Improvements Are Needed in the Automated Underreporter
Program to Ensure That Taxpayer Information Sent to External Tax Agencies Is
Accurate
September 2002
Reference
Number: 2002-10-169
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.
September
18, 2002
MEMORANDUM FOR
COMMISSIONER ROSSOTTI
FROM: Pamela J. Gardiner /s/ Pamela J. Gardiner
Acting Inspector General
SUBJECT: Final Audit Report - Improvements Are
Needed in the Automated Underreporter Program to Ensure That Taxpayer
Information Sent to External Tax Agencies Is Accurate (Audit # 200110043)
This
report presents the results of our review of the Internal Revenue Service’s
(IRS) Automated Underreporter (AUR) program exchanging information with
external tax agencies. The overall objectives of this audit were to
determine if the IRS is protecting taxpayer rights and providing accurate
information when disclosing AUR case information to external tax agencies.
In
summary, the IRS matches information
reported on individual tax returns against information reported by banks,
employers, and other payers. In
instances where the matching process identifies discrepancies, the taxpayer may
have an additional tax amount due.
Information about these additional tax assessments is exchanged with
external tax agencies through the AUR program.
Based on our statistical sample case review, we estimate 469 of 533
partially agreed cases sent to external tax agencies in the first 2 shipments
of Tax Year 1999 contained inaccurate AUR case information. We also found that information disseminated
to external tax agencies was inaccurate regarding the tolerance criteria. In addition, the IRS did not change
tolerances requested by 4 of 43 external tax agencies participating in the AUR
program so that inaccurate information based on unchanged tolerances was
disseminated to external tax agencies.
These conditions could increase taxpayer burden and reduce the
effectiveness of tax administration at the external tax agencies.
We
recommended the IRS stop shipping certain AUR information to external tax
agencies until programming and procedural changes are made to prevent the shipment
of inaccurate information, and the IRS should inform external tax agencies of
this situation. We also recommended
that information disseminated by the IRS to external tax agencies about
tolerance criteria be clarified, including an outreach effort to clarify the
application of the tax tolerance criteria.
In addition, we recommended that changes to tax tolerance criteria be
updated by the IRS when requested by external tax agencies.
Management’s
Response: The IRS management agreed with our results
and recommendations. The IRS plans to stop sending potentially
incorrect AUR information to external tax agencies until computer programming
changes are made and will inform the external tax agencies of the inaccurate
AUR information. The IRS has disseminated
information to external tax agencies to clarify the tolerance criteria and has
established procedures and responsibilities when external tax agencies request
a tolerance change. Management’s
complete response to the draft report is included as Appendix V.
Copies
of this report are also being sent to the IRS managers who are affected by the
report recommendations. Please contact
me at (202) 622-6510 if you have questions, or your staff may contact Daniel R.
Devlin, Assistant Inspector General for Audit (Headquarters Operations and
Exempt Organizations Programs), at (202) 622-8500.
External
Tax Agencies Are Provided Inaccurate Automated Underreporter Case Information
Procedures Were Not Always
Followed When Changes to Tolerances Were Requested
Appendix
I – Detailed Objectives, Scope, and Methodology
Appendix
II – Major Contributors to This Report
Appendix
III – Report Distribution List
Appendix
IV – Outcome Measures
Appendix V
– Management’s Response to the Draft Report
The Internal Revenue Service’s (IRS) Automated Underreporter
(AUR) program matches information reported on individual tax returns against
information reported by banks, employers, and other payers. When the AUR matching process identifies a
discrepancy, an AUR case is established and an Underreporter Notice (CP2000) is
issued to the taxpayer. The CP2000
notice may contain proposed changes to income, payments, credits, and/or
deductions. Depending on the taxpayer’s
response to the CP2000 notice, the AUR caseworker will make necessary tax
adjustments to the taxpayer’s account.
Tax information can be exchanged with external tax agencies
per the Internal Revenue Code and is provided free of charge. These external tax agencies can include
state, city, county, and foreign government tax agencies. Currently, AUR case information is provided
to 42 state tax agencies and the Puerto Rico tax agency. For Tax Years 1996 through 1999, the IRS has
provided external tax agencies with magnetic media tapes containing over 12
million pieces of AUR case information.
