Professional Documents
Culture Documents
Learning From WorldCom Implications For Fraud Dete PDF
Learning From WorldCom Implications For Fraud Dete PDF
net/publication/220010876
CITATIONS READS
43 2,342
2 authors, including:
Steve Sutton
University of Central Florida
122 PUBLICATIONS 2,291 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Steve Sutton on 27 October 2014.
INTRODUCTION
W
ith massive corporate failures sending shock waves throughout the stock markets,
the last few years have seen a heightened focus on enterprise-risk management
through stronger corporate governance, improved internal-control systems, more
transparent corporate reporting, and broadening of the assurance scope to encompass all of
these areas. Continuous assurance has accordingly received substantially greater attention
The authors thank participants at the 10th World Continuous Auditing and Reporting Symposium, and in particular
Miklos Vasarhelyi, for their helpful comments and suggestions on an earlier version of this paper.
Corresponding author: Steve G. Sutton
Email: [email protected]
61
62 Kuhn and Sutton
as it is increasingly being viewed as a potential tool for helping minimize the risk of
corporate fraud—particularly on the scale of what occurred at Enron, WorldCom, and Tyco.
Vasarhelyi (2005a) provides an examination of the Enron fraud and demonstrates how
continuous assurance would have helped detect the fraud surrounding the special-purpose
entities (SPEs) that were used to hide debt and prop up a positive outlook presented in the
corporate financial statements. However, the continuous audit metrics that would have been
most useful in detecting fraud at Enron are fairly specialized metrics that would have
addressed a fairly unique fraud issue—and one that was reasonably understood by the
auditors a priori to the fall of Enron.
WorldCom presents a very different situation, albeit a fraud of very similar magnitude.
In 1999, revenue growth at WorldCom (the then second largest telecommunications com-
pany in the U.S.) began to slow quite dramatically, expenses became a steadily increasing
percentage of revenue, and accordingly WorldCom’s stock price began to drop. In an effort
to meet earnings projections, management effected several fraudulent cost-reducing and
revenue-enhancing mechanisms.
The purpose of this study is three-fold. The first objective is to examine the key methods
of fraud utilized by the management at WorldCom as a basis for demonstrating the rea-
sonableness by which a continuous assurance strategy could be formulated to detect such
fraudulent behavior through the use of established principles of analytic monitoring
(Groomer and Murthy 1989; Vasarhelyi and Halper 1991; Kogan et al. 1999; Woodroof
and Searcy 2001; Vasarhelyi et al. 2004). The study focuses on a set of analytics that are
not complex, but rather demonstrate the ease at which the fraudulent behavior could have
been detected through the application of continuous assurance. As SAP implementations
are highly prevalent in industry, the second objective is to clearly articulate a design for
integrating the prescribed continuous assurance strategy with an SAP-based enterprise sys-
tem similar to WorldCom’s. The third objective is to explore the barriers to continuous
assurance applications that arise in highly complex system environments by illustrating how
the myriad of loosely connected legacy systems under WorldCom’s enterprise software
created an intractable monitoring problem that would have limited the completeness of any
continuous monitoring system.
The demonstration of how a reasonable and practical implementation of continuous
assurance would have detected a major fraud is a critical next step in the support of efforts
to make continuous assurance a frequently used audit tool. The system detailed is concep-
tually simple and readily implementable in an enterprise systems-driven environment. On
the other hand, the complexities of the continued use of widespread, independent legacy
systems creates an intractable control and monitoring problem and should similarly be
recognized as a major risk factor in any financial statement audit—not just for future
continuous audit implementations. In total, a contingency model exists whereby the com-
plete support of business processes through standardized enterprise systems software pre-
sents a feasible environment for continuous assurance implementation, but as the number
of legacy systems feeding data into the enterprise software increases, the complexity of
implementing continuous assurance escalates rapidly to the point of reaching infeasibility
rather quickly.
