TREASURY INSPECTOR GENERAL FOR TAX ADMINISTRATION

 

 

Actions Have Been Taken to Address Deficiencies in the Questionable Refund Program; However, Many Concerns Remain, With Millions of Dollars at Risk

 

 

 

May 31, 2007

 

Reference Number:  2007-10-076

 

 

This report has cleared the Treasury Inspector General for Tax Administration disclosure review process and information determined to be restricted from public release has been redacted from this document.

 Redaction Legend:

1 = Tax Return/Return Information

2(a) = Law Enforcement Criteria

2(c) = Law Enforcement Tolerance(s)

2(e) = Law Enforcement Procedure(s)

2(f) = Risk Circumvention of Agency Regulation or Statute (whichever is applicable)

7 = Predecisional Staff Recommendations or Suggestions to Agency Decision Makers

Phone Number   |  202-927-7037

Email Address   |  Bonnie.Heald@tigta.treas.gov

Web Site           |  http://www.tigta.gov

 

May 31, 2007

 

 

MEMORANDUM FOR CHIEF, CRIMINAL INVESTIGATION

 

FROM:                            Michael R. Phillips /s/ Michael R. Phillips

                                         Deputy Inspector General for Audit

 

SUBJECT:                    Final Audit Report – Actions Have Been Taken to Address Deficiencies in the Questionable Refund Program; However, Many Concerns Remain, With Millions of Dollars at Risk (Audit # 200610003)

 

This report presents our review of the Questionable Refund Program.[1]  The overall objective of this review was to determine the effectiveness of the Internal Revenue Service’s (IRS) procedures for detecting fraudulent and potentially fraudulent refund income tax returns. 

We began this review as a follow-on audit to our recent limited-scope review regarding the effectiveness of the IRS’ processes to detect fraudulent prisoner refund returns and to our prior reviews regarding the effectiveness of the controls placed on taxpayer accounts identified as fraudulent or potentially fraudulent.[2]  However, due to the release of the National Taxpayer Advocate’s (Advocate) 2005 Annual Report to Congress and the revelation that the Electronic Fraud Detection System would not be operational during Processing Year 2006, we added several tests to determine how the IRS had implemented a limited fraud detection program for that year and how it plans to change processing procedures for Processing Year 2007 and beyond.

Impact on the Taxpayer

The IRS estimates fraudulent refund claims exceed $500 million a year.  The Criminal Investigation (CI) function Questionable Refund Program, established to identify and prevent the issuance of fraudulent refunds, received harsh criticism from the Advocate in January 2006 as being a Program that was inefficient and ineffective and that did not afford taxpayers their rights.  The IRS immediately reevaluated its processes, establishing new procedures and discontinuing others in an effort to address deficiencies cited by the Advocate.  However, we believe several of these procedural changes may adversely affect the IRS’ ability to prevent potentially fraudulent refunds from being issued in the future, possibly placing millions of dollars at risk.

Synopsis

In January 2006, the Advocate released a Report that brought to light deficiencies within the CI function’s Questionable Refund Program and caused Congress to express concerns about this Program to the IRS Commissioner and the Secretary of the Treasury.  Shortly after the release of the Advocate’s Report, the IRS formed an Executive Steering Committee to implement a notification process to afford taxpayers’ their rights and to revise the Questionable Refund Program.  While we are encouraged by the IRS’ actions to address stakeholder concerns and restore balance between taxpayer rights and effective administration of the tax laws, we are concerned that some of the changes will not stop millions of dollars of potentially fraudulent refunds from being issued.

For example, the CI function’s prior policy of placing a freeze on the future year’s refund return is an effective means for protecting revenue and should still be considered, with some modifications.  In addition, the IRS needs to revalidate its study for establishing the time period for imposing an automatic release of frozen refunds.  Data maintained by the Office of Refund Crimes for Processing Year 2006 show almost $15.9 million (22.3 percent) in potentially fraudulent refunds were issued because the returns were not verified before the automatic releases were imposed.

In addition to protecting revenue, the Fraud Detection Centers (Centers) are responsible for referring fraudulent accounts to the civil functions for resolution or to the CI function field offices for criminal investigation.  However, few criminal investigations result from refund fraud referrals, and civil actions are not aggressively pursued.  The CI function advised us of its plans to change the tax scheme referral criteria in Processing Year 2007 and refer only those that meet a certain dollar loss amount.  In addition, the Office of Refund Crimes revised its procedures to include a refund tolerance for referring returns claiming refundable credits to the Examination function.  While we understand these decisions were made to balance available resources with workload, we are concerned they could have a negative impact on effective tax administration. 

Further, the IRS allowed potentially fraudulent refunds to be issued because it did not verify refund returns with ****2(a), 2(f)**** during Processing Year 2006.  We also have concerns about the adequacy of the prerefund screening criteria that will be used during Processing Year 2007 to identify potentially fraudulent ****2(a), 2(f)**** returns.  The criteria are based on a limited number of factors, are applied on individual returns, and in our opinion do not address the many variables and patterns of filing multiple fraudulent returns found in refund schemes.

Through case reviews, we determined that, although there were incorrect determinations of fraud by Center personnel, the problem was not as widespread as originally indicated in the Advocate’s Report.  The Report stated that 46 percent of the 142 returns in the sample of taxpayer inquiries about current year frozen refunds were subsequently determined to be not fraudulent and the refunds were released.  Only 15 percent of the 27 cases in our sample (that also received a Taxpayer Advocate Service inquiry) were subsequently determined to be not fraudulent.  The CI function determined that 5.8 percent of the 6,831 returns originally determined to be fraudulent during Processing Year 2004 were later determined to be valid refunds.  We believe the primary reason for the differing rates is the Advocate’s sample consisted only of taxpayers who filed a complaint with its office.  The CI function believes these taxpayers are much more likely to contact the Advocate’s office than those who have filed false or fraudulent returns.  The Advocate’s office recognized this and advised us it did not intend to imply its results were indicative of the entire population of frozen refunds, only that there were some innocent taxpayers within the CI function’s population of fraudulent returns.

Finally, some problems we had previously reported still existed,[3] including that Center personnel were not timely removing controls placed on taxpayers’ accounts to allow refunds to be issued and were not always taking timely actions to resolve accounts.  In addition, the Centers did not always document their determination that a return was fraudulent.  While concerns relating to timely account resolution have been reduced, the Centers need to improve the documentation of fraud determinations. 

Throughout our review, we discussed the issues described in this report, and the following recommendations, with IRS executives and staff to alert them of issues that may also affect Processing Year 2007.

Recommendations

We recommended the Chief, CI, (1) consult with other IRS functions and reconsider placing a CI freeze on the subsequent year’s returns of those accounts identified in the current year as fraudulent; (2) revalidate the optimal time needed to verify wage/withholding information prior to automatically releasing a refund; (3) continue working with the United States Attorney’s Offices on the referral criteria for refund schemes, to ensure there is a balance between workload and effective tax administration; (4) consult with the IRS Office of Chief Counsel to initiate a legislative proposal to exempt the IRS from having to issue statutory notices of deficiency for disallowance of the ****2(a), 2(f)**** and ****2(a), 2(f)**** when the deficiencies and credits are the result of fraudulent returns; (5) consult with the Executive Steering Committee and consider lowering the tolerances for sending cases to the Examination function; (6) reconsider the decision to exclude certain returns from the Center screening process; and (7) reemphasize the requirement to maintain documentation to adequately support fraud determinations. 

Response

IRS management agreed with Recommendations 2, 3, and 7 and advised us of plans to review 2007 data to obtain more current time periods for completing verification; to collaborate with the Department of Justice and the Executive Office of United States Attorneys on referral criteria, to ensure a balance between workload and effective tax administration; and, through a task force, to consider revising review procedures, including the adherence to maintaining documentation.  Management partially agreed with Recommendation 1 to freeze the subsequent year’s returns of those accounts identified in the current year as fraudulent, stating that, to work the additional inventory that would result from implementation of the recommendation, they would have to consider other enforcement priorities.  Management neither agreed nor disagreed with Recommendations 4, 5, and 6 and, as a result, was noncommittal on the corrective actions.  Management plans to work with the IRS Offices of Legislative Affairs and Chief Counsel but believes legislative proposals are the responsibility of the Assistant Secretary for Tax Policy; stated that, while they would like to address all cases of potential noncompliance, enforcement resources are limited; and plans to conduct an analysis to determine if the current selection process (use of the Dependent Database Program, a prerefund Earned Income Tax Credit program) for certain returns should be modified.  Management’s complete response to the draft report is included as Appendix VII.

Office of Audit Comment

For the one partially agreed to recommendation and the three recommendations that management neither agreed nor disagreed with, the IRS proposed different types of actions.  For Recommendation 1, instead of reconsidering placement of a freeze on the subsequent year’s returns, management stated they would wait for a cross-functional team to evaluate the results of the 2007 Questionable Refund Program and make recommendations on any adjustments before the 2008 Filing Season.  For Recommendation 4, instead of working with the IRS Office of Chief Counsel to initiate a legislative proposal, IRS management stated they would work with relevant parties to discuss initiating a legislative proposal.  For Recommendation 5, instead of considering lowering the tolerance for sending cases to the Examination function, management stated that criteria will be established each filing season based on business decisions about competing priorities.  Finally, for Recommendation 6, instead of reconsidering the decision to exclude certain returns from the screening process, management stated they used a different screening process for the 2007 Filing Season and plan to conduct an analysis to determine if the current selection process should be modified.

While we see these actions as a start, we believe the Executive Steering Committee should consider the factors in our report and the potentially egregious nature of the questionable refund fraud identified when comparing the priority of the Questionable Refund Program and its impact on tax administration with that of other civil programs.  Specifically, we believe freezing the subsequent year’s returns as we recommended is a more efficient and effective means of identifying repeated fraud, protecting Federal Government revenue, and protecting innocent taxpayers who are victims of identity theft.  In addition, we did not determine if the IRS’ actions regarding the identity theft indicator would protect innocent taxpayers’ refunds.

We continue to believe the IRS should take a leading role in pursuing a legislative change, while working with the Assistant Secretary for Tax Policy.  A legislative proposal has the potential to improve tax administration while still protecting taxpayer rights.  In addition, implementing the recommended legislative change would allow the IRS to lower its tolerances, drastically reduce the number of cases that would need to be referred to the Examination function, and reduce the number of already limited resources needed to process these cases.  Finally, approximately a month prior to responding to this report, the CI function concluded that not everything was being identified by the Dependent Database Program.  This admission came after we expressed our concerns about the effectiveness of this Program on several occasions.  As a result, IRS management was already aware the 2007 Filing Season screening process was ineffective for certain returns, but they did not indicate in their response what changes were needed.

Copies of this report are also being sent to the IRS managers affected by the report recommendations.  Please contact me at (202) 622-6510 if you have questions or Nancy A. Nakamura, Assistant Inspector General for Audit (Headquarters Operations and Exempt Organizations Programs), at (202) 622-8500.