For Tax Year 1999, this program involved approximately 1.7 million
taxpayers with approximately $1.86 billion in additional federal
assessments. External tax agencies may
use this information to verify the accuracy of reported state or local income
taxes.
The IRS Wage and Investment Division (W&I) and the Small
Business/Self-Employed Division (SB/SE) administer the AUR program within six
AUR sites. The six AUR sites and the
Modernization, Information Technology and Security (MITS) Services compile
closed AUR case information onto magnetic media tapes. These tapes are sent to external tax
agencies based on need and use requirements of the external tax agencies and
stated tolerances. To provide timely
information, the IRS generally sends six shipments for each tax year under
review. At the time of our review, the
AUR program had processed five of the six shipments for Tax Year 1999.
When tax information is exchanged, the Office of
Governmental Liaison and Disclosure, under the Chief, Communications and
Liaison (C&L), provides oversight and education to the IRS and external tax
agencies. The Office of Governmental
Liaison and Disclosure will enroll the external tax agencies and ensure
safeguards are adequate. To receive AUR
case information, each external tax agency must enroll in the magnetic media
tape exchange program, provide updated safeguard documentation, and be reviewed
by the IRS for safeguard measures. We
found that the participating external tax agencies generally had the proper
documentation in place in order to be enrolled in the program.
This review focused on whether the IRS was minimizing taxpayer
burden by providing accurate AUR case information to external tax
agencies. We obtained documents from,
and held discussions with, employees in the C&L, W&I, SB/SE, and MITS
organizations. We conducted our review
during July 2001 through July 2002 in accordance with Government Auditing Standards. Detailed
information on our audit objectives, scope, and methodology is presented in
Appendix I. Major contributors to the
report are listed in Appendix II.
The IRS provides external tax agencies with AUR information,
and these external tax agencies rely on accurate information for their tax
administration. There are two types of
AUR cases exchanged, fully agreed and partially agreed. A case is considered fully agreed when the
taxpayer agrees with the entire proposed tax change or when the taxpayers’
response agrees with part of the proposed tax change and the remaining
difference is resolved. A partially
agreed case results when the taxpayer agrees with part of the proposed change,
as with a fully agreed case, but the taxpayer’s recalculation differs from the
AUR caseworker’s recalculation. If this
difference is minimal, the IRS will use the taxpayers’ recalculation; otherwise,
a revised CP2000 notice is issued.
For information accuracy, we reviewed statistical random
samples of fully agreed and partially agreed AUR cases. These samples were taken from the first two
shipments for Tax Year 1999 AUR information being exchanged with external tax
agencies. These two shipments had 90,490
fully agreed and 533 partially agreed AUR cases. Only 5 of the 271 (2 percent) fully agreed sample cases we
reviewed contained inaccurate data.
However, 202 of the 230 (88 percent) partially agreed cases we reviewed
contained inaccurate data. We estimate
469 of 533 taxpayers involved with partially agreed AUR cases contained
inaccurate data.
Some of the partially agreed error cases we reviewed
contained significant differences between the tax increase amounts reported to
external tax agencies and the actual tax increases assessed. For example, the IRS reported to an external
tax agency that 1 taxpayer owed federal taxes of approximately $49,000 when the
actual amount owed by the taxpayer was approximately $3,000. In another case, a taxpayer was reported as
owing an additional $6,000 in federal taxes, but the actual assessed amount was
a refund to the taxpayer of approximately $10,000. In the 202 error cases, the IRS inaccurately reported that
$903,937 was owed in federal tax increases when the actual assessed amount
totaled only $108,898. This represents
an increase of 730 percent between the tax increases reported to external tax
agencies and the actual amounts assessed by the IRS.
Since we selected our samples, 3 more shipments of AUR cases
for Tax Year 1999 have been sent to external tax agencies containing 4,625
partially agreed cases. The last
shipment for Tax Year 1999 was scheduled for September 2002, and we were not
able to analyze it in time to be included in this review.
We found two main causes for the errors. The first cause was that the AUR computer
program selected tax information from the most recently issued CP2000 notice
rather than the most current data within the case file. This amount can differ because of
adjustments made by the AUR caseworker due to a taxpayer’s response. An AUR computer programmer said he was
unaware of this situation and that the program has been this way since the AUR
computer program began in 1993. In our
sample, 56 (28 percent) of the 202 errors were due to this AUR computer
programming.
