Auditing an XBRL Instance Document:

1
2007 International Conference on XBRL Taxonomies
and Assurance May 11 12, 2007
Auditing an XBRL Instance Document
The Case of United Technologies Corporation
University of Waterloo Efrim Boritz Won Gyun No
2
Agenda

• Introduction
• XBRL and Assurance
• Case Research Method
• Reporting Framework for Electronic Filings
• Assurance Framework for Electronic Business
  Reporting
• Findings
• Recommendations and Conclusion

3
Introduction

• Increased XBRL implementation for regulatory
  filings across the world.
• U.S. SEC XBRL voluntary program on EDGAR and
  5.5M to XBRL-US to develop taxonomies.
• U.K. Plans to make XBRL mandatory for company
  tax filings from 2010.
• Canada CSA XBRL voluntary filing program on
  January 19, 2007.
• Japan TSE XBRL reporting system in 2006.
• XBRL 3rd in FEIs top 10 financial reporting
  challenges for 2007(Heffes, 2007)
• Most companies currently providing their
  information using XBRL are doing so without
  assurance.
• Very limited guidance and experience on XBRL
  instance document preparation
• SECs voluntary XBRL filing program
• Some errors in SEC filings
• Case study of the EDGAR filing of United
  Technologies Corporation
• Perform mock audit procedures.
• Interview preparers and auditors.
• Address a number of issues that an auditor might
  confront if/when provides assurance on XBRL
  instance document.

4
XBRL and Assurance Three Phases in Electronic
Financial Reporting
5
SECs Voluntary XBRL Filling

• SECs voluntary XBRL filing on EDGAR
• February 3, 2005
• File paper format and furnish XBRL format
  statements.
• MDA, notes, and accountants report are
  optional.
• Companies create customized taxonomies to enable
  the XBRL filing to parallel the paper format
  filing as closely as possible.
• Some insights into SEC filings (at February 28,
  2007)
• 36 companies and total 126 fillings
• Significant parts of the filed instance documents
  are based on companies own customized taxonomy
  extensions.
• Instance Validation 51 (49.5)
• Taxonomy Validation 114 (90.5)
• Financial Reporting Instance Standards (FRIS) and
  Financial Reporting Taxonomies Architecture (FRTA)

6
Case Research Methodology

• Adopt case research methodology due to
• The context in which the filings under the
  voluntary SEC program are made.
• The limited information available on applying the
  new XBRL technology in a useful way
• UTXs Q3 2005 8K filing represents a critical
  incident The first XBRL filing assured by a
  public accounting firm
• United Technologies Corporation
• Multinational corporation based in Hartford,
  Connecticut, USA
• 40 billion company that provides high-technology
  products and services to the building systems and
  aerospace industries worldwide
• Eight business unitsCarrier heating and air
  conditioning systems, Hamilton Sundstrand
  aerospace and industrial systems, Otis elevators
  and escalators, Pratt Whitney aircraft engines,
  Sikorsky helicopters, Fire Security protection
  services, Power, and United Technologies Research
  Center.

7
Research Objectives

• Compare the XBRL instance document in detail to
  the paper format filing.
• Interview preparer to obtain explanations.
• Identify issues that an auditor might confront if
  performing audit procedures on an XBRL instance
  document.
• Learn about and report on the nature and extent
  of the work that would be required to determine
  whether the translation of the paper format
  document to the XBRL document was complete and
  accurate.
• Provide recommendations for both auditors and
  XBRL instance document preparers.