 

 

Table of Contents

 

Background

Results of Review

The Growth of the Questionable Refund Program Led to Its Eventual Management and Inventory Problems

The Internal Revenue Service Revised Procedures to Afford Taxpayers’ Their Rights

The Internal Revenue Service Should Reconsider How It Freezes Taxpayer Accounts

Recommendation 1:

Recommendation 2:

The Internal Revenue Service Needs to Be More Aggressive in the Criminal and Civil Pursuit of Fraudulent Refunds

Recommendations 3 and 4:

Recommendation 5:

Changes During Processing Year 2006 Had a Detrimental Impact on Identifying Fraudulent Returns and Will Have an Undeterminable Effect on Processing Year 2007

Recommendation 6:

Actions Taken and Decisions Made by the Fraud Detection Centers Were Generally Proper but Not Always Timely or Supported

Recommendation 7:

Appendices

Appendix I – Detailed Objective, Scope, and Methodology

Appendix II – Major Contributors to This Report

Appendix III – Report Distribution List

Appendix IV – Outcome Measures

Appendix V – Related Treasury Inspector General for Tax Administration Audit Reports

Appendix VI – Glossary of Terms

Appendix VII – Management’s Response to the Draft Report

 

 

Abbreviations

 

CI

Criminal Investigation

EFDS

Electronic Fraud Detection System

EITC

Earned Income Tax Credit

IRS

Internal Revenue Service

QRP

Questionable Refund Program

 

 

Background

 

The mission of the Internal Revenue Service (IRS) Criminal Investigation function (hereinafter referred to as CI function or function) is to serve the American public by investigating potential criminal violations of the Internal Revenue Code and related financial crimes in a manner that fosters confidence in the tax system and compliance with the law.  The CI function’s Questionable Refund Program (QRP)[4] is a nationwide, multifunctional program whose mission is to detect refund fraud, prevent the issuance of fraudulent refunds, and provide support to the CI function field offices.  During our exit conference, the CI function advised us it had begun to refer to these refunds as “potentially fraudulent” because an actual fraud determination could not be made without the taxpayer’s explanation.  In this report, we retained the CI function’s prior terminology of “fraudulent” because the scope of our case review involved evaluating prior years’ procedures, and there was inconsistent use of the terms “fraudulent” and “potentially fraudulent” during 2006.  

Responsibility for coordinating the QRP resides with the CI function’s 10 Fraud Detection Centers (hereinafter referred to as the Centers), which are located at each of the 10 IRS campuses.  In their early years, the Centers primarily used manual procedures to identify fraudulent returns by looking for similar characteristics.  This process gradually evolved into an automated identification of fraudulent returns, with the Electronic Fraud Detection System (EFDS) being prototyped in 1 Center in 1994 and implemented in all 10 Centers in 1996.  Beginning in 2002, the Centers introduced data mining techniques to more effectively identify fraudulent returns. 

When a Center identifies a potentially fraudulent return, it contacts the employer or third party to verify wage information on the return.  If the verification process is not completed within a certain time period, the Center places a temporary freeze on the account to prevent the refund from being issued.  When a Center verifies fraud, it should replace the temporary freeze with a permanent freeze and take the appropriate action to either reverse the frozen refund or pursue recovery of the fraudulent refund if it was issued.

As detection techniques and data mining evolved and improved, the Centers became more proficient and efficient in identifying fraudulent returns and stopping fraudulent refunds.  At the same time, during the late 1990s and into early 2000, taxpayers became bolder in filing returns claiming fraudulent refunds.

The CI function reported that, as of December 1, 2005, the Centers had identified 132,945 fraudulent refund returns claiming $515.5 million in refunds during Processing Year 2005.  Compared to the prior year, this was a 12.6 percent increase in the number of fraudulent returns and a 16.7 percent increase in the amount of fraudulent refunds identified.  In contrast, through September 29, 2006, the CI function had identified only 44,788 fraudulent returns claiming $232.3 million in refunds during Processing Year 2006.  The dramatic decrease occurred because the redesigned EFDS web-based application (Web EFDS) was not implemented due to a lack of adequate oversight and monitoring of the project, as we had previously reported.[5]

The IRS estimates fraudulent refund claims exceed $500 million a year.  Congress has held hearings urging the IRS to devote additional resources to and improve its detection of fraudulent refunds, particularly those claimed by prisoners.  At the same time, the Advocate reports that actions taken by the Centers adversely affected taxpayers’ rights.

We began this review as a follow-on audit to our 2005 limited-scope review regarding the effectiveness of the IRS’ processes to detect fraudulent prisoner refund returns and to our prior reviews regarding the effectiveness of Center controls placed on taxpayer accounts identified as fraudulent or potentially fraudulent.[6]  However, due to the release of the National Taxpayer Advocate’s (hereinafter referred to as the Advocate) 2005 Annual Report to Congress and the revelation that the EFDS would not be operational during Processing Year 2006, we added several tests to determine how the IRS had implemented a limited fraud detection program during Processing Year 2006 and how it plans to change processing procedures for Processing Year 2007 and beyond.  In addition, we are continuing our efforts, through a separate review, to evaluate the validity of the scoring methodology used by the CI function to identify potentially fraudulent returns and are compiling demographic profiles of taxpayers to determine the effectiveness of the IRS’ screening process.[7]

This review was performed at the Office of Refund Crimes in Washington, D.C., and in the Atlanta, Georgia; Brookhaven, New York; Fresno, California; and Kansas City, Missouri, Centers during the period December 2005 through November 2006.  In addition, we telephonically contacted the Andover, Massachusetts; Austin, Texas; Cincinnati, Ohio; Memphis, Tennessee; Ogden, Utah; and Philadelphia, Pennsylvania, Centers.  The audit was conducted in accordance with Government Auditing Standards, except for the Field Work Standard for Performance Audits governing sufficient, relevant, and competent evidence.  Although we requested that the Centers send us information directly, there were instances when the Office of Refund Crimes requested that the Centers provide it with the information to determine if all documents provided were responsive to our request.  We are unaware if there were any instances in which the Office of Refund Crimes eliminated documents sent to it by the Centers in responding to the audit.  Nonetheless, because we could not always examine documents at the source, we could not independently confirm that all examples of pertinent documents submitted to the CI function Headquarters office from the field were, in turn, forwarded to us.  Detailed information on our audit objective, scope, and methodology is presented in Appendix I.  Major contributors to the report are listed in Appendix II. 

 

 

Results of Review

 

The Growth of the Questionable Refund Program Led to Its Eventual Management and Inventory Problems

While we recognize there is no one system that can identify 100 percent of fraud, we believe the Centers are accomplishing the primary mission of the QRP to identify and stop fraudulent refunds.  Since its inception in 1977, the QRP has identified over $4.3 billion in fraudulent refunds and prevented the issuance of over $3.6 billion.[8]  However, the growth in refund fraud, combined with the CI function’s priority of protecting revenue by freezing refunds, may have contributed to the function’s inability to effectively manage the QRP.

Several symptoms surfaced during the last few years that, if adequately and timely addressed, may have prevented the QRP from receiving harsh criticism by the Advocate and Congress in early 2006.  First, we have issued five reports since 1999 repeatedly citing inventory and management control deficiencies in the QRP.  Second, the volume of frozen accounts has shown a steady increase over the years, from 100,205 accounts in Fiscal Year 2002 to 238,466 accounts in Fiscal Year 2005.  Finally, the Advocate reported concerns about the Taxpayer Advocate Service’s dramatically increasing volume of workload pertaining to CI function account freezes.  The number of Taxpayer Advocate Service receipts increased from 5,587 in Fiscal Year 2002 to 28,639 in Fiscal Year 2005, becoming the second most serious problem highlighted in the Advocate’s Report.

The IRS did not respond to warning signs

There have been several warning signs over the years indicating the QRP was facing problems and becoming unmanageable.  During 1999 to 2005, we issued five audit reports highlighting some of the same areas for improvement, as shown in Figure 1.

Figure 1:  Office of Audit Summary of Prior Audit Issues

Report Number and Date

Report Title

Findings

Planned Corrective Action by the IRS

2000-40-018

December 1999

The Internal Revenue Service Can Improve the Effectiveness of Questionable Refund Detection Team Activities

Refund freeze controls not always removed, and refunds were not timely released.

Issue national guidance to
CI function Branch Chiefs and reissue established guidelines.

2001-40-025

January 2001

Letter Report:  Revised Questionable Refund Program Procedures Were Not Consistently Implemented

QRP cases were not transferred to the Examination function for adjustment, depriving taxpayers of proper notification.

Stress the importance of following procedures during training.

2003-10-094

March 2003

Improvements Are Needed in the Monitoring of Criminal Investigation Controls Placed on Taxpayers’ Accounts When Refund Fraud Is Suspected

Centers were not always reviewing and reconciling quarterly control reports and taking timely actions to resolve accounts and issue legitimate refunds.

In the CI function, pursue multiple alternatives to gain access to automated databases, provide timely guidance and training to Centers, and include QRP procedures in Center visitations.

****2(f)****

****2(f)****

****2(f)****

****7, 2(f)****

****2(f)****

****2(f)****

****2(f)****

In the CI function, delay refunds on returns meeting certain criteria to identify and prevent the issuance of fraudulent refunds.

Source:  Our prior audit reports.

Although the IRS planned to take corrective action after each audit, in many cases the corrective action was not taken or was ineffective.

In addition, the number of Taxpayer Advocate Service inquiries resulting from refund freezes began increasing at alarming rates.  While the volume of fraudulent returns identified increased 63 percent from Processing Year 2002 to Processing Year 2005, the volume of Taxpayer Advocate Service inquiries related to CI function cases increased from 5,587 in Fiscal Year 2002 to 28,639 in Fiscal Year 2005, an increase of 413 percent.  In Fiscal Year 2002, refund inquiries were the twelfth leading cause of a Taxpayer Advocate Service inquiry.  They were third in Fiscal Year 2003 and became the number one cause in Fiscal Year 2004.  Figure 2 shows the volumes of all Taxpayer Advocate Service inquiries and inquiries related to fraudulent refunds.

Figure 2:  Comparison of All Taxpayer Advocate Service Inquiries to Inquiries Related to Refund Fraud

Figure 2 was removed due to its size.  To see Figure 2, please go to the Adobe PDF version of the report on the TIGTA Public Web Page.

 

Figure 3 shows a further comparison of the number of fraudulent refunds identified to the number of Taxpayer Advocate Service inquiries.

Figure 3:  Comparison of Fraudulent Refunds Identified to Taxpayer Advocate Service Inquiries

Figure 3 was removed due to its size.  To see Figure 3, please go to the Adobe PDF version of the report on the TIGTA Public Web Page.

Further, the volume of potentially fraudulent frozen refund accounts increased 138 percent, while the amount of refunds frozen increased 180 percent, from 2002 to 2005.  Figure 4 shows the numbers and amounts of tax accounts frozen at the end of these years.

Figure 4:  Comparison of Fraudulent Frozen Refund Accounts

 

Year Ended

Accounts Frozen

Amounts Frozen

2002

100,205

$367,682,083

2003

169,106

$598,568,111

2004

200,520

$836,276,193

2005

238,466

$1,028,606,118

Source:  The CI function for Fiscal Year 2005 and our analysis of the IRS’ frozen refund file for other years.
 The year ended 2002 is the calendar year; the other years are fiscal years
.

As mentioned above, in two of five prior audit reports, we had raised concerns about the CI function’s inaction in timely reviewing and releasing freezes.[9]  We believe the significant increase in Taxpayer Advocate Service inquiries is directly related to the volume of frozen refunds.  These factors, when combined, should have signaled to the IRS there were workload problems that needed to be addressed within the QRP.

The Advocate and Congressional stakeholder concerns

The Advocate’s Report released in January 2006 brought to light several problems with the QRP.  As a result, Congress expressed concerns about the QRP to the IRS Commissioner and the Secretary of the Treasury.  While all stakeholders acknowledged the value of the QRP in preventing tax fraud and abuse and promoting compliance among taxpayers, they indicated there needs to be a balance between protecting taxpayer rights and enforcing the laws.  The Advocate’s Report highlighted several significant problems, such as:

  • Taxpayers were not notified their returns were frozen as potentially fraudulent.
  • Procedures in resolving accounts subjected taxpayers to excessive delays.
  • Future refund returns were frozen until a taxpayer filed a certain number of legitimate returns, even though there is little evidence to suggest that a taxpayer is likely to repeat refund fraud after the initial attempt.
  • Identity theft victims’ returns were automatically classified as fraudulent for successive years, and their refunds were frozen.
  • Frozen refunds were not timely reviewed due to insufficient CI function resources.
  • Other IRS functions, especially the Examination function, did not have resources to address refund freezes.

