Tax Administration: IRS Should Take Steps to Improve the Accuracy
of Schedule K-1 Data (30-SEP-04, GAO-04-1040).			 
                                                                 
Over a trillion dollars in income was distributed in tax year	 
2002 by flow-through entities, such as partnerships, subchapter S
corporations, and trusts, to their partners, shareholders, or	 
beneficiaries, respectively. The Internal Revenue Service (IRS)  
estimates that from 6 to 15 percent of such income is unreported 
on individual tax returns. This income is reported to both IRS	 
and to the recipients on a Schedule K-1 (K-1). IRS uses K-1 data 
in its document-matching program to identify noncompliance and	 
for other purposes. GAO was asked to (1) assess the accuracy of  
K-1 data, specifically transcription errors and taxpayer	 
identification numbers (TIN); (2) determine whether any 	 
limitations in the availability or accuracy of K-1 data have	 
affected IRS's ability to identify noncompliance; and (3)	 
identify the benefits and challenges of increasing e-filing of	 
K-1s.								 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-04-1040					        
    ACCNO:   A12854						        
  TITLE:     Tax Administration: IRS Should Take Steps to Improve the 
Accuracy of Schedule K-1 Data					 
     DATE:   09/30/2004 
  SUBJECT:   Information systems				 
	     Performance measures				 
	     Systems evaluation 				 
	     Tax administration 				 
	     Tax administration systems 			 
	     Income taxes					 
	     Tax returns					 
	     Computer matching					 
	     Noncompliance					 
	     Data integrity					 
	     Electronic forms					 
	     Electronic government				 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-04-1040

                 United States Government Accountability Office

              GAO	Report to the Committee on Finance, U.S. Senate

September 2004

TAX ADMINISTRATION

       IRS Should Take Steps to Improve the Accuracy of Schedule K-1 Data

                                       a

GAO-04-1040

[IMG]

September 2004

TAX ADMINISTRATION

IRS Should Take Steps to Improve the Accuracy of Schedule K-1 Data

                                 What GAO Found

The accuracy of paper-filed K-1 data is reduced by transcription errors;
paper and e-filed K-1s have inaccurate TINs. IRS estimates that
transcription errors for tax year 2002 ranged from 5 and 9.5 percent and
is taking steps to address such errors. Although e-filed K-1s do not
require transcription, for tax year 2002, the percentage of invalid TINs
for e-filed K-1s and paper-filed K-1s were comparable (7 and 6 percent,
respectively). Due to potential burden on flow-through entities and
resource constraints, IRS does not notify the entities of invalid TINs on
K-1s for correction. If IRS did so, this would likely give e-filing
entities enough time to correct invalid TINs before IRS runs its
document-matching program.

Inaccurate or limited K-1 data have created problems for IRS researchers
and examiners. IRS research staff indicated that inaccurate TINs adversely
affected their analysis of flow-through entity networks. Further, because
IRS captures limited data from flow-through entity returns, including the
K-1, IRS staff lack data they consider helpful, such as "Other Income" to
help identify tax shelters. In at least 40 percent of closed examination
cases we sampled, IRS examiners found errors with return line items not
entered into IRS's databases when returns are received. If these lines
were available up front, researchers say they would be able to better
identify returns with potential noncompliance.

Increased e-filing of K-1s would provide benefits and challenges to both
IRS and taxpayers. Benefits for IRS include faster, more complete
information and millions in annual cost reductions. Benefits for taxpayers
include fewer IRS contacts with them because IRS would have more accurate
information in its systems. The primary challenge for IRS is its current
inability to electronically process all flow-through entity returns and
related forms, including the K-1. For taxpayers, the primary challenge is
the cost of converting from paper to e-filing.

 Paper K-1s: Lengthy Processing Yields Incomplete Data United States Government
                             Accountability Office

Contents

     Letter                                                                 1 
                                          Results in Brief                  3 
                                             Background                     5 
                            Data Transcription Errors and Erroneous TINs   
                                         Reduce the Accuracy               
                                             of K-1 Data                    9 
                               K-1 Data Accuracy and Availability Pose     
                                      Problems in Research and             
                                         Examination Efforts               15 
                              Increasing E-Filing of K-1s Would Provide    
                                       Benefits and Challenges             
                                        for IRS and Taxpayers              20 
                                             Conclusions                   24 
                                 Recommendation for Executive Action       25 
                                 Agency Comments and Our Evaluation        25 
Appendixes                                                              
               Appendix I:       Objectives, Scope, and Methodology        27 
              Appendix II:   Comments from the Internal Revenue Service    32 
                             Table 1: Percentage of Invalid Schedule K-1   
     Table                               TINs That Were Not                
                               Corrected and Corrected by IRS for Tax Year 
                                                              2002 by Type 
                                          of Entity Return                 
                            Figure 1: IRS's Program to Match Flow-Through  
    Figures                  Income Reported by Partnerships with Income   
                                       Reported on a Partner's             
                                      Federal Income Tax Return             6 
                            Figure 2: Processing Flowchart for Paper and   
                                          E-Filed Schedule                 
                                                 K-1                        8 
                            Figure 3: How Inaccurate TINs May Prevent IRS  
                                            from Tracking                  
                             Flow of Income through a Chain of Financial   
                                            Transactions                   17 

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

A

United States Government Accountability Office Washington, D.C. 20548

September 30, 2004

The Honorable Charles Grassley Chairman The Honorable Max Baucus Ranking
Minority Member Committee on Finance United States Senate

Over a trillion dollars in income was distributed in tax year 2002 by
entities such as partnerships, subchapter S corporations (S-Corp), and
trusts that distribute net income-as well as losses-to partners,
shareholders, and beneficiaries.1 Such entities, called flow-through
entities, are required to report distributed income annually to the
Internal Revenue Service (IRS) on a Schedule K-1, thus providing IRS with
a way to determine whether individuals are reporting the income on tax
returns.2 IRS estimates that from 6 to 15 percent of such income is
unreported on individual tax returns. In addition, IRS is concerned about
flow-through entities claiming improper expenses or otherwise being used
to evade tax liabilities. Thus, IRS is concerned both with potential
noncompliance by individuals who fail to properly report income from
flow-through entities and by flowthrough entities themselves, which
ultimately leads to improper income reporting by individuals as well.

Schedule K-1s are one of several forms and schedules included in the
annual return filed by a flow-through entity. Partnerships with more than
100 partners are currently required by law to file their annual returns,
including all Schedule K-1s,3 electronically. Those with 100 partners or
less, as well as S-Corps and trusts, are not required to do so.

The annual returns of flow-through entities are used to report the income,
deductions, gains, losses, and so forth of the respective entities. These
returns, including the K-1s, enable IRS to identify compliance issues that

1 An S-Corp is a domestic corporation with no more than 75 shareholders,
all of which are individuals, estates, exempt organizations, or certain
trusts. A trust is an arrangement by which trustees take title to property
for the purpose of protecting or conserving it for beneficiaries.

2 Although for simplicity we refer only to individuals' returns,
corporations, partnerships, and trusts may also be recipients of
flow-through entities' income or losses.

3 For the remainder of this report, we will refer to Schedule K-1s simply
as K-1s.

may require an examination of the flow-through entities' books and
records. Further, during 2001, IRS began matching K-1 information with
information included on individual tax returns to identify potential
underreporting of income by individuals. To use K-1 information not filed
electronically in its document-matching program,4 IRS must first
transcribe the data for use in its computer systems. As reported in 2003,
this is a costly and error-prone process that can result in taxpayer
burden.5

Because of concerns about IRS's ability to effectively use K-1 data to
detect noncompliance, you asked us to assess IRS's current use of this
information. Specifically, our objectives were to (1) evaluate the
accuracy of K-1 data used by IRS, specifically transcription errors and
invalid TINs; (2) determine whether any limitations in the availability or
accuracy of K-1 data have affected IRS's ability to identify
noncompliance; and (3) describe the benefits and challenges of increasing
electronic filing (e-filing) of K-1s.6

To meet our first objective, we discussed and obtained estimates from IRS
staff concerning the type of IRS transcription errors for paper-filed
K-1s. We also assessed IRS's procedures related to the processing and
transcription of K-1 information. In addition, we obtained and analyzed
the IRS K-1 database for tax year 20027 to identify, among other things,
the size and type of entities that file K-1s, the frequency of inaccurate
taxpayer identification numbers (TIN)8 included on these schedules, and
the amount of income associated with inaccurate TINs. On the basis of our
data reliability review of IRS's K-1 database, we determined that the data
were

4 The document-matching program matches information concerning selected
tax issues reported on tax returns by individual taxpayers and reported on
information returns by third parties, such as employers, banks,
flow-through entities, and other payers of income. Transcription involves
manually keypunching the information for use in IRS's computer systems.