The research presented in this paper contributes to the growth of continuous assurance
research in three important ways. First, it provides a detailed understanding of how contin-
uous assurance techniques that have been explored in the literature can effectively identify
fraud in a known fraud situation. Second, it moves the literature on continuous assurance
models forward by addressing the complexities of implementation within a standardized
enterprise software environment (e.g., SAP). Third, it addresses the realities and risks as-
sociated with large numbers of disparate legacy systems.
The remainder of the paper consists of four major sections. The first section details the
nature of the underlying fraud techniques used by WorldCom’s management and describes
a set of analytic monitoring features that would have detected the fraud. The second section
addresses the implementation of the analytic monitoring features in an SAP enterprise
software system such as that used by WorldCom. The third section addresses where the
continuous monitoring system would break down at WorldCom due to the widespread use
of disparate billing systems that continued to exist in legacy form without directly feeding
the data into the central SAP system. The fourth and final section summarizes the impli-
cations of the research and outlines opportunities for future research.
TABLE 1
Financial Effects
Fraud Scheme Expenses Assets E / R Ratio Cite
Categorizing operating expenses as ↓ ↑ ↓ Feder and
capital expenditures Schiesel
(2002)
Reducing the book value of acquired ↓ Balanced — Eichenwald
MCI assets and simultaneously Effect (2002)
increasing goodwill for a balancing
amount
Including future expenses in write- ↓ ↓ ↓ Eichenwald
downs of acquired assets (2002)
Revising bad debt reserve calculations ↓ ↑ — Sender (2002);
to boost the valuation of Feder and
receivables that were subsequently Schiesel
factored (2002)
may be capitalized as assets and depreciated over their useful life.1 Deferment of these
costs artificially inflated reported net income and misled financial statement users. The
payments to lease phone network lines from other companies (i.e., allowing access to their
networks) are commonly referred to as ‘‘line costs’’ and represent the numerator in the
E/R ratio. The fraudulent capital expenditures resulted from manual reclassification of
existing operating expense account balances and inappropriate recording of future line-cost
transactions. As an example, the first quarter 10-Q report filed with the SEC in 2001
reported $4.1 billion of line costs and revenue of $9.8 billion resulting in an E/R ratio of
42 percent. Restated financial statements (i.e., post-fraud) disclosed actual line costs of $4.9
billion. The initial reclassification by WorldCom of $771 million of line costs to capital
expenditures was used to reduce the E/R ratio from 50 percent back to the desired level
of 42 percent (Kaplan and Kiron 2004).
The MCI merger provided WorldCom another opportunity to defer expenses. Manage-
ment reduced the book value of MCI assets by several billion dollars and simultaneously
increased the value of goodwill by a balancing amount. The related goodwill would be
amortized over a significantly longer period than that for normal asset depreciation based
on estimated useful lives of the respective assets. At the time of the fraud, GAAP permitted
amortization of goodwill over 40 years. This scheme enabled WorldCom to spread the cost
of the MCI assets over a longer period of time by recognizing a smaller amount of expense
each year and overstating net income.
In addition to the aforementioned MCI goodwill scheme, WorldCom employed fraud-
ulent accounting to numerous other corporate acquisitions. The company wrote-down mil-
lions of dollars in acquired entities’ assets resulting in excess charges against current earn-
ings—generally referred to in the earnings management literature as ‘‘taking a big bath’’
(Healy 1985) by absorbing costs at one point where they are not unexpected nor predictable
by the markets in order to improve future financial reports. The excess charges related to
costs other than those related to the assets that were being written down. The net effect
1
The Financial Accounting Standards Board (FASB) Concept Statement No. 5, Recognition and Measurement in
Financial Statements of Business Enterprises, discusses when various types of expenses should be recognized.