Unfortunately, it took the release of the Advocate’s Report for the IRS to realize that the CI function could not and should not fix these problems alone.  Thus, shortly after the release of the Report, the IRS Commissioner directed a review of the QRP, to include establishing a process for notifying taxpayers.  An Executive Steering Committee, consisting of members from the Advocate’s office, the CI function, the Small Business/Self-Employed Division, the Wage and Investment Division, and the Modernization and Information Technology Services organization, was formed to establish a process to notify taxpayers that their refunds were being held and to revise the QRP.  We are encouraged by the IRS’ actions to address the concerns raised by the Advocate; however, we have reservations about some of the changes that have been implemented.  We discuss our concerns throughout this report.

The Internal Revenue Service Revised Procedures to Afford Taxpayers’ Their Rights 

Previous procedures did not allow a taxpayer with a CI freeze on his or her account to be given any information about the freeze by any part of the IRS, including the Taxpayer Advocate Service, until 180 days had elapsed from the date of the taxpayer’s initial contact with the IRS.  Even after expiration of the 180 days, the CI function might have designated a case as a “no contact” case, which means the IRS, including the Taxpayer Advocate Service, was not permitted to provide account information to the taxpayer.

Due to the sensitive nature of criminal investigations, the CI function generally did not notify taxpayers that their refunds were frozen.  The function defended its policy by stating the delay in providing information is needed to avoid tipping off targets of fraud investigations and these are ongoing investigations in which an individual could be filing multiple returns for refunds throughout the filing season.

CI function officials conceded that they could do a better job of communicating with taxpayers whose refunds had been frozen.  To that end, the Executive Steering Committee drafted a notice to advise taxpayers whose refunds had been frozen and provide them with a time period for contacting the IRS.  Taxpayers started to receive these initial notices in February 2006.  In addition, the Committee drafted a second set of notices to use when it was determined a refund is fraudulent.  These notices include an explanation and description of the taxpayer’s rights to challenge the claim.  The IRS issued these notices beginning in May 2006.  We are encouraged by the IRS’ actions to address the Advocate’s concerns regarding notifying taxpayers and its actions to implement this process so quickly.  We will discuss these notices in further detail when we discuss plans for referring fraudulent refund cases to other IRS functions for resolution. 

The Internal Revenue Service Should Reconsider How It Freezes Taxpayer Accounts

The CI function uses three basic types of freezes during the process of identifying, verifying, and controlling fraudulent refund returns.  The identification process begins when refund returns are passed through data mining models.  Each return receives a score based on certain return characteristics.  The higher the score, the greater the likelihood the refund could be fraudulent.  The Centers select returns that meet predetermined tolerances for further review.  There is an automatic resequencing (delays) in the processing of these returns for 1 week to prevent the potentially fraudulent refunds from being issued.  This also provides the Centers additional time to determine if the returns are fraudulent.

Resequencing has minimal impact on taxpayers

According to the CI function, it resequenced about 869,000 refund returns during Processing Year 2004.  Of those, about 96 percent were released after 1 week and the taxpayers received their refunds timely.  In Processing Year 2005, the CI function resequenced almost 526,000 refund returns and released about 92 percent after 1 week.

As part of the CI function’s ongoing efforts to combat refund fraud, the Director, Refund Crimes, initiated a programming change to delay refunds for 2 weeks, to allow additional time to screen returns in Processing Year 2006.  The CI function plans to have the EFDS operational during Processing Year 2007 and will continue with its plans to delay by 2 weeks any refund meeting the data mining tolerance.  We believe this processing change will have minimal impact on taxpayers and is an appropriate balance to allow the Centers time to screen refunds.

Changes in temporary freezes may not give the IRS sufficient time to verify refunds

The CI function will place a temporary freeze on an account when it cannot complete the verification process within the time allotted during resequencing or if it identifies a potentially fraudulent return through other means.  This temporary freeze should be removed when wage and/or withholding verification[10] supports the release of the refund.  However, in many cases, these temporary freezes became more permanent in nature because workload and staffing did not always allow the Centers to make a fraud determination. 

According to the CI function, it placed a temporary freeze on about 542,000 accounts during Processing Year 2004 and about 561,000 accounts during Processing Year 2005.[11]  However, it did not notify taxpayers of this or afford them the courtesy of knowing how long their refunds would be delayed and a right to challenge a claim disallowance if necessary.  The Advocate’s Report acknowledged this created an undue burden on taxpayers whose refunds were legitimate.  As a result, the IRS implemented new procedures in Processing Year 2006 to notify taxpayers that their refunds have been delayed.  In addition, the Centers will have only a certain number of days for completing the verification of wages, withholding, and/or claims for tax credits instead of leaving the freeze on the account permanently.  If the Centers do not complete the verification process and make the determination there is fraud during this time, the refund will be automatically released. 

When we asked how the Office of Refund Crimes decided on the criteria for automatically releasing refunds that cannot be verified within a certain number of days, we were told the criteria were established by a consensus of the Executive Steering Committee and the Advocate.  Office of Refund Crimes staff further advised us they conducted a study using prior years’ data to assess the length of time between when a return is temporarily frozen and when the Center makes a fraud determination.  This study determined the Centers made a fraud determination on 90 percent of their cases within a certain number of days.  Our review of 385 fraudulent refund returns processed during Processing Year 2004 identified 108 returns that were temporarily frozen.  The Centers made a fraud determination on 90 percent of these returns within the time period established by the Committee, similar to the Office of Refund Crimes study results.  However, these statistics were based upon work done by the Centers in prior years. 

Statistics maintained by the Office of Refund Crimes for Processing Year 2006 show 19,399 fraudulent refund returns were identified for review as a result of prior years’ freezes or freezing the current year’s accounts of prisoners.  Refunds of almost $15.9 million (22.3 percent) were issued because the returns were not verified before the automatic release was imposed.  Thus, only about 78 percent of the returns for Processing Year 2006 were verified within the established time, a 13 percent decrease from the verification rate achieved during the prior year.  Office of Refund Crimes personnel advised us they did not analyze this information to revalidate the time periods for automatically releasing refunds.  In light of the new procedures being implemented in the QRP, we believe the Office of Refund Crimes should revalidate its study using the most current data, to ensure it has sufficient time to complete the verification process.

Permanent freezes have the biggest impact on taxpayers and may still be necessary

When the CI function verifies a refund return as fraudulent, it places a CI freeze on the account.  Procedures in effect during Processing Years 2004 and 2005 required the Centers to place the CI freeze on a taxpayer’s account when a return was verified as fraudulent and to keep the account frozen until the taxpayer had filed a certain number of legitimate returns in future years.  During Processing Years 2004 and 2005, the Centers placed freezes on 161,855 and 171,531 taxpayer accounts, respectively.  The freeze status remained on the taxpayers’ accounts indefinitely until the Centers took action to resolve the accounts or release the freezes.  As a result, the volume of frozen accounts grew to almost 240,000 by the end of Fiscal Year 2005.  In some cases, refunds were frozen for several years. 

The Advocate’s Report highlighted the “automatic” freezing of future years’ refund returns as a significant problem with the QRP.  As a result, the Office of Refund Crimes discontinued placing a freeze on future years’ refund returns and instead identified certain high-risk categories as exceptions to this process.  This revised procedure concerns us because we believe the future year freeze is an effective means for protecting revenue, when considered along with the 2006 changes to notify taxpayers and minimize the time refunds are frozen.

Our analysis of 385 fraudulent refund returns filed during Processing Year 2004 identified 159 taxpayers who also filed a return during Processing Year 2005; returns for 149 of the 159 taxpayers were frozen, with refunds totaling about $917,000.  The Centers subsequently reviewed these 149 returns, determined 63 (42.3 percent) with refunds totaling $716,232 were fraudulent, and entered them into the Scheme Tracking and Referral System.  This demonstrates that freezing the subsequent year’s returns achieved a higher rate of success in identifying fraud, as only 12.7 percent of all returns selected for review by data mining were determined to be fraudulent during Processing Year 2005.  In addition, this differs from the Advocate’s belief that the CI function’s policy to automatically freeze future years’ refunds is overly restrictive and administratively inefficient.  Further, we believe use of the CI freeze on future years’ refunds was more effective in protecting revenue because 99 percent of the refunds were stopped in these 63 cases, versus 80 percent for all fraudulent refunds identified during Processing Year 2005.  We project that 20,078 returns totaling $71.7 million would have been determined to be fraudulent during 2005.[12]

In the 385 Processing Year 2004 sample cases reviewed, we identified 90 taxpayers who filed a return during Processing Year 2006; accounts for 72 of the 90 taxpayers were frozen.  We subsequently provided a list of the 72 returns to the CI function and asked for assistance in determining if these returns were deemed to be fraudulent.  Because of the changes and problems encountered during Processing Year 2006, the Centers had released 9 refunds totaling $25,200 on 10 returns with an attached ****2(a), 2(f)**** without making a fraud determination.  For the remaining 62 returns, the CI function determined 11 (17.7 percent), with refunds totaling $51,400, were fraudulent.  Although we noted a declining rate of fraudulent refunds going into the second subsequent year, we believe the subsequent year freeze is a viable tool for identifying and stopping fraudulent refunds in the immediate subsequent year.  Further, we believe the impact on the taxpayer will be reduced by the new procedures to notify the taxpayer of the freeze and to limit the time the IRS has to resolve the freeze.

In an October 2, 2006, memorandum to the Secretary of the Treasury, the Treasury Inspector General for Tax Administration outlined the management and performance challenges facing the IRS for Fiscal Year 2007, specifically expressing concerns about erroneous and improper payments.  Although the memorandum recognized the Advocate’s recent criticisms and the IRS’ subsequent response to improve the QRP, we continue to have concerns and believe a necessary balance must be struck between allowing sufficient time to detect fraudulent claims and issuing timely refunds.

When properly identified and timely worked, frozen refunds involving identity theft protect the innocent taxpayer and revenue

Identity theft is a growing problem, with both consumer fraud and refund fraud.  Of the 44,788 refunds verified as fraudulent during Processing Year 2006 through September 29, 2006, the CI function indicated 7,957 (17.8 percent) involved identity theft.

The Advocate’s Report stated 27 cases in its sample (6 percent) involved identity theft and took exception to the CI function’s policy of automatically freezing the current and future years’ refunds of identity theft victims.  The Advocate expressed concern that this policy is overly broad and causes significant and continuing inconvenience.  The Report indicated a need for an IRS-wide system that identifies which taxpayer is the victim of identity theft.  In response to the Report, the IRS will no longer place a CI freeze on subsequent years’ returns of accounts involving identity theft.

We believe the biggest problem associated with identity theft cases was the CI function’s inability to identify victims whose tax returns are frozen in future years and to timely determine if the taxpayers are again the victims of identity theft.  In our opinion, the policy to not freeze the subsequent years’ returns for these accounts will not significantly reduce taxpayer inquiries and could result in additional lost revenue and significant taxpayer burden for the following reasons:

·         The CI function indicated that, generally in identity theft cases, the ****2(a), 2(f)**** taxpayer files first.  Because subsequent years’ returns for these accounts will no longer be frozen, if the ****2(a), 2(f)**** taxpayer files first, the refund will be issued.  When the ****2(a), 2(f)**** taxpayer files, the return will not be processed because the system recognizes this as a duplicate return.  This will likely result in the ****2(a), 2(f)**** to resolve the status of the requested refund. 