5 GAO, Tax Administration: Changes to IRS's Schedule K-1 Document Matching
Program Burdened Compliant Taxpayers, GAO-03-667 (Washington, D.C.: May
30, 2003).

6 Because the K-1 is part of the flow-through entity return, increased
e-filing of the K-1 would require increased e-filing of the flow-through
entity return.

7 At the time of our review, this was the most recent year that complete
K-1 data were available from IRS.

8 The TIN is a unique nine-digit number, usually a Social Security number
(SSN) for an individual, an employer identification number (EIN) assigned
by IRS for a partnership or corporation, and either an SSN or an EIN for a
sole proprietor.

sufficiently reliable to enable us to identify the extent to which K-1s
have invalid TINs.

To address our second objective, we determined what portion of flowthrough
entities' returns, including K-1s, is transcribed by IRS. We also
discussed with IRS staff and research consultants from MITRE Corporation9
how IRS currently uses K-1 information and whether and how data
availability and accuracy affect their ability to use these data. We also
reviewed a projectable sample10 of partnership and S-Corp closed
examination files to determine the compliance issues IRS identified, the
related line items that were adjusted, and whether having additional K-1
line items available would help to identify potential noncompliance. We
subsequently discussed our findings with IRS classification11 and
examination staff.

To meet our third objective, we discussed with both IRS officials and
officials from seven organizations that represent the taxpayer community
the costs, benefits, and challenges of requiring increased e-filing of
K-1s. We selected the organizations based on prior GAO knowledge and
referrals from some of the organizations that we contacted. We also
contacted the software companies that offer e-filing for flow-through
entities to obtain their current fees for preparing and e-filing
flow-through entity returns, including K-1s.

Our work was done from May 2003 through August 2004 in accordance with
generally accepted government auditing standards. (App. I describes our
overall objectives, scope, and methodology.)

Results in Brief	The accuracy of K-1 data is reduced by IRS transcription
errors on paperfiled K-1s and by flow-through entities submitting invalid
TINs on both

9 We talked with MITRE because IRS has a contract with MITRE to conduct
flow-through entity data analysis.

10 The population from which the sample is drawn is agreed 2002 closed
case examinations of partnerships and S-Corps listed in IRS's Audit
Information Management System with tax years ending in 2000 or 2001.
Agreed cases are cases where the IRS has made adjustments to a taxpayer's
tax return, and the taxpayer has actively or passively accepted the
adjustment.

11 Classification staff, called classifiers, review returns and related
documentation and select returns to be examined.

paper and e-filed K-1s. IRS estimates that the overall transcription error
rate for the almost 18 million tax year 2002 paper-filed K-1s ranged from
5 to 9.5 percent based on its quality reviews. E-filed returns are not
transcribed and thus do not have these errors. IRS is taking steps to
reduce the transcription error rate, such as implementing a bar-coding
process that bypasses the transcription process and taxpayer education and
outreach efforts. In tax year 2002, IRS processed almost 1.5 million K-1s
with invalid TINs. The combined income on these K-1s totaled $57.3
billion. IRS was able to correct the TINs on about half of the K-1s with
income totaling $20.6 billion. The remaining 50 percent that had income of
$36.6 billion could not be corrected and thus could not be used in the
documentmatching program. Regarding TIN accuracy, the percentage of
invalid TINs for tax year 2002 e-filed K-1s was 7 percent, which was
comparable to the 6 percent invalid TIN rate on paper-filed K-1s. IRS is
not notifying flowthrough entities of invalid TINs so they can take
corrective actions due to concern over taxpayer burden on flow-through
entities and resource constraints. If this were done, it would more likely
give e-filing entities enough time to correct many invalid TINs before IRS
runs its documentmatching program because IRS could send error notices to
e-filers more quickly than to paper filers.

Inaccurate or limited K-1 data have created some problems in IRS's
research and examination efforts. IRS research staff studying flow-through
entity relationships indicated that missing or inaccurate TINs have
affected their ability to build more effective computer models to analyze
flowthrough networks that may be used for tax evasion. In addition,
because of the limited number of line items captured from the flow-through
entity return, including the K-1, research and examination staff lack
certain data fields, such as "Other Income/(Loss)," that would be helpful
to identify compliance issues and better target resources. Based on our
sample of closed examination cases, in at least 40 percent of the
examinations, IRS auditors corrected line items on entities' returns that
are not currently being transcribed.12 If these line items were available
before IRS classifiers select returns for examination, IRS researchers
could use the data to support more effective computer modeling and thereby
focus examination classifiers' attention on returns that are more likely
to involve noncompliance.

12 Estimates from our sample are subject to sampling error. We are 95
percent confident that corrected line items not transcribed occur in at
least 40 percent of the 2002 closed case examinations based on our sample
evidence.

Increasing e-filing of K-1s from the current rate of about 24 percent
would provide benefits and challenges for IRS and taxpayers. For IRS, the
benefits include faster and more comprehensive information as well as cost
reductions due to the lack of transcription costs for e-filed returns. For
example, in fiscal year 2002, IRS spent over $13 million for processing
and transcribing paper-filed K-1s, much of which would be eliminated. For
taxpayers, the benefits of e-filing include the receipt of an IRS
acceptance acknowledgment; the quicker receipt of rejection notices that
would allow the taxpayers to correct problem returns faster; and more
accurate information in IRS databases due to the lack of transcription
errors, thus reducing the potential for erroneous and burdensome taxpayer
notices. IRS's primary challenge in mandating increased e-filing of K-1s
is its current inability to electronically process all the other forms
that may accompany the flow-through entity return. IRS is scheduled to
have this capability by 2007. The primary challenge for taxpayers is the
cost of converting from paper filing to e-filing. However, based on a
limited review of flow-through entity returns, most K-1s are currently
computer generated, which is a prerequisite for e-filing. Also, all of the
software companies that offer efiling that disclosed their fees (about
half of those we contacted) do so for less than a dollar per K-1 or at no
additional cost. Both IRS and Congress are considering various
administrative and legislative proposals to increase mandatory e-filing of
information and tax returns, including those filed by flow-through
entities.

We are making a recommendation that IRS implement a pilot study to
determine the benefits and costs of requiring flow-through entities to
correct invalid TINs on K-1s as soon as it has been determined that the
TINs cannot be "perfected" via IRS's TIN validation program. The
Commissioner of Internal Revenue agreed with our recommendation and said
that IRS plans to study a number of options to ensure that TINs included
on Schedule K-1s are accurate.

Background	Partnerships, S-Corps, and trusts are commonly referred to as
flow-through entities, as they do not generally pay taxes on income.
Instead, they distribute net income-as well as losses-to partners,
shareholders, and beneficiaries, respectively, who are subsequently
required to report the net income or loss on their individual tax returns
and to pay any applicable taxes. Distributed income is reported to IRS on
a K-1, which is included in the annual return filed by the flow-through
entity. Copies of the Schedule K1 are provided to partners, shareholders,
and beneficiaries for use when filing their respective annual returns.
Partners receive a Form 1065

Schedule K-1, "Partner's Share of Income, Credits, Deductions, etc.";
shareholders receive a Form 1120S Schedule K-1, "Shareholder's Share of
Income, Credits, Deductions, etc."; and beneficiaries receive a Form 1041
Schedule K-1, "Beneficiary's Share of Income, Deductions, Credits, etc."

As shown in figure 1, as part of its overall underreporter program, IRS
has a specific K-1 document-matching program in which selected K-1
information reported by flow-through entities is compared to information
reported by individuals on their tax returns in order to determine whether
distributed income has been reported as required.

Figure 1: IRS's Program to Match Flow-Through Income Reported by
Partnerships with Income Reported on a Partner's Federal Income Tax Return

Source: GAO analysis.

In like manner, income reported to IRS on a K-1 by S-Corps and trusts can
be matched with income reported on tax returns by shareholders and

beneficiaries, respectively.13 The purpose of this program is to increase
voluntary reporting of flow-through income by taxpayers and to target K-1
related underreporter notices to noncompliant taxpayers. IRS identified
about $4.1 billion in underreported income for tax years 2000 and 2001 via
the K-1 matching program and assessed about $110 million in additional
taxes.14

In addition to use in the matching program, IRS can also use K-1
information to aid in selecting flow-through entity returns for
examination. For example, IRS can use K-1 information to aid in
identifying flow-through entities involved in potential tax evasion
schemes and to develop computer models that may enable IRS to more
effectively select returns for examination with the greatest likelihood
for a tax change.15

In order for IRS to use K-1 information in its matching program, the
information must either be e-filed by a flow-through entity or, if filed
via paper, transcribed by IRS staff for use in its computer systems.
Currently, only partnerships with over 100 partners are required by law to
e-file their annual returns, including any related K-1s. As a result, for
tax year 2002, less than one-quarter of 1 percent of partnerships was
required to e-file.

Figure 2 illustrates that an e-filed K-1 goes through two basic steps
before the information is input into the Information Returns Master File
(IRMF). At Step A, the K-1 undergoes up-front checks prior to final
acceptance by IRS, whereby the K-1 data must pass specific checks or the
entire flowthrough entity return is to be rejected until corrected by the
entity. The upfront checks include verifying the tax year and proper
formatting of names, addresses, and TINs. For example, the partner's TIN
on a K-1 filed by a partnership must be within a specific range
established by IRS; if not, the entire partnership return is to be
rejected. The only other step for an e-filed K-1 prior to its going
through IRS's document-matching program is the TIN validation process, in
which the TIN and name on the K-1 are electronically

13 For this report we focused upon the K-1 and did not assess the accuracy
of the dollar amounts reported by the flow-through entity.

14 As of May 2004, the final results for tax year 2000 were complete,
while the final results for tax year 2001 were still to be determined.

15 An IRS examination of a taxpayer's books and records that results in no
tax change is generally inefficient for IRS because it has spent resources
to audit an accurate return. No tax change audits also burden taxpayers,
who are forced to go through the audit process even though they are
compliant.

matched with information in IRS's files to determine whether the TIN is
valid. Generally, this validation occurs several months after IRS accepts
the e-filed return.

       Figure 2: Processing Flowchart for Paper and E-Filed Schedule K-1

Source: GAO analysis of IRS data.

Note: Does not reflect the processing changes to paper K-1s that IRS began
testing in April 2004 with the introduction of bar coding and scanning.

In contrast, a paper-filed K-1 goes through several manual steps,
including some of the up-front checks conducted electronically for e-filed
K-1s, before TIN validation takes place and the information can be input
into the IRMF. These steps, particularly transcription, can take up to 6
months to complete, with transcription beginning in May. For example, at
Step 4, IRS staff are to edit the flow-through entity return and contact
the taxpayer if a required K-1 is missing. At Step 8, IRS staff are to
transcribe selected K-1 line items. During the transcription process, the
computer conducts checks on select aspects of the keypunched data, such as
correlating zip code and state information, and creates an error record
for correction. Subsequently, other IRS staff are to compare a sample of
the transcribed K-1 data to the original paper-filed K-1 to determine
whether the data were accurately transcribed. The TIN on a paper-filed
K-1, as on an e-filed K-1, is not

computer validated until it reaches the stage where electronic TIN
validation occurs, generally several weeks or months after the return was
filed.

IRS's program to electronically validate TINs matches the TIN and name on
the K-1 to taxpayer identity information in its files. If there is no
match, IRS will attempt to "perfect" or correct an incorrect TIN/name
combination via a TIN validation process, which entails matching the TIN
and name control-the first four characters of an individual's last name or
the first four characters of a business name-with (1) a file which
contains all Social Security numbers (SSN) ever issued and all name
controls ever associated with them and (2) a file that contains all
employer identification numbers (EIN) ever issued and all name controls
associated with them. This TIN validation process occurs four times per
year, beginning about a month and a half after the end of the filing
season.

  Data Transcription Errors and Erroneous TINs Reduce the Accuracy of K-1 Data

Data transcription errors made by IRS on paper-filed K-1 data and invalid
TINs submitted by flow-through entities on both paper-filed and e-filed
K-1s lower the accuracy of K-1 data. IRS transcription errors, which occur
only for paper-filed K-1s, ranged from 5 to 9.5 percent for tax year 2002,
and IRS is taking steps to reduce these errors. The percentage of invalid
TINs for efiled K-1s is comparable to that for paper-filed K-1s. However,
due to potential taxpayer burden and resource constraints, IRS is not
notifying flow-through entities of invalid TINs so they can take
corrective actions, a step which would likely give e-filing entities
enough time to correct many invalid TINs before IRS runs its
document-matching program. Paper-filing entities may not have sufficient
time to correct invalid TINs before document matching occurs.

    Data from Paper-Filed K-1s Contain Transcription Errors, but IRS Is Taking
    Steps to Reduce These Errors

According to IRS K-1 quality reviews conducted at two IRS locations, the
overall K-1 transcription error rate for tax year 2002 ranged from 5 to
9.5 percent-errors that by definition are not made in e-filed returns. The
most frequent errors dealt with names and addresses. IRS also found
transcription errors in dollar amounts and TINs. Errors detected during
quality reviews are corrected before the K-1s are posted to the IRMF,
which IRS uses to detect potential underreporters and nonfilers. However,
less than 2 percent of all K-1s are selected for the K-1 quality review.
Transcription errors on all other K-1s are included when the data are
posted to the IRMF. Consequently, data from an estimated 18 million tax

year 2002 paper K-1s that were entered into databases used by IRS for
research and enforcement purposes have transcription error rates from 5 to
9.5 percent.16 For example:

o 	IRS's K-1 database for tax year 2002 included 16 paper-filed K-1s each
of which showed interest income of over $1 billion. These interest income
amounts appeared to be transcription errors. One partnership filing paper
K-1s had 73 partners. For 72 of the partners, the K-1 interest recorded in
the IRMF was under $200,000. The remaining partner's interest as recorded
in the IRMF was $85.3 billion.17

o 	According to an IRS data quality review of tax year 2001 K-1
documentmatching cases, about 5 percent of the cases that were either
screened out before taxpayers were contacted or resulted in no change to
taxpayers' tax liabilities after an erroneous underreporter notice was
sent to the taxpayer were due to transcription errors. The transcription
errors included misplaced decimal points and positive money amounts that
were transcribed as negative numbers and vice versa.

According to IRS officials, it would be too costly to do more data
transcription quality review of paper-filed K-1s, such as reentry of K-1
data. Instead, IRS is taking other measures to improve K-1 data accuracy.
For example:

o 	For tax year 2003, IRS began scanning all K-1s using optical character
recognition (OCR) equipment. Also, for tax year 2003, IRS is accepting
K-1s with bar codes that contain all the K-1 data. If the bar code is
present, the system will pick up the information from the bar code,
otherwise the system will image the K-1 and read the line entries using
OCR. Portions of the K-1 or bar code that cannot be read by OCR are
manually transcribed. Although IRS originally projected 30 percent of K1s
would be bar coded in tax year 2003, as of July 2004 only 8 percent of
K-1s submitted were bar coded. For the 92 percent of K-1s without bar

16 Some transcription errors are corrected during processing after the
point of transcription. IRS did not have data on what percentage of
transcription errors are later corrected.