8
Longitudinal Analysis of UTX Filings
9
Analysis of Selected Other Filings
10
Reporting Framework for Electronic Filings
11
Assurance Framework for Electronic Business
Reporting
1. Acceptance
8. Using the work of an expert
9. Management representations

• From Assurance Working Group of XBRL
  International (2006)INTERACTIVE DATA THE IMPACT
  ON ASSURANCE, NEW CHALLENGES FOR THE AUDIT
  PROFESSION
• From Hayes, R., Dassen, R., Schilder, A.,
  Wallage, P. (2005)Principles of Auditing An
  Introduction to International Standards on
  Auditing

12
Audit Process Guidelines and Client Acceptance
Phase

• Guidelines
• Assurance framework proposed by the AWG
• Staff QA Regarding XBRL Financial Reporting
  (PCAOB)

• Client acceptance phase
• A auditor examining an engagement involving XBRL
  instance documents should not only have adequate
  knowledge of regulatory requirements, XBRL
  taxonomies, and XBRL specifications to perform
  the examination, but also have sufficient
  understanding of the companys financial
  statements and underlying financial records to
  evaluate the risk of misstatement in the instance
  document (PCAOB 2005, p. 3).
• Acceptance
• Have professional competency and sufficient
  knowledge and skills required to perform the
  engagement.
• Terms of engagement
• Prepare a report regarding the XBRL instance
  document.

13
Audit Process Planning Phase

• Users are most likely to be assured that the data
  presented in the XBRL instance document is a
  complete and accurate reflection of the data in
  the (audited or reviewed) financial statements,
  and whether the appropriate taxonomies are used
  (AWG 2006, p. 15).
• Planning the engagement Assessing the
  appropriateness of the subject matter

• The subject matter of the XBRL engagement10-Q
  instance document
• Primary focus of our substantive process
• Assess whether the elements in the XBRL instance
  document of UTXs 10-Q are the same as the
  financial facts in UTXs paper format 10-Q.
• Examine whether the appropriate taxonomies are
  used for mapping the facts in the financial
  statements to XBRL elements.
• Assessing the suitability of the criteria
• XBRL Specification v2.1, approved XBRL taxonomy,
  and stand alone add-on taxonomies would be
  considered as suitable criteria (PCAOB 2005, p.
  4).

14
Audit Process Testing and Evidence Phase

• The objectives of examination procedures
  regarding XBRL instance documents are to assess
  whether the XBRL-tagged data comply with the
  appropriate XBRL specifications and taxonomies
  and to evaluate whether the data elements in the
  XBRL instance documents reflect the same
  information as the corresponding official EDGAR
  filings (PCAOB 2005, p. 5).

• Obtaining evidence
• Examine whether the 10-Q instance document is a
  valid document that complies with XBRL
  Specification v2.1 using Fujitsu Instance
  Creator.
• Examine whether the UTX taxonomy, the company
  extension of the approved XBRL taxonomy, is a
  suitable taxonomy using Fujitsu Taxonomy Editor.
• Assess whether the 10-Q instance document is the
  complete and accurate reflection of the financial
  facts stated in UTXs 10-Q by performing a 100
  manual substantive audit consisting of two-way
  tracing.

15
Audit Process Manual Two-way Tracing Step I

• Contexts

16
Audit Process Manual Two-way Tracing Step II

• Financial Facts and XBRL Elements

17
Audit Process Manual Two-way Tracing Step III

• Notes and MDA

18
Audit Process Evaluation and Reporting Phase

• Evaluate the 10-Q instance document based on the
  evidence obtained through the audit process.

• Evaluate whether PWCs report regarding the XBRL
  instance document is consistent with the PCAOBs
  recommendation and includes the basic elements
  suggested by the AWG.

19
Findings Taxonomy

• Found that only 44.3 of the 10-Q instance
  document was based on the approved XBRL
  taxonomies.
• Could not determine whether this use of custom
  extensions is appropriate due to the lack of
  assurance standards or practices with respect to
  this matter.
• A custom taxonomy was used instead of the
  approved XBRL taxonomy that was available.
• Accountants Information Name ltAccountantsInform
  ationNamegt vs. ltAccountantNamegt
• Used different elements for the same names,
  signatures, job titles, and date contained in
  different exhibits.
• James E. Geisler and Gregory J. Hayes in the
  signature section of UTXs 10-Q were not used in
  the section 1350 certifications. Instead
  different taxonomy elements were used.

20
Findings Instance Document

• Contexts
• 145 contexts
• Not in any logical order
• Could cause problems in subsequent periods to
  determine inter-period consistency (e.g., quarter
  to quarter).
• Labels
• Titles and subtitles were not always presented in
  the instance document in a consistent manner, or
  were missing altogether.
• Some labels were missing or not exactly the same
  in the label linkbase as in the taxonomy.
• Totals
• Some totals which appear in the notes in the 10-Q
  are omitted in the instance document.
• Sometimes totals are in the instance document and
  sometimes they are left as implicit items defined
  in the calculation linkbase.

21
Findings Notes and MDA

• Disaggregation of textual narratives in both
  notes and MDA

XBRL Elements
22
Findings Accountants Report

• Used own taxonomy for Report of Independent
  Registered Public Accounting Firm instead of
  using usfr-ar (US Financial Reporting -
  Accountants Report) taxonomy.

23
Findings Accountants Report Regarding XBRL
Engagement

• AWG (2006, p. 24) 9 basic elements should be
  included in the report.
• PCAOB (2005, p. 6)
• An audit report if the XBRL instance document
  has been audited.
• A review report if it was reviewed and the
  report was filed with the SEC.
• Do not express opinion if it was reviewed, but
  the review report was not filed with the SEC.
• Disclaim an opinion if it was not covered by an
  audit report or review report.
• The AWGs recommended elements were
  satisfactorily addressed in the report.
• PWC stated that the underlying information in the
  10-Q instance document has not been audited and
  did not express an opinion on that information in
  the report since the review report prepared by
  PWC was not filed with the SEC.
• The accountants opinion on the instance document
  was created in HTML format rather than XBRL and
  was filed as a separate document.

24
Findings Accountants Report Regarding XBRL
Engagement
25
Findings Summary

• Took an XBRL expert about 63 hours to complete
  (not counting other steps that would be
  required).
• Had high assurance that the instance document was
  a complete and accurate reflection of UTXs 10-Q
  paper format filing.
• Cannot form a conclusion on the fairness ...in
  accordance with GAAP of the instance document.
• No assurance standards or guidelines for making
  such an assessment
• Limited knowledge for evaluating the MDA,
  regulatory information, and the appropriateness
  of the companys own extensions to the XBRL
  taxonomies

26
Recommendations

• Auditors effort, time, and cost could be
  reduced.
• A deliberate structuring of an XBRL instance
  document
• A logical ordering of elements in the XBRL
  instance document (e.g., by financial statement,
  notes, and MDA)
• An embedded description of the structure
• Explanation of why custom taxonomies are required
  or are preferred to approved taxonomies.
• Create rules for naming attributes
  (Sgt_Ottis_2004 vs. Context11).
• Organize contexts systematically and document
  rules used to create them.
• Assurance on XBRL instance documents.
• Some form of comfort/assurance will be necessary.
• Need guidance for both preparers and auditors.
• If an external report is provided, then it should
  be provided as an XBRL instance document to
  utilize the advantage of XBRL.
• Need a taxonomy for an auditor's assurance report
  on XBRL instance.

27
Conclusion

• Report on a case study in which we conducted a
  mock audit on the XBRL instance document of UTXs
  10-Q to identify issues that an auditor might
  confront if performing audit procedures on an
  XBRL instance document.
• Identified several significant issues that will
  need to be addressed by preparers,
  standard-setters, regulators, and auditors.
• Provided recommendations.
• Since we were independent of the preparation and
  assurance processes used by both UTX and PWC, our
  findings may not fully reflect some of the
  problems that occurred during the preparation and
  review of the XBRL instance document.
• For near future
• Understanding the process and internal controls
  over creation of XBRL instance documents will be
  essential because a purely substantive approach
  may not be feasible.
• Auditors will need knowledge of the XBRL
  taxonomies and specifications.

28
Questions Suggestions