·         On the other hand, if previous identity theft cases were frozen, and the ****2(a), 2(f)**** taxpayer files first, the refund will be stopped, likely resulting in the ****2(a), 2(f)**** about the refund. 

While there would likely be taxpayer contact in both cases, by freezing the refund, the IRS is protecting potential lost revenue.  Further, because taxpayers are now notified when their refunds are frozen and if the CI function identifies and timely works identity theft cases, we believe any adverse effects will be reduced.  Our review of 104 frozen credit cases included 14 cases with credits over $500,000 (totaling $40.3 million) that resulted either from estimated tax payments, payments included with requests for extensions to file, and/or a transfer of credits from a previous year belonging to taxpayers whose Social Security Numbers were improperly used to receive fraudulent refunds in a prior year.  The credits were later determined to belong to the ****2(a), 2(f)**** taxpayers.  The current policy of not freezing these accounts could result in these credits being issued if the ****2(a), 2(f)**** taxpayers filed first.

In our opinion, if the CI function properly identifies identity theft freezes, notifies the taxpayers of the freezes, and timely resolves the freezes, the IRS will be providing a valuable service to the taxpayer while at the same time protecting Federal Government revenue.

The Office of Refund Crimes staff acknowledged that identity theft is an area it still needs to address.  In addition, the Office of Refund Crimes staff advised us the identity theft program area will be transferred from the Wage and Investment Division to the Mission Assurance and Security Services organization.

Recommendations

The Chief, CI, should:

Recommendation 1:  In light of the 2006 changes to reduce burden on taxpayers, consult with other IRS functions and reconsider placing a CI freeze on the subsequent year’s returns of those accounts identified in the current year as fraudulent, including those returns involving identity theft.

Management’s Response:  The IRS partially agreed with the recommendation.  Under the auspices of the QRP Executive Steering Committee, a cross-functional team plans to evaluate the results of the 2007 QRP and make recommendations as to whether adjustments are necessary before the 2008 Filing Season.  The IRS plans to consider the Executive Steering Committee’s and our recommendations as well as any tradeoffs that would have to be made against other enforcement priorities to work the additional inventory that would result from the implementation of this recommendation.  The IRS is developing an IRS-wide indicator to be used on those accounts on which identity theft has been documented.  The QRP plans to follow the IRS-wide decisions on identifying victims of identity theft.

Office of Audit Comment:  We believe freezing the subsequent year’s returns as we recommended is a more efficient and effective means of identifying repeated fraud, protecting Federal Government revenue, and protecting innocent taxpayers who are victims of identity theft.  Freezing these returns for a limited time would be preferable to attempting to collect fraudulent refunds that have already been issued.  We did not determine if the IRS’ actions regarding the identity theft indicator would protect innocent taxpayers’ refunds.

Recommendation 2:  Revalidate the optimal time needed to verify wage/withholding information prior to automatically releasing a refund.

Management’s Response:  The IRS agreed with this recommendation.  The Director, Refund Crimes, plans to review the 2007 data to obtain more current time periods for completing verification and is working to incorporate the use of Department of Health and Human Services employment data into the EFDS in this fiscal year.  This would automate the verification process, reduce the volume of manual verifications made, and reduce the time required to verify wage/withholding information.  The IRS plans to reconsider this change taking into account the time required for verification in this filing season, implementation of the Department of Health and Human Services employment data, and its commitment to expedite the review of returns so legitimate refunds can be released earlier.

The Internal Revenue Service Needs to Be More Aggressive in the Criminal and Civil Pursuit of Fraudulent Refunds

The CI function has been the primary stakeholder in the QRP, with support from the civil functions.  As mentioned in the Background section of this report, the mission of the Centers is to identify and detect refund fraud, prevent the issuance of fraudulent refunds, and provide support to the function’s field offices.  However, the process does not end with freezing refunds.  Basically, every fraudulent return identified requires some subsequent action.  Schemes need to be evaluated to determine if they should be referred to a field office for potential criminal investigation.  Fraudulent returns that do not meet the field office referral criteria should be monitored and accounts adjusted to reverse the fraudulent information.  In addition, Center employees may need to respond to refund or Taxpayer Advocate Service inquiries.  Further, if fraudulent refunds are issued, actions should be taken to pursue recovery of the refunds.

Referrals to CI function field offices

As part of their mission to provide support to the CI function field offices, the Centers analyze fraudulent returns to identify common return characteristics and patterns and a possible perpetrator.  The Centers will refer schemes to the CI function field offices for further evaluation if the amount of refunds claimed exceeds a certain dollar amount.  When a refund fraud investigation is referred for prosecution, there is great success in obtaining a conviction.  In Fiscal Year 2006, the function reported the results of QRP investigations obtained an overall conviction rate of 96.8 percent and received publicity on 81 percent of its QRP investigations. 

However, we noted the majority of the refund schemes are not being referred for potential criminal prosecution.  For example, we previously reported[13] that, as of March 2005, almost 91 percent of the 18,343 fraudulent prisoner returns identified during Processing Year 2004 were not referred for investigation or the investigations were closed without a prosecution recommendation.  In this review, we analyzed all of the approximately 118,000 fraudulent returns identified during Processing Year 2004 and determined about 85 percent were not referred for a primary investigation, the primary investigation was closed without a subject criminal investigation, or the subject criminal investigation was closed without a prosecution recommendation.  Figure 5 further illustrates this point. 

Figure 5:  Status of Investigations on Fraudulent Returns

Status of Schemes

Schemes

Related Fraudulent Returns

Refund Amount Claimed

Refund Amount Deleted (Stopped)

 

 

Schemes Without a Primary Investigation

2,624

87,255

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Schemes With a Primary Investigation

523

30,632

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Total Schemes Processed in
Processing Year 2004

3,147

117,887

$424,761,167

$295,882,147

 

Primary Investigations Closed Without a Resulting Subject Criminal Investigation

214

9,491

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Primary Investigations Open Without a Subject Criminal Investigation

137

4,586

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Primary Investigations With a Subject Criminal Investigation Opened

172

16,555

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Subtotal of Schemes With a Primary Investigation

523

30,632

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Subject Criminal Investigations Closed Without Prosecution

48

3,796

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Subject Criminal Investigations Referred for Prosecution

29

2,676

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Subject Criminal Investigations Still Open

95

10,083

**2,(c), 2(f)**

**2,(c), 2(f)**

 

Subtotal of Schemes With a Subject Criminal Investigation

172

16,555

**2,(c), 2(f)**

**2,(c), 2(f)**

Source:  Our analysis of the Scheme Tracking and Referral System and the Criminal Investigation
Management Information System as of March 31, 2006.

Our current review of 385 fraudulent refund returns, consisting of 194 different schemes, identified 152 schemes (78 percent) that met the criteria established by the CI function for referring cases to a CI function field office.  We compared this information to the CI function’s management information system and determined only 16 schemes had been referred for criminal investigation, ****1****

Some United States Attorney’s Offices may not be inclined to accept many tax fraud cases for prosecution because of the small dollar loss.  We had previously identified this as a cause in our report of the CI function’s efforts to increase legal source cases.[14]  In addition, according to three Review and Program Evaluation reports we reviewed, several of the function’s field offices received approval for an increased referral threshold (some as much as three times higher) because of the overwhelming volumes of fraudulent schemes versus the available resources to prosecute the schemes.  Further, Center staff advised us and CI function executives acknowledged some United States Attorney’s Offices prefer a high actual dollar loss before they will accept a case for prosecution.  Ironically, the fact that the Centers are proficient in stopping fraudulent refunds before issuance may have contributed to cases not being referred for prosecution because the dollars were not actually lost.

The CI function recently advised us it plans to conduct a review of the overall criteria for referring cases to its field offices for criminal investigation.  In addition, the Office of Refund Crimes staff acknowledged plans to change the QRP scheme referral criteria for Processing Year 2007 from the amount of refunds claimed to amount of refunds lost and will refer only those that meet a certain higher dollar loss amount.  While we understand the need to balance workload with available resources, we are concerned that this change of referring fewer cases for criminal prosecution could negatively affect tax administration by not holding perpetrators of smaller valued schemes accountable for their fraudulent actions.

Referrals to the Examination function

Not only are very few refund schemes referred for criminal investigation, the IRS also does not provide the necessary account maintenance to properly reflect the status of accounts or aggressively pursue the recovery of fraudulent refunds that were issued.  Those returns not meeting the criminal prosecution potential are retained at the Centers, and the Centers should refer the returns to either the Accounts Management organization or the Examination function to adjust the accounts to reflect the correct balances. 

Our review of 385 fraudulent refund returns determined 226 (59 percent), with refunds totaling about $2.2 million, should have been sent to the Examination function for either a fraud determination, an adjustment to reverse the ****2(a), 2(f)**** or ****2(a), 2(f)**** or an adjustment to an account for which a refund was issued.  At the time of our review, only 102 (45 percent) of the 226 fraudulent refund returns, totaling about $1.5 million, had had examination activity.  The examinations were still open on 20 cases and had been completed on 82 cases.  Adjustments had been made to the accounts to reverse the fraudulent refunds in 79 of the 82 cases, with $53,280 eventually being collected.  In the other three cases, the examinations were closed with no change to the accounts.  We did not determine if the remaining 124 of 226 returns had been referred to the Examination function or remained at the Centers.

Procedures in effect during Processing Years 2004 and 2005 required the CI function to refer a fraudulent return claiming a refundable credit(s) to the Wage and Investment Division for resolution because the law required that the taxpayer receive a statutory notice of deficiency whenever an assessment was made to increase the tax on the account.  The conference report accompanying the Taxpayer Relief Act of 1997[15] summarized the procedural rule as follows:

The IRS must follow deficiency procedures when investigating other types of questionable EITC claims.  Under these procedures, contact letters are first sent to the taxpayers.  If the necessary information is not provided by the taxpayers, a statutory notice of deficiency is sent by certified mail, notifying the taxpayer that the adjustment will be assessed unless the taxpayer files a petition in Tax Court within 90 days.  If a petition is not filed within that time, and there is no other response to the statutory notice, the assessment is made and the EITC is denied.

The following statistics, based on our analysis of the Scheme Tracking and Referral System and the EFDS, illustrate the impact of the current law on the Examination function to issue a statutory notice of deficiency to resolve frozen fraudulent refunds with the EITC: 

  • During Processing Year 2004, the Centers stopped 41,660 fraudulent returns claming over $133.1 million in EITC.  Each of these accounts would require a deficiency notice to reverse the EITC.
  • Also in Processing Year 2004, the Centers did not stop 35,134 returns claiming $124 million in fraudulent refunds, including $66 million in EITC that would require the IRS to issue a statutory notice of deficiency, to begin the step to collect the erroneous refunds that were issued.
  • During a previous audit, we reported that over 18,000 prisoners incarcerated during the entire Calendar Year 2003 filed returns as ****2(a), 2(f)**** or ****2(a), 2(f)**** and claimed over $19 million in ****2(a), 2(f)****[16]  Because these prisoners were incarcerated the entire year, they would not have had ****2(a), 2(f)**** or ****2(a), 2(f)****  The Centers did not identify these as potentially fraudulent returns; however, if they had, each would have required issuance of a statutory notice of deficiency to resolve the account.

The CI function has recognized the significance of this issue and has worked toward a change in the regulations or legislation that would treat fraudulent returns differently, referring to them as “nullity”[17] returns.  In June 2002, the IRS drafted a legislative proposal that would in effect allow the IRS to (1) assess as a math error an overstatement of any refundable credit based on a false information return (e.g., a Wage and Tax Statement [Form W-2]) and (2) assess as a math error any case in which a refund of tax is paid to any person other than to the person whose name is on the return and the return was signed by a person who was not authorized to sign the return.  These changes would have eliminated the need for the IRS to issue a statutory notice of deficiency; however, the legislative proposal was not adopted.  The IRS is once again seeking a regulatory change that will give it the ability to adjust as claim disallowances any refundable credits that were based on using false income documents.  We are encouraged by this and believe this type of change would allow the Examination function to use its resources to better address areas of noncompliance.

Throughout Processing Year 2006, the IRS implemented several changes to enhance the QRP, including improvements to the flow of work between the Centers and the civil functions.  In general, the Centers will transfer fraudulent returns that do not contain refundable credits to the Accounts Management organization for resolution and will transfer fraudulent returns containing the ****2(a), 2(f)**** or ****2(a), 2(f)**** to the Examination function.  There are other criteria for referring returns to either the Accounts Management organization or the Examination function.  The most notable of these is the difference in dollar threshold; there is no dollar tolerance for the Accounts Management organization, while only refunds greater than a specified tolerance will be forwarded to the Examination function ****2(a), 2(f)****. 

In addition, the IRS established time periods that would help in expeditiously resolving these accounts.  Taxpayers who receive a Wage and Withholding Verification Request (Letter 4115-C) are given 30 days to respond.  If the taxpayer provides insufficient information or does not respond by this deadline, the Accounts Management organization will disallow the claim.  Similarly, taxpayers who receive a CI Examination Referral (Letter 4116-C) are advised they will be contacted within 60 days. 

Although the improvements in notifying taxpayers and establishing time periods for referring accounts are a step in the right direction, we are concerned that fraudulent returns with claimed refunds below a certain dollar amount will not be forwarded to the Examination function for consideration.  The revised QRP procedures appear to be in contrast to lower tolerances used for other Examination function programs (e.g., National Research Program and Campus Examination Operations).  In those programs, the tolerances are based on an expected tax loss.  In the case of a fraudulent refund, the loss is often already known.  The revised QRP procedures indicate that returns will be forwarded to the Examination function if the refunds involved are greater than the tolerance, ****2(a), 2(f)****  The freezes will be released on returns with refunds below the tolerance and the refunds will be issued.

Using the Processing Years 2004 and 2005 Scheme Tracking and Referral System data provided to us by the CI function, we identified returns claiming the EITC and with a refund below the tolerance that the Centers stopped.  A further analysis of these data by stratum appears in Figure 6.

Figure 6:  Stratification of EITC Refund Returns

Stratification of Refund Amount

Processing Year 2004 Returns

Total Dollar Amount of Refunds Stopped

Processing Year 2005 Returns

Total Dollar Amount of Refunds Stopped

****2(c)****

321

****2(c)****

309

****2(c)****

****2(c)****

416

****2(c)****

1,069

****2(c)****

****2(c)****

164

****2(c)****

200

****2(c)****

****2(c)****

365

****2(c)****

464

****2(c)****

****2(c)****

636

****2(c)****

1,669

****2(c)****

****2(c)****

1,624

****2(c)****

2,701

****2(c)****

Totals

3,526

****2(c)****

6,412

****2(c)****

 

Source:  Our analysis of the Scheme Tracking and Referral System.

According to our analysis, the CI function stopped 3,526 returns with refunds below the tolerance totaling ****2(c)**** million during Processing Year 2004.  About 64 percent of the returns with refunds accounting for about 82 percent of the dollars were in the refund range between ****2(c)****  During Processing Year 2005, 6,412 returns with refunds below the tolerance were stopped, totaling about ****2(c)**** million.  Similarly, 68 percent of the returns with refunds accounting for about 85 percent of the dollars were in the range between ****2(c)****.  This shows that considerably more revenue could be protected if the dollar tolerance is lowered.  While we understand the need to balance resources with workload, we believe the IRS should consider lowering the tolerance to ensure it is maximizing its resources to protect the most possible revenue.  During our exit conference, Wage and Investment Division management reiterated the need for tolerances to ensure coverage in the EITC program area, given the limited resources available to work these cases.  IRS management also indicated that the return on investment associated with reviewing QRP cases is lower than that for other EITC work. 

Actions are needed to recover fraudulent refunds issued

During Processing Year 2004, over $124 million in refunds identified as fraudulent were issued.  Our sample of 385 fraudulent refund cases included 104 for which fraudulent refunds of about $1.5 million were issued, including over ****1**** issued to prisoners.  Collection action can begin only if the case is referred to the Examination function and an adjustment is made.  However, cases generally were not referred to the Examination function if fraudulent refunds were issued.  For example, 88 (85 percent) of the 104 cases, with refunds totaling $1.4 million, met the criteria for referral to the Examination function, but 71 were not referred by the Centers. 

Figure 7 shows the number of fraudulent refunds in our sample that were issued and actions taken by the IRS to recover the refunds.

Figure 7:  Status of Efforts to Collect Fraudulent Refunds Issued

All Refunds

Prisoner Refunds

Status of Refund

Refunds Issued

Total Refund Amount

Refunds Issued

Total Refund Amount

Not Referred for Examination

71

$489,886

****1****

****1****

Adjustment Made by the Examination Function

27

$938,245

****1****

****1****

Pending Adjustment by the Examination Function

6

$24,733

****1****

****1****

Total Fraudulent Refunds Issued

104

$1,452,864

****1****

****1****

Amount Collected

20

$53,28016

****1****

****1****

Total Uncollected[18]

99

$598,627

****1****

****1****

Source:  Our analysis of sample cases.

As of the date of our review, only 33 of the 104 cases had had examination activity; 27 cases were adjusted and $53,280 was collected.[19]  In addition, there has been little action taken to collect over $44,000 in refunds issued on 13 prisoner returns, as only 8 percent of the amount erroneously refunded has been collected.  We project that the IRS took no action to recover refunds issued in Processing Year 2004 totaling $81.5 million for 22,556 returns.[20]

While we realize a commonly held belief is that many of the fraudulent refunds, once issued, likely will not be recovered, the IRS has collected delinquent accounts through the notice process and by offsetting future refunds.  However, to offset a future refund, the IRS needs to make an adjustment on the taxpayer’s account to reverse the erroneous refund.  This currently requires a referral to the Examination function to issue the statutory notice of deficiency.

We further analyzed the 104 cases in our sample for which the fraudulent refunds were issued and identified 62 taxpayers who had filed either Tax Year 2004 and/or 2005 tax returns and had refunds of $285,477 frozen by the CI freeze.  As of the date of our review, $154,902 had been refunded to taxpayers who still owed $289,096 on 60 accounts previously determined to be fraudulent.  Had the IRS taken action to adjust the fraudulent accounts of these 60 taxpayers in prior years, it could have collected $97,601 through refund offsets.  For the estimated $81.5 million in fraudulent refunds issued during Processing Year 2004 that the IRS took no action to recover, we project the IRS could have collected $27.5 million from 11,013 taxpayer accounts though refund offsets alone.[21]  

Recommendations

The Chief, CI, should:

Recommendation 3:  Continue working with the United States Attorney’s Offices on the referral criteria for refund schemes, to ensure there is a balance between workload and effective tax administration.

Management’s Response:  The IRS agreed with this recommendation, stating the CI function continues to proactively collaborate with Department of Justice Tax Division and the Executive Office of United States Attorneys on the referral criteria for refund schemes, to ensure balance between workload and effective tax administration.

Recommendation 4:  In consultation with the IRS Office of Chief Counsel, initiate a legislative proposal to exempt the IRS from having to issue statutory notices of deficiency for disallowance of the ****2(a), 2(f)**** and ****2(a), 2(f)**** when the deficiencies and credits are the result of fraudulent returns, if its current efforts through a regulatory change are not successful.  This includes ****2(a), 2(f)**** claims made by inmates with a filing status of ****2(a), 2(f)**** or ****2(a), 2(f)**** who are incarcerated the entire calendar year.

Management’s Response:  The IRS did not agree or disagree with this recommendation.  The CI function plans to work with the IRS Offices of Legislative Affairs and Chief Counsel to discuss initiating a legislative proposal regarding disallowance of the ****2(a), 2(f)**** and ****2(a), 2(f)**** on fraudulent returns.  Legislative proposals are the responsibility of the Assistant Secretary for Tax Policy, who must weigh tax administration issues with other tax policy considerations such as simplicity and fairness.

Office of Audit Comment:  While the CI function agreed to work with the IRS Offices of Legislative Affairs and Chief Counsel to discuss such a proposal, we are disappointed that the IRS is not vigorously pursing this recommendation.  We believe the current legal requirement to issue statutory notices in certain cases and IRS procedures that require the Examination function to issue the notices are the driving factors behind the IRS’ use of higher tolerances in these cases and its failure to initiate civil action to recover fraudulent refunds that were issued.  Although legislative proposals are the responsibility of the Assistant Secretary for Tax Policy, the IRS had previously drafted a legislative proposal in June 2002 that would in effect have eliminated the need for the IRS to issue a statutory notice of deficiency for certain types of cases.  We believe this recommendation has the potential to improve tax administration while still protecting taxpayer rights.  We continue to believe the IRS should take a leading role in pursuing a legislative change, while working with the Assistant Secretary for Tax Policy.

Recommendation 5:  Consult with the Executive Steering Committee and consider lowering the tolerances for sending cases to the Examination function.

Management’s Response:  The IRS did not agree or disagree with this recommendation, stating that, while the IRS would like to address all cases of potential noncompliance, its enforcement resources are limited.  Each year, top IRS management makes carefully thought-out business decisions about how to deploy these resources.  Tradeoffs must be made among competing priorities, and criteria will be established each filing season based on these business decisions.

Office of Audit Comment:  We agree IRS resources are limited and the IRS is not able to address all cases of noncompliance.  We recognize it is important to have processes and procedures that allow the IRS to use those resources efficiently and effectively.  In general, examination cases involve one taxpayer filing one or two returns on which the refunds have already been issued.  Refund fraud, on the other hand, sometimes involves one or two taxpayers filing multiple (sometimes hundreds of) fraudulent returns.  In most cases, the refunds are stopped.  Both types of cases require adjustments and issuance of notices to the taxpayers.  However, due to the greater impact on revenue loss, we believe it is appropriate for the IRS to have a lower individual return tolerance in cases of refund fraud, to prevent the issuance of tens of thousands of dollars in fraudulent refunds.  Implementing the legislative change in Recommendation 4 would allow the IRS to lower its tolerances, drastically reduce the number of cases that would need to be referred to the Examination function, and reduce the number of already limited resources needed to process these cases.

Changes During Processing Year 2006 Had a Detrimental Impact on Identifying Fraudulent Returns and Will Have an Undeterminable Effect on Processing Year 2007

Because of the IRS’ failure to implement the EFDS during Processing Year 2006, the Office of Refund Crimes had to rely on other means to identify fraudulent refunds.  The Centers relied on four primary sources for identifying fraudulent refunds during Processing Year 2006:  returns that were frozen due to a prior year fraud, certain prisoner returns, internal leads, and external leads. 

During Processing Year 2005, the Centers identified $515.5 million in fraudulent refunds and stopped $412.8 million.  However, during Processing Year 2006, as of September 29, 2006, the Centers had identified only $232.3 million in fraudulent refunds and stopped $164.2 million.  Figure 8 shows the number of refund returns received, reviewed, and verified as fraudulent and the related refund amounts by source.

Figure 8:  Fraudulent Returns Identified During Processing Year 2006

Source of Receipt

Returns Received

Returns Verified As Fraudulent

Fraudulent Refunds Claimed

Fraudulent Refunds Stopped

External Leads

21,108

10,767

$50,403,184

$23,344,017

Internal Leads

55,846

14,622

$110,772,137

$85,610,613

2006 Prisoner List

43,064

4,235

$18,962,765

$11,513,967

2006 Unpostables

125,565

15,164

$52,174,848

$43,760,092

Totals

245,583

44,788

$232,312,934

$164,228,689

Source:  The Office of Refund Crimes Program Analysis Report, dated September 29, 2006.

We believe the following two changes could have had a material impact on identifying fraudulent returns and stopping fraudulent refunds during Processing Year 2006 and will have an undeterminable impact during Processing Year 2007.

The Centers did not verify tax returns with ****2(a), 2(f)**** during Processing Year 2006

If the Centers identified a suspicious tax return claiming only ****2(a), 2(f)**** they would ****2(e)****  However, many tax returns without ****2(a), 2(f)**** appear suspicious but include a ****2(a), 2(f)**** that can be ****2(e)****  In prior years, if the Centers identified a pattern with these ****2(a), 2(f)**** they would freeze the accounts and place them into a scheme.  These schemes may be referred to a CI function field office for criminal investigation or to the Examination function for a civil audit. 

Because tax returns with a ****2(a), 2(f)**** require a referral to the Examination function, the IRS decided not to verify potentially fraudulent ****2(a), 2(f)**** returns[22] and instead issued the refunds in Processing Year 2006.  As Figure 8 indicates, the Centers verified only 15,164 (12 percent) of the unpostable returns as fraudulent.  However, because the Centers did not, and could not, verify returns with a ****2(a), 2(f)****, the Office of Refund Crimes did not report how many returns had a ****2(a), 2(f)****.  At our request, the Office of Refund Crimes staff extracted data from their system to identify the number of tax returns with a ****2(a), 2(f)**** that were unpostable.  We chose the unpostable category because this represents returns that were sent to the Centers for review because the prior year’s returns were determined to be fraudulent.  The Office of Refund Crimes reported that, as of June 25, 2006, 24,464 (23 percent) of the 125,565 unpostable returns had a ****2(a), 2(f)****  If this percentage of unpostables and the average fraudulent refund amount of $3,441 as of September 29, 2006, were representative of the total applicable population, we calculate about $99.4 million in potentially fraudulent refunds were issued on about 28,880 returns with a ****2(a), 2(f)**** that were frozen due to a prior year fraud and were not verified.

In addition, we recently reported[23] that one Center had identified tens of thousands of returns on which individuals used valid but stolen Social Security Numbers to file fraudulent tax returns and claim dependents on Forms 1040 with a ****2(a), 2(f)****  About 70,000 tax returns were not accepted for criminal prosecution by the CI function field offices after $100 million in potential refunds had been temporarily stopped.  The taxpayer accounts on which the refunds were stopped needed to be reviewed by other IRS functions to ensure the Center had not stopped any refunds that should have been issued to legitimate taxpayers and to reverse erroneous credits that would have caused fraudulent refunds. 

The CI function does not plan to include returns with a ****2(a), 2(f)**** in its normal screening process during Processing Year 2007.  Instead, it will rely on a prerefund screening process during normal returns processing to identify and stop fraudulent refunds with a ****2(a), 2(f)****  Under this process, returns meeting specific criteria are identified and the refunds delayed while the IRS attempts to verify the returns.  However, we are concerned the criteria used during the prerefund screening will not be as effective in detecting fraudulent returns.  The criteria are based on a limited number of factors, are applied on individual returns, and in our opinion do not address the many variables and patterns of filing multiple fraudulent returns found in refund schemes.  We are also concerned the volume of potentially fraudulent ****2(a), 2(f)**** returns will overwhelm the IRS and prevent it from auditing many of the returns identified during prerefund screening before the refunds are issued.  During Processing Year 2004, the Centers identified 30,500 fraudulent returns with a ****2(a), 2(f)**** that claimed over $84 million in refunds. 

If the IRS cannot audit returns during the prerefund screening, it plans to audit some of the returns after the refunds have been issued.  However, IRS statistics show only 3 percent of this type of return is audited, and, even if the returns were audited, the refunds would already have been issued.  In a recent USA Today article, the IRS Commissioner was quoted as saying, “If you are looking at stopping fraud on refunds, you either stop it upfront, or it’s very hard to get the money back.” 

The IRS’ decision not to screen ****2(a), 2(f)**** returns is disturbing in light of recent comments by the former Director, Refund Crimes, that a tax fraud trend is an increase in the number of returns listing fraudulent ****2(a), 2(f)****  The CI function advised us it plans to analyze the data mining scores of ****2(a), 2(f)**** returns early in Processing Year 2007 to determine the effectiveness of these alternate prerefund screening procedures.  We will evaluate these procedures during our follow-on review.

While we recognize the difficulty in verifying ****2(a), 2(f)**** returns and acknowledge there is a risk of improperly freezing an innocent taxpayer’s account, we believe these risks are reduced based upon the new procedures to notify taxpayers’ their refunds are frozen.  Further, we believe the potential loss of revenue is of the magnitude that the IRS needs to reevaluate how it processes returns that include a ****2(a), 2(f)****

Prisoner returns claiming the ****2(a), 2(f)**** and/or with a ****2(a), 2(f)**** were not subject to screening

In April 2005, the House Ways and Means Subcommittee on Oversight expressed concerns about the increase in refund fraud committed by individuals who are incarcerated in Federal and State prisons.  The Oversight Subcommittee members asked us to assess the IRS’ procedures to detect fraudulent and potentially fraudulent prisoner refund returns.  We issued a report[24] in response to that request and agreed to conduct follow-on work in response to the Congressional concerns.

Due to the failures of the EFDS in Processing Year 2006, the Centers were unable to identify prisoner returns through data mining techniques.  Instead, the CI function used various criteria to freeze prisoner refunds for tax returns on which the identifying information matched the same information from a prisoner file obtained each year from the Federal, State, and local prisons.  However, these criteria did not include returns claiming the ****2(a), 2(f)**** or returns with a ****2(a), 2(f)**** 

CI function management advised us they had ****2(a), 2(f)**** in part, because they believed this category was declining.  Also, we were advised the CI function had little time to implement a system to identify prisoner returns because the EFDS was not operational.  We are also concerned that the function established a tolerance for prisoner returns and then increased that tolerance due to the high volume of frozen prisoner returns.

During a previous audit,[25] we analyzed a database containing fraudulent returns identified by the Centers during Processing Year 2004 and identified 4,554 prisoner returns with the ****2(a), 2(f)**** that claimed $11.9 million in refunds.  Prisoners also filed 1,314 fraudulent returns with a ****2(a), 2(f)****, claiming $5 million in refunds (894 of the refunds for $2.4 million also involved claims for the ****2(a), 2(f)****).

In addition to the 4,554 fraudulent prisoner returns claiming the ****2(a), 2(f)**** during Processing Year 2004, we reported that over 18,000 prisoners incarcerated during the entire Calendar Year 2003 filed returns with a filing status as ****2(a), 2(f)**** or ****2(a), 2(f)**** and claimed over $19 million in ****2(a), 2(f)****  Because these prisoners were incarcerated for the entire year, they would have had neither ****2(a), 2(f)**** to qualify for the ****2(a), 2(f)**** nor a ****2(a), 2(f)****

Figure 8 showed only 4,235 prisoner returns claiming approximately $19 million in refunds were identified as fraudulent in Processing Year 2006 and only $11.5 million in refunds were stopped.  In contrast, during Processing Year 2004 year, 18,159 prisoner returns claiming $68.2 million in fraudulent refunds were identified and 14,033 refunds totaling $53.5 million were stopped.  This shows the potential magnitude of the IRS’ lost ability to detect and stop fraudulent prisoner refunds during Processing Year 2006.

The CI function plans to have the EFDS operational during Processing Year 2007.  However, we remain concerned about how fraudulent prisoner returns will be identified.  The CI function has requested programming changes to the EFDS that will effectively eliminate a certain category of prisoner refunds from the screening process, believing prisoners in this category are less likely to commit fraud.  However, during Processing Year 2004, the Centers identified 521 prisoner returns in this category that claimed almost $1.5 million in fraudulent refunds.  Although this is not a large volume, we believe extra efforts are needed to prevent prisoners from obtaining fraudulent refunds.  Further, it is possible that prisoners could eventually learn this type of return is not screened and begin to file more returns of this type.

Recommendation

Recommendation 6:  The Chief, CI, should reconsider the decision to exclude certain returns from the Center screening process.

Management’s Response:  The IRS did not agree or disagree with this recommendation, stating that, for the 2007 Filing Season, rather than using the Center screening process, the IRS used its Dependent Database Program (a prerefund ****2(a), 2(f)**** program) for screening and selecting for examination those returns referenced in the recommendation with the ****2(a), 2(f)**** In addition, returns submitted by prisoners were identified and forwarded to the Small Business/Self-Employed Division for examination.  The CI function and the Small Business/Self-Employed Division plan to conduct an analysis to determine if the current selection process should be modified.

Office of Audit Comment:  On several occasions, we have expressed concern about the effectiveness of using the Dependent Database Program.  The CI function assured us this Program would identify certain types of returns and advised us it plans to analyze the data mining scores of these returns early in Processing Year 2007, to determine the effectiveness of these alternate prerefund screening procedures.  On March 30, 2007, prior to the date of IRS management’s response to this report, we were advised the CI function had identified a large gap between the cases identified by the function and those identified by the Dependent Database Program.  The CI function concluded that not everything is being identified by the Dependent Database Program.  As a result, IRS management was already aware the 2007 Filing Season screening process was ineffective for certain returns, but they did not indicate in their response what changes were needed. 

Actions Taken and Decisions Made by the Fraud Detection Centers Were Generally Proper but Not Always Timely or Supported

As previously mentioned, the Centers were successful in protecting revenue by preventing fraudulent refunds from being issued.  However, this process, combined with the CI function’s emphasis on freezing returns to protect the Federal Government’s interest and a rapidly growing number of fraudulent returns identified each year, resulted in the QRP’s management and inventory problems.  We previously reported that Center personnel were not timely removing controls placed on taxpayers’ accounts and were not always taking timely actions to resolve accounts that lacked criminal prosecution potential.[26]

In this audit, we reviewed 385 fraudulent refund returns identified during Processing Year 2004 and 104 taxpayer accounts with a frozen refund as of September 30, 2005, and on which the account balance was $0 as of December 31, 2005 (indicating the refund situation for these 104 accounts had been resolved). 

Incorrect determinations of fraud exist but are not as widespread as previously indicated

One of the Advocate’s concerns was that a large number of taxpayers whose refunds were frozen and who complained to the Taxpayer Advocate Service eventually received their refunds.  This was especially troublesome because the CI function had previously concluded that these were fraudulent refunds.  The Advocate’s sample of 398 “decided” cases (taken from the Taxpayer Advocate Service’s closed inventory during Fiscal Year 2004 and the first 6 months of Fiscal Year 2005) identified 66 percent that resulted in taxpayers receiving refunds.[27]  Included in the 398 cases were 142 returns categorized as involving current year fraud.  The Taxpayer Advocate Service determined that 65 (46 percent) of the 142 returns were not fraudulent and the taxpayers received full refunds; 43 of the 65 cases were from 1 Center. 

The CI function responded to the Advocate’s conclusion regarding the pool of innocent taxpayers negatively affected by the function’s account freezes as follows:

[The] Taxpayer Advocate Service’s conclusion was based on an extrapolation from a limited sample of [the] Taxpayer Advocate Service’s cases and innocent taxpayers are much more likely to contact [the] Taxpayer Advocate Service than those who have filed false or fraudulent returns.  To extrapolate their conclusions from Taxpayer Advocate Service cases to the whole universe of stopped refunds inflates the universe of innocent taxpayers affected by CI [function] fraud detection programs.

Our case review (taken from a sample of the approximately 118,000 fraudulent refund returns identified during Processing Year 2004) determined there were incorrect determinations of fraud, but they were not as widespread as originally indicated by the Advocate’s Report.  We identified only 8 of 385 returns that were later determined to be not fraudulent.  In addition, we obtained a computer file of 6,831 returns that were originally determined to be fraudulent during Processing Year 2004 but were subsequently determined to be valid refunds.  This represents 5.8 percent of the returns originally determined to be fraudulent during Processing Year 2004.

We believe several factors contributed to the inaccurate determinations of fraud.  First, although the Centers confirm employment with an employer, the employer may have provided an incorrect response because the employee may have given the employer an incorrect Social Security Number or the employer recorded the information incorrectly.  In addition, due to the high volume of workload, one Center did not have time to determine if fraud existed and permanently froze accounts intending to review them later.  However, a fraud determination never occurred and the accounts remained frozen, giving an impression that the returns were fraudulent.

We also compared our sample of 385 fraudulent refund returns to the Advocate’s management information system and identified 50 cases (13 percent) that had an inquiry.  This is nearly the same percentage (14 percent) as that for all inquiries the Taxpayer Advocate Service received during 2004 (see Figure 3).  In 27 (54 percent) of the 50 cases, the Taxpayer Advocate Service inquiry related to our sample year; in 24 cases, the inquiry was related to a subsequent year.[28]  The taxpayers received full refunds in 4 (15 percent) of the 27 cases involving frozen refunds for the current year’s returns.  This percentage is significantly less than the results identified by the Advocate’s sample.  We believe the primary reason for the differing rates is the Advocate’s sample consisted only of taxpayers who filed a complaint with its office.  The CI function believes these taxpayers are much more likely to contact the Advocate’s office than those who have filed false or fraudulent returns.  The Advocate’s office recognized this and advised us it did not intend to imply its results were indicative of the entire population of frozen refunds, only that there were some innocent taxpayers within the CI function’s population of fraudulent returns.

Our sample also had taxpayer complaints on 24 (16.1 percent) of the 149 cases for which the subsequent year’s return was filed and frozen.  The taxpayers received full refunds in 14 (58.3 percent) of the 24 cases.  Although lower, the percentage of taxpayers receiving refunds is closer to the 79 percent from the Advocate’s sample.  We believe the higher refund rate in subsequent years can be attributed to the CI function’s previous practice of freezing refunds and not notifying taxpayers or not timely verifying the frozen refunds.  If the CI function reinstates the freezing of subsequent refunds, we believe the number of Taxpayer Advocate Service inquiries will be reduced because of the new procedures for notifying the taxpayer of the freeze and establishing time limits to resolve the freeze.

Account resolution is proper; however, it is not timely

Our review of 104 accounts with frozen refunds as of September 30, 2005, and for which the account balance was $0 as of December 31, 2005, identified the following:

·         The Centers released 42 of 104 refunds in whole or in part because they determined the refunds were not fraudulent.  The Centers or the Examination function later determined that 62 refunds were fraudulent and made the appropriate adjustments to the accounts to reverse the refunds.

·         In our opinion, the Centers did not timely resolve 42 of the 104 cases.[29]  Specifically, the Centers took an average of 262 days to release the refunds on 22 accounts for which they verified the refunds were good and an average of 726 days to make adjustments to resolve 20 accounts for which the refunds were verified as fraudulent. 

Improvements are needed in documentation of fraud determinations

Our sample of 385 fraudulent refund cases included 287 returns with wages or other income documents and another 98 returns with a ****2(a), 2(f)****  In 41 (14.3 percent) of the 287 returns for Processing Year 2004, we were not provided sufficient documentation to support the determination that the return was fraudulent.  In some of these instances, the Centers provided us with documentation of fraud based on documents that did not exist at the time the fraud was determined.[30]  Although this information confirmed that the Center had made the correct determination, the initial determination was based on information not available at the time.  In addition to these 41 returns, we identified 4 other cases in which the Centers indicated the filer was a prisoner; however, documentation was not provided supporting the assertion the filer was incarcerated throughout the entire tax year.  Further, the Centers did not provide us with documentation to show how the fraud determination was made on 10 of 104 frozen credit cases and provided us with incomplete documentation in another 14 cases.  It is good business practice and important for the Centers to maintain documentation of how they arrived at the determination that a return is fraudulent.

We believe any concerns that existed relative to the timely actions of the IRS to resolve accounts have been reduced by the taxpayer notification process implemented this past year and the revised procedures implemented by the Office of Refund Crimes.  However, it is necessary for the Centers to improve documenting how they make fraud determinations.  This can be achieved by reemphasizing the requirement to maintain sufficient documentation and/or notes of how fraud was determined.

Recommendation

Recommendation 7:  The Chief, CI, should reemphasize the requirement to maintain documentation to describe how Centers determined fraud.  The CI function’s Review and Program Evaluation office staff should include adherence to this requirement in their periodic reviews of the Centers.

Management’s Response:  The IRS agreed with this recommendation.  The CI function has established a task force to consider revising review procedures, including the adherence to maintaining documentation and/or notations.  The function plans to conduct formal reviews in the Centers beginning in the fourth quarter of Fiscal Year 2007. 

 

Appendix I

 

Detailed Objective, Scope, and Methodology

 

The overall objective of this review was to determine the effectiveness of the IRS’ procedures for detecting fraudulent and potentially fraudulent refund income tax returns.  To accomplish this objective, we planned to evaluate the controls over and procedures for properly managing the inventory of fraudulent refund returns.  Specifically, we:

I.                        Evaluated the controls over and procedures for ensuring the Fraud Detection Centers (the Centers) correctly manage inventory, including the proper freezing[31] of fraudulent refund returns and the timely and proper release of freezes no longer needed.

A.    Evaluated new procedures and controls developed by the Office of Refund Crimes and the Executive Steering Committee as a result of the National Taxpayer Advocate’s 2005 Annual Report to Congress.  These procedures included, but were not limited to:

1.      Freezing subsequent years’ returns.

2.      Notifying taxpayers that their refunds are delayed.

3.      Releasing existing freezes.

4.      Adjusting taxpayer accounts.

B.     Through assessment of the electronic data sources used in this audit, concluded the data were of undetermined reliability.  However, answering the audit’s objective would not be feasible if the data were not used, and it was our opinion that using the data would not weaken the analysis or lead to an incorrect or unintentional message.  Additional steps to determine data reliability prior to testing were not feasible.  For samples selected, the electronic data were validated to source documents or to the Master File.

C.     Selected a statistically valid sample of 385 fraudulent refund returns from the universe of about 118,000 fraudulent returns identified by the 10 Centers during Processing Year 2004 and placed in the Scheme Tracking and Referral System.  The sample had a confidence level of 90 percent, a precision level of +3 percent, and an estimated error rate of 15 percent.

1.      Analyzed Master File account transcripts to determine if the CI function had input a freeze on the account, the type of freeze, when it was input, and when it was removed.[32]  We also looked for indications that the taxpayer was involved in a prior year fraud and that adjustments were made and/or cases were referred to the Examination function or Collection function, as appropriate.

2.      Determined if the fraud determination was properly supported (e.g., wage verification was conducted) by reviewing documentation provided by the Centers and discussing with applicable Center personnel.[33]

3.      Analyzed the Master File account transcripts for each sampled case and determined if the taxpayer filed tax returns during Processing Years 2005 and/or 2006.  We made the following assessments by reviewing the account transcripts, analyzing documentation provided by the Centers, and discussing with applicable Center personnel:[34]

a.       Was there a freeze on the account?

b.      Did the Center timely review the account?

c.       What is the status of the refund?

4.      Determined if schemes were referred to the CI function field offices for potential prosecution.

a.       Obtained from the Scheme Tracking and Referral System a computer file of about 118,000 returns determined to be fraudulent by the function during Processing Year 2004.

b.      Compared the file to the Criminal Investigation Management Information System as of March 30, 2006, and determined the status of the schemes in the Scheme Tracking and Referral System.

c.       Compared each of the schemes represented in our sample to the Criminal Investigation Management Information System data and reviewed documentation provided by the Centers.

5.      Obtained data from the Scheme Tracking and Referral System for Processing Years 2004 and 2005 and identified the number of refunds with the EITC that were determined to be fraudulent by the CI function.

6.      Determined if a Taxpayer Advocate Service inquiry was received by reviewing data from the Taxpayer Advocate Management Information System and documentation provided by the Centers.

D.  Determined the potential impact of the new procedures on Processing Years 2006 and 2007.

1.      Obtained information from the CI function on the source; the number of returns received, returns verified as fraudulent, and refunds claimed; and the amount of refunds stopped during Processing Year 2006.

2.      Obtained data from the CI function data to determine the number of returns from Step I.D.1. that were frozen (unpostable) due to a prior year freeze and had a ****2(a), 2(f)****

3.      Obtained from the Scheme Tracking and Referral System data for fraudulent returns identified by the CI function during Processing Year 2004 and determined the number of:

a.       Returns with a ****2(a), 2(f)****

b.      Prisoners that claimed the ****2(a), 2(f)**** or filed a return with a ****2(a), 2(f)****

4.      Discussed procedures with various IRS personnel and obtained information about the IRS’ plans to identify and stop refunds on returns of prisoners or with a ****2(a), 2(f)****

E    Obtained from the CI function data extracts of its frozen credit file as of September 30, 2005, containing 238,466 frozen credits, and December 31, 2005, containing 207,632 frozen credits. 

1.      Compared the 2 files to identify the number and amount of refunds released during the 3 months.  We identified 40,663 accounts[35] totaling about $176 million in refunds that were no longer frozen as of December 31, 2005.

2.      From the universe of 40,663 accounts identified in Step I.E.1., selected a judgmental sample of 30 cases from each of the Centers in Atlanta, Georgia; Fresno, California; and Kansas City, Missouri, for a total of 90 cases.  We selected a judgmental sample because we did not intend to project our results to the universe.  Further, we selected all 14 accounts with a credit balance exceeding $500,000, for a total of 104 cases.

3.      Through interviews with Center personnel, case reviews, and review of Master File account transcripts, determined:

a.       How and why the refund was released.

b.      If the release was proper and appropriately supported.

c.       If the Center timely reviewed the account.

d.      If the freeze was the result of a prior year or current year fraud.

 

Appendix II

 

Major Contributors to This Report

 

Daniel R. Devlin, Assistant Inspector General for Audit (Small Business Programs)

Nancy A. Nakamura, Assistant Inspector General for Audit (Headquarters Operations and Exempt Organizations Programs)

John R. Wright, Director

Diana M. Tengesdal, Audit Manager

Michael J. Hillenbrand, Lead Auditor

Joseph F. Cooney, Senior Auditor

Jeff K. Jones, Senior Auditor

Janice A. Murphy, Senior Auditor

Ahmed Tobaa, Senior Auditor

Robert Carpenter, Information Technology Specialist

 

Appendix III

 

Report Distribution List

 

Commissioner  C

Office of the Commissioner – Attn:  Chief of Staff  C

Deputy Commissioner for Services and Enforcement  SE

Commissioner, Small Business/Self-Employed Division  SE:S

Commissioner, Wage and Investment Division  SE:W

Chief Counsel  CC

Division Counsel/Associate Chief Counsel (Criminal Tax)  CC:CT

Director, Office of Governmental Liaison and Disclosure  CL:GLD

Director, Campus Compliance Services  SE:S:CCS

Director, Compliance  SE:W:CP

Director, Operations Policy and Support, Criminal Investigation  SE:CI:OPS

Director, Refund Crimes, Criminal Investigation  SE:CI:RC

National Taxpayer Advocate  TA

Director, Office of Legislative Affairs  CL:LA

Director, Office of Program Evaluation and Risk Analysis  RAS:O

Office of Internal Controls  OS:CFO:CPIC:IC

Audit Liaisons:

Commissioner, Small Business/Self-Employed Division SE:S

Commissioner, Wage and Investment Division  SE:W

Chief, Criminal Investigation  SE:CI:S:PS

 

Appendix IV

 

Outcome Measures

 

This appendix presents detailed information on the measurable impact that our recommended corrective actions will have on tax administration.  These benefits will be incorporated into our Semiannual Report to Congress.

Type and Value of Outcome Measure:

Revenue Protection – Potential; $71.7 million in fraudulent refunds would have been prevented from issuance due to the subsequent year freeze (see page 9).

Methodology Used to Measure the Reported Benefit:

We obtained an extract from the Scheme Tracking and Referral System[36] containing about 118,000 fraudulent refund returns identified by the CI function Fraud Detection Centers during Processing Year 2004.  We stratified the population of fraudulent refund returns by refund amount into five categories.  We did not select fraudulent returns with a refund amount less than $0 (i.e., balance-due returns) because we did not want our sample projections to be distorted. 

Our estimates are based on review of a statistically valid sample of 385 fraudulent refund returns handled by the CI function in Processing Year 2004.  The statistical samples were selected using attribute and variable sampling methods with a 95 percent confidence level and precision of +24 percent (returns) and +27 percent (dollars).[37] 

We analyzed account information for the 385 sample cases and identified 159 taxpayers who also filed returns during Processing Year 2005; returns for 149 of the 159 taxpayers were frozen, and 63 (42.3 percent) were determined to be fraudulent.  Based on these results, we project that 20,078 returns with refunds totaling $71.7 million would have been determined to be fraudulent during Processing Year 2005. 

Type and Value of Outcome Measure:

Increased Revenue – Potential; $54 million in fraudulent refunds could have been recovered if these cases had been referred to the Examination function for an adjustment (see page 15).

Methodology Used to Measure the Reported Benefit:

Using the sample described above, we concluded the CI function issued fraudulent refunds to 71 taxpayers and did not take subsequent action to recover the money (see Figure 7 in the report). 

Based on these results, we project the IRS could have taken action to recover refunds totaling $81.5 million for 22,556 returns.  Our projection is based upon a statistically valid sample with a 95 percent confidence level and a precision of +27 percent (dollars) and +22 percent (returns).  As reported in the following outcome measure, we project the IRS could have collected $27.5 million of the $81.5 million through refund offsets, leaving a net amount of $54 million.

Type and Value of Outcome Measure:

Increased Revenue – Potential; $27.5 million in fraudulent refunds could have been recovered through the use of refund offsets (see page 15).

Methodology Used to Measure the Reported Benefit:

Using the sample information described above, we reviewed 385 fraudulent refund returns and concluded the IRS issued 104 fraudulent refunds.  We analyzed the 104 cases in our sample and identified 62 taxpayers who filed either Tax Year 2004 and/or 2005 tax returns and had refunds of $285,477 frozen by the CI freeze.  As of the date of our review, $154,902 had been refunded to taxpayers who still owed $289,096 on 60 accounts previously determined to be fraudulent.  Had the IRS taken action to adjust the accounts of these taxpayers in prior years, it could have collected an additional $97,601 through refund offsets from these 60 taxpayers.

Based on these results, we project the IRS could have collected an additional $27.5 million from 11,013 taxpayer accounts.  Our projection is based on a statistically valid sample with a 95 percent confidence level and a precision of +39 percent (dollars) and +33 percent (returns).

 

Appendix V

 

Related Treasury Inspector General for Tax Administration Audit Reports

 

  1. The Internal Revenue Service Can Improve the Effectiveness of Questionable Refund Detection Team Activities (Reference Number 2000-40-018, dated December 1999).
  2. Improvements Are Needed in the Monitoring of Criminal Investigation Controls Placed on Taxpayers’ Accounts When Refund Fraud Is Suspected (Reference Number 2003-10-094, dated March 2003).
  3. The Criminal Investigation Function Has Made Progress in Investigating Criminal Tax Cases; However, Challenges Remain (Reference Number 2005-10-054, dated March 2005).
  4. ****2(f)****
  5. ****2(f)****
  6. The Electronic Fraud Detection System Redesign Failure Resulted in Fraudulent Returns and Refunds Not Being Identified (Reference Number 2006-20-108, dated August 9, 2006).

 

Appendix VI

 

Glossary of Terms

 

Accounts Management Organization – This is the organization within the Wage and Investment Division responsible for taxpayer relations by answering tax law/account inquiries and adjusting tax accounts.  In addition, it is responsible for providing taxpayers with information on the status of their returns/refunds and for resolving the majority of issues and questions to settle their accounts.

Campus – The data processing arm of the IRS.  The campuses process paper and electronic submissions, correct errors, and forward data to the Computing Centers for analysis and posting to taxpayer accounts.

Criminal Investigation Freeze/Freezing – When a Fraud Detection Center determines that a refund is potentially fraudulent, it places a computer control on the taxpayer’s account.  This control freezes the account and prevents a future refund from being issued.

Criminal Investigation Management Information System – A database that tracks the status and progress of criminal investigations and the time expended by special agents.  It is also used as a management tool that provides the basis for decisions of both local and national scope.

Data Mining – The process of automatically searching large volumes of data for patterns.

Electronic Fraud Detection System – The primary source for the identification of leads on fraudulently filed tax returns; however, leads are also received from sources internal to and external from the IRS.

Filing Season – The period from January through mid-April when most individual income tax returns are filed.

Master File – The IRS database that stores various types of taxpayer account information.  The database includes individual, business, and employee plans and exempt organizations data.

Primary Investigation – The development of an investigation on individuals or entities when it appears there is prosecution potential.

Processing Year – The year in which taxpayers file their returns with the IRS.  For example, most Tax Year 2004 returns were filed in Processing Year 2005.

Questionable Refund Program – A nationwide program established to detect and stop fraudulent claims for refunds on income tax returns.

Refund Offset – A computer program that will automatically apply a refund due a taxpayer to another account on which the taxpayer owes money to the IRS.

Review and Program Evaluation Report – A report prepared by CI function staff located in the Headquarters office containing the results of periodic operational reviews of CI function field offices.

Scheme – A scheme could include only one return but generally includes numerous returns.  In addition, many small fraudulent refunds that do not have common characteristics may be placed in a “dump” scheme.

Scheme Tracking and Referral System – The system of records maintained at each Fraud Detection Center for QRP and Return Preparer Program schemes.  It was designed to store information, for multiple processing years, that is used for tracking and historical purposes.

Subject Criminal Investigation – An investigation developed when an individual or entity is alleged to be in noncompliance with tax laws and there is prosecution potential.  The objective of a Subject Criminal Investigation is to gather evidence to prove or disprove the existence of a violation of the laws enforced by the IRS.

Taxpayer Advocate Management Information System – A database of the Taxpayer Advocate Service that is exclusively dedicated to the recordation, control, and processing of Taxpayer Advocate Service taxpayer cases and to the capturing and analysis of core tax issues, laws, policies, and internal IRS functional processes that are the sources of significant taxpayer hardship and other critical problems.

Taxpayer Advocate Service – An independent organization within the IRS that helps taxpayers resolve problems with the IRS and recommends changes to prevent problems.

Unpostable Transaction – Those transactions that cannot be posted to the Master File because of a specific freeze code (e.g., CI freeze).  A transaction that fails to post to an account is returned to the Submission Processing site and is referred to the appropriate area for corrective action (e.g., cases with a CI freeze will go to the Fraud Detection Centers for review).

 

Appendix VII

 

Management’s Response to the Draft Report

 

The response was removed due to its size.  To see the response, please go to the Adobe PDF version of the report on the TIGTA Public Web Page.



[1] Appendix VI includes a glossary of terms used in this report.

[2] For more information, see Appendix V, reports 6, 2, and 5.

[3] For more information, see Appendix V, reports 1 and 2.

[4] Appendix VI includes a glossary of terms used in this report.

[5] For more information, see Appendix V, report 6.

[6] For more information, see Appendix V, reports 2 and 5.

[7] Questionable Refund Program Phase II (Audit # 200710024).

[8] We excluded 14 returns claiming almost $2 billion from this number, including ****1****

[9] For more information, see Appendix V, reports 1 and 2.

[10] During the verification process, Center staff will contact employers or payers to verify income or withholding information submitted by taxpayers with their tax returns.

[11] There is no correlation between the number of returns resequenced and the number of returns with a temporary freeze.  Most resequenced returns never have a temporary freeze.  Likewise, returns with a temporary freeze that were identified through other means were not resequenced.

[12] See Appendix IV for details.

[13] For more information, see Appendix V, report 5.

[14] For more information, see Appendix V, report 3.

[15] Pub. L. No. 105-34, 111 Stat. 788 (codified as amended in scattered sections of 5 U.S.C., 19 U.S.C., 26 U.S.C., 29 U.S.C., 31 U.S.C., 42 U.S.C., and 46 U.S.C. app.).

[16] For more information, see Appendix V, report 5.

[17] Webster’s dictionary defines nullity as anything that is null and has no legal force.

[18] A case could have an amount collected with an amount remaining to be collected.  In these instances, the cases would be identified in both the “Amount Collected” and “Total Uncollected” categories. 

[19] ****1****

[20] See Appendix IV for details. 

[21] See Appendix IV for details.

[22] An exception occurred if the return was related to a current investigation open in a CI function field office and the field office wanted to include the return in the investigation.

[23] For more information, see Appendix V, report 4.

[24] For more information, see Appendix V, report 5.

[25] For more information, see Appendix V, report 5.

[26] For more information, see Appendix V, reports 1 and 2.

[27] A “decided” case is one for which the Taxpayer Advocate Service, in conjunction with the CI function, could determine there was either fraud (the taxpayer’s refund was either reversed or reduced) or no fraud (the taxpayer received a full refund).

[28] One of the 27 inquiries involved both the sample year and a subsequent year; we counted the inquiry in both years. 

[29] These 42 cases are not necessarily the same 42 cases referred to in the prior bullet.  We considered a case untimely if (1) the refund was determined to be legitimate and the Center took more than 2 months to release the refund or (2) a refund was deemed to be fraudulent and the Center had not adjusted the fraudulent information from the account by the end of the calendar year in which the fraud was identified.

[30] ****2(a)****

[31] Appendix VI includes a glossary of terms used in this report.

[32] The error attributes for this test included the following characteristics:  (1) the taxpayer’s account was frozen and the return was not verified as fraudulent or (2) the taxpayer’s account was not frozen and the return was determined to be fraudulent.  If the return was not frozen or timely frozen, we determined the amount of any subsequent refund(s).  We were unable to project the total amount because that would have required a larger variable sample.

[33] The error attribute for this test was that the Centers incorrectly determined the return was fraudulent (e.g., account frozen and no wage verification present). 

[34] The error attributes for this test included the following characteristics:  (1) the subsequent year return was frozen and the return was not timely verified as correct or (2) the subsequent year return was not frozen and there was no indication of a fraud determination.

[35] The sample universe was identified through a query that compared the December frozen credit file to the September frozen credit file and extracted those records that did not match, indicating they were no longer frozen. 

[36] Appendix VI includes a glossary of terms used in this report.

[37] Because we did not have information for the universe of fraudulent returns identified during Processing Years 2005 and 2006, we could project our results only onto the universe of returns identified as fraudulent during Processing Year 2004.