The second cause of errors was due to AUR caseworkers
incorrectly closing cases. AUR
caseworkers were not always following Internal Revenue Manual (IRM) procedures
when closing partially agreed cases.
The errors occurred when the taxpayer and the AUR caseworker agreed to
an amount that was different from the initial proposed CP2000 notice amount. In this situation, the IRM procedures
require that cases be closed as fully agreed cases, assigned specific
processing codes, and recomputed to match the amount to which the taxpayer and
AUR caseworker agreed. If cases are not
recomputed and assigned the proper processing codes, the system will report the
originally proposed amount and not the agreed to amount. In our sample, 146 (72 percent) of the 202
errors were due to AUR caseworkers not following IRM case processing
procedures. Managerial reviews
conducted by AUR Operations Managers, and Headquarters Quality Reviews
conducted by the Program Analysis System have not been effective in disclosing
these processing errors in past reviews.
Based on our statistical sample case review, 469 of the 533
partially agreed cases sent to external tax agencies in the first 2 shipments
of Tax Year 1999 contained inaccurate AUR case information. Taxpayers involved in this situation may be
burdened by having to resolve their tax cases for a second time. This may be simple or complicated and may
even require confirmation from the IRS.
The IRS could minimize taxpayer burden by providing accurate AUR case
information and possibly prevent delays with taxpayers’ resolving their tax
issues with these external tax agencies.
1.
The Commissioner, W&I, the Commissioner, SB/SE, and the
Deputy Commissioner for Modernization & Chief Information Officer should
coordinate to stop sending partially agreed AUR cases to external tax agencies
until reliable computer data can be sent.
Management’s Response: The Director, Compliance, W&I, coordinated with the other IRS
functions to discontinue sending partially agreed AUR cases to external tax
agencies.
2. The Commissioner, W&I, the Commissioner, SB/SE, and the
Deputy Commissioner for Modernization & Chief Information Officer should coordinate
the design and implementation of a computer programming change that selects and
sends accurate information when partially agreed AUR cases are involved.
Management’s Response: The Director, Compliance, W&I, coordinated with the other IRS
functions to amend the AUR computer
program for Tax Year 2001 to exchange only fully agreed case information with
external tax agencies. The IRS indicated that these changes would not affect
the overall volume of cases disbursed to external tax agencies.
3.
The Chief, C&L, should alert the external tax agencies of
the inaccurate AUR case information being sent by the IRS. He should describe the type of case involved
(partially agreed AUR cases) and keep the external tax agencies informed of the
IRS’ corrective actions until this situation is resolved.
Management’s Response: The Deputy Chief, C&L, coordinated the IRS efforts to
distribute information explaining the past and future process for exchanging
partially and fully agreed AUR cases with external tax agencies. In addition, this information was posted on
the Federation of Tax Administrators’ web site.
4.
The Commissioner, W&I, and the Commissioner, SB/SE, should
coordinate to inform and educate the AUR caseworkers about partially agreed
case processing and the impact on external tax agencies. Both should also institute quality control
measures for AUR casework to ensure that partially agreed cases are processed
and closed according to established procedures.
Management’s Response: The Director, Compliance, W&I, coordinated the IRS plans to
emphasize the IRM procedures and guidelines through Continuing Professional
Education and procedure alerts. In
addition, during Fiscal Year 2003 operational reviews will include reviews of
partially agreed AUR cases.
Information about the IRS exchange program must be
adequately described to external tax agencies in order for these external tax
agencies to effectively administer their tax laws. One of the documents necessary to receive computer data extracts
from the IRS is the Magnetic Media Extract Program Enrollment Agreement. This enrollment agreement provides
information about the external tax agency requesting the information as well as
details about the external tax agency’s participation in various computer data
extract programs, including AUR case information.
Within the AUR case information area, the enrollment
agreement must indicate a dollar tolerance to be used to exclude AUR case
information that is not useful to the external tax agency. The enrollment agreement contains a line
item titled “tolerance to AGI” in which the external tax agency must stipulate
a dollar amount that is used to determine the dollar criteria of cases sent to
the external tax agency. For example,
an external tax agency can set its tolerance at $500. Any case with a change below $500 would not be sent to the
external tax agency, and cases at or above the $500 would be sent.
The enrollment agreement does not adequately disclose how
the tolerance is applied to AUR case information and leaves the potential for
misinterpretation by the external tax agencies. The enrollment agreement indicates that the tolerance is applied
to the Adjusted Gross Income (AGI) amount (line 33 of Individual Income Tax
Form 1040). We asked external tax
agencies about their understanding of how the tolerance is applied. One external tax agency thought the
tolerance was applied to the change in AGI amount. A second external tax agency thought it was applied to the change
in taxable income amount (line 39 of Form 1040) and a third external tax agency
thought it was the total AGI amount.
From discussions with AUR computer programmers, we
determined that the tolerance was actually applied to the change in taxable
income amount and not the AGI amount or the total AGI amount. We were told the current AUR programming was
inherited from the previous Underreporter program and no changes have been made
concerning how the tolerance is applied.
There is not a “change to AGI amount” field in the current extract
program, so the program uses the “change to taxable income amount field.”
Another document provided to the external tax agencies also
did not accurately reflect the tolerance criteria. The CP2000 Extract Specification Book, provided to the external
tax agencies annually, contains the file and record structure of the computer
data extract for AUR case information but did not clearly explain how the
tolerance was applied. This document
stated the tolerance field is “the amount to be used for a participating state
for a tolerance check” and refers to “increases to Federal Adjusted Income.” Again, the tolerance was actually applied to
the change in taxable income amount.
The CP2000 Extract Specification Book and the Magnetic Media
Extract Program Enrollment Agreement used by the external tax agency did not
reflect how the tolerance was actually applied to AUR case information. The C&L Division was not aware of the
incorrect tolerance definitions.
The effect of the incorrect and inconsistent explanation of
the tolerance could cause some taxpayers to be unnecessarily included in
external tax agency programs, while other taxpayers are excluded. External tax agencies are affected because
they are uncertain about the usage and application of the tolerance. External tax agencies could further be
affected because the number of cases, on which their revenue assessments are
made, fluctuates based on how the tolerance is applied, and their resources may
not be effectively used.
5.
The Chief, C&L, should revise future versions of the
Magnetic Media Extract Program Enrollment Agreement and the CP2000 Extract
Specification Book to clarify the tolerance criteria for AUR cases. In addition, an outreach effort by the
Chief, C&L, should be made to each external tax agency to communicate the
correct tolerance criteria to ensure a clear understanding of the tolerance
application.
Management’s Response: The Deputy Chief, C&L, coordinated with the IRS and corrected
the wording for the tolerance selection in the CP2000 TY 2000 Extract
Specification Book and the Governmental Liaison Data Exchange Enrollment Form. In addition, the IRS revised and distributed
the specification book to external tax agencies and posted a notice concerning
the revision to the Federation of Tax Administrators’ web site.
Annually,
local Governmental Liaisons and local Disclosure Officers within the C&L
Division obtain and review the Magnetic
Media Extract Program Enrollment Agreements from participating external tax
agencies. This review includes
confirmation of the programs in which the external tax agencies want to
participate during the coming year, confirmation of the dollar tolerance
specified by the external tax agency, and a comparison to the prior year to
determine if the tolerance changed.
Based on information in the enrollment agreements, C&L prepares a
Request for Information Services (RIS) that is used to inform MITS on how to
maintain the AUR exchange program. The
procedures for preparing a RIS are outlined in the IRS’ IRM.
The
Detroit Computing Center (DCC) within MITS receives copies of the enrollment
agreements from C&L. In the past,
personnel at the DCC have prepared a spreadsheet summarizing the enrollment
agreements. This spreadsheet summarizes
information from the enrollment agreements including the exchange programs the
external tax agencies want to participate in as well as the AUR tolerance
established by each external tax agency.
The spreadsheet is sent to the programmers at the Western Development
Center (WDC) within MITS, along with the RIS prepared by C&L. The WDC programmers will input necessary
changes to the AUR computer program to extract the information. The extracted AUR information is sent to the
DCC to separate and forward to the appropriate external tax agency.
During
Tax Year 1999, 4 of the 43 participating external tax agencies indicated on
their enrollment agreements a request to change their prior years’
tolerances. C&L prepared a RIS that
was sent, with the enrollment agreements, to the DCC. The RIS did not specifically outline the four changes requested by
the external tax agencies. However, the
RIS specifically instructed computer programmers to verify tolerances and to
make changes as appropriate.
A
DCC manager stated that they did not forward the RIS to the WDC because they
were new to the process and were unfamiliar with the procedures for processing
the RIS. The DCC manager also stated
that the WDC programmers acknowledged receipt of the spreadsheet summarizing
the enrollment agreements. The DCC
assumed the changes would be made, but a WDC programmer stated that changes are
made to the AUR computer program only when requested through a RIS. It was confirmed that a RIS had not been
received by the WDC for Tax Year 1999 to change AUR tolerances. As a result, tolerance changes requested by
the four external tax agencies were not made by the WDC programmers.
Incorrect
tolerances in the AUR computer data extracts program cause external tax
agencies to receive more or fewer AUR cases than they have requested. In the four instances of Tax Year 1999 tolerance
changes that were not made, the requested tolerance was lower than that for the
prior year. This means these four
external tax agencies should have possibly received more cases than they did. We were unable to determine how many cases
these external tax agencies should have received because the data were not
available.
6. The
Chief, C&L, and the Deputy Commissioner for Modernization & Chief
Information Officer should coordinate to establish and communicate new written
procedures that will ensure external tax agencies’ requests for tolerance
changes via the Magnetic Media Extract Program Enrollment Agreements are
properly identified, controlled, and processed. For the AUR computer program, C&L should clearly list each
external tax agency with its requested tolerance on an annual RIS. This RIS should be sent to the DCC and
WDC. A summarized spreadsheet should
not be used in place of the RIS for changes to the AUR computer program.
Management’s Response: The Deputy Chief, C&L, and the Director, Business Systems
Development, established procedures and responsibility for tolerance
changes. The WDC application
development team will assume the duties of the RIS. The Director, Business System Development, will provide the
C&L office with the name of a contact to facilitate the coordination with
any external tax agency that may have changed tolerance levels since the last
enrollment. Any such changes will be
included in the RIS.
Appendix I
Detailed Objectives, Scope, and Methodology
The overall objectives of this review were to determine if
the Internal Revenue Service (IRS) is protecting taxpayer rights and providing
accurate information when disclosing Automated Underreporter (AUR) case
information to external tax agencies.
To achieve these objectives, the following tests were performed:
I. Determined if
there were concerns, issues, or problems with the exchange of AUR information
by contacting various disclosure officers, government liaisons, computer
programmers, and external tax agencies.
II. Determined if AUR
case information disclosed to external tax agencies by the IRS is accurate.
A.
Confirmed the schedule of when AUR case information is
exchanged.
B.
Obtained and validated the first two shipments of computer
data extracts for Tax Year 1999.
Validated the extracts for content and timing. Clarified data fields with the IRS.
C.
Obtained the third, fourth, and fifth Tax Year 1999 extract
shipments, which occurred in October 2001, January 2002, and May 2002 and
determined the numbers and types of AUR cases exchanged.
D.
Selected two statistically valid random samples from the April
and July 2001 computer data extracts of AUR information exchanged with external
tax agencies for Tax Year 1999. The
sample cases were researched using source documents including taxpayer tax
account transcripts, taxpayer returns ordered from the storage files, and AUR
case files maintained by the local AUR sites.
Exceptions were confirmed with IRS personnel.
The first sample
involved exchanged cases marked as fully agreed. Initially we selected a sample of 300 fully agreed cases using
the discovery sampling technique (population of 90,490 fully agreed cases, 95
percent confidence level, and estimated error rate not to exceed 1
percent). This sample was converted to
the attribute sampling method when errors were identified. In addition, source documents could be
obtained for only 271 of the 300 initially selected cases. This sample was reevaluated and found to be
statistically acceptable (population 90,490 cases, 95 percent confidence level,
error rate of 2 percent from actual case review, and precision of plus or minus
1.61 percent).
The second sample
involved exchanged cases marked as partially agreed. Initially we selected a sample of 230 partially agreed cases using
the discovery sampling technique (population of 533 partially agreed cases, 95
percent confidence level, and estimated error rate not to exceed 1
percent). This sample was also
converted to the attribute sampling method when errors were identified. Adequate source documents were obtained for
all 230 of the initially selected cases.
This sample was reevaluated and found to be statistically acceptable
(population 533 cases, 95 percent confidence level, error rate of 88 percent
from actual case review, and precision of plus or minus 3.19 percent).
E.
From Modernization, Information Technology and Security
Services records, determined the numbers of AUR records exchanged with external
tax agencies that have occurred over the past 3 years to date.
F.
From AUR statistical reports, determined the numbers, types,
and dollar amounts of AUR closures that have occurred over the past 3 years to
date.
G.
Estimated the number of taxpayers affected by unreliable AUR
case information by comparing historical AUR information with our case reviews.
H.
Confirmed with IRS management the actual and estimated number
of unreliable AUR records sent to external tax agencies and discussed causes
and recommendations.
I.
Contacted all six AUR site Operations Managers and discussed
the Quality and Management Review process.
Appendix
II
Major Contributors to This Report
Daniel R. Devlin, Assistant Inspector General for Audit
(Headquarters Operations and Exempt Organizations Programs)
Mary V. Baker, Director
Aaron R. Foote, Audit Manager
Jeff K. Jones, Senior Auditor
Yasmin B. Ryan, Auditor
Appendix III
Deputy
Commissioner N:DC
Chief, Communications and Liaison CL
Commissioner, Small
Business/Self-Employed Division S
Commissioner, Wage and Investment
Division W
Deputy Commissioner for Modernization
& Chief Information Officer M
Chief, Information Technology Services
M:I
Deputy
Chief, Communications and Liaison CL
Director,
Business Systems Development M:I:B
Director,
Compliance, Small Business/Self-Employed Division S:C
Director,
Compliance, Wage and Investment Division
W:CP
Director, Office of Governmental Liaison and Disclosure
CL:GLD
Chief
Counsel CC
National Taxpayer
Advocate TA
Director,
Legislative Affairs CL:LA
Director,
Office of Program Evaluation and Risk Analysis
N:ADC:R:O
Office of
Management Controls N:CFO:F:M
Audit
Liaisons:
Chief,
Communications and Liaison CL
Commissioner,
Small Business/Self-Employed Division S
Commissioner,
Wage and Investment Division W
Deputy
Commissioner of Modernization & Chief Information Officer M
National Taxpayer Advocate TA
Director, Legislative Affairs CL:LA
Appendix IV
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 the Congress.
Type and Value of Outcome Measure:
·
Taxpayer Burden – Potential; 469 taxpayer accounts with
inaccurate Automated Underreporter (AUR) case information sent by the Internal
Revenue Service (IRS) to external tax agencies during the first 2 shipments of
Tax Year 1999 AUR case information (see page 2).
Methodology Used to Measure the Reported Benefit:
We
selected two statistically valid random samples from the April and July 2001
computer data extracts of AUR information exchanged with external tax agencies
for Tax Year 1999. The sample cases
were researched using source documents including taxpayer tax account transcripts,
taxpayer returns ordered from the storage files, and AUR case files maintained
by the local AUR sites. Exceptions were
confirmed with IRS personnel.
1.
The
first sample involved exchanged AUR cases marked as fully agreed. It was determined from our review of this
sample that no corrective action could be recommended to improve our identified
error rate of 2 percent; therefore, this type of case will not be considered in
calculating the above outcome measure.
2.
The
second sample involved exchanged cases marked as partially agreed. Initially we selected a sample of 230
partially agreed cases using the discovery sampling technique (population of
533 partially agreed cases, 95 percent confidence level, and estimated error
rate not to exceed 1 percent). This
sample was converted to the attribute sampling method when errors were
identified. Adequate source documents
were obtained for all 230 of the initially selected cases. This sample was reevaluated and found to be
statistically acceptable (population 533 cases, 95 percent confidence level, error
rate of 88 percent from actual case review, and precision of plus or minus 3.19
percent). We made recommendations to
improve this error rate of 88 percent; therefore, this type of case was used in
calculating the above outcome measure.
Using this statistically valid sample, we estimate 469 inaccurate
partially agreed AUR cases were sent to external tax agencies during the first
2 shipments of Tax Year 1999.
Appendix V
The response was removed due to its size. To see the complete response, please go to the Adobe PDF version of the report on the TIGTA Public Web Page.