FASB Concept Statement No. 6, Elements of Financial Statements, further defines the characteristics of expenses
and assets and treatment of related costs.
was to create larger losses for the current quarter (i.e., the big bath) in order to improve
the picture presented through financial reports in future quarters. WorldCom wished to
create the false impression that expenses were declining over time in relation to revenue
(i.e., reducing the E/R ratio and increasing net income from operations). Users of the
statements were essentially deceived into believing there would be better performance in
future periods, maintaining upward pressure on the value of WorldCom’s stock.
From 1998 to 2000, WorldCom aggressively managed accounts receivable assumptions
and related bad debt reserves. When a company extends credit to a customer with the
promise by the customer to pay in the future, invariably a portion of these customers will
ultimately not pay all or part of their bill. An estimate of the portion of receivables that
will not ultimately be collected must be derived and a related expense accrued on the
financial statements. Companies estimate the amount that should be expensed as uncollec-
tible on the income statement and the net amount of accounts receivable (i.e., gross accounts
receivable less the allowance for uncollectible accounts) to report on the balance sheet.
Altering the percentage of accounts receivable in the estimation of the uncollectible portion
directly impacts the bad debt expense reported on the income statement and the reserve
amount netted against gross accounts receivable on the balance sheet.
WorldCom’s revenue and total accounts receivable decreased $3.9 billion and $1.95
billion, respectively, in 2001. As a percentage of total accounts receivable, the allowance
for doubtful accounts (i.e., the reserve) decreased from 18.35 percent in 2000 to 16.98
percent in 2001. Applying the same reserve calculation as utilized in the previous year,
earnings would have decreased even further by another $87 million. Either the company
drastically improved collection efficiency or under-reserved thus artificially inflating both
reported revenue and the value of receivables to be factored in the future.2 The latter seems
most likely given that the company recorded a $322 million adjustment to write-off addi-
tional bad debt in 2002, restated the 2001 allowance account by an increase of $751 million,
and ultimately filed for bankruptcy protection on July 21, 2002.
The methods of manipulating financial information described in the preceding para-
graphs may not comprise all the ways in which WorldCom committed fraud. The four do
represent the core techniques employed by WorldCom that ultimately resulted in the most
significant adverse consequences. The challenge for the auditor is to identify a feasible
means for detecting such behavior in order to avoid a replication of the WorldCom fraud
at another company in the future.
2
Factoring accounts receivable (i.e., selling) at higher values results in increased reported income based on faulty
assumptions.
Consistent with the description by Alles et al. (2002), the remainder of this paper will focus
on the technique of continuous auditing as a tool in a continuous assurance framework.
The Continuous Process Auditing Methodology (CPAM) put forth by Vasarhelyi and
Halper (1991) offers a framework from which to explore the development and refinement
of a continuous auditing approach. In their methodology, transactional and system data are
monitored and analyzed continuously based on a rule-set predefined by the auditors in the
continuous audit application. Exceptions to the rules trigger alarms that automatically notify
the auditors of potential irregularities. The nature of audit work transcends from primarily
a substantive-based test of details approach to a focus on auditing by exception.
3
Substantive audit procedures are activities performed by the auditor that gather evidence to test the completeness,
validity, and / or accuracy of account balances and underlying classes of transactions. Tests of detail represent
the collection of certain types of evidence (e.g. vouching supporting documentation, physical examination of
assets, recalculation of estimates and entries, third party confirmation of assets / liabilities, inquiry of client
employees, etc.).
FIGURE 1
Monitoring Framework
Alarms
To Other Stakeholders
Monitoring
Analytics and
Exception Reporting Audit Exceptions
External Information
Internal and External
Monitoring Metrics To Operations
Corporate Strategic and
Tactical Me trics
Scorecard
Monitoring IT Structure
Internal Information
Replicated from Vasarhelyi (2005b).
three phases: (1) Measurement, (2) Monitoring, and (3) Analysis. Measurement consists of
key management reports (e.g., financial statements, security settings) from which metrics
are derived for comparison to standards. During the monitoring phase, the audit system
performs constant comparison of metrics to the standards programmed into the system (i.e.,
analytics) and triggers alarms to notify auditors of any inconsistencies. Vasarhelyi and
Halper (1991, 117) define analytics as ‘‘functional (natural flow), logical (key interaction),
and empirical (e.g., it has been observed that ...) relationships among metrics’’ that can be
derived from auditor, management, or user experience and historical data in the system. In
the final phase, the auditors review the nature of the alarms and investigate as appropriate.
Vasarhelyi et al. (2004) further expand on the theme of continuous auditing as an
analytic tool. They contend that market demand will shift assurance requirements away
from the traditional ‘‘ex post evaluation’’ to a more ‘‘close-to-the event’’ review. The new
role of auditors will be more invasive as they monitor day-to-day business operations
through the use of technology and auditing by exception. To distinguish from the traditional
auditor role, Vasarhelyi et al. (2004) refer to the new process as ‘‘analytic monitoring.’’
The new assurance technologies described by Vasarhelyi et al. (2004) allow analytic
monitors to observe events as they transpire, trigger alarms when exceptions occur, drill
down to transactional detail, integrate data across multiple and distinct processes, and per-
form repeated tests at minimal cost. Vasarhelyi et al. (2004) also examine several tools that
underlie analytic monitoring such as continuity equations, tagging data accuracy, time series
analysis, dynamic reconciliation of accounts, data taps, confirmatory extranets, and invisible
tags. These analytic tools could have been strategically implemented at WorldCom, facili-
tating early identification of fraudulent activity.
Recent legislation and reporting trends, such as the Sarbanes-Oxley Act of 2002 (SOX)
and initiatives for continuous disclosure of financial information, provide greater incentives
for the development and use of continuous assurance in the financial reporting environment.
Section 404 of SOX requires external auditor evaluation of the effectiveness of internal
controls over financial reporting. Continuous assurance of key technology-related controls
can be an integral component of the auditor’s overall 404-compliance audit with the results
used as evidence to assist in the determination of the auditor’s opinion (e.g., Alles et al.
2006). SOX also speaks to the need for continuous financial reporting which is only feasible
if continuous assurance can be provided over system’s reliability and through event trans-
action monitoring. Continuous assurance necessitates activities that monitor both internal
controls and transaction processing.
complexity of SAP databases, Alles et al. (2006) elected to extract data from the application
layer using application program interfaces, termed BAPIs in SAP terminology. BAPIs are
typically supported by system vendors and well documented—providing a practical ap-
proach. This approach works well with the Siemens prototype where control processes are
largely embedded within the application layer.
Given the focus of this research study on financial statement information, the rationale
for aggregating data through the application layer no longer exists. For the required ana-
lytics in the WorldCom example, the focus is on analyzing consolidated financial accounting
information from an analytic perspective. This requires obtaining transaction and account
detail from the SAP general ledger that resides in two SAP tables: the GLPCT (summary
table of account balances) and the GLPCA (individual transaction data). Extracting all
necessary data from two tables within the database layer significantly reduces the com-
plexity faced by Alles et al. (2006), offering a more simplistic approach than accessing
data through the application layer via BAPIs.
The CA Analyzer operates external to the SAP application and performs test procedures
against extracted data residing in a relational database. To obtain the necessary data, an
extractor program executes a remote function call (RFC) requesting access to SAP. RFC is
a standard programming interface that can be used to send and receive data. The RFC
connection must be given a unique SAP user ID and password with permission to display
tables (authorization object S TABU DIS) and to execute an RFC (authorization object
S RFC). Table display access can be further restricted to view only GLPCT and GLPCA.
SAP software provides an application-level gateway program called SAProuter that
serves as a proxy controlling communication with other systems (i.e., access to and from
SAP). SAProuter contains a routing table that defines the IP addresses for approved external
connections. If utilized, SAProuter can provide an additional layer of security for interaction
between the RFC and the client’s SAP environment. The auditor therefore must coordinate
with the client to establish and configure user access rights for the RFC and connection
permission in the SAProuter table. Figure 2 illustrates the interaction between the external
program called an extractor that executes an RFC connection to SAP and is validated by
SAProuter.
FIGURE 2
Continuous Audit Data Flow
(MCL←→SAP)
CA Analyzer Internet
Industry Client
(with rule-set)
Data Portfolio
Exception
Report
Internet
Third Party
Alerts Storage
Auditor Black Box
Database
‘‘feed forward’’ into the planning, testing, and review phases of the audit. Likewise, inter-
faces to the CA Analyzer from the client portfolio application can continually update the
rule-set configuration within the CA analyzer with client-specific risk factors.
The ‘‘black box log file,’’ as recommended by Alles et al. (2004), provides a mechanism
for recording and retaining the continuous audit evidence allowing tertiary monitoring of
the auditor (i.e., audit of the auditor). The maintenance of the monitoring data provides a
motivation for the auditor to explore risks highlighted in alerts rather than ignoring them.
The illustration herein will address the four previously discussed fraudulent schemes
employed at WorldCom by demonstrating the manner in which a properly configured con-
tinuous auditing application would identify the issues in a timely manner. For purposes of
the illustration, assume a single SAP instance containing the general ledger account bal-
ances and consolidated financial information (i.e., the information reported to the SEC and
investors) versus a company operating multiple SAP applications for various business
processes/units.
The key factors dictating the type of relational database required for the continuous
assurance structure include the size of the extraction, number of ERP systems evaluated,
length of retention of the data, and frequency of downloads (Alles et al. 2006). For pro-
totyping purposes, the Siemens proof-of-concept stored the extracted data in an MS Access
database. However, the team recognized that a robust relational database with database
management and support capabilities (e.g., Oracle or MS SQL Server) would be required
in a production environment.
The key component of the continuous assurance system entails the configuration
and operation of the CA Analyzer. The rule-set defined in the application determines the
identification of exceptions and subsequent auditor notification. The rule-set is the driving
force enabling monitoring.
Alles et al. (2006) developed a CA Analyzer in Visual Basic for the Siemens project
as a test environment to evaluate the technical research questions regarding continuous
assurance. The Siemans project focused on continuous monitoring of business process con-
trols. The configuration of the CA Analyzer addressed the SAP controls as defined in the
Siemens IT audit plans (e.g., comparison of password and system parameter settings to best
practice standards).
The focus of this paper differs from Alles et al. (2006) in the nature of the analytic
monitoring to be performed by the CA Analyzer. The Siemens project standardized and
automated certain tasks and procedures within existing audit programs developed by the
internal IT audit department to review SAP business process controls and system parameter
settings replicable across hundreds of SAP instances. The objective of the approach devel-
oped herein concentrates on developing an automated mechanism for external auditors to
detect financial irregularities and material misstatements of the financial statements in a
timely fashion through substantive-based audit procedures. The illustration in this research
is alternatively focused on how WorldCom management conducted financial fraud and the
rule-set configuration necessary to detect such behavior. To reiterate, the key WorldCom
fraudulent activities included inappropriate reclassification of operating expenses as capital
expenditures, reclassification of acquired MCI assets as goodwill, inclusion of future com-
pany operating expenses in the write-down of acquired assets, and manipulation of bad
debt reserve calculations. The rule-set developed in this study focuses on these specific acts
of fraud. Development of the full set of analytics that would be desirable for complete
monitoring is beyond the scope of this study.
Identification of material reclassifications of operating expenses as capital expenditures
could occur through several monitoring analytics. The CA Analyzer can store multiple years
worth of financial data (i.e., general ledger account balances) and compare the company’s
historical ratios of operating expenses and capital expenditures to sales revenue. A decline
in the proportion of operating expenses to sales revenue and corresponding increase in the
ratio of capital expenditures to sales revenue exceeding a configured threshold would trigger
one alarm. Related industry and firm risk data can also be incorporated into the CA Ana-
lyzer to facilitate variations of the metric. An alert is communicated if the company’s ratios
differ substantially from expected ranges.
Analytic monitoring to identify inappropriate reclassifications (based on industry av-
erages) can be configured as:
IF the operating expenses to sales ratio is ⬎ 2% below .93 AND the capital expenditures to sales
ratio is ⬎ 5% above .15, THEN create an alert.
WorldCom’s 2001 operating to sales and capital expenditures to sales ratios were .90 and
.22, respectively, and exceeded the sample thresholds by $946 million and $585 million,
respectively.
WorldCom wrote down assets acquired in the MCI merger transferring the cost to
goodwill in the same amount. The sheer magnitude of the entries, in and of themselves,
should raise concern. All material journal entries, regardless of account, should be examined
by the auditors. The auditors can configure the rule-set to identify journal entries above a
certain dollar threshold to trigger such an alert. More specific to the assets and goodwill
account, the CA Analyzer can contain logic identifying significant changes in key account
balances over a short time period. The excessive growth of WorldCom through merger and
acquisition, asset additions, write-downs, and purchased goodwill accounts warranted spe-
cial attention. The CA Analyzer can be configured to focus attention on accounts containing
higher audit risk.
A sample monitoring analytic to identify inappropriate asset reclassifications to good-
will can be configured as:
IF the property, plant, and equipment and goodwill account balances increase or decrease by ⬎ .01%
from the last extraction, THEN create an alert.
WorldCom’s goodwill balance at the 2001 year-end was $50.5 billion. The 01 percent
threshold equated to a change of $5 million. Actual change from the previous year was
$3.9 billion which would trigger an alert for the group of entries during 2001 that reclas-
sified MCI assets to goodwill.
The acquisition activities of WorldCom resulted in additional fraudulent acts. The com-
pany wrote-down assets and in doing so, included future company operating expenses. This
allowed the E/R ratio (i.e., line cost expenditure/revenue) to remain steady at 42 percent
over a sustained period. The telecommunications industry, as a whole, experienced lesser
margins during the same period as a result of steadily increasing E/R ratios (i.e., the
industry average was 50 percent—see Kaplan and Kiron [2004]). Entering key industry
indicators into the CA Analyzer and configuring the application to identify significant de-
viations from those trends, would flag such an irregularity. A number of vendors such as
the United States Telecom Association, Moody’s, BizMiner, etc. sell key telecom industry-
specific information such as financial statement line items and key ratios, historical and
forecasted growth, specific sector analyses, and other information that auditors could au-
tomatically load into the underlying database of the CA Analyzer. Comparison of current
line cost expenditures and revenues to industry trend ratios by the CA Analyzer would have
raised alerts.
Analytic monitoring to identify the inappropriate inclusion of future expenses in current
write-downs of acquired assets can be configured to compare the trend of the client’s E/R
ratio over both a five-year period and the course of the most recent 12 months to that of
the industry.
IF the slope of the trend (where x ⫽ time period, y ⫽ E / R ratio) falls below that of the industry
trend by ⬎ 1%, THEN create an alert.
As previously noted, the industry experienced progressively worse E/R ratios (i.e., positive
slope) whereas WorldCom repeatedly reported constant E/R ratios (i.e., a flat trend line
with zero slope). The difference in slope between the industry and WorldCom would have
created an alert.
Historically, bad debt reserve calculations represent an area susceptible to accounting
fraud. The reserves can be manipulated in either direction by the company’s management,
expensing more bad debt in good times and less in leaner periods. During the years of
fraudulent activity, WorldCom exhibited behavior consistent with under-funding of bad debt
reserves in an apparent effort to lower expenses and inflate receivables. Industry related
data integrated into key metrics within the CA Analyzer, along with logic to perform
comparisons of current bad debt ratios to historical estimations and actual expenses in-
curred, would have triggered an alert.
Analytic monitoring to identify bad debt reserves manipulation could be configured as:
IF the change in the ratio of bad debt allowance to total accounts receivable is ⬎ 1% below last
month’s figure, THEN create an alert.
The bad debt reserve estimate dropped by 1.4 percent in 2001 from the prior year ‘‘saving’’
the company $87 million in bad debt expense. An alert to the auditors would have been
created when the reserve calculation dropped below the threshold.
As shown in the Siemens project (Alles et al. 2006) and our systems design for financial
statement information monitoring, the MCL strategy for continuous assurance can be ef-
fectively implemented in an SAP environment while minimizing the impact on the system
and its users. Automation of common analytical procedures through the configuration of
the CA Analyzer rule-set offers a valuable tool to quickly process and analyze the extensive
volume of data typically residing in a major ERP system. The external auditor has the
additional benefit that data necessary for financial transaction and account balances moni-
toring can be extracted directly from the SAP database, precluding the client from mon-
itoring the specific information being examined whereby the auditor’s tests might be cir-
cumvented. The auditor also avoids any manipulations of the client’s system, thereby
alleviating the independence concerns that have been raised in relation to continuous as-
surance. In summary, the technology exists to successfully design, implement, and operate
an effective continuous assurance application within a standard enterprise systems
environment.
Texas (Schultz and Watt 1998). The article reports staggering statistics related to the com-
plexity of the WorldCom back-end systems: thousands of mainframes (Amdahl, Hitachi,
and IBM) and minicomputers (Digital, Hewlett-Packard, Sun Microsystems, and from com-
panies no longer in business) located in data centers across the country (Colorado Springs,
Pentagon City, Cedar Rapids, etc.). With this network, WorldCom provided $1.6 billion in
outsourcing services and managed more than 200 customer networks in addition to their
own operations. The engineer joked about the Year 2000 problem, ‘‘We have so many
systems, it would take us until 3000 to write custom applications for them.’’ Figure 3 depicts
the data flow of telephony traffic from origination (telephone switches) to legacy mainframe
traffic and billing systems to ultimately the SAP consolidated financial reporting application.
The acquisition frenzy during the 1990s (more than 60 companies) created significant
challenges for WorldCom management. Rather than requiring the acquired companies to
migrate to WorldCom applications, management routinely elected to retain the acquired
systems and create complex interfaces. KPMG identified over 30 billing and accounts re-
ceivable systems outside SAP during the restatement audit in 2003.4 A CA Analyzer that
only reviews data from the SAP general ledger and subledgers fails to consider the audit
assertions of completeness and valuation.5 A serious question arises: ‘‘Does the consoli-
dating financial reporting application contain all the data and are the amounts correct?’’
The question cannot be answered unless the CA Analyzer receives information from all
legacy applications. With the data, the CA Analyzer can confirm the successful transmission
of the interfaces to SAP to address completeness, and perform analytics such as recalcu-
lating bad debt reserves to confirm proper valuation. Receiving that data may not be a
realistic expectation.
The necessary monetary and human capital costs as well as the overall time required
to establish links to each of the billing applications drastically increases the scope. Consider
also that the billing and accounts receivable cycle represents only one of the many
WorldCom business processes. The implementation of a comprehensive, fully integrated
continuous assurance process at WorldCom may have been too cost-prohibitive and time-
consuming simply because of the myriad of legacy systems providing data to the SAP
general ledger.
4
Regulators required MCI WorldCom to restate the financial statements from 1999–2002 to reflect accurate
reporting (i.e., backing out of the fraud). KPMG performed the audit of the restated statements.
5
Statement of Accounting Standard (SAS) No. 31 identifies the five general classes of assertions about which
auditors are required to collect enough relevant information to lend credence to the items reflected in the financial
statement: existence, completeness, valuation, rights and obligations, and presentation and disclosure. SAS No.
31 states ‘‘Assertions about completeness deal with whether all transactions and accounts that should be presