
Chairman, Subcommittee on Oversight
Committee on Ways and Means
House of Representatives

The Honorable Earl Pomeroy
Ranking Minority Member, Subcommittee on Oversight
Committee on Ways and Means
House of Representatives

The Earned Income Credit (EIC), a tax credit available to the working
poor,
has generally been considered a successful antipoverty program by
researchers. In recent years, the Internal Revenue Service (IRS) has paid
about $30 billion annually to about 20 million EIC recipients. However,
the
EIC program has long experienced high rates of noncompliance. For tax
year 1999, IRS estimated the EIC overclaim rates1 ranged between 27 and
32 percent of dollars claimed or between $8.5 billion and $9.9 billion.
Unlike many benefit programs, EIC recipients generally receive payments
without a prior, complete review of their eligibility; IRS checks some
aspects of taxpayers' eligibility before and after the credit is granted.

To help combat the high rates of noncompliance, IRS plans to test a new
program, beginning in December 2003.2 Referred to as the qualifying child
certification program, some taxpayers will be asked to verify EIC
"residency" requirements3 for their qualifying children before getting
that
portion of their refund or reduction in tax liability. IRS plans to test
the EIC
program with 25,000 taxpayers whose residency cannot be confirmed

1Overclaim rates are calculated based on erroneous claim amounts less any
amounts IRS recovered or expects to recover, such as through examinations.
IRS also has limited information on underclaim rates, or instances in
which taxpayers claimed less than they were entitled to receive. This
report primarily focuses on IRS's efforts to address noncompliance related
to overclaims using the qualifying child certification program.

2The program discussed in this report is one part of a strategy to target
three major known sources of EIC noncompliance. The other two parts
involve the improper reporting of filing status, such as among married
taxpayers who report as single or head of household to avoid reporting
their spouse's income, and income misreporting, such as underreporting
earned income.

3Taxpayers must meet multiple criteria in order to claim the EIC, as shown
in table 1. Residence is one criterion for a taxpayer with a qualifying
child.

through other means. Future plans for the program largely depend upon the
results of this test.

Because IRS's plans surrounding the program have garnered much attention,
you asked us to respond to questions about the qualifying child
certification program. These questions cover various topics, such as the
status of the program, understandability of letters and forms going to
taxpayers, certification requirements and taxpayers' ability to comply,
taxpayer burden, impact on compliance rates, impact of recent legislative
changes, and data from other federal or state benefit programs. We grouped
these questions into three objectives: (1) describe the design and basis
for the EIC qualifying child certification program as proposed by the EIC
task force, (2) describe the current status of the program, including
significant changes since program approval, and (3) assess whether the
program is adequately developed to (a) prevent unreasonable burdens on EIC
taxpayers and (b) improve compliance so that the test should proceed. In
addition, you asked us to provide readily available information on the (1)
significant noncompliance rates other than for the EIC and (2) overclaim
rates and administrative costs of comparable benefit programs administered
by states or the federal government and any verification process used by
these programs. This information is presented in appendixes I and II,
respectively.

This report is based primarily on our previous work and analysis of IRS
and Department of the Treasury documents and reports, new letters and
forms that will be sent to taxpayers, and interviews with senior officials
at IRS and Treasury. We did not verify the accuracy of reports or data
obtained. Rather, we reviewed the steps IRS officials had taken to
implement the program and determined, to the extent possible, how they
assured themselves that the program had been adequately developed to
prevent unreasonable burden and improve compliance. We did not determine
the adequacy of various other preparations for the qualifying child test,
such as staffing and training of staff. Appendix III provides more detail
on the scope and methodology used in conducting our work.

Results in Brief	Due to persistently high EIC noncompliance, among other
factors, the Assistant Secretary of the Treasury and IRS Commissioner
convened a task force in February 2002 to find ways of reducing EIC
overclaim rates while minimizing the burden to taxpayers and maintaining
the EIC's relatively high participation rate. The task force considered
the likely effect of recent legislative changes on EIC compliance in
formulating its proposal to

combat noncompliance. It also considered various options, such as
partnering with other federal or state agencies to verify EIC taxpayers'
eligibility. The task force analyzed data and reviewed studies to design
the program to focus on known sources of noncompliance. In August 2002,
the Treasury Secretary approved the task force recommendation that IRS
certify the eligibility of taxpayers' qualifying children. Only taxpayers
most likely to make errors and whose qualifying child eligibility cannot
be verified from available information would be asked to certify.

Since taking its broad charge from the task force, IRS obtained input from
external and internal stakeholders, and made key changes to the
certification program, including (1) concentrating on residency
certification and postponing the relationship certification for an
undetermined period of time, (2) delaying program implementation until
later this year, and (3) reducing the test sample from 45,000 to 25,000.
According to IRS officials, the relationship portion of the program was
postponed indefinitely for several reasons, including concerns raised
about the proposed relationship certification form and studies that have
shown meeting requirements for relationship is less of a compliance issue
than meeting residency requirements. IRS had previously changed the start
date of the test, and, as we were finalizing this report, announced in
August 2003, that it was again delaying the program and now plans to send
letters about the qualifying child certification program to taxpayers in
December 2003. With this change, taxpayers will now have to provide proof
that qualifying children meet the residency test when they file their 2003
individual income tax returns; thus, any unresolved issues could result in
frozen refunds. As part of the August announcement, IRS also reduced the
number of taxpayers that will be included in the test from 45,000 to
25,000, in part, in response to comments received during a 30-day public
period. However, these changes create additional challenges for IRS and
taxpayers. For example, the test will no longer be a direct test of the
original concept of certifying taxpayer eligibility in advance of the
filing season. At the same time, the process for selecting taxpayers, what
taxpayers will receive from IRS, and what taxpayers will be asked to
provide to prove the residency of a qualifying child remains basically the
same as originally planned. Further, IRS officials have emphasized that
program expansions, if any, will depend on the results of the test.

The changes made in the proposed certification program-including the ones
announced in August-help achieve a better balance between preventing
unreasonable taxpayer burden and addressing the EIC's high overclaim rate,
and support IRS's plans to move forward with the residency

test. The qualifying child program is based on analyses of the leading
sources of EIC errors, thus focusing attention and burden on the subset of
taxpayers making those errors as opposed to all EIC recipients. In
addition, IRS has taken steps to address the burden taxpayers will
experience as participants in the test this year. However, IRS's plan for
evaluating the results of the test is not yet complete. Although an
evaluation plan does not have to completely identify all issues or how
they will be evaluated prior to a program's start, the more completely a
plan is developed, the more likely that the evaluation will be sufficient
to support future decisions. The draft evaluation plan presents only some
information on how IRS will show whether certification's objectives-reduce
the overclaim rate, minimize burden, and maintain the EIC's relatively
high participation rate-will be achieved. However, the plan proposes
potential options for identifying how and when some critical data will be
obtained and analyzed, but does not provide further details on when, how,
and by whom decisions will be made on the specific data that will be
collected. Officials have developed preliminary drafts identifying data to
be obtained and have begun considering how to use contractors to gather
the data. Since the data relates to taxpayers' actions that will occur
next spring, IRS has some time to finalize the evaluation plan.

We are recommending that the Commissioner of Internal Revenue, to the
extent possible, accelerate the development of the evaluation plan, and
have the plan demonstrate how each program objective will be evaluated,
including milestones for critical steps such as how data will be obtained
and analyzed in time to support decisions about the future of IRS's
certification program.

We requested comments on a draft of this report from the Commissioner of
Internal Revenue. We received written comments, which are reprinted in
appendix VI. In his comments, the Commissioner said that IRS would be
including the components we suggested in their evaluation plan and said
that IRS is working to incorporate these components well before the
certification test begins. We further discuss the Commissioner's comments
in the "Agency Comments and Our Evaluation" section of the report.

Background	Congress enacted the EIC in 19754 with the goal of offsetting
the Social Security taxes paid by the working poor and creating a greater
work incentive for low-income taxpayers. According to data cited in the
task force report, an estimated 4.3 million individuals were lifted out of
poverty in 1998 by the EIC, including 2.3 million children.

The EIC is a refundable tax credit, meaning that qualifying working
taxpayers may receive a refund greater than the amount of income tax paid
during the year. Taxpayers can qualify for the credit in one of two ways:
with a "qualifying child" or by "income only," if they do not have a
qualifying child. For example, for tax year 2002, the amount of EIC that
could be claimed with a qualifying child or children ranged from $0 to
$4,140. EIC payments have a phase-in range in which higher incomes yield
higher EIC amounts, a plateau phase in which EIC amounts remain the same
even as income rises, and a phase-out range in which higher incomes yield
lower EIC payments or tax liability.

EIC requirements for tax year 2002 include rules for everyone, additional
rules for taxpayers with qualifying children, and additional rules for
taxpayers without qualifying children, as shown in table 1.

426 U.S.C. Sec. 32.

                  Table 1: EIC Requirements for Tax Year 2002

Additional rules for Additional rules for Rules for all taxpayers
taxpayers with a qualifying taxpayers without a claiming the EIC child
qualifying child

Must have a valid Social Income limitations: If one Income limitations:
$11,060

Security number	child: $29,201 (or $30,201 if (or $12,060 if married
filing married filing jointly). If more jointly). than one child: $33,178
(or $34,178 if married filing jointly).

Cannot use married filing Child must meet age, Must be at least 25 years
separately status relationship, and residency old, but under 65

tests

Must be a U.S. citizen or Child can be claimed by one Taxpayer cannot be
the

resident alien all year taxpayer only 	dependent of another person

Cannot file form 2555 or Taxpayer cannot be a Taxpayer cannot be a

2555-EZ	qualifying child of another qualifying child of another taxpayer
taxpayer

Investment income must be Must have lived in U.S. more $2,550 or less than
half of a year

                            Must have earned income

Source: IRS.

IRS has periodically measured EIC compliance for overclaims and
underclaims. The most current data available, for tax year 1999, show EIC
overclaim rates estimated to be between 27 and 32 percent of dollars
claimed or between $8.5 billion and $9.9 billion. IRS has limited data on
underclaims, which for tax year 1999 were estimated to be between $710
million and $765 million.5 IRS has tried to reduce noncompliance through
various means, including education and outreach to taxpayers and tax
return preparers. In addition, Congress has enacted legislation aimed at
resolving some concerns with EIC rules. Because a new analysis of EIC
compliance using 2001 tax return information is not expected to be
complete until late in 2004, IRS does not know whether compliance has
significantly changed since 1999, but officials do not think it has
improved

5Underclaims refers to the amount in which taxpayers claimed less than
what they were entitled to receive.

substantially. Because of the persistently high rates of noncompliance, we
have identified the EIC program as a high-risk area for IRS since 1995.6

Currently, taxpayers claim the EIC by filing an individual income tax
return (e.g., a Form 1040 or 1040A) and including a Schedule EIC-a
procedure similar for claiming other tax credits. Unlike with other
benefit programs such as Supplemental Security Income, however, EIC
taxpayers are not required to be found qualified before claiming the
credit or file any other documents with their return to establish
eligibility. Instead, IRS uses four primary means to evaluate EIC
eligibility and check for noncompliance after the return is filed and
checks some aspects of taxpayers' eligibility before and after the credit
is granted: (1) the math error program, (2) correspondence and
face-to-face examinations (also called audits), (3) the document matching
program, and (4) criminal investigations. Some of these means, such as the
math error program, check all EIC returns, but only for limited aspects of
eligibility. Other means, such as examinations, only check a small subset
of EIC returns, but the review is more expansive. In general, IRS subjects
all returns to its math error program and takes corrective action on
errors found. Depending on the resources IRS has available, IRS works only
a small portion of cases identified as potentially meriting follow-up
under its examination, document matching, and criminal investigations
efforts.

While processing all tax returns, IRS uses its automated math error
program to identify and correct the simpler errors found in claiming the
EIC. For example, the math error program can identify invalid Social
Security numbers and taxpayers who fail to follow recertification
requirements.7 As a result, some inappropriate EIC claims are stopped
before refunds are issued. During fiscal year 2001, IRS stopped more than
371,000 incorrect EIC claims using its math error authority. After
identifying errors, IRS corrects them so the tax return can be processed
and sends a computerized notice to the taxpayer identifying the error and
stating that IRS disallowed or reduced the EIC claim. The notice tells

6Prior to 2001, EIC was part of a broader IRS tax filing fraud high-risk
area. Beginning in 2001, the focus of that designation was narrowed to EIC
specifically. U.S. General Accounting Office, High-Risk Series: An Update,
GAO-01-263 (Washington, D.C.: January 2001).

7Taxpayers are required to meet recertification requirements when they
have been denied the EIC in a previous year. Recertification involves
taxpayers providing documentation, such as a birth certificate, to support
their claim of a relationship to a qualifying child.

taxpayers that if they can correct the error, the EIC claim will be
allowed and any refund related to the EIC claim will be issued.

Two types of examinations--correspondence and face-to-face-are used when
EIC noncompliance is suspected, in most cases before refunds are issued.
IRS uses various systematic means to "score" the likelihood of
noncompliance on any return and uses experienced staff to manually
identify the specific items on returns for examination. Most EIC
examinations occur shortly after a return is filed, largely because of the
difficulty in recovering refunds. IRS stops refunds on these returns until
examinations are completed. This contrasts with IRS's normal examination
practice of performing examinations many months after tax returns have
been processed and any refunds paid. The EIC examinations usually rely on
correspondence with taxpayers rather than face-to-face contacts. IRS
completed about 368,000 EIC related correspondence exams during fiscal
year 2002. IRS tends to use face-to-face meetings with taxpayers to
examine tax returns with EIC claims on a very limited basis and primarily
when examinations are initiated for other reasons. As part of either type
of examination, however, IRS would describe the potential noncompliance in
a computerized notice to taxpayers claiming the EIC. IRS requests
documentation, such as a school record or birth certificate, to establish
EIC requirements. Depending on whether IRS officials accept or reject the
support, they may make changes to the return and refund related to the EIC
claim. If taxpayers disagree with IRS's decisions, they have the right to
appeal administratively and/or through the courts.

IRS also uses its document matching programs to identify potentially
misreported income on tax returns claiming the EIC. By comparing the tax
return to wage and income statements provided by third parties such as
employers and financial institutions, the document-matching program
identifies whether a taxpayer appears to have misreported income. Given
the phase-in and phase-out ranges of the EIC, some taxpayers may claim too
much EIC by overreporting or underreporting their income. This program
notifies such taxpayers months after returns are filed and refunds are
issued. Similar to audits, a notice is issued telling a taxpayer that an
error appears to have been made, that he or she may disagree and provide
any support for income reported, and that he or she may appeal IRS's
decision about additional taxes owed. Unlike audits, the program is highly
automated and is designed to require less contact with taxpayers by IRS
staff.

IRS also uses criminal investigations to stop the payment of false
refunds, identify refund scams/schemes, and prosecute perpetrators,
including those with fraudulent EIC claims. For EIC, IRS uses a specific
computer program that looks for questionable refund claims and for return
preparers known to have prepared questionable returns. IRS also has teams
that scan returns and receive referrals from other parts of IRS and
informants. IRS stops many returns as they are being processed so that
criminal investigators can review the claims before the refund is paid or
after the return has been processed.

Task Force Considered Much Information Related to EIC Compliance before
Recommending Qualifying Child Certification Program

When IRS's study of EIC compliance rates for 1999 was released, the
Assistant Secretary of the Treasury and IRS Commissioner convened a task
force in February 2002 to find ways of reducing EIC overclaims while
minimizing the burden to taxpayers and maintaining the EIC's relatively
high participation rate. The task force considered whether changes in
statutes recently enacted by Congress or proposed by Treasury may have
lessened the need for new EIC compliance initiatives and concluded that,
while statutory changes addressed some sources of noncompliance, they
likely would not reduce other leading sources of noncompliance. The task
force also considered a range of new methods, including partnering with
other federal or state agencies or programs and developing a new database
to verify EIC eligibility before issuing tax refunds, but decided that
these options were not viable. Ultimately, the task force recommended the
qualifying child certification program. The task force reviewed IRS's EIC
compliance study results and other data, as well as other studies, to
identify the sources and develop new methods of addressing noncompliance.

Task Force Convened to Address Long-Standing Compliance Problem

The joint Treasury and IRS task force addressed a long-standing problem of
high EIC overclaim rates. Although the release of IRS's 1999 compliance
study precipitated the formation of the EIC task force in February 2002,
the study results were generally consistent with high overclaim rates
reported in prior IRS studies. While some stakeholders view the 1999 study
as having some methodological weaknesses, it showed that of the
approximately 20 million taxpayers that claimed the EIC in 1999, 46 to 50
percent of their tax returns had errors that led to claiming too much of
the credit (IRS often refers to this as the error rate). IRS also
estimated that the total dollars overclaimed on those returns represented
between 27 and 32 percent of total EIC dollars claimed in 1999, or between
$8.5 billion and $9.9 billion.

IRS also has some data on underclaims-instances where taxpayers claimed
less than they were entitled to receive. For tax year 1999, underclaims
were estimated to be between $710 million and $765 million.

IRS has conducted EIC compliance studies for several years and the
overclaim rate, which is the percentage of total dollars paid out in
error, was estimated to be about 24 percent in tax year 1994. According to
IRS officials, because different methodologies were used in the subsequent
studies, changes in estimated overclaims found in other studies do not
support conclusions about trends in the overclaim rate over time. However,
IRS officials also acknowledged the overclaim rate has not improved
significantly. Overclaim rates for tax years 1997 and 1999 are shown in
table

2. The information in table 2 does not reflect the current compliance
situation; for example, it does not reflect the presumably positive impact
of new legislation that has taken effect since 1999 aimed at improving
compliance.

    Table 2: EIC Overclaim Rates for 1997 and 1999 Tax year Lower-bound rate
                                Upper-bound rate

                                1997 23.8% 25.6%

                                1999 27.0% 31.7%

Source: IRS data.

Note: Because not all individuals responded to audit contacts, IRS uses
certain assumptions to estimate the overclaim rate range. The lower-bound
rate assumes that the overclaim rate for the nonrespondents is the same as
for the respondents, while the upper-bound assumes that all nonrespondents
are overclaims.

Although IRS's studies have shown high EIC overclaim rates for many years,
other studies had shown that EIC's participation rate was fairly high. For
example, in 2001 we reported that an estimated three of every four
eligible participants received the EIC in tax year 1999.8 For taxpayers
with one or two qualifying children, we estimated that participation rates
exceeded 90 percent. Individuals with no children, who receive a much
smaller credit than taxpayers with qualifying children, had a much lower
participation rate that we estimated to be about 45 percent. Although at
the time we reported that available data did not enable us to determine
the

8U.S. General Accounting Office, Earned Income Tax Credit Participation,
GAO-02-290R (Washington, D.C.: Dec. 14, 2001).

reasons for these differences, IRS officials attributed these differences,
in part, to the lower EIC amounts allowed for individuals and because the
program did not include individuals without children when it first began.

Task Force Reviewed Recent Legislative Changes Likely to Improve
Compliance

The EIC task force reviewed whether recent statutory changes have the
potential to reduce the major sources of EIC noncompliance, either by
changing the rules or providing IRS new enforcement options. Because the
study of tax year 1999 compliance was the most recent available, the task
force lacked data on the effect of the recent changes and relied on other
analyses that showed whether the changes would affect compliance. Of the
recent changes, the task force estimated that one change, to the Adjusted

Gross Income (AGI) tiebreaker rule,9 would likely reduce noncompliance.
The task force judged that the other legislative changes, including those
proposed by Treasury, while potentially helping reduce noncompliance from
other sources, would not be enough to reduce noncompliance without further
IRS efforts.

Three key pieces of legislation, which have been recently enacted or taken
effect, were at least partially aimed at improving EIC compliance, as
shown in table 3. They may eventually help reduce noncompliance after
taxpayers and tax preparers become familiar with the new laws. The
statutory changes were to serve several purposes, including improving
compliance and simplifying tax laws associated with the EIC.

9The new AGI tiebreaker rule applies when two taxpayers can claim the same
qualifying child. If one of the taxpayers claiming the credit is the
child's parent (or parents who file a joint return), then the child is
considered the qualifying child of the parent or parents. If both parents
claim the child and parents do not file a joint return, then the child is
considered a qualifying child first of the parent with whom the child
resided for the longest period during the year, and second of the parent
with the highest adjusted gross income. If none of the taxpayers claiming
the child as qualifying is the child's parent, the child is considered a
qualifying child with respect to the taxpayer with the highest adjusted
gross income.

Table 3: Recent Statutory Changes Law Change

Economic Growth and Tax Relief This act made several changes, including
(a) effective Reconciliation Act of 2001(P.L. for tax years after December
31, 2001, simplifying the 107-16, March 7, 2001) "Adjusted Gross Income
tiebreaker rule," (b) for tax

years beginning after December 31, 2001, establishing a new definition for
earned income by eliminating non-taxable earned income and by having the
Earned Income Credit (EIC) based on adjusted gross income, (c) for tax
years beginning after December 31, 2001, amending the definition of a
foster child by reducing the residency requirement to over half a year,
and (d) effective January 1, 2004, allowing the IRS to use math error
authority to deny EIC claims if the Federal Case Registry (FCR) indicates
that the taxpayer is the noncustodial parent of the qualifying child with
whom the credit is claimed.

Ticket to Work and Work Effective after December 31, 1999, part of this
act	Incentives Improvement Act of simplified the definition of a foster
child.	1999 (P.L. 106-170, December	17, 1999)	

Tax Relief Act of 1997 (P.L. 105-After October 1,1998, this act requires
that each

34, August 5, 1997)
record in the state registry include the Social Security number of any
child for whom support has been ordered. This information is included in
the FCR database. It also requires an applicant for a Social Security
number who is under age 18 to provide his or her parent's Social Security
numbers, in addition to other required evidence, such as age, identity,
and citizenship.

Source: GAO analysis of legislation.

A Treasury study showed that the change in the AGI tiebreaker rule
effective for tax years after December 31, 2001, would likely have
eliminated about $1.4 billion of the nearly $2 billion in tax year 1999
EIC overclaims that were due to tiebreaker errors. Accordingly, the task
force decided that this source of EIC overclaims did not need to be
further addressed by a new compliance initiative.

Although officials recognized the benefits of these recent legislative
changes to help improve EIC compliance, they concluded that additional
initiatives were still needed. For example, officials recognized the value
of IRS being able to use math error authority to deny EIC claims on and
after January 1, 2004, when the Department of Health and Human Services'
Federal Case Registry (FCR) indicates that the taxpayer is the
noncustodial parent of the qualifying child. However, officials told us
that this authority

was limited and not applicable to a significant number of taxpayers whose
compliance may be problematic. IRS has a study in process to determine the
effectiveness of using FCR data to deny EIC claims using its math
authority. The study was scheduled for completion by July 30, 2003, but as
of August 20, 2003, was not yet completed.

Task Force Considered Three Alternatives to Improve Qualifying Child
Compliance

The EIC task force considered three key options to verify taxpayers'
qualifying children: (1) partnering with other federal or state agencies
or government programs to verify EIC taxpayers' eligibility, (2) creating
a federal database that would automatically match and detect questionable
or erroneous EIC claims, and (3) certifying taxpayers' eligibility for
certain EIC criteria. Ultimately, in August 2002, the Secretary of the
Treasury approved the qualifying child certification program, which at the
time was to include providing proof of eligibility in advance of the
filing season (July-December), and was referred to as "precertification."

The first two options were expected to impose little or no documentation
requirements on taxpayers. The task force was trying to determine for both
options whether sufficient information was already available from others,
or that little additional information would need to be collected by
others, to verify a taxpayer's qualifying children. However, the task
force found that there was little overlap between the EIC population and
verification criteria used to administer other federal or state programs.
In addition, although some databases existed, the task force found that
they could not be used to effectively verify EIC eligibility, largely for
the same reason. Consequently, the task force judged that these options
were not likely to be useful in addressing EIC compliance problems.
Similarly, the task force also found that if a federal database were
created to facilitate EIC verification, IRS would have to gather the bulk
of the information itself, thus imposing a burden on taxpayers, which
would also be costly and time-consuming for IRS. The third option, which
the task force selected, required taxpayers to demonstrate EIC eligibility
for certain criteria, namely residency and relationship tests for
qualifying children, prior to receiving the credit.

The relative cost of the options the task force considered did not drive
the decision to select the qualifying child program because the other two
alternatives were not considered viable. The task force did compare IRS's
EIC administrative costs to those of other federal benefit programs and
found them to be much smaller. IRS has had a special appropriation for EIC
compliance initiatives since 1998-and has received about $875 million

total through fiscal year 2003. It requested a total of about $250 million
in fiscal year 2004, which included $100 million for the EIC compliance
initiatives, including the qualifying child program, and about $150
million for the special appropriation. IRS estimated that this $250
million10 total was about 0.8 percent of the total annual EIC benefits
distributed, and therefore much smaller than the 9 to 13 percent
administrative costs the task force had found for other benefit programs.
See appendix II for more information we obtained on administrative costs
for other benefit programs.

Task Force Reviewed Studies and Data to Design Initiatives Focused on
Known Sources of Compliance Problems

The EIC task force reviewed IRS studies, other IRS data, and studies by
other parties to better understand the sources of EIC noncompliance and
devise new initiatives to address those known sources. In reviewing IRS
studies and data, the task force found that the three leading sources of
EIC errors resulting in overclaims in 1999 were (1) claiming nonqualifying
children incorrectly, accounting for about $3 billion, (2) using the wrong
filing status, accounting for about $2 billion, and (3) misreporting
income, also accounting for about $2 billion. Three administrative
proposals resulted, involving (1) qualifying child certification, (2)
improper filing status, and (3) income misreporting.11 In 1999, another
leading source of EIC overclaims involved taxpayers with lower modified
adjusted gross income claiming a child when another person with a higher
income should have done so. The task force did not propose an initiative
dealing with these errors, primarily because the "AGI tiebreaker"
legislation was specifically enacted to decrease this source of
noncompliance, as previously discussed.

10In addition to the $250 million that IRS requested in its fiscal year
2004 budget request to administer the EIC, IRS incurs some additional
costs. For example, IRS incurs costs to process the EIC tax returns.
Therefore, the full cost of administering the EIC is not known.

11Improper filing status claims and income misreporting are other common
problems associated with the EIC. IRS plans to verify the filing status
for about 41,650 cases in fiscal year 2004, but the criteria for selecting
the cases have not yet been finally determined. In fiscal year 2004, IRS
plans to use document matching to verify the income reported by about
300,000 EIC filers who have a history of misreporting income for 2
consecutive years in order to increase (or receive) the EIC. Depending on
how well these efforts work in fiscal year 2004, they would be expanded in
future years. Also see U.S. General Accounting Office, May 20 Oversight
Hearing on the Internal Revenue Service - Questions for the Record,
GAO-03-962R (Washington, D.C.: June 27, 2003).

To deal with the error attributable to claiming children who are not EIC
qualifying children, the task force proposed a qualifying child
certification program. Based on analyses of past compliance data, IRS
found that taxpayers who overclaimed the EIC, most frequently claimed
children who did not meet the residency or relationship criteria.12 As a
result, the task force proposed the qualifying child program that was to
include an annual residency certification and a one-time relationship
certification.

Under this program, during the period from July through December,
taxpayers would have been asked to document that the children they intend
to claim under the EIC, meet the EIC relationship and residency criteria.
The task force proposed targeting the program to those taxpayers with
qualifying children for whom IRS could not establish residency or
relationship through other available means and it proposed that this
concept be tested on a sample of EIC taxpayers for the tax year 2003. The
task force envisioned that ultimately all EIC claimants whose eligibility
could not be verified through available means would be asked to provide
additional eligibility documentation prior to the filing season. Taxpayers
who successfully certified qualifying children's eligibility in advance of
the filing season would have their claims processed and paid expeditiously
during the filing season, absent any other problems with their tax return
or EIC claim. Having taxpayers certify between July and December was also
intended to allow IRS to process the taxpayers' documents outside of the
filing season when IRS processing systems are in highest demand. Taxpayers
who did not respond and/or were unable to document their eligibility
during the certification period, but then claimed the EIC when they
submitted their tax returns, would have the EIC portion of their tax
refund frozen. Then they would be required to provide the same
documentation during or after the filing season as they were asked to
provide during the certification period. When and if they document their
eligibility, the EIC portion of their refunds would be released.

12For 2002 returns, taxpayers who claim the EIC with a qualifying child
must meet certain tests, including residency and relationship. To meet the
residency test, the qualifying child had to live with the taxpayer in the
United States for more than half of the year. To meet the relationship
test, the qualifying child had to be a son, daughter, adopted child,
stepchild of the taxpayer, or a descendent of any such individual.
Sisters, brothers, stepsisters, stepbrothers, and descendents of any such
individual also qualify if the taxpayer cares for the individual as they
would their own child. In addition, a foster child can qualify for the
relationship test if certain conditions are met. Internal Revenue Service,
Earned Income Credit (EIC), Publication 596 (Washington, D.C.: 2002).

Process for Identifying Taxpayers When IRS Planned to Certify for
Relationship

Process for Identifying Taxpayers to Certify for Residency

According IRS officials, as the task force neared its end and before the
Secretary of the Treasury approved the program, IRS developed a means for
using existing data to determine whether each taxpayer likely would meet
the relationship or residency test for children they had claimed for EIC
for tax year 2002. For relationship, IRS developed a plan to match
taxpayers to several databases that show the parents of children. For
instance, one database IRS planned to use was the Social Security
Administration's database (which IRS refers to as KIDLINK) that ties
parent's and children's Social Security numbers for children born after
1998 in U.S. hospitals. For tax year 2003, IRS had planned to match 1.6
million, or 10 percent of the approximately 16 million EIC taxpayers with
a qualifying child, to the databases. Under this scenario, any taxpayer
who was not shown to be the parent of a qualifying child claimed for tax
year 2002 would then be part of the population from which IRS would
randomly select taxpayers to test for relationship.

IRS considered the work of the task force in developing a comparable means
for using available data to identify those who have met the residency
criterion. The task force had analyzed data from the 1999 compliance study
and information in other reports.13 It found that residency errors related
to qualifying children were often correlated with the taxpayer's
relationship to the child and the taxpayer's filing status and gender. The
analysis showed that, overall, parents who filed married filing jointly
were the most compliant when compared to taxpayers filing single or head
of household in claiming a qualifying child who meets the residency test.
Married filing jointly parents had the fewest qualifying child residency
errors-1.5 percent-compared to any other combination of taxpayers by
relationship to the child, gender, or tax filing status. Among taxpayers
who file single or head of household, mothers were the most compliant (see
figure 1).

13Andrew J. Cherlin and Paula Fomby, "Welfare, Children, and Families: A
Three-City Study," A Closer Look at Changes in Children's Living
Arrangements in Low-Income Families, Working Paper 02-01(Baltimore, MD.:
Johns Hopkins University, Feb. 20, 2002), and Allen Dupree and Wendell
Primus, Declining Share of Children Lived with Single Mothers in the Late
1990s: Substantial Difference by Race and Income (Washington, D.C.: Center
on Budget and Policy Priorities, June 15, 2001). We did not review these
studies in detail. Instead, we relied upon IRS's assessment.

Figure 1: Estimated Percentage of EIC Taxpayers (with Children) Who Made
Residency Errors and Filed Single or Head of Household in Tax Year 1999

80 Percentage

70

60

50

                                      44.1

40

30

20

10

0 mothers male fathers female nonparent nonparent

Source: IRS data.

Note: This figure reflects residency errors only. Mothers who filed as
single or head of household, for example, made qualifying child residency
errors 3.4 percent of the time, while 96.6 percent of the time, they
either made no errors or errors other than residency-related errors.

The task force also found other reports that reinforced the results of its
analysis. Specifically, an independent study of low-income households in
three urban areas estimated that children resided with biological mothers
90 percent of the time. Another study estimated that 89 percent of
children in low-income households lived with both parents or their mother.

IRS used this information to propose a process for identifying taxpayers
to include in the population that would be subject to the residency
certification requirement. IRS proposed that the 1.6 million taxpayers, or
10 percent of the 16 million taxpayers with a qualifying child, would be
matched to the FCR database. IRS officials considered the FCR to be the
most useful database for identifying those meeting the residency
requirements. This database compiles court and other records that indicate
who is the custodian for a child (which could be a parent or nonparent).
IRS assumes that children live with the custodian of record. According to
IRS, the FCR database contains custodial information for about 40 percent
of the EIC population. If a taxpayer matched as the custodian of the child
claimed for the EIC for 2002, the taxpayer would not

be among those needing to certify. When the FCR database showed someone
was the custodian of a child other than the EIC taxpayer who had claimed
that child for the EIC in 2002, those taxpayers would be among the group
from which the residency certification sample would be drawn.

When the FCR contains no information about the child a taxpayer had
claimed for EIC in 2002, the IRS would attempt to establish the
relationship of taxpayers to qualifying children by comparing information
in several databases. Those taxpayers IRS could identify from databases as
the child's mother would be excluded from the sample, if they filed
married filing jointly, single, or as head of household. Mothers would be
excluded on the basis of the task force analyses showing mothers to be
among the most compliant on the residency criterion. Also excluded from
the sample would be fathers who filed as married filing jointly.
Otherwise, all males who were shown to be a child's father filing single
or head of household would be included in the group from which the
certification sample would be drawn because of the data showing a high
level of noncompliance on the residency criterion for these taxpayers.
Finally, all nonparents who are not shown in the FCR to be the custodian
would go into the group from which taxpayers would be selected for
residency certification, also due to information showing nonparents to be
among the less compliant taxpayers on the residency criterion. The
selection processes for relationship and residency would have, therefore,
yielded a group of taxpayers that would include some needing to certify
for relationship only, some for residency only, and some to certify for
both.

IRS Has Made Key Changes to Its Initial Qualifying Child Certification
Program and More May Occur

Since adopting the EIC task force recommendations in August 2002, IRS has
made key changes to the qualifying child certification program in response
to input received and additional analyses done. Some of these changes
include (1) postponing relationship certification for an undetermined
period of time, (2) delaying program implementation, and (3) reducing the
test sample from 45,000 to 25,000. However, these changes create
additional challenges for IRS and taxpayers. Despite these challenges, the
process for selecting taxpayers, what taxpayers will receive from IRS, and
what taxpayers will be required to provide remains basically the same as
originally planned. According to officials, the same factors were
considered when setting the new sample size, which is still designed to
allow IRS to achieve the same goals as the original sample size, albeit to
a lesser extent.

In addition, IRS has emphasized that program expansions, if any, will
depend on the results of this year's test. Concerns we identified in our
report on recertification14 were considered and taken into account by IRS
in designing the new qualifying child certification program.

IRS Broadly Adopted Task Force Recommendations in Its Initial Design of
the Qualifying Child Certification Program

IRS took the broad charge from the EIC task force and designed the
qualifying child certification program. Its focus was to decrease the EIC
overclaim rate while striving to maintain the high rate of participation
and minimize taxpayer burden. Initially, IRS decided that the
certification program would involve

o 	testing of 45,000 taxpayers for both relationship and residency
beginning in July 2003, and

o 	immediately expanding the certification program for relationship to 2
million taxpayers in 2005 and to both relationship and residency in
substantial numbers in future years.

However, as IRS obtained input on the program, it modified these plans.

Certification for Residency Only Is to Begin in December 2003 for 25,000
Taxpayers

Since initially formulating plans for the qualifying child certification,
IRS has made multiple changes to the program. First, IRS postponed
relationship certification for an undetermined period for a number of
reasons. IRS had developed a draft form for certifying relationships and
obtained input on the form from external and internal stakeholders. Some
stakeholders raised concerns about whether taxpayers would be able to
provide some of the types of documentation IRS planned to request, such as
marriage certificates, within the time envisioned. IRS officials said that
testing the relationship certification this year was postponed, in part,
because these concerns were unresolved. The officials also noted that
Treasury studies have shown relationship requirements to be a lesser
compliance issue than residency, and taxpayers that were found to be

14 The "certification" program is different from the IRS "recertification"
program, which is required by statute. Recertification was implemented in
1998 and requires taxpayers who have been disallowed the EIC through an
IRS examination to substantiate their qualification for the EIC, i.e.,
recertify before they receive the credit again. U.S. General Accounting
Office, Opportunities to Make Recertification Program Less Confusing and
More Consistent, GAO-02-449 (Washington, D.C.: Apr. 25, 2002).

noncompliant with relationship requirements were also often noncompliant
due to residency errors. Since both residency and relationship
requirements have to be met, if taxpayers fail certification on residency
there would be no need to test on relationship. Consequently, officials
gave a higher priority for testing residency certification.

Second, IRS has changed the start date of the test twice. Originally, IRS
planned to start the test in July 2003, but postponed implementation until
mid-August. As we were finalizing this report, IRS announced in August
that it now plans to begin the test in December 2003, in conjunction with
the 2004 tax filing season. (Appendix IV shows key milestones from 2002
through 2005.) According to IRS officials and documents, implementation
was postponed from July to August to allow time to conduct focus group
testing, request and obtain public comments during a 30-day period, and
make changes to the program as a result of those efforts. Thereafter, IRS
postponed implementation a second time from August to December to ensure
(1) taxpayers have better access to tax practitioners since many only
operate during the filing season and (2) more time for outreach and
education.

However, as a result of the delays, taxpayers will be providing proof of
residency documentation during the filing season and not "precertifying"
before the filing season as originally envisioned. This change is
important and creates additional challenges for both IRS and taxpayers, as
follows:

o 	Taxpayers will no longer have the opportunity to provide proof of
qualifying child residency, correspond with IRS in advance of the filing
season, and resolve any potential issues before filing their tax returns.
Because all correspondence will take place during the filing season,
selected taxpayers could experience a delay in receiving the EIC portion
of any refund, if the EIC portion is frozen because of any problems until
certification is successfully completed.

o 	IRS will no longer be able to spread out its workload and processing
may be slower since certification will occur during the filing
season---IRS's busiest time of year.

o 	IRS will not have the opportunity to assess taxpayers' ease or
difficulty in obtaining required documentation in advance of the filing
season and whether taxpayers would do so. This is important because
taxpayers may be given the opportunity to certify in advance of the filing
season in future years.

Third, IRS reduced the number of taxpayers included in the test from
45,000 to 25,000, in part, in response to comments received during the
30-day public period. According to IRS officials, despite reducing the
number of taxpayers included in the test, the sample size should still
allow IRS to make statistically valid measurements of results in addition
to helping IRS meet its desired goals of protecting revenue and testing
the process for conducting the certification program. In addition, the
smaller sample should help mitigate the challenge related to processing
the certification forms during the filing season.

Process for Selecting Taxpayers, What They Will Receive, and What They
Will Provide to IRS Remains Basically the Same

Despite these changes, how IRS selected taxpayers for the test, what
taxpayers will receive from IRS, and what taxpayers will be asked to
provide as proof of residency for qualifying children will remain
fundamentally the same. IRS's process for selecting the taxpayers for the
test is shown in figure 2.15 Using this process, IRS selected 25,000
taxpayers in August. The 25,000 represents about 0.16 percent of the
approximately 16 million EIC claimants with a qualifying child in tax year
2002 and about 0.13 percent of the approximately 20 million EIC recipients
overall.

15IRS will exclude from its sample taxpayers who are subject to an
examination, investigation, or other treatment at the same time.

Figure 2: IRS's Process to Identify the Population of Taxpayers from Which
the 25,000 Person Test Sample Was Drawn

aThe FCR contains custodial and welfare assistance data. If there is a
match showing that the taxpayer is the child's custodian, regardless of
the taxpayer's gender, filing status, or relationship to the child, IRS
assumes the child resides with the taxpayer.

bThese taxpayers are excluded, based on (1) IRS compliance data that show
married filing jointly parents are 20 times more compliant on residency
than nonparents, and single or head of household mothers are 10 times more
compliant on residency than single or head of household fathers and (2)
private studies that show about 90 percent of children in low income
households live with their mother or both parents.

cIRS estimates that using the FCR and other databases will exclude almost
75 percent of the 1.6 million taxpayers. Therefore, about 402,000
taxpayers were included in the population from which IRS drew the
certification sample.

According to agency officials, IRS will now send the 25,000 taxpayers
forms and instructions about the program in December instead of this
summer. IRS plans to send Notice 84-A, a letter informing them about the
new program; Form 8836, "Qualifying Children Residency Statement;"
Publication 3211M, "Earned Income Tax Credit Question and Answers;" and
Publication 4134, "Free/Nominal Cost Assistance Available for Low Income
Taxpayers." However, officials are changing these documents based on the
public comments received. Appendix V has the most current copies of these
documents.

Once taxpayers receive this information from IRS, they would obtain
documentation to prove the qualifying child's residency and send it back
to IRS. IRS examiners would review the documentation and send a letter
back to the taxpayer either accepting or rejecting the claim, as shown in
figure 3.

Figure 3: The EIC Certification Process as Envisioned

IRS currently envisions that the 25,000 taxpayers selected for
certification will be required to provide proof that the qualifying child
meets residency requirements before getting the EIC portion of their
refund. IRS officials

say that taxpayers who are able to establish eligibility when filing their
tax return should receive their refunds more expeditiously than those who
do not. Taxpayers selected for certification but who are not able to
provide the necessary documentation will be treated essentially the same
as taxpayers undergoing a correspondence audit. The EIC portion of their
refund-if they are to get one-will be frozen until proof of eligibility is
established.

Same Factors Considered When Setting Smaller Sample Size

According to IRS's draft evaluation plan for the certification test and
our discussions with officials, three factors were considered in setting
the original sample size of 45,000: (1) show that certification would
"protect revenue," (2) determine whether the test will succeed, and (3)
test its processes and systems. According to IRS officials, the smaller
sample size of 25,000 is designed to allow IRS to achieve the same goals
as the original sample size, albeit to a lesser extent.

One factor considered by IRS for the certification test was to stop as
large an amount of EIC overclaims due to ineligible qualifying children as
possible during the 2003 tax year. To determine how many taxpayers to
include in the certification test to achieve this goal, officials said
they determined the maximum number of staff that could be assigned to and
adequately supported by the planned central unit in Kansas City that would
be responsible for the certification program. Based on the maximum number
of staff that could be assigned and assumptions about how many cases staff
could handle, IRS calculated that 45,000 taxpayers could be included in
the test. IRS estimated that $114.5 million in protected revenues could be
realized from including 45,000 taxpayers in the test. Based on the revised
sample size of 25,000, IRS now estimates that $63.6 million in protected
revenues could be realized.

A second factor considered was to have a large enough sample to support
analyses of whether the test succeeds. For instance, IRS is interested in
how many taxpayers provide the information needed for IRS to determine
qualifying child eligibility, whether taxpayers in the sample population
who are actually qualified to claim the EIC do not do so with their 2003
tax return and why, and whether taxpayers found the certification process
burdensome. IRS's draft plan for evaluating the certification test notes
that the original 45,000 sample size was much larger than needed to obtain
statistically valid measures of test results. The draft plan indicated
that a sample size of about 3,600 taxpayers, which would have provided an
estimate at 95 percent confidence levels plus or minus 5 percent, was the

number of taxpayers needed for IRS to determine qualifying child
eligibility. According to the draft plan, the 45,000 sample size would
allow very precise estimates for the population as a whole and should
provide statistically valid information about sub-sets of claimants.
Despite the reduction to 25,000, IRS officials still believe that this
sample size will allow for precise estimates for the universe as a whole
and smaller subsets as well.

Finally, a third factor in selecting both the 45,000 and 25,000 sample
sizes was to have a large enough sample to test the processes and systems
that would be required if IRS were to expand certification in the coming
year. IRS had been preparing to work on approximately 25,000 certification
cases during the filing season under its original plan for 45,000
taxpayers. It based the 25,000 on worst case assumptions about how many of
the 45,000 would not opt to submit proof of eligibility in advance of the
filing season and, instead, would have submitted their documentation
during the filing season. In addition, for this goal, the draft plan
preceded IRS's current thinking that the certification program likely will
not be expanded as rapidly next year, if expanded at all. However,
according to IRS officials, based on the number of cases IRS estimates can
be worked on and what it plans to achieve under this goal, the 25,000
sample size is appropriate to help test the systems and processes.

Future Expansion of Certification Program Depends on Test Results

Although IRS has consistently referred to the certification effort as a
test, officials recently have stressed this point. For example, officials
have recently referred to the efforts for this year as a "pilot or proof
of concept." Furthermore, as a result of the most recent changes, the
program will no longer take place in advance of the filing season, but
instead, during the filing season. IRS officials told us that it is
unlikely that the certification program will be expanded to cover 2
million claimants in the summer of 2004, as originally anticipated.
Instead, IRS officials plan to assess the program's overall effectiveness
and make any necessary modifications before expanding it to additional EIC
claimants in the future. Thus, particularly in light of IRS's most recent
announcement and according to IRS officials, the program may be expanded
more slowly, if at all, depending upon the evaluation results. Officials
also said that test results will contribute to a future decision about
whether certification, if continued, will precede the filing season or be
part of the filing season as it will be this year.

Input from Focus Groups and Others Has Resulted in Changes and May Result
in More

On the basis of stakeholder input, focus groups, and other input, IRS has
made several changes to the planned certification test in addition to
those discussed previously. IRS held informal meetings with external and
internal stakeholders, focus group meetings with taxpayers and paid
preparers, and one-on-one interviews with third parties to share the
certification letters, forms, and/or instructions and obtain views on
aspects of the new process. In response, IRS took several actions,
including revising the forms. As of August 2003, IRS was evaluating
comments received during the 30-day public comment period, which IRS
officials said may result in additional changes.

IRS held several informal meetings with external and internal parties with
an interest in the qualifying child certification program. In March 2003,
the Stakeholder Partnership, Education, and Communication Organization16
and the National Taxpayer Advocate held four informal meetings with
various external stakeholders, such as representatives of the National
League of Cities, the Boston EIC Coalition, and the American Institute of
Certified Public Accountants. Similarly, IRS officials coordinated the
certification initiative with internal stakeholders, such as
representatives from the Compliance unit, Wage and Investment operating
division, Small Business/Self Employed (SB/SE) operating division, and
Forms and Publications unit. The purpose of these meetings was to discuss
the EIC certification proposal and share the drafts of the two new
certification tax forms-Form 8836, "Qualifying Children Residency
Statement," and Form 8856, "Qualifying Child Relationship Statement."
Officials told us they received comments from these groups of stakeholders
and revised and improved the forms based on the feedback received. For
example, for the residency form IRS added a list of community-based
organizations and a list of acceptable third parties from which IRS would
accept affidavits.

After incorporating the recommendations from these informal meetings,
officials said they felt comfortable with testing the Form 8836, its
instructions, and the accompanying letter in other ways, including focus
groups, one-on-one interviews, and a 30-day public comment period.17

16 The Stakeholder Partnership, Education, and Communication Organization
is a unit with IRS's Wage and Investment operating division. Its role is
to educate and assist taxpayers before their returns are filed.

17Before the focus groups commenced, IRS had decided to move forward only
with the residency certification.

In June 2003, a contractor conducted nine focus groups, five with
taxpayers who claimed EIC in tax year 2002, and four with tax preparers
who had prepared returns for taxpayers claiming EIC. In addition to the
focus groups, the contractor also conducted nine one-on-one interviews
with a cross section of the third parties listed on Part IV of the Form
8836 (the participants were landlords, employers, and child care
providers). The goal of the testing was to determine whether individuals
understood the documents and thought they could obtain the requested
supporting documents and whether the suggested third parties would be
willing to sign the affidavit.

The focus groups and interviews were held in Philadelphia, Chicago,
Dallas, and Los Angeles. These cities were selected because of their high
EIC population. The participants were selected using screening guidelines
developed by IRS in conjunction with the contractor. Taxpayers were
selected on the basis that they claimed the EIC for tax year 2002 with a
qualifying child. Similarly, preparers were selected on the basis that
they worked as a tax preparer on federal tax returns for 2002 and prepared
tax returns for clients claiming EIC with a qualifying child. Those
selected for one-on-one interviews represented a cross section of the
types of individuals IRS deemed credible to provide affidavit information
about the EIC claimant. In total, 816 people were contacted and 109 agreed
to participate in the focus group testing. Of the 109, 88 persons arrived
for the testing and 81 actually participated in the focus groups. For the
one-on-one interviews, 12 individuals were qualified to participate and 9
actually participated in the interviews.18

Key IRS officials were on site during the focus groups and one-on-one
interviews to observe the participants' comments. A variety of comments
were received and some changes were made. For example, IRS highlighted
where taxpayers and third parties were to sign the forms. Although the
contractor's report of these meetings was not available before the public
comment period, IRS officials who attended the meetings concluded that
they had not received any feedback that would preclude moving forward with
getting comments from the public.

18According to IRS officials, approval from the Office of Management and
Budget is required in situations where the agency conducts interviews with
more than nine private sector participants. However, because conducting
such interviews was suggested later in the process and because OMB had
been briefed on the development of the certification test, IRS did not
seek that approval.

During the comment period, anyone could write or go to IRS's Web site and
provide any comments or opinions about the qualifying child certification
program, including the form IRS expected to use and the data it planned to
request to prove eligibility. According to IRS, during the 30-day public
comment period, IRS received about 200 communications containing comments.
Any other comments about the certification program are due by December 31,
2003. In addition, individuals can comment on the certification process
during the filing season until April 15, 2004. As of August 2003, IRS
officials were reviewing the comments received and anticipated making
additional changes to the forms and publications shown in appendix V.

Our Concerns with IRS's Recertification Were Considered as IRS Designed
the Qualifying Child Certification Program

IRS officials told us they considered the recommendations in our
recertification report19 when planning their certification program. We
agree that our applicable recommendations have been considered. Whether
the strategies IRS adopted to deal with the concerns that led to our
recertification report recommendations are successful will not be known
until IRS evaluates the certification test.

Our recertification report described three aspects of the recertification
process that caused problems for taxpayers. Specifically,

o 	one form used for recertification was of questionable value to IRS and
another form was potentially confusing to taxpayers;

o 	taxpayers were asked to submit information that was difficult for them
to obtain or inconsistent with what many IRS examiners considered
acceptable; and

o 	IRS examiners' inconsistent assessment of documentation submitted by
taxpayers could result in different recertification decisions for
taxpayers in similar circumstances.

IRS has taken steps to deal with all of these concerns in designing the
certification process. Regarding the problems with recertification forms,
the form that was of questionable value to IRS, which was essentially a
means for taxpayers to tell IRS that they wished to be considered for

19GAO-02-449.

recertification, is not applicable to the certification program. The other
recertification form told taxpayers what they had to submit to establish
their eligibility for the EIC. We found that this form could confuse
taxpayers into believing they had to show that a qualifying child was also
their dependent, a criterion not applicable to EIC eligibility. We also
found that the form provided insufficient guidance to taxpayers on what
information they needed to provide to prove that qualifying children met
the EIC eligibility requirements. We made several recommendations,
including that IRS should clarify taxpayers do not need to demonstrate
that qualifying children are also dependents, help taxpayers better
understand what documentation they need to provide to establish their
relationship with any qualifying children, eliminate a requirement that
statements from child care providers be notarized, and encourage taxpayers
to submit more than one type of documentation.

Regarding our concerns about taxpayers who were recertifying being asked
to submit documentation that was difficult for them to obtain and that tax
examiners did not all accept, we found, for example, that EIC taxpayers'
living arrangements could make providing various documents difficult. We
also found taxpayers did not always understand that school records were
for a calendar year and therefore needed to cover the spring and fall of
separate school years. We also found situations in which IRS examiners
would not accept a document even though the recertification form listed
the document as being acceptable. This concern overlapped with our finding
that IRS examiners' were inconsistent in their assessment of whether
documentation provided by the taxpayers was sufficient to establish their
qualifications for the EIC.

Regarding our concerns about inconsistent documentation, IRS again took
actions intended to deal with our concerns in developing the certification
program. By introducing a new option-obtaining an affidavit affirming that
a qualifying child resided with the taxpayer for more than half the
year-IRS intended to give taxpayers another means of showing that the
residency requirement is met, which would prevent the taxpayer having to
obtain the other types of documents that the draft certification form
lists. Although IRS did not, as we recommended, create a new form to be
used by taxpayers when seeking school records, it did follow our
alternative recommendation that IRS clearly remind taxpayers that they
need records for part of 2 school years. The information is included in
the certification form's instructions, which contain an example where a
taxpayer must provide records from 2 school years. Finally, by
centralizing the EIC certification processing in one location-Kansas
City-and providing

training to those who will be involved, IRS is seeking to ensure a higher
level of consistency in how tax examiners judge whether taxpayers
adequately establish that qualifying children meet the residency criterion
for EIC.

Whether the manner in which IRS took our recertification recommendations
into account when designing the certification program will be successful
will not be known until IRS evaluates the certification test.

Qualifying Child Certification Program Developed to Improve Compliance
While Considering Taxpayers' Burden, but Plan for Evaluating Test Is
Incomplete

The certification program appears to be adequately developed to
potentially improve EIC compliance with consideration for minimizing
taxpayer burden so that testing should proceed, particularly in light of
IRS's recent announcement further delaying the program's start and
reducing the sample size. For example, the EIC task force and IRS have
taken steps that directly minimized the number of taxpayers who will be
burdened by the certification program. That is, the certification proposal
is based on analyses of the leading sources of EIC errors detected in
earlier studies, thus focusing attention and burden on the subset of
taxpayers making those errors, as opposed to all EIC recipients. In
addition, IRS has taken steps to address the burden taxpayers will
experience as participants in the certification test this year.

Although IRS has made and is continuing to make progress in defining its
plan to evaluate the certification test, the plan is incomplete. For
example, the draft plan does not indicate how and when some information
that will be needed to evaluate whether certification achieves its
objectives will be obtained and analyzed. However, officials recognize the
draft plan needs to be further developed and the importance of doing so
quickly.

Initial Design and Subsequent Changes May Improve EIC Compliance and Have
Helped to Minimize Burden

In initially designing and subsequently modifying the EIC certification
program, officials took into account the burden that taxpayers may
experience while attempting to improve compliance. Officials designed the
program to include, and thus burden, only the taxpayers most likely to
make the errors that contribute most to the EIC's overclaim rate. By
focusing on these noncompliant taxpayers, IRS expects to improve EIC
compliance. In addition, officials took a number of steps, such as
obtaining input from external and internal stakeholders that resulted in
changes and delaying the program while considering comments received
during the

comment period, which should reduce the burden on those taxpayers who are
identified to certify.

To help improve compliance, the task force focused on known sources of
noncompliance including claiming nonqualifying children, filing status,
and misreporting income. To deal with errors attributable to claiming
nonqualifying children, the task force proposed a program for certifying
the eligibility for qualifying children and envisioned targeting taxpayers
most likely to make those errors. In contrast, other benefit programs that
we reviewed generally require all applicants to provide documentation
before receiving assistance. For example, to receive Supplemental Security
Income, an individual must visit a Social Security office, meet with a
representative, and provide documentation including birth certificates and
payroll information. The Social Security Administration then matches this
information to determine eligibility in advance of benefits being
received. IRS's certification effort, even if fully implemented, would
require only a subset of all EIC taxpayers to provide documentation to
support their eligibility and only when IRS is unable to verify
eligibility from other sources of information.

After the proposal was formally adopted, IRS took a number of steps in
developing plans for implementation that have been intended at least in
part to minimize the burden that taxpayers actually asked to certify would
experience, including the following.

o 	IRS has undertaken more activities than usual to ensure the residency
form and other explanatory documents related to the certification program
have been reviewed by those who would use them. 20 IRS sought feedback
from focus groups and stakeholders on various aspects of the certification
test and the draft letter, form, and instruction proposed for the
residency test, such as whether taxpayers will be able to obtain and
provide documents within the time available, and made some changes to the
proposed form due to that feedback. As previously described, IRS held
focus groups with taxpayers, paid tax preparers, and other parties to
obtain feedback on certification. Officials also interviewed a small
number of third parties who would be called upon

20 IRS generally tests few forms and instructions with taxpayers before
using them. See U.S. General Accounting Office, Tax Administration: IRS
Should Reassess the Level of Resources for Testing Forms and Instructions,
GAO-03-486 (Washington, D.C.: Apr. 11, 2003).

to provide requested documents. IRS also held a 30-day open period to
receive comments from any interested party and expects to revise
certification materials due to comments received. Finally, IRS officials
say they will again revisit, among other things, the appropriateness of
the forms and explanations going to taxpayers after evaluating the results
of this year's test of certification.

o 	IRS considered the issues we raised in our report about the
recertification program21 when planning for certification. For example, as
discussed previously in this report, IRS developed a standard form that
includes an affidavit, which taxpayers can provide to third parties, such
as an employer, as an alternative to obtaining other documents to prove
residency. We also noted in our report that examiners inconsistently
accepted or declined supporting documentation for recertification
purposes. To address this concern, IRS officials conducted special
training and have all certification examiners in one location, Kansas
City, where EIC claims will be processed.

o 	IRS provided taxpayers with a variety of documentation choices in order
to prove eligibility for their qualifying children. To certify for
residency, taxpayers will need to provide Form 8836, "Qualifying Children
Residency Statement," with one or more of the following supporting
documents:

o 	school records, medical records, day care provider records, leases, or
social service agency records that show the parent's name and the child's
name and address, and the dates that the child lived with the parent; or

o 	a letter on official letterhead for a qualifying child from the child's
school, health care provider, landlord, or member of the clergy that shows
the parent's name and the child's name and address, and dates that the
child lived with the parent; or

o 	a third party affidavit from a clergy member, community-based
organization official, health care provider, landlord or property manager,
school official, or day-care provider.

21 GAO-02-449.

o 	IRS dropped for an undetermined period of time, its plan to ask
taxpayers to certify their relationship to qualifying children. IRS
officials do not know whether or if they will test certification of
relationships in the future. Various external stakeholders had expressed
concerns about whether taxpayers would be able to provide the type of
documents, such as marriage and birth certificates, which IRS had planned
to request to document relationships to qualifying children on time. Also,
IRS officials said that the relationship portion of the program was
dropped for other reasons, including that (1) studies that have shown
relationship requirements to be less of a compliance issue than residency
and (2) taxpayers found to be noncompliant because of relationship
requirements often were also noncompliant due to residency errors. As a
result, certification will only include residency this year.

As part of its effort to balance burden with ensuring compliance, IRS made
the changes listed above. As we drafted this report, it had not yet
determined what additional changes it would make to the forms on the basis
of comments received during the 30-day public comment period, but
officials said more changes will likely result. As previously discussed,
the initial design of the residency form was responsive to concerns we
raised in our earlier report on IRS's recertification program. Additional
changes, especially dropping relationship recertification, were responsive
to the concerns that stakeholders raised before the public comment period.
Accordingly, the current draft residency certification form addresses many
burden concerns.

IRS's Plan for Evaluating the Test Is Incomplete

Although IRS has made and is continuing to make progress in defining its
plan to evaluate the certification test, the plan is not yet complete.
From its inception, the certification program was intended to: (1) reduce
the EIC's overclaim rate, (2) minimize burden on taxpayers and (3)
maintain the EIC's relatively high participation rate. Although there are
many ways to organize an evaluation, determining whether the major
objectives of a program are accomplished should help policymakers
determine whether and how to proceed with the program. The draft plan is
not explicitly organized to show whether certification's objectives are
achieved, but does present some information on how IRS would evaluate
these objectives. However, the plan proposed potential options for
identifying how and when some critical data will be obtained and analyzed,
but does not provide further details on when decisions will be made on
specific data that will be collected, how, and by whom. Officials
recognize that the draft plan

needs to be further developed and the importance of doing so quickly. They
have, for instance, developed preliminary drafts identifying additional
data needed and have begun considering how to use contractors to gather
the data. Because evaluating these objectives will depend in part on
actions of EIC certification participants that will now occur as part of
next year's filing season, IRS appears to have some time before it must
make final decisions on how it will determine whether the objectives were
met.

Although an evaluation plan may not have to completely identify all issues
that need to be evaluated and precisely how they will be evaluated before
a program begins, the more completely such a plan is developed before a
program is implemented, the more likely that the evaluation will be
sufficient to support future decisions. For example, identifying key
questions that need to be answered before a project's implementation
increases the chances that necessary data will be collected to answer
those questions. IRS's Internal Revenue Manual22 recognizes the
desirability of having evaluation plans in place before a project is
implemented. For instance, it requires such plans before reorganizations.

IRS has been preparing an evaluation plan for the certification test and
has a draft plan, dated April 22, 2003. That draft describes how IRS
expects to evaluate the program and the process IRS used to select
taxpayers for the test.

The draft plan identifies one "threshold" question for evaluating the
certification program: whether the claimant selected for the test provided
the required information to allow IRS to determine the eligibility of
qualifying children regardless of whether the claimant was determined to
be eligible or not. The plan lists data that are to be gathered throughout
the certification program to answer this question.

The threshold question is part of what must be included in determining
whether certification for residency helps lower EIC overclaims, but
additional information is needed. Although not tying methodologies or
planned data collection specifically to whether certification lowers the
EIC overclaim rate, the draft plan has a combination of approaches that
should contribute to answering this question. For example, the draft plan
identified data that IRS would gather throughout the test on how many
taxpayers in the test sample certify or fail to do so in advance of the
filing

22See Internal Revenue Manual 1.1.4 and 1.2.1.

season as well as how many prove that children meet the EIC residency test
during the filing season.23

The extent to which certification may reduce overclaims due to qualifying
children not meeting the residency requirement, however, will depend
significantly on why some taxpayers will not attempt to certify and why
some will fail to claim the EIC, or as much EIC, for tax year 2003 as they
did for 2002. The draft plan takes into account that some taxpayers may
receive the certification materials, determine that a child does not meet
the residency test, and therefore not attempt to certify and not claim the
EIC with their 2003 tax return or claim EIC on another basis. IRS expects
to use available data to help assess whether the taxpayer had a filing
requirement and whether the taxpayer may have been eligible to claim the
EIC (e.g., did the taxpayer's income fall within the appropriate range?).
In addition, the draft plan proposes that IRS use a contractor or another
third party to gather information from taxpayers about why they did not
claim the EIC. A subsequent document cites ideas for the types of
questions that could be asked.

Regarding the burden certification imposes on taxpayers who participate,
the plan is not organized to show how this will be evaluated, but it
recognizes that participants' burden should be evaluated. For example, the
data IRS plans to collect throughout the process will have some utility in
answering this question. IRS will keep track of the number of
communications back and forth between IRS and the taxpayer before a tax
examiner makes a final certification decision. IRS's plan also recognizes
that some information on burden will need to be collected directly from
taxpayers. The plan includes a general description of a potential opinion
survey that would gather burden-related information from certification
participants. Little or no detail is provided on how taxpayers would be
selected for such a survey, what types of questions would be asked, and
when the survey would be done; however, some of this information is shown
in a subsequent draft document. Because taxpayers will not have completed
their certification experience until sometime next filing season, IRS has
some time to decide whether to do such a survey and how to define its
parameters.

23IRS's draft evaluation plan preceded its decision to test certification
during the 2004 filing season. These data will now need to be collected
during the filing season.

Regarding the objective of maintaining the EIC's relatively high
participation rate, the draft plan proposes to obtain information from
those taxpayers who are asked to certify, do not, and then fail to claim
the EIC. The plan proposes to use a contractor or other third party to
gather information from these taxpayers about why they did not claim the
EIC. The plan does not amply describe how and when final decisions would
be made about selecting contractors or another third party to do this,
when the contractor would contact taxpayers, or what data they would
attempt to obtain from the taxpayers. Because the population IRS will need
to contact for these surveys will not be known until during and after the
spring of 2004, IRS has a number of months to further develop and
implement an approach.

Recognizing that its plans need to be further developed, IRS officials
have continued to explore how the evaluation will be done. For example,
officials have drafted ideas for the type of survey questions a contractor
or other party would ask of EIC taxpayers to help IRS assess why taxpayers
take, or do not take, various actions (such as why they may stop claiming
the EIC after being asked to certify eligibility) and to assess taxpayers'
experiences under the certification program, including the burdens they
experience. In addition, officials have begun identifying potential
contractors who would perform the surveys and considering contracting
options. According to IRS officials and documents, some discussions have
been held with potential contractors to gain a better understanding of
ways to test the survey instruments, techniques available to ensure the
best possible response rate, and the number of taxpayers needing to be
contacted to have useful results.

Finally, because IRS would like to undertake some version of the
qualifying child program next year, possibly including certification
during the latter part of 2004, the timely production of evaluative data
for this year's test will be critical for supporting decisions about what
form future efforts will take. IRS is aware of the tight schedule.
Officials note that while they will not have complete information on which
to base some decisions about whether and how to continue with
implementation in 2004 before those decisions must be made, they expect to
have preliminary data in a timely fashion. For example, IRS will not be
able to completely answer whether, and if so, why, taxpayers who are
legitimately qualified to receive the EIC do not claim it when they file
their 2003 tax return until the end of the 2003 tax filing season, or
later if taxpayers request a filing extension. IRS does expect that its
contractor will have contacted many, if not most, of the taxpayers who
file returns before the end of the filing season and do not

claim EIC. Thus, IRS expects to know during the fall of 2004 why many
taxpayers in the certification test stop claiming the EIC.

Some Implementation Issues Not Reviewed

We did not evaluate some implementation issues because they were outside
the scope of our review, still under development, or the Treasury
Inspector General for Tax Administration had audits planned in these
areas. Nonetheless, implementation issues could affect whether IRS is able
to fully implement the certification test and ultimately improve
compliance. We did not assess (1) whether IRS assigned an appropriate
number of staff to assist taxpayers with questions and process the forms
and documents relating to certification, (2) the adequacy of training
materials for staff or the procedures put in place to help examiners
consistently accept or decline taxpayers' supporting documentation, (3)
the design or reliability of the databases that will be used to capture
and evaluate program information, and (4) supporting tools, which
examiners will use to do their job.

IRS has developed broad plans for processing the certification workload.
Officials identified about 30 different offices that will be affected by
the new certification program. As a key part of its processing strategy,
IRS plans to dedicate employees at its Kansas City campus to process
cases, answer a special toll-free number, and make updates to a
certification database based on responses from the test of the 25,000
taxpayers. The Kansas City site will have about 180 staff, the bulk of
whom will come on-board between September and December 2003. Approximately
40 staff took initial training between April and June 2003.

Conclusions	Given the persistently high EIC overclaim rates, that the
certification program is a test, and that IRS has taken key steps to
address burden issues and focus the test on individuals least likely to
meet the qualifying child residency requirements, we believe IRS has
struck a reasonable balance between preventing unreasonable burden on EIC
taxpayers and balancing the need to obtain information on whether
certification can be a useful approach to improving EIC compliance. In
addition, with the recent program changes announced in August, it appears
that IRS is taking even more steps to be mindful of these concerns.
Although certification during the 2004 filing season gives IRS somewhat
more time to modify the forms and take other actions to potentially
further reduce the burden on taxpayers subject to the test, it also
creates new challenges for IRS. The

test will no longer be a direct test of the original concept of certifying
taxpayer eligibility in advance of the filing season. Instead, testing
will occur during the filing season-IRS's busiest time of year-and gives
IRS only indirect evidence on how well certification may work before the
filing season as originally envisioned. Further, because IRS currently
plans for taxpayers to have to successfully provide proof of eligibility
when they file their individual income tax return or have a refund frozen
until they do, a greater portion of the taxpayers chosen for the test may
have their refunds delayed than if certification had been done before the
filing season. Finally, like virtually all aspects of the qualifying child
certification program, IRS's future plans have yet to be determined and
are largely dependent of the results and subsequent evaluations of this
test.

For various reasons, we did not review in detail some implementation
issues, such as staffing and procedures for handling taxpayer responses,
which could affect whether IRS is able to successfully implement the
certification test. Thus, our opinion on whether IRS is ready to proceed
is based only on whether it has adequately developed the test to prevent
unreasonable burden and to improve compliance.

Although the balance IRS has struck supports proceeding with the test,
IRS's plan for evaluating the certification program test is incomplete.
IRS recognizes the need to evaluate the test and is developing its plan to
do so. For some key test objectives, IRS has preliminarily identified some
data that it believes must be collected to determine whether
certification's objectives are achieved and has broadly identified when
and how that information will be collected. Because the data are related
to taxpayers' actions that will occur later this year or next spring, IRS
appears to have some time to finalize its evaluation plan.

Recommendations for Executive Action

Given that the qualifying child certification program is a key part of
IRS's plans for reducing EIC overclaims and that certification is intended
to help reduce overclaims while minimizing the burden on taxpayers and
maintaining the EIC's participation rate, the Commissioner of Internal
Revenue should, to the extent possible, accelerate development of the
evaluation plan for the test. The plan should demonstrate how each of the
certification's objectives will be evaluated, including milestones for
such critical steps as defining the specific data that will be collected,
who will collect the data, and how the data will be analyzed in time to
support decisions about the future of the program.

Agency Comments and Our Evaluation

While not explicitly agreeing with our recommendation, in his September
22, 2003, letter, the Commissioner of Internal Revenue said that IRS would
be including the components we suggested in their evaluation plan and said
that IRS is working to incorporate these components well before the
certification test begins. The Commissioner said that our discussion of
the evaluation plan is essentially accurate, but provided an enclosure to
his letter that noted supplemental information on the plan. We are aware
of the information described in the enclosure to the Commissioner's
letter, and considered it when drafting our report.

The Commissioner also raised concerns about the comparability of EIC error
rates to the error rates in taxpayers' reporting of certain types of
income. We concurred that, by-and-large, the compliance data on reporting
of these types of income are not comparable to the EIC error rate. As a
result, we no longer show those comparisons in our final report.

We are sending copies of this report to the Chairmen and Ranking Minority
Members of the Senate Committee on Finance and the House Committee on Ways
and Means. We are also sending copies to the Secretary of the Treasury;
the Commissioner of Internal Revenue; the Director, Office of Management
and Budget; and other interested parties. We will make copies available to
others on request. In addition, the report will be available at no charge
on the GAO Web site at http://www.gao.gov.

This report was prepared under the direction of Joanna Stamatiades,
Assistant Director. Other major contributors are acknowledged in appendix
VII. If you have any questions about this report, contact Ms. Stamatiades
at (404) 679-1900 or me on (202) 512-9110.

Michael Brostek Director, Tax Issues

Appendix I

Significant Noncompliance Rates Other Than for the EIC

The IRS has compliance data on some taxpayer groups such as individuals
and small businesses and some tax items such as income and credits.
By-and-large, the compliance data IRS currently has are not comparable to
the EIC. IRS is implementing its National Research Program (NRP), which
will provide new compliance data in 2004. In the meantime, IRS is using
its Strategic Planning and Performance Management process to prioritize
compliance issues.

The compliance data that IRS has available for some taxpayer groups and
tax items are largely based on the Taxpayer Compliance Measurement Program
(TCMP),1 which was last conducted in 1988. However, these data cannot be
compared to the EIC overclaim or error rates, in part because these data
are 15 or more years old and reliable inferences cannot be drawn because
much of the tax system and the economy have changed during that time. In
addition, the methods used to calculate compliance rates for TCMP are
different than those used to calculate EIC.

In late 2002, IRS began implementing its new NRP, a detailed study of
individual taxpayers' compliance.2 As part of NRP, IRS has identified a
random sample of approximately 47,000 returns from tax year 2001 and is in
the process of verifying the information on the returns through reviews of
IRS and third-party data. Where necessary to confirm the accuracy of
taxpayer-reported information, IRS is conducting either correspondence or
face-to-face examinations. IRS intends to conduct additional NRP reviews
of additional types of taxpayers, such as small corporations, and use the
NRP periodically to measure compliance of individual taxpayers.

The NRP sample of 47,000 returns includes about 7,300 EIC returns. These
EIC returns are subject to the same processes as the other returns in the
sample, and will include a review of the taxpayers' eligibility for the
EIC. In order to determine whether the NRP review of these returns will
yield results methodologically similar to those of the 1999 EIC compliance
study, IRS is also comparing the results of the 1999 compliance study with
NRP by putting a sample of returns from the 1999 study through NRP
processes

1For almost 30 years, the Taxpayer Compliance Measurement Program was
IRS's method for statistically estimating the voluntary compliance of
taxpayers in reporting their tax obligations.

2Among other things, IRS plans to use NRP data to update the formulas used
to select tax returns for examination and allow it to design programs that
will help taxpayers comply with tax laws.

Appendix I
Significant Noncompliance Rates Other Than
for the EIC

(not including examinations). According to IRS officials, this should
allow them to see the impact of the methodological differences between the
compliance study and NRP review. IRS expects the results of the comparison
study by September 2003. IRS plans to have preliminary NRP results in May
2004 and final results in November 2004.

Until better compliance measurement data are available, IRS's
organizational divisions use the Strategic Planning Budgeting and
Performance Management process to prioritize the compliance problems IRS
faces. Through this process, IRS says that it (1) identifies and explores
critical trends, issues, and problems, (2) develops operational priorities
and improvement projects to address existing or emerging problems, (3)
explores drivers of program resources in order to develop resource
allocation targets for carrying out the proposed strategies, and (4)
enables division commissioners and the senior leadership teams to
prioritize the strategies and projects and determine the resource
requirements to apply to each strategy, operational priority, and
improvement project.

Based on managers' judgments made during this process, the Small
Business/Self Employed (SB/SE) operating division, for example, identified
its top six compliance priorities for fiscal year 2003 and 2004:

o  high income nonfilers (income greater than $100,000),

o  abusive offshore financial transactions,

o  promoter investigations (those selling tax schemes to others),

o  abusive tax avoidance transactions,

o  high income taxpayers (income greater than $1 million), and

o  returns with a high probability of unreported income.

SB/SE, which conducts few examinations of EIC claims, did not consider EIC
in this prioritization exercise since EIC has its own dedicated
appropriation.3 Because IRS used different means to identify and
prioritize these potentially noncompliant taxpayer groups, their
identification as

3Personnel in IRS's Wage and Investment operating division conduct the
majority of EIC examinations.

Appendix I
Significant Noncompliance Rates Other Than
for the EIC

SB/SE priorities does not mean their noncompliance rate is comparable to
noncompliance rates established for EIC or rates, which will be determined
through the current or future NRPs or other EIC compliance studies.

Appendix II

Overclaim Rates, Administrative Costs, and Eligibility Verification
Processes of Benefit Programs

In addition to the data we complied on IRS's EIC and qualifying child
certification program, we also compiled overclaim rate and administrative
cost data, as well as information on the eligibility verification
processes, for nine other federal or state benefit programs. We selected
the nine benefit programs because each requires some type of certification
for benefits, similar to the EIC, and because the EIC task force reviewed
the same programs. We did not do a comprehensive analysis to determine
which programs, if any, are most comparable to the EIC, nor did we
determine whether the information reported is comparable across programs.

The overclaim rates, administrative costs, and the eligibility
verification processes for the EIC and the other nine benefit
programs--Unemployment Insurance, Supplemental Security Income, Social
Security Disability Insurance, the National School Lunch program, the Food
Stamp program, Housing and Urban Development rental assistance, Medicaid,
Medicare, and Temporary Assistance for Needy Families--are shown in table
4.

Overclaim rates for programs other than the EIC for which data were
available ranged from 0.2 to 10.7 percent. These overclaim rates reflect
the percentage of total dollars paid out in error, not, for example, the
percentage of claimants who made errors.1 To calculate the overclaim
rates, most of the nine agencies selected a sample of program participants
and conducted a detailed analysis of the cases. This can involve
collection of additional supporting documentation, personal contacts with
employers and other third parties, or home visits to program recipients. A
description of how the overclaim rates were calculated is in table 5.

Administrative costs range from $123 million to $11.9 billion for the nine
programs. Administrative costs reported by federal agencies are likely not
comparable across programs and may not include all of the costs involved
in administering the programs. For example, various agencies and entities
at the federal, state, and local levels have administrative
responsibilities under the National School Lunch program. However, while
the federal budget provides funds separate from program dollars to pay for
administrative processes at the federal and state level, officials at the
local

1We use the term overclaim rate to be the total excess payments made in
error. This is generally referred to as the error or overpayment rate in
the other programs we reviewed.

                                  Appendix II
                   Overclaim Rates, Administrative Costs, and
                 Eligibility Verification Processes of Benefit
                                    Programs

level pay for administrative costs from program dollars that include
federal and state funding and student meal payments.2

The process used to determine and validate eligibility varies
significantly. Some programs, such as the school lunch program, rely
primarily on self-reported information and verification is limited. Other
programs, such as the Food Stamp program, require program staff to conduct
extensive verification.

Table 4: Overclaim Rates, Administrative Costs, and Eligibility
Verification Processes for EIC and Other Programs

Program Overclaim ratea Administrative cost Eligibility verification
process
Earned Income Credit 27-32 percent $145 millionb Currently, no
certification requirement exists. An EIC claimant's

                        (tax year 1999) (tax year 1999)

return is selected for examination when it meets certain selection
criteria that points to potential overclaim. When this occurs,
documentation is generally requested to establish requirements for EIC and
related issues using document request forms including for qualifying
child, filing status, and dependency issues.

Unemployment  8 percentc  $2.3 billiond       Eligibility for unemployment 
                                             insurance benefits is determined 
                (fiscal year  (fiscal year            under state law, so the 
    Insurance          2001)     2001)             information applicants are 
                                                                  required to 
                                            provide varies by state. States   
                                            rely heavily on self-reported     
                                            information; in some states       
                                            applicants can apply over the     
                                            telephone or on-line. Only a      
                                            limited number of states          
                                            independently verify claimants'   
                                            identity by using the Social      
                                              Security Administration's State 
                                               Online Query system, which can 
                                            match a claimant's name, date of  
                                            birth, and Social Security        
                                                       number. States may use 
                                                   independent automated data 
                                                                      sources 
                                                  to verify other eligibility 
                                                    factors such as wages and 
                                                                   employment 
                                                         status.              

2See, for example, U.S. General Accounting Office, School Meal Programs:
Estimated Costs for Three Administrative Processes at Selected Locations,
GAO-02-944 (Washington, D.C.: Sept. 25, 2002).

                                  Appendix II
                   Overclaim Rates, Administrative Costs, and
                 Eligibility Verification Processes of Benefit
                                    Programs

                         (Continued From Previous Page)

  Program Overclaim ratea Administrative cost Eligibility verification process

Supplemental 7.2 percente $2.8 billione Security Income (fiscal year 2001)
(fiscal year 2002) Supplemental Security Income is available to
individuals who are blind or disabled and poor. The amount of Supplemental
Security Income an individual receives depends on several factors
including, but not limited to, other sources of income and living
arrangements. To apply for Supplemental Security Income, applicants must
visit a Social Security office and meet with a Social Security
representative. Examples of documentation required under the program
include: (1) social security card; (2) birth certificate or other proof of
age; (3) mortgage or lease; (4) payroll slips, bank books, insurance
policies, and other records about income and assets; (5) names and
addresses of doctors (if disabled); and (6) proof of United States
citizenship or noncitizen status. To verify this information, the Social
Security Administration uses computer matches to compare Supplemental
Security Income records against recipient information contained in records
of third parties, such as other federal and state government agencies. The
Social Security Administration periodically conducts "redetermination"
reviews to verify eligibility factors such as income, resources, and
living arrangements. Recipients' eligibility is to be reviewed at least
every 6 years.

                                                 Social Security Disability   
Social Security  0.2 percente  $2.0 billione  Income provides income       
                                                 support                      
      Disability    (fiscal year   (fiscal year  benefits to former workers   
      Insurance     2001)             2001)      who have suffered a          
                                                 long-term                    
                                                   disability. Applicants can 
                                                      start their application 
                                                           online, but before 
                                                 they are approved for        
                                                 benefits, applicants must    
                                                 provide the                  
                                                 following documentation to a 
                                                  social security office: (1) 
                                                                     a Social 
                                                 Security number; (2) birth   
                                                 certificate or other proof   
                                                 of age;                      
                                                 (3) medical information,     
                                                 such as names and addresses  
                                                 of                           
                                                 doctors and medical records; 
                                                     (4) work history for the 
                                                                     prior 15 
                                                 years; and (5) a W-2 or tax  
                                                           return.            

School Lunch Not known $123 millionf   Households submit applications with 
                                                   self-reported information, 
                           (fiscal year      including the names of household 
                              2002)                members and all sources of 
                                         income for each household member,    
                                         and the Social Security              
                                         number of the adult that signs the   
                                         application. Alternatively,          
                                           children with Temporary Assistance 
                                                   for Needy Families or Food 
                                         Stamp case number can be certified   
                                         directly. Generally, no              
                                          documentation or additional support 
                                                  is requested at the time of 
                                         application. Local authorities       
                                         verify eligibility for free and      
                                          reduced price meals for a sample of 
                                                    applications using either 
                                              random or focused sampling      
                                                     techniques.              

                                  Appendix II
                   Overclaim Rates, Administrative Costs, and
                 Eligibility Verification Processes of Benefit
                                    Programs

                         (Continued From Previous Page)

  Program Overclaim ratea Administrative cost Eligibility verification process

Food Stamps	8.66 percentg $2.4 billionh (fiscal year 2001) (fiscal year
2002) Food Stamp applicants are asked to provide documentary evidence of
household assets, income, and allowable deductions, as well as proof of
noncitizen status. Allowable deductions may include housing and utility
expenses, medical expenses, and child support payments. If documentary
evidence is not available, state agencies may use other means of verifying
information provided by applicants, including contacting third parties.
State agencies are required to verify certain information on income and
deductions and states may opt to require verification of additional
information. State agencies may establish their own standards for the use
of verification subject to the parameters specified in federal
regulations.

                                                      Applicants must provide 
Housing and Urban  10.7 percenti  $945 millionj   third party verification 
                                                             of the following 
                      (fiscal year    (fiscal year  factors: (1) family       
Development rental 2001)              2001)      annual income, (2) value  
                                                    of assets, (3)            
       assistance                                   expenses related to       
        programs                                    deductions from annual    
                                                    income, and               
                                                    (4) other factors that    
                                                    affect the determination  
                                                    of adjusted               
                                                     income. Housing agencies 
                                                    may require documentation 
                                                                      or they 
                                                     may verify self-reported 
                                                    information by telephone. 
                                                                   Individual 
                                                             housing agencies 
                                                          determine their own 
                                                     verification procedures. 

Medicaid Not known	11.9 billionk (fiscal year 2001) Medicaid provides
health insurance coverage to certain low income adults and children in a
program jointly administered and funded by the federal government and the
states. While the Social Security Administration sets the income threshold
for Supplemental Security Income-related Medicaid eligibility annually,
most Medicaid eligibility requirements and verification procedures are
determined at the state level and vary by state. For example, some states
require applicants to provide documentation of both income and allowable
deductions, such as child care expenses and child support payments. Other
states require applicants to provide information on income and deductions
but do not require further documentation. Federal law requires that
Medicaid applicants provide a Social Security number, unless the applicant
refuses to obtain a number because of well established religious
objections. While states are required to establish procedures for the
periodic redetermination of a recipient's Medicaid eligibility at least
annually, the procedures may vary by state.

Medicare	6.3 percentl 4.4 billionm (fiscal year 2002) (fiscal year 2001)
The Centers for Medicare and Medicaid Services administers the Medicare
program; however, the Social Security Administration determines
entitlement to Medicare benefits. Retired recipients who are receiving
retirement benefits from Social Security or the Railroad Retirement Board
are automatically enrolled in Medicare the month they turn 65. Individuals
who are under age 65 and disabled are also automatically enrolled in
Medicare after they have been receiving Social Security or Railroad
Retirement disability payments for 24 months. All other individuals (e.g.,
retired individuals who are not eligible for Social Security and
individuals who have end stage renal disease) are required to file a
Medicare application at a Social Security office.

                                  Appendix II
                   Overclaim Rates, Administrative Costs, and
                 Eligibility Verification Processes of Benefit
                                    Programs

                         (Continued From Previous Page)

Program Overclaim ratea Administrative cost Eligibility verification
process

Temporary Not yet Available $2.3 billionn Eligibility requirements are
determined at the state or local level,
Assistance for Needy and therefore vary by state and locality.
Families

Source: GAO analysis of multiple agency data.

aSee table 6 for information on how overclaim rates were calculated for
each program.

bIRS does not know the full cost of administering the EIC; for example,
the $145 million does not include the cost of processing EIC returns.

cUnited States Department of Labor Benefit Accuracy Measurement data.

dUnited States Department of Labor.

eSocial Security Administration, Performance and Accountability Report,
fiscal year 2002.

fUnited States Department of Agriculture Food and Nutrition Service.
Includes state expenses for administering the School Breakfast Program,
the Child and Adult Care Food Program, and the Special Milk Program.

gUnited States Department of Agriculture Food and Nutrition Service,
Quality Control Division.

hUnited States Department of Agriculture Food and Nutrition Service.

iHousing and Urban Development's Audit of Financial Statements Fiscal Year
2002 and 2001, 2003-Fiscal Year 2004 (January 31, 2001), Note 17, p.
104ff.

jDepartments of Veterans Affairs and Housing and Urban Development, and
Independent Agencies Appropriations for 2003, Hearings. 552-070-28357-9,
p.106.

kHealth and Human Services' Financial Management Report, fiscal year 2001.

lHealth and Human Services' Office of Inspector General, Improper Fiscal
Year 2002 Medicare Fee-For-Service Payments A-17-02-0220, (January 2003).

mThe Boards of Trustees, Federal Hospital Insurance and Federal
Supplementary Medical Insurance Trust Funds, 2003 Annual Report of the
Boards of Trustees of the Federal Hospital Insurance and Federal
Supplementary Medical Insurance Trust Funds. Administrative cost figure
includes administrative costs for both Federal Hospital Insurance and
Federal Supplementary Medical Insurance Trust Funds. Administrative costs
for Hospital Insurance include the costs of experiments and demonstration
projects as well as fraud and abuse control expenses.

nHealth and Human Services' Administration for Children and Families,
Temporary Assistance for Needy Families Program Fifth Annual Report to
Congress. (February 2003).

To the extent known, how the overclaim rates are calculated, for the nine
benefits program we reviewed, including EIC, is shown in table 5.

                                  Appendix II
                   Overclaim Rates, Administrative Costs, and
                 Eligibility Verification Processes of Benefit
                                    Programs

Table 5: How Overclaim Rates Are Calculated for Selected Benefit Programs

                   Program How overclaim rates are calculated

Earned Income Credit	Compliance study based on audits of a statistically
representative sample of taxpayers and adjustments made to their tax
returns.

Unemployment Insurance	The quality assurance system is used to estimate
overpayments based on a statistically valid sample of Unemployment
Insurance claims from each state. Investigators conduct detailed,
comprehensive analyses of each case by personally contacting employers,
claimants, and third parties. Investigators typically spend 5 to 8 hours
examining each case.

Supplemental Security Income	The payment accuracy rate is based on a
detailed analysis of a sample of Supplemental Security Income cases.
However, the Social Security Administration's Office of Inspector General
reported that not all types of overpayments are counted as errors so
payment accuracy rates do not correspond to overpayments reported in the
Social Security Administration's financial statements.

Social Security Disability Insurance 	The overpayment overclaim rate is
based on a monthly sample selection from the payment rolls consisting of
beneficiaries in current payment status. For each sampled case, the
recipient or representative payee is interviewed, collateral contacts are
made as needed, and all factors of eligibility are redeveloped as of the
current sample month.

School Lunch	The United States Department of Agriculture does not
routinely collect data on the percentage of ineligible children receiving
free or reduced price school lunches.

Food Stamps	Under the food stamp quality control system, states draw a
statistical sample, review case information, and make home visits to
determine whether households were eligible for benefits and received the
correct benefit payment. Regional offices validate the results by
reviewing a subset of each state's overpayment and underpayment errors as
necessary.

Housing and Urban Development rental The overclaim rate is based on an
in-depth analysis of a statistical sample of cases. The

assistance programs	overclaim rate showed an increase in 2001 over
previous years largely because Housing and Urban Development expanded its
methodology for measuring error to cover three types of program errors -
incorrect reporting of income by tenants; mistakes by public housing
agencies, owners, and renting agents in calculating income and rent
amounts; and mistakes made by public housing agencies, owners, and renting
agents in completing appropriate paperwork and billing Housing and Urban
Development for rental assistance.

a

Medicaid GAO has found that few states measure the overall accuracy of
their payments.

Medicare	This overclaim rate for Medicare fee-for-service claims is based
on a review of claims conducted by the Housing and Human Service's
Inspector General's office. The Inspector General used a multistage
stratified sample design. For each claim, the provider was contacted and
asked to provide copies of all medical records supporting services billed.
These records were assessed to determine whether the services billed were
reasonable, adequately documented, medically necessary, and coded in
accordance with Medicare reimbursement rules and regulations. Billing
practices were also reviewed. Starting in fiscal year 2003, the Centers
for Medicare and Medicaid Services will publish a national overclaim rate
developed through the Comprehensive Error Rate Testing Program and the
Hospital Payment Monitoring Program. These programs build on the Inspector
General's methodology.

Temporary Assistance for Needy Families	Payment accuracy overclaim rates
have not yet been calculated for the Temporary Assistance for Needy
Families' program.

Source: GAO analysis of multiple agency data.

aU.S. General Accounting Office State, Efforts to Control Improper
Payments Vary, GAO-01-662 (Washington, DC: June 7, 2001).

Appendix III

                       Objectives, Scope, and Methodology

Objectives	We were asked to respond to 12 questions about IRS's
certification program, as shown in table 6.

                        Table 6: Questions We Were Asked

1. 	What is the status of the EIC certification program, including the
timing and number/types of taxpayers to be contacted?

2. 	Has IRS "tested" or conducted a "focus-group" of any related letters,
forms, or documents for understandability and other issues; and, if so,
what have been the results of such efforts?

3. 	(a) What is the appropriateness of the draft certification forms and
explanations? (b) What are IRS's plans for processing them? (c) What types
of documents will EIC taxpayers need to provide the IRS? (d) Will
taxpayers generally be able to obtain the required documentation to
otherwise establish eligibility within the required time frame, such as
for marriage certificates, school transcripts, and rental agreements?

4. 	What is the range of alternatives considered by IRS for obtaining
similarly reliable documentation, including the cost of alternatives, and
taking into account the cost of EIC noncompliance?

5. 	What is the percentage of EIC claimants that would be required to
precertify prior to the 2004 filing season?

6. 	What information does IRS have regarding differences in the EIC
overclaim rate among EIC claimants that are positively correlated with
filing status, relationship to the qualifying child, or other factors?

7. 	To what extent are the issues of concern in GAO's report on the
current recertification program of similar concern in the new
certification program, including probable solutions to problem areas?

8. 	Does GAO believe the certification program has been adequately
developed to prevent unreasonable burdens on EIC taxpayers and to improve
compliance?

9. What is the current process for evaluating EIC eligibility?

10. 	(a) What is the current EIC error rate? (b) Have recent statutory
changes had an impact on the error rate or on the rate of overpayments?
(c) Were these statutory changes for the purpose of deterring
noncompliance?

11. 	What are the error rates of non-EIC taxpayer groups having
significant compliance issues?

12. 	What are the error rates of comparable benefit programs administered
by states or the federal government and do these programs use any
verification process?

Source: Subcommittee on Oversight, House Committe on Ways and Means.

In consultation with our requesters' offices, we grouped these questions
into three objectives, as follows: (1) describe the design and basis for
the EIC qualifying child certification program as proposed by the EIC task
force, (2) describe the current status of the program, including
significant changes since program approval, and (3) assess whether the
program is

                                  Appendix III
                       Objectives, Scope, and Methodology

adequately developed to (a) prevent unreasonable burdens on EIC taxpayers
and (b) improve compliance so that the test should proceed. In addition,
we were asked to provide readily available information on (1) significant
noncompliance rates other than for the EIC and (2) the overclaim rates and
administrative costs of comparable benefit programs administered by states
or the federal government and any verification process used by these
programs.

Scope and Methodology

To respond to all of the questions, we reviewed and analyzed relevant IRS
and other documentation, such as compliance reports, EIC task force
reports, draft letters and forms, testing and focus group records,
implementation plans, evaluation plans, and our prior products, and
interviewed Department of the Treasury and IRS officials involved in the
EIC certification program, including the Assistant to the Commissioner;
the National Taxpayer Advocate; Research, Analysis, and Statistics
officials; and members of the qualifying child certification
implementation team. We did not verify the accuracy of the data shown in
the various reports that we reviewed. Rather, we reviewed the steps IRS
had taken to implement the certification program and determined, to the
extent possible, how IRS ensured that the program had been adequately
developed to prevent unreasonable burden and improve compliance. We did
not evaluate whether IRS's preparations for implementing the certification
test, such as staffing and training, were sufficiently developed to
support proceeding with the test, because they were outside the scope of
our review, still under development, or the Treasury Inspector General for
Tax Administration had audits planned.

The first objective includes, in order, our response to questions 10, 4,
and 6. To determine the current EIC error rates and whether any studies
had been done on the impact of recent statutory changes on error rates, we
reviewed IRS's most recent compliance study, the Treasury Inspector
General for Tax Administration reports and our previous reports, and
interviewed IRS officials. In addition, we reviewed the legislative
history of recent statutory changes-effective since 1999-that pertained to
EIC. We analyzed these data and IRS and Treasury reports to determine
whether an analysis on the impact of the legislative changes on EIC error
or overclaim rates had been conducted. To determine the range of
alternatives considered by the task force, we reviewed documents and
interviewed members of the EIC task force. To determine the correlation
between overall EIC error rates, filing status, and gender, we interviewed
officials from Research, Analysis, and

Appendix III
Objectives, Scope, and Methodology

Statistics and analyzed their data, and reviewed the EIC task force
reports and Treasury's past compliance studies.

The second objective included, in order, our response to questions one,
five, two, and seven. To determine the status of the EIC certification
program, including the number and types of taxpayers to be contacted, we
interviewed IRS and Treasury officials and reviewed documents showing
timelines and key milestones. We reviewed plans for the certification
program, such as IRS's Concept of Operations and the 2004 Increment
Evaluation Plan, in conjunction with IRS's current process for evaluating
EIC eligibility. To calculate the percentage of EIC claimants subject to
certification in 2004, we divided the planned sample size by the number of
EIC claimants with qualifying children. To obtain information on IRS's
testing of letters, forms, and documents for understandability, we
observed the focus group testing that IRS conducted in Dallas, Tex., with
EIC taxpayers, tax preparers, and other parties to understand how IRS
assured itself that such persons understood the forms and thought they
could obtain the required documentation. For whether items of concern we
found within the recertification program could have similar concerns in
the new initiative, we analyzed our prior reports on IRS's recertification
program and IRS's progress in implementing our recommendations, then we
compared our analysis to the certification plans.

The third objective includes, in order, our responses to questions eight
and three. To make our determination as to whether the program had been
adequately developed to improve compliance with minimal burden to
taxpayers, we asked IRS officials to describe and provide documentation to
support the steps they took to assure that the program was adequately
developed. This included interviews and a high-level review of key steps
and decisions found in various documents, such as the EIC task force
reports, the Concept of Operations, staffing plans, training materials,
and the evaluation plan. To determine the potential extent of the burden
on taxpayers, we reviewed reports from outside groups that analyze
programs and policies for low-income groups. We obtained the opinions of
IRS officials and discussed those of outside stakeholders, such as
representatives from the Annie E. Casey Foundation, low income taxpayer
clinics, and large tax preparation organizations, that IRS had met with
about any problems taxpayers might have in complying with the
documentation requirements to establish EIC eligibility. We also
interviewed IRS officials and reviewed EIC task force documents to learn
about the range of alternatives taxpayers have available to obtain
similarly

Appendix III
Objectives, Scope, and Methodology

reliable documentation, if they were unable to comply with the
certification documentation requirements.

Our responses to questions 11 and 12 are in appendixes I and II,
respectively. To determine the error rates of non-EIC taxpayer groups
having significant compliance issues, we reviewed compliance research
reports, interviewed officials about IRS's National Research Program, and
reviewed information contained in the Strategy and Program Plan. We
discussed our analysis with key IRS officials, including representatives
of the Assistant to the Commissioner. To determine the overclaim rates,
administrative costs, and verification process of comparable benefit
programs administered by states or the federal government, we researched
our prior reports and contacted our staff knowledgeable about the selected
programs. We selected nine programs to review: Unemployment Insurance,
Supplemental Security Income, Social Security Disability Insurance, school
lunch, food stamps, Housing and Urban Development rental assistance,
Medicaid and Medicare, and Temporary Assistance for Needy Families. We
chose these programs largely because they were the same programs the EIC
task force reviewed and because each of them had some sort of
precertification program. We did not do a comprehensive analysis to
determine which programs, if any, were most comparable to the EIC, nor did
we determine whether the information reported for each program was
consistent and could be compared across programs. We did not do additional
analyses to determine how administrative costs compared to program
outlays.

Our response to question nine is in the background section of this report.
To determine the current process for determining EIC eligibility, we
reviewed relevant IRS documents and our prior reports. We verified the
accuracy of this information in interviews with IRS officials.

We conducted our work in Atlanta, Ga., Dallas, Tex., and Washington, D.C.,
from May 2003 through September 2003 in accordance with generally accepted
government auditing standards.

Appendix IV

Key Milestones for the Qualifying Child Certification Program

Key milestones for the certification program for fiscal years 2002 through
2005, are shown in figure 4.

Figure 4: Key Milestones for the Certification Program

2002

2003

2004

2005

Source: GAO analysis of IRS data.

Appendix V

Documents Related to the Precertification Program

According to agency officials, IRS will send each of the 25,000 taxpayers
subject to precertification four documents, including (1) Notice 84-A, a
letter informing taxpayers about the new program; (2) Form 8836,
"Qualifying Children Residency Statement;" (3) Publication 3211M, "Earned
Income Tax Credit Question and Answers;" and (4) Publication 4134,
"Free/Nominal Cost Assistance Available for Low Income Taxpayers." Copies
of these documents, current as of September 2003, follow.

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

Appendix V
Documents Related to the Precertification
Program

                                  Appendix VI

                   Comments from the Internal Revenue Service

Appendix VI
Comments from the Internal Revenue Service

Appendix VI
Comments from the Internal Revenue Service

Appendix VI
Comments from the Internal Revenue Service

Appendix VI
Comments from the Internal Revenue Service

Appendix VI
Comments from the Internal Revenue Service

Appendix VII

                     GAO Contacts and Staff Acknowledgments

GAO Contacts	Michael Brostek (202) 512-9110 Joanna Stamatiades (404)
679-1984

Acknowledgments	In addition to those named above, Tiffany Brown, Evan
Gilman, Veronica Mayhand, Kathryn Larin, David Lewis, Donna Miller, Libby
Mixon, Cheryl Peterson, and Tom Short made key contributions to this
report.

GAO's Mission	The General Accounting Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through the Internet. GAO's Web site (www.gao.gov) contains abstracts
and full-text files of current reports and testimony and an expanding
archive of older products. The Web site features a search engine to help
you locate documents using key words and phrases. You can print these
documents in their entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its
Web site daily. The list contains links to the full-text document files.
To have GAO e-mail this list to you every afternoon, go to www.gao.gov and
select "Subscribe to e-mail alerts" under the "Order GAO Products"
heading.

Order by Mail or Phone	The first copy of each printed report is free.
Additional copies are $2 each. A check or money order should be made out
to the Superintendent of Documents. GAO also accepts VISA and Mastercard.
Orders for 100 or more copies mailed to a single address are discounted 25
percent. Orders should be sent to:

U.S. General Accounting Office 441 G Street NW, Room LM Washington, D.C.
20548

To order by Phone: 	Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

To Report Fraud, 	Contact: Web site: www.gao.gov/fraudnet/fraudnet.htm

Waste, and Abuse in E-mail: [email protected]

Federal Programs Automated answering system: (800) 424-5454 or (202)
512-7470

Public Affairs	Jeff Nelligan, Managing Director, [email protected] (202)
512-4800 U.S. General Accounting Office, 441 G Street NW, Room 7149
Washington, D.C. 20548

                               Presorted Standard
                              Postage & Fees Paid
                                      GAO
                                Permit No. GI00

United States
General Accounting Office
Washington, D.C. 20548-0001

Official Business
Penalty for Private Use $300

Address Service Requested
*** End of document. ***