17 Because each partner's distributive share of income is determined by
the partnership agreement (and this will be respected for tax purposes
unless the allocations agreed to lack "substantial economic effect"),
income may not always be distributed evenly. However, this example was
selected because $85.3 billion appears to be an excessive dollar amount
for interest income.

codes that OCR read, almost 20 percent required no transcription, 60
percent required less than half of normal transcription, and 20 percent
were entirely transcribed. Although bar coding and OCR bypass most of the
manual data transcription, which reduces some data transcription cost and
errors, IRS officials still prefer e-filing because bar coding is a paper
process with accompanying processing costs.

o 	To improve the accuracy of transcription, IRS has implemented new
software and improved transcription training. At two IRS locations, IRS is
using a new transcription software intended to increase transcription
productivity and accuracy, compared to the current transcription software.
In addition, transcription training for the K-1 program has evolved. Each
year, feedback is funneled to the IRS transcription trainers to improve
the K-1 transcription process.

o 	 IRS is redesigning the K-1s for both partnerships and S-Corp returns
so that IRS can scan them into the computer instead of having to
transcribe the data manually. Although the redesigned partnership and
S-Corp K-1s are expected to be ready by tax year 2004, the redesigned
trust K-1 will not roll out until tax year 2005 because trust law makes
the trust K-1 different from the other two K-1s.

o 	IRS is conducting educational outreach to increase accurate K-1 filing
and provide updates to changes in K-1 design. In April 2004, IRS issued a
news release to provide tips for businesses, individuals, and tax
professionals on accurate K-1 filing. For example, flow-through entities
are instructed to ensure the correct TINs are used on K-1s. In addition,
the six IRS Tax Forums in 2004 include a session on reporting flowthrough
items, which addresses the redesign of K-1 forms and K-1 reporting
reminders. IRS has also included updates on the K-1 matching program and
K-1 redesign in external speeches to stakeholder groups. Finally, in late
2004, IRS plans to implement a multifaceted communication plan to
publicize the release of the redesigned K-1s.

IRS's K-1 Data Contain For IRS to use K-1 data in its document-matching
program, the TINs and Incorrect Taxpayer names on K-1s need to be accurate
so they can be linked to individuals' tax Identification Information
returns and other tax documents. In tax year 2002, about 94 percent of 24

million K-1s that IRS processed contained valid TINs. The remaining 6
percent, or approximately 1.5 million K-1s, had invalid TINs because
either IRS made transcription errors or the flow-through entities
submitted invalid data. The 1.5 million K-1s with invalid TINs had
combined income

gains of $57.3 billion and combined income losses of $84.1 billion. IRS
was able to correct the invalid TINS on about 750,000 of the K-1s, with
income gains totaling $20.6 billion and incomes losses totaling $6.8
billion, so that they could be used in IRS's document-matching program or
for other compliance and research purposes. However, the remaining 740,000
K-1s with invalid TINs, with income gains of $36.6 billion and incomes
losses of $77.2 billion, could not be perfected and thus were unmatchable.
IRS did not have data on the number of K-1s that had either corrected or
unmatchable TINs in the IRMF that resulted from transcription errors.

    IRS Does Not Notify Flow-Through Entities of Invalid TINs That It Was
    Unable to Correct

After IRS checks the validity of TINs provided on K-1s, it does not notify
either paper-filing or e-filing flow-through entities of the invalid TINs
it finds so the entities can take steps to correct the TINs, due to
concerns about the potential burden on the entities and resource
constraints. Because e-filed returns do not go through time-consuming
paper processing steps, including transcription, if IRS were to notify the
originating entities of invalid TINs, they should have time to correct
invalid K-1s before IRS performs its document matching in the fall
following a tax filing year. For paper-filed K-1s, many entities likely
could not respond before the document matching occurs.

Because e-filed K-1s are not subject to transcription errors, none of the
keypunching errors associated with paper returns are in e-filed data.
However, as table 1 shows, in tax year 2002 the overall percentage of
invalid K-1 TINs IRS found with its TIN validation program was comparable
for e-filed (about 7 percent) and paper (6 percent).

Table 1: Percentage of Invalid Schedule K-1 TINs That Were Not Corrected
and Corrected by IRS for Tax Year 2002 by Type of Entity Return

E-filed K-1s Paper K-1s

             Percentage                       Percentage            
                     of                               of            
             TINs found Percentage            TINs found Percentage 
                                of                               of 
             invalid by invalid    Percentage invalid by invalid    Percentage 
                  IRS's TINs IRS       of     IRS's      TINs IRS           of 
Type of   validation was unable invalid    validation was unable    invalid 
entity                       to TINs IRS                      to   TINs IRS 
return       program    correct  corrected    program    correct  corrected 
 Partnership        8.7       62.1       36.8        6.5       50.8 
    Trust           3.7       24.3       75.7        7.1       56.3 
S-Corp          n/aa       n/aa       n/aa        5.0       36.0 
  Total all                                                         
    K-1s            7.0                              6.1            

Source: GAO analysis of IRS information.

Note: Numbers may not add up to 100 percent because of rounding.
aNot applicable because e-file was not available for S-Corp returns until
tax year 2003.

Factors that may be contributing to e-filed K-1s having TIN errors
comparable to those of paper K-1s include (1) large partnerships, which
are mandated to file K-1s electronically, submitting such large volumes of
K-1s that many may unknowingly submit one or more K-1s with invalid TINs18
and (2) IRS not applying one of its up-front checks for e-filed
partnership K1s.

According to our analysis of IRS's K-1 database, partnerships that submit
a higher volume of K-1s are more likely to submit a K-1 with an invalid
TIN compared to partnerships that submit only a few K-1s. In tax year
2002, efiled partnerships' K-1s had the highest rate of invalid TINs (8.7
percent). That same year, 97 percent of the partnerships with more than
100 partners, which are required to e-file, submitted at least one K-1
with an invalid TIN. In contrast, 18 percent of partnerships with 100 or
fewer partners submitted at least one K-1 with an invalid TIN.

To encourage electronic filing of partnership returns, IRS is not applying
its up-front check that would reject an e-filed partnership's return if it
has even one TIN on a K-1 that falls outside the range of numbers
associated with SSNs and EINs. If IRS applied this validation criterion in
tax year 2002, 12 percent of the e-filed partnership K-1s with unmatchable
TINs would have

18 Each partner has a unique TIN.

been rejected and the originating entities would have been asked to take
steps to correct the TINs. However, some partnerships have hundreds or
thousands of partners, making it more challenging for them to ensure that
all partners' TINs are correct. IRS officials have determined that
accepting an e-filed return when the vast majority of the K-1 TINs fall
within the range of numbers associated with SSNs and EINs, rather than
rejecting the entire entity return due to one or a few TINs that fall
outside that range, promotes e-filing.

In addition, IRS does not notify either e-filing or paper-filing
flow-through entities if submitted TINs are found to be invalid during the
TIN validation checks it performs subsequent to accepting entities'
returns. In tax year 2002, more than half of the K-1s submitted by 2
percent of the flow-through entities contained invalid TINs. The total
number of unmatchable K-1s submitted by these entities represented about
29 percent of the total number of K-1s with unmatchable TINs. IRS
officials said that requiring flow-through entities to correct invalid
TINs could be a burden because the entities rely on information supplied
by individual taxpayers and the correct TINs may not be readily available,
particularly for those entities submitting a large number of K-1s. In
contrast, IRS does notify filers of missing or invalid TINs submitted on
other types of information returns, which then may require the filers to
contact third parties for corrected information. For example, for tax
years 2000 and 2001 combined, IRS proposed just over $204 million in
penalties against nonfederal payers for information returns with invalid
TINs. IRS officials acknowledged that flow-through entities may have made
mistakes themselves that resulted in invalid TINs or may have the correct
information on hand. They also stated that sending such notices would
entail some additional cost to IRS and that they currently face resource
constraints. However, IRS officials do not have estimates of either the
potential benefits, such as increased revenue obtained from document
matching utilizing accurate TINs, or the cost to IRS of obtaining valid
TINs from flow-through entities.

If IRS were to notify flow-through entities of invalid TINs and ask that
they take steps to correct the TINs, it likely would be able to receive
many corrected TINs, particularly from e-filers, in time for its annual
documentmatching program. IRS does its document matching generally from
November of the calendar year through January of the following year. The
time when document matching occurs changes somewhat from year to year. IRS
corrects TINs, including K-1 TINs, four times a year: at the end of June,
early September, mid-November, and late November. Based on the IRS's 2001
Statistics of Income samples, at least 97 percent of partnerships

and S-Corps filed calendar year returns.19 Consequently, all of these
returns were due to be filed prior to IRS's first TIN validation check in
June.20 Since IRS accepts e-filed returns within 2 days of their
submission, all e-filed returns for which filers have not requested
extensions should be available for IRS's June TIN validation program. In
this case, IRS would be able to notify the flow-through entities of the
invalid TINs and the entities would have several months to correct the
TINs and get them back to IRS before IRS posts the corrected K-1s to the
IRMF in time for use in the documentmatching program. Even if a
flow-through entity did not submit the corrected TIN in time, the entity
would be aware of the error and could correct the TIN for the following
year.

For paper-filing flow-through entities, fewer entities likely would be
able to correct invalid TINs in time for inclusion in the
document-matching program. Transcription of paper-filed entity returns,
including K-1s, begins in May. Because transcription can take up to 6
months, a significant portion of paper-filed entity returns and associated
K-1 TINs likely would not be available for the June TIN validation. For
those not available until the early September TIN validation, the entities
would have much less time to correct TINs and get back to IRS in time for
IRS to include the corrected TINs in the document-matching program.
Because IRS's new return scanning and bar-coding efforts should make
paper-filed return data available more quickly, IRS may be able to include
more of them in the June TIN validation and thus provide entities
sufficient time to provide corrected TINs if it sends notices of invalid
TINs to flow-through entities.

  K-1 Data Accuracy and Availability Pose Problems in Research and Examination
  Efforts

In addition to using K-1 data in its document-matching program, IRS is
using K-1 data in its research programs to better understand flow-through
relationships. When data such as TINs are unavailable or inaccurate,
researchers are unable to establish a complete understanding of the
network of related entities and taxpayers. Data limitations have also
affected IRS's efforts to identify potentially noncompliant taxpayers for
examination. IRS researchers and examination staff indicated that more

19 These figures are based on sample data and subject to sampling error.
We are 95 percent confident that the percentages exceed the value shown
for the tax year 2001 partnerships and S-Corps.

20 In general, for calendar year filers, S-Corp returns are due to be
filed by March 15 and partnership returns are due by April 15.

complete and accurate data would enhance their efforts to detect
noncompliance.

    Inaccurate TINs Affect Research Efforts to Link Related Entities

IRS researchers are using K-1 data to visualize how taxpayers are related
to different entities and to evaluate whether compliance issues may exist
with flow-through entities. However, inaccurate TINs have sometimes
prevented researchers from establishing all relevant links in a network of
related entities. As a result, IRS is less able to track the flow of
income and losses among entities and could be missing opportunities to
address areas of noncompliance.

Figure 3 illustrates how inaccurate TINs may prevent IRS from tracking the
flow of income through a chain of financial transactions. In this example,
an S-Corp distributes losses to an individual shareholder, possibly to
allow the shareholder to offset other gains, and distributes income to
another shareholder, a trust. Since trusts are flow-through entities and
may be nontaxable, the individual shareholder may be using the trust to
reallocate income (perhaps to someone in a lower tax bracket) that would
otherwise need to be reported and taxed on that individual's return.

Figure 3: How Inaccurate TINs May Prevent IRS from Tracking Flow of Income
                   through a Chain of Financial Transactions

                       Source: GAO analysis of IRS data.

In our example, the S-Corp transfers income to Trust A, which in turn
transfers the income to Trust B. In both transactions, the S-Corp and
Trust A submit K-1s with accurate TINs to IRS, so IRS can track the flow
of income between the entities. Trust B then transfers the income again to
Trust C and submits a K-1 with an inaccurate TIN to IRS. Because of the
inaccurate TIN on the K-1, IRS would likely be unable to identify that
Trust C is related to the other entities or track the flow of income to
its final destination and ultimately determine whether any income was
underreported.

Limited Transcription Lines IRS transcribes limited line items from K-1s
that accompany partnership
Do Not Fully Meet the and S-Corp returns. According to IRS staff, at least
some of the
Needs of Examination and nontranscribed lines would provide useful
information. Similarly, IRS does
Research Programs not transcribe many of the lines from the flow-through
entity's return to

which the K-1s are attached. Since e-filing of the full entity's return is
part and parcel of achieving e-filing of K-1s, e-filing of K-1s would have
the additional effect of making the complete entity's tax return
information available to IRS examiners and researchers. Complete entity
data also would provide useful information for research and examination
purposes. For K-1s, IRS identified the line items to be transcribed based
on the needs of its document-matching program and not on other potential
uses.

Regarding K-1s, IRS transcribes about 14 percent of partnership K-1 line
items and about 17 percent of S-Corp K-1 line items. Research and
examination staff indicated that the nontranscribed information would
provide useful information. For example:

o 	The "Other Income/(Loss)" line is not captured from the K-1 because it
is not useful for document matching, but it can be helpful to researchers
in identifying abusive shelters, for example, where the gain is allocated
to a tax haven country and the loss is allocated to a domestic investor.

o 	Transcribing the shareholder's ownership percentage from the K-1 would
more easily allow classifiers to determine if the taxpayer has a
controlling interest in the S-Corp and if income/losses are distributed
evenly.

Regarding the flow-through entity's return, IRS currently transcribes
about 23 percent and 20 percent of the line items for partnership and
S-Corp returns, including the K-1, respectively. As discussed below,
additional data from the full entity return would potentially benefit
examination and research programs. For example:

o 	The IRS Examination Guide for Abusive Tax Shelters and Transactions
lists partnership (Form 1065) and S-Corp (Form 1120S) tax return lines
that when examined with other information, may indicate tax shelter
transactions. For partnership returns, about 80 percent of the line items
listed in the guide as useful to detect tax shelters are not captured in
IRS's database. For S-Corp returns, about 87 percent of line items listed
are not captured. When selecting returns to be examined, IRS classifiers
who focus on tax shelter issues lack information that may identify
taxpayers most likely to be using tax shelters.

o 	IRS researchers developed a computer model to better identify S-Corp
returns for examination by helping classifiers separate accurate returns

from those needing further investigation. The model uses data from IRS's
database of business returns, including Form 1120S and accompanying
schedules. From the approximately 23,000 returns that the model analyzed,
IRS identified about 58 percent of the returns as having low potential for
noncompliance and therefore eliminated them from the universe that might
be examined. The remaining 42 percent of returns could not be classified
because the model did not have enough data to evaluate them. As a result,
IRS has continued to rely on examination staff who focus on S-Corp
compliance issues to manually review the returns that the model has not
been able to classify. To help address the lack of 1120S data, IRS will be
capturing 10 additional line items from the 1120S for tax year 2003. One
IRS researcher estimated that the 10 additional 1120S line items would
enable the computer model to identify 15 S-Corp compliance issues compared
to its current capability of identifying 2 compliance issues.

From our file review of closed S-Corp and partnership tax return
examination cases and discussions with IRS examination and research staff,
we also found that additional line items from the K-1 and other parts of
the entity's return may assist IRS in selecting tax returns to examine.
Based on our sample of closed examination cases,21 in at least 40 percent
of the examinations, IRS corrected line items that are currently not
transcribed. IRS examination and research staff we interviewed indicated
that if IRS captured this information and made it available to them, it
would help them identify those returns with errors or omissions that IRS
should examine. Most of the nontranscribed line items that were corrected
were from Schedule A (Cost of Goods Sold) and Schedule K22 (Partners'
Shares of Income, Credits, Deductions, etc., or Shareholders' Shares of
Income, Credit, Deductions, etc.) for both partnerships and S-Corps. IRS
identified two of these line items, both from Schedule K and K-1, as
important for improving the effectiveness of computer modeling and
mentioned other nontranscribed lines, such as "Short Term Capital Loss,"
from the entity return that would be useful.

21 See app. I for information regarding the sample population.

22 Schedule K is a summary schedule of all partners' or shareholders'
shares of the partnership's or S-Corp's income, credits, deductions, and
so forth. Schedule K-1 shows each partner's or shareholder's separate
share. In our file review, we noted adjustments on Schedule K, unless the
adjustment was specific to Schedule K-1.

  Increasing E-Filing of K-1s Would Provide Benefits and Challenges for IRS and
  Taxpayers

Increasing e-filing of K-1s provides benefits and challenges for IRS and
taxpayers. The benefits for IRS are faster and more comprehensive
information as well as cost reductions. The benefits to taxpayers are the
receipt of acknowledgment notices, faster rejection notices that allow
taxpayers to resolve problems faster, and more accurate information.
Currently IRS's main challenge is the lack of complete e-filing capacity,
but IRS is scheduled to have this capacity by 2007. The main challenge for
taxpayers is the cost of converting from paper to e-filing. However,
limited data indicate that most K-1s are computer generated, which is a
prerequisite for e-filing. Also, all of the software companies that offer
efiling that disclosed their fees (about half of those we contacted) do so
for less than a dollar per K-1 or for no additional charge. Congress has
mandated that IRS increase e-filing to at least 80 percent of all tax and
information returns by 2007. Currently, both IRS and Congress are
considering increasing mandatory e-filing of flow-through entity returns.

    Increasing E-Filing of K-1s Would Provide Better Information and Cost
    Reductions for IRS

Currently, IRS electronically receives about a quarter of the K-1s filed,
although only partnerships with more than 100 partners are mandated to
efile. Increasing e-filing of K-1s would benefit IRS because of the
following:

E-filing K-1s provides IRS with faster and more complete information for
use in compliance and research programs. A recent Treasury Inspector
General for Tax Administration (TIGTA) report stated that the savings in
processing time resulting from e-filing would significantly affect IRS's
attempt to reduce its lengthy corporate examination process. In addition,
the TIGTA report stated that comprehensive electronic information would
minimize the number of no change audits by enabling IRS to better target
resources to issues that have the greatest compliance risk.23

o 	E-filing K-1s would save IRS millions of dollars a year because it
would eliminate the processing and transcription costs of paper K-1s.
According to IRS, the cost to process e-filed K-1s is minimal once the
systems are in place, while processing and transcribing paper K-1s cost
IRS $14.6 million in fiscal year 2001 and $13.1 million in fiscal year
2002. If IRS was able to re-allocate these cost savings, IRS could, for
example,

23 Treasury Inspector General for Tax Administration, New Regulations Are
Needed to Take Full Advantage of the Opportunities Offered by Filing Large
Corporate Income Tax Returns Electronically, 2003-30-123 (Washington,
D.C.: May 30, 2003).

pay the salaries of 284 additional field collection revenue officers.
While, as noted earlier, some of the processing and transcribing costs
will be reduced because of bar coding and scanning, IRS regards bar coding
as a lesser alternative to e-filing. In addition, bar coding results in
incomplete information because only the transcribed lines are scanned into
the computer systems, and the K-1s are the only part of the entity return
that is bar coded. Also, there is limited availability of software that
has bar-coding capacity; only four software companies provide barcoding
capability compared to 19 software companies that provide efiling.

The primary benefits for taxpayers of increasing e-filing are as follows:

o 	Taxpayers that e-file will receive electronic acceptance or rejection
notices within 2 days of submitting tax returns. The tax form is
electronically transmitted to the software company and then the software
company transmits the tax return to IRS. IRS sends the software company an
electronic acceptance or rejection notice within 2 days, and the software
company then sends the notice to the taxpayers. Taxpayers that file paper
returns do not receive acceptance notices and thus do not have proof that
the returns were filed on time in case the tax returns are lost. E-filing
taxpayers also receive rapid rejection notices and are thus informed of
problems much faster than paper-filing taxpayers, who may wait for 6
months for IRS to process tax returns.

o 	The information on an e-filed tax return should be more accurate
because of the lack of IRS transcription errors. More accurate information
would reduce the potential for burdensome taxpayer notices resulting from
transcription errors. To respond to IRS notices, taxpayers and preparers
are required to collect, organize, and submit information to IRS to
explain any discrepancies cited in the notices, which requires an
investment of both the taxpayers' time and money. In recent reports, TIGTA
noted that e-filing would eliminate transcription errors that result in
erroneous and burdensome taxpayer notices.24

24 Treasury Inspector General for Tax Administration, The Internal Revenue
Service Could Reduce the Number of Unnecessary Notices Sent to Taxpayers
Regarding Unreported Income From Schedule K-1, 2003-30-071 (Washington,
D.C.: Mar. 14, 2003), and New Regulations Are Needed to Take Full
Advantage of the Opportunities Offered by Filing Large Corporate Income
Tax Returns Electronically.

    Increasing Electronic Filing of K-1s Requires Complete IRS E-Filing Capacity
    and Taxpayers' Conversion from Paper Filing to E-Filing

The primary challenge for IRS of mandating increased e-filing is to
implement computer systems that can electronically process the complete
set of tax documents that flow-through entities may file with K-1s.
Although IRS currently has the capacity to electronically process K-1s
that accompany flow-through entity returns, IRS is unable to
electronically process all the forms that accompany those of trusts and
partnerships. This impedes e-filing of the flow-through returns with
accompanying K-1s because taxpayers that submit partnership and trust
returns (that include three-fourths of K-1s) have to submit both paper and
electronic documents-a disincentive for e-filing. For example, signature
forms have to be sent in on paper. In contrast, IRS currently has complete
e-filing capacity for the entire S-Corp return, so no forms have to be
filed on paper. IRS is scheduled to have complete e-filing capacity for
partnerships and trusts, but has pushed the completion date for this
effort from 2006 to 2007 due to limited resources.

The main challenge to expanded use of e-filing for taxpayers is the cost
of converting from paper filing to e-filing. In separate reviews of
flow-through entity returns by IRS and GAO, the majority of the tax
returns were found to be computer generated, prepared by a paid preparer,
or both, which might make the conversion to e-filing easier. Based on our
sample of Audit Information Management System agreed closed case
examinations of partnerships and S-Corps with tax years ending in 2000 or
2001, paid preparers prepared at least 84 percent of the returns, and at
least 90 percent of the returns were computer generated.25 Of the
nonprojectable sample of 200 partnership and trust returns that IRS
reviewed, a paid preparer prepared 169 returns, and 173 were computer
generated.26

Since the above-mentioned reviews of flow-through entity returns indicate
that a paid preparer prepares the majority of returns that accompany K-1s
by computer, the cost to convert to e-filing may be marginal or
nonexistent. If a paid preparer is using software that has e-filing
capacity, then taxpayers can simply choose to use this option, which can
entail a marginal cost increase. According to our survey of the software
companies that offer efiling of partnership, trust, and S-Corp returns,
all of the software

25 These figures are subject to sampling error. We are 95 percent
confident that the percentages exceed the values shown.

26 IRS reviewed a nonprojectable sample of tax year 2000 partnership and
trust returns from the Austin Service Center.

companies that disclosed their fees (about half of those we contacted)
either charge from $0.30 to $0.90 per e-filed K-1 or include the option to
efile in the price of the software. In order to e-file, flow-through
entities have to buy the software and e-file the entire flow-through
entity return, at costs that vary from $3.50 per return to over $15,000
for comprehensive support for partnership returns, corporate income tax
returns, and affiliated forms. For partnerships, if the paid preparer does
not use software with e-filing capacity and the data are formatted
according to IRS's specifications, paid preparers can send the partnership
return electronically to a software company that will then electronically
transmit the partnership return. One software company stated that it would
generally charge $0.40 per K-1.

    IRS and Congress Are Considering Increasing E-Filing

According to IRS officials, IRS is considering mandating increased
e-filing of information and tax returns, including those of flow-through
entities. In recent reports, TIGTA has recommended that IRS should work
with the Department of the Treasury to mandate increased e-filing of
flow-through entity returns, either through current regulatory provisions
or through legislative action.27 As a result, IRS is currently studying
the possibility of increasing mandated e-filing of flow-through entities'
returns with accompanying K-1s under Internal Revenue Code (I.R.C.)
Section 6011 as part of an agencywide initiative to increase e-filing to
meet a congressionally mandated goal of having at least 80 percent of all
tax and information returns filed electronically by 2007.28 IRS's study
includes the cost for taxpayers to convert from paper filing to e-filing,
the cost for IRS to initiate and administer increased mandated e-filing,
the perspective of the paid tax preparer and business communities, and how
to implement increased mandated e-filing. In addition, according to IRS
officials, IRS is also considering mandating e-filing for those returns
for which IRS has complete e-filing capacity.

27 Treasury Inspector General for Tax Administration, The Internal Revenue
Service Could Reduce the Number of Unnecessary Notices Sent to Taxpayers
Regarding Unreported Income From Schedule K-1 and New Regulations Are
Needed to Take Full Advantage of the Opportunities Offered by Filing Large
Corporate Income Tax Returns Electronically.

28 The TIGTA report noted that although there are restrictions under
current law prohibiting IRS from requiring individuals, estates, or trusts
to file electronically, there are no restrictions against IRS requiring
through regulations that corporate and other types of returns be filed
electronically.

In addition, Congress is currently considering the Tax Administration Good
Government Act of 2004,29 which would permit IRS to mandate increasing
e-filing of flow-through entity returns and accompanying forms, such as
K1s, in two new ways. First, the law would remove the present restrictions
in I.R.C. Section 6011 that prevent IRS from mandating individuals,
estates, and trusts to e-file. Since the law would remove the restriction
on mandating e-filing of individuals, IRS would then be able to mandate
efiling by paid preparers that prepare individual tax returns. Second, the
law would lower the threshold at which IRS could mandate e-filing of
information and tax returns for any taxpayer to 5 returns. Currently, the
threshold is 250 returns. Thus, IRS could mandate e-filing by paid
preparers who file 5 or more flow-through entity returns or individual tax
returns.

Conclusions	Although there are some costs to taxpayers to e-file and to
IRS in processing e-filed flow-through entity returns and related K-1s, in
general efiled K-1s offer substantial advantages for both IRS and
taxpayers. We are not making a recommendation for further action to expand
e-filing of flow through-entities' returns, including K-1s, because IRS
agreed to take steps to do so pursuant to a TIGTA recommendation and is
currently studying the costs of increasing e-filing to IRS and taxpayers.
One step, upgrading its overall capability to accommodate an increase in
e-filed flow-through entity returns, including K-1s, is under way.
However, we are concerned that IRS's estimated date for having this
capacity has been pushed back to 2007 due to limited resources. The sooner
this can be accomplished, the sooner IRS can reap the potential benefits
of an increase in e-filed Schedule K-1s while moving closer to achieving
the congressionally mandated goal of having 80 percent of all federal tax
returns and information returns filed electronically by 2007.

Regardless of whether e-filing is expanded, IRS is missing an opportunity
to improve the accuracy of TINs associated with K-1s and thereby is
undermining the benefits that can be realized from its document-matching
program, efficient targeting of examination resources, and new research to
identify noncompliance. Although IRS officials expressed concern about the
possible burden on flow-through entities of dealing with TIN error notices
and about IRS's ability to deal with the costs of sending such notices
given its resource constraints, IRS does not have information on

29 H.R.1528.

the likely benefits and costs of sending TIN error notices to flow-through
entities. Given the high concentration of TIN errors among a small portion
of flow-through entities, even if costs are high compared to the benefits
of sending notices to some flow-through entities, the situation may be
much different for error-prone flow-through entities.

Recommendation for 	To improve the availability and usefulness of Schedule
K-1 data to IRS for detecting noncompliance, we recommend that IRS conduct
a pilot study to

Executive Action	determine the benefits and costs of requiring
flow-through entities to correct invalid TINs on K-1s as soon as it has
been determined that the TINs cannot be "perfected" via IRS's TIN
validation program.

  Agency Comments and Our Evaluation

We received written comments on a draft of this report from the
Commissioner of Internal Revenue, which are reprinted in appendix II. The
Commissioner agreed with our assessment of Schedule K-1 TIN accuracy and
that a pilot project would be useful in identifying ways to improve TIN
accuracy. He said that IRS plans to study a number of options to ensure
that TINs included on Schedule K-1s are accurate, including our
recommendation that IRS conduct a pilot study to determine the benefits
and costs of obtaining corrected TINs from flow-through entities. The
Commissioner said that IRS's Flow-Through Compliance Committee recently
initiated a project to study invalid TINs on Schedule K-1s to determine
their potential compliance impact. In addition, he also mentioned other
initiatives, such as form redesign, outreach efforts, and scanning
Schedule K-1s, to improve the overall effectiveness of flowthrough
compliance efforts.

As agreed with your offices, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days
after its date. At that time, we will send copies to the Chairman and
Ranking Minority Member, House Committee on Ways and Means, and the
Chairman and Ranking Minority Member, Subcommittee on Oversight, House
Committee on Ways and Means. We will also send copies to the Secretary of
the Treasury, the Commissioner of Internal Revenue, and other interested
parties. The report will also be available at no charge on GAO's Web site
at http://www.gao.gov.

If you have any questions concerning this report, please contact me at
(202) 512-9110 or [email protected] or Jonda Van Pelt at (415) 904-2186 or
[email protected]. Key contributors to this report were Ralph Block, Maya
Chakko, Keira Dembowski, Elizabeth Fan, Robert McKay, and Samuel
Scrutchins.

Michael Brostek Director, Tax Issues

Appendix I

                       Objectives, Scope, and Methodology

Our objectives were to (1) evaluate the accuracy of K-1 data used by the
Internal Revenue Service (IRS), specifically transcription errors and
invalid taxpayer identification numbers (TIN); (2) determine whether any
limitations in the availability or accuracy of K-1 data have affected
IRS's ability to identify noncompliance; and (3) describe the benefits and
challenges of increasing electronic filing of K-1s.

To evaluate the accuracy of K-1 data used by IRS, we requested, obtained,
and analyzed data from IRS's K-1 database for tax year 2002.1 We examined
two versions of the database, both of which had been modified from the
original K-1 database by IRS research analysts. One, called the K-1
"cleaned database," has original K-1s removed when possible where
duplicate or amended K-1s for the same taxpayer were subsequently
submitted by the parent flow-through entity. The second, called the
"money-cleaned database," also has all amounts that were obvious
transcription errors removed. Generally, these were amounts in excess of
$900 million and that exceeded the total amount reported on the parent
flow-through entity's Schedule K for the particular line item. We used the
"cleaned database" to identify one such transcription error.

We analyzed the "money-cleaned" database to identify the number of K-1s
that were filed with inaccurate TINs by type of flow-through entity
(partnership, subchapter S corporation (S-Corp), or trust) and by type of
submission (e-filed versus paper filed). We subsequently analyzed this
subset of K-1s to determine the number and total income of K-1s with
invalid TINs that IRS (1) was able to perfect via its TIN Perfection
Program and (2) could not perfect and thus remained invalid and, in
effect, unusable for compliance or research purposes. Because specific
data were unavailable from the K-1 database concerning transcription
errors, we conferred with IRS analysts to identify the type of
transcription errors found during K-1 product reviews they conducted from
July through November 2003.

To identify whether any limitations in the availability or accuracy of K-1
data have affected IRS's ability to identify noncompliance, we obtained
from IRS the line items the agency transcribes from the K-1 and related
flow-through entity returns. When we calculated the percentage of line
items transcribed from the entity return, we included the K-1 as part of
the

1 At the time of our review, tax year 2002 was the most recent year for
which complete K-1 data was available from IRS.

Appendix I
Objectives, Scope, and Methodology

return. To count line items, we included all labeled lines and sub lines,
but excluded certain fields, including calendar or tax year, name and
address, supplemental information/attachments, signature and date,
preparer's signature and date, preparer's self-employment, and preparer's
firm name and address.

We also interviewed IRS examination and research staff, as well as outside
research consultants from the MITRE Corporation, with whom IRS contracted
to analyze flow-through entities. Specifically, we discussed how IRS
currently uses K-1 data to select flow-through entity returns for
examination, how IRS research staff and research consultants are using K-1
data to develop analytical tools to aid IRS in better targeting returns
for examination, and how data limitations affect their ability to
effectively use K-1 information.

To determine the compliance issues IRS identified and the related line
items that were adjusted, we reviewed a stratified probability sample of
partnership and S-Corp tax returns. We selected these returns from the
population of 253 partnership and 1,121 S-Corp agreed closed examination
cases listed in the IRS Audit Information Management System 2002 Closed
Case database with tax years ending in 2000 or 2001. We reviewed a sample
of 107 returns of which 91 returns, consisting of 52 of the partnership
and 39 of the S-Corp returns, could be analyzed. The remaining 16 returns
could not be analyzed, generally because the examination workpapers were
not available or the case adjustments were based on unusual circumstances,
such as amended return submissions from taxpayers. We used this sample of
91 returns to estimate several characteristics2 of this population of all
1,374 agreed partnership and S-Corp cases.

Because these estimates are based on a probability sample, our sample is
only one of a large number of samples that we might have drawn. Since each
sample could have provided different estimates, we express our confidence
in the precision of our particular sample's results as a one-sided 95
percent confidence interval. For example, paid preparers prepared an
estimated 93 percent of the returns, and a one-sided 95 percent confidence
interval for this estimate has a lower bound of 84 percent. Since the
actual population value would be contained in this interval for 95 percent
of the

2 These estimates are the percentage of returns with adjusted line items
not currently transcribed, percentage of paid preparer prepared returns,
and percentage of computergenerated returns.

Appendix I
Objectives, Scope, and Methodology

samples we could have drawn, we are 95 percent confident that the
proportion of paid preparer returns in the study population exceeds 84
percent. Similarly, for the adjusted lines found in the file review, we
are 95 percent confident that adjustments made to nontranscribed line
items occur in at least 40 percent of the examinations.

We subsequently discussed our file review findings with IRS research and
examination staff to obtain their views regarding whether having
additional K-1 data available, such as line items not currently
transcribed, would increase their ability to identify returns with
compliance issues.

To describe the benefits and challenges of increasing e-filing of K-1s, we
discussed this issue with IRS officials and officials from seven
organizations that represent the taxpayer community. We selected the
organizations based on prior GAO knowledge and referrals from some of the
organizations that we contacted. From IRS officials, we obtained estimates
of the cost to transcribe K-1 information, to help identify the potential
cost savings if K-1s were e-filed. We also discussed IRS's current
requirements for mandating e-filing K-1s and IRS's experience in enforcing
these requirements, and obtained data on penalties levied for failure to
efile required K-1s. Finally, we discussed IRS's current and future
ability to electronically process an increase in the number of e-filed
flow-through entity returns, including K-1s. We also contacted all 19 of
the software companies that offer e-filing for flow-through entities and
received e-mail responses from just over half of the companies. From
officials with the software companies, we obtained their current fees for
preparing and efiling flow-through entity returns and K-1s.

To determine how many flow-through entities filed on a calendar year basis
we used the 2001 Partnership and Corporation Statistics of Income (SOI)
samples.3 The SOI partnership data we used included the entire sample, but
the SOI corporation data we used were limited to the flow-through S-Corps.
Because these are probability samples, the SOI estimates are subject to
sampling error. We produced estimates from these samples using SOI's
sampling weights and methods that are appropriate for stratified
probability samples. In this report we present these estimates as

3 SOI last published information for estate and trust returns for tax year
1997.

                                   Appendix I
                       Objectives, Scope, and Methodology

intervals, reporting the lower bound on one-sided 95 percent confidence
intervals.4

We did our work at IRS headquarters in Washington, D.C., as well as at the
Ogden, Utah, Processing Center and the Oakland, California, Area Office
from April 2003 through July 2004 in accordance with generally accepted
government auditing standards.

Data Reliability	We assessed whether the information contained in the K-1
databases, the two SOI databases, and the Audit Information Management
System (AIMS) database were sufficiently reliable for the purposes of this
report. We ensured that the copies of all five databases we received from
IRS were identical to the original databases based on record counts and
analyses of control totals, comparison to published data, or both. In
addition, we performed electronic tests on the each database to search for
missing data and obvious errors.

For the original K-1 database from which the two cleaned versions we used
originated, we assessed IRS's procedures for processing and transcribing
Schedule K-1 data. We also assessed other procedures and methodologies IRS
research analysts used to remove duplicate records and obvious errors from
transcribed monetary fields. We anticipated the K-1 database would have
some reliability issues because our engagement was designed in part to
assess the sufficiency of the data transcription effort. While large
monetary transcription errors were removed from the "money-cleaned"
database, additional, undetectable transcription errors of amounts within
normal ranges may remain.

For the AIMS 2002 Closed Case database, we relied exclusively on variables
that allowed us to identify agreed closed case partnerships and S-Corps,
as this was the population of cases from which we drew our sample. We
interviewed IRS personnel who manage the AIMS databases and found that
groups in IRS conducting examinations are required to validate annually
that completed examination cases are actually shown as having been closed.
In addition, we collected data from original returns during our data
collection effort and compared those data to data contained in the

4 Additional information regarding the IRS's SOI program and its
stratified probability samples is available on the Internet at
www.irs.gov/taxstats/.

Appendix I
Objectives, Scope, and Methodology

AIMS database. We found no indication that our sample contained ineligible
cases.

SOI samples are widely used for research purposes. We have documented for
recent reports5 that IRS performs a number of quality control steps to
verify the internal consistency and completeness of SOI sample data. The
agency uses similar quality control procedures for all types of SOI
samples. For example, the agency performs electronic tests to verify the
relationships between values on the returns selected as part of the SOI
samples and manually edits data items to correct for problems, such as
inaccurate and missing items. Because we used the partnership and
corporate samples only to determine the percentage of partnerships and
S-Corps that were calendar year filers, we needed no more than four
variables from each database to make this analysis. We checked these
variables for completeness and accuracy and found no missing or out of
range values.

On the basis of our data reliability reviews of the five IRS databases, we
believe all five contain data that are sufficiently reliable for the
purposes of this report.

5 GAO, Tax Administration: Comparison of the Reported Tax Liabilities of
Foreign-and U.S.-Controlled Corporations, 1996-2000, GAO-04-358
(Washington, D.C.: Feb. 27, 2004). GAO, International Taxation: Tax Haven
Companies Were More Likely to Have a Tax Cost Advantage in Federal
Contracting, GAO-04-856 (Washington, D.C.: June 30, 2004).

                                  Appendix II

                   Comments from the Internal Revenue Service

Appendix II Comments from the Internal Revenue Service

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

Obtaining Copies of The fastest and easiest way to obtain copies of GAO
documents at no cost

is through GAO's Web site (www.gao.gov). Each weekday, GAO postsGAO
Reports and newly released reports, testimony, and correspondence on its
Web site. To Testimony have GAO e-mail you a list of newly posted products
every afternoon, go to

www.gao.gov and select "Subscribe to Updates."

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

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548

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

  To Report Fraud, Contact:
  Waste, and Abuse in Web site: www.gao.gov/fraudnet/fraudnet.htm

E-mail: [email protected] Programs Automated answering system: (800)
424-5454 or (202) 512-7470

Congressional	Gloria Jarmon, Managing Director, [email protected] (202)
512-4400 U.S. Government Accountability Office, 441 G Street NW, Room 7125

Relations Washington, D.C. 20548

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

                               Presorted Standard
                              Postage & Fees Paid
                                      GAO
                                Permit No. GI00

United States
Government Accountability Office
Washington, D.C. 20548-0001

Official Business
Penalty for Private Use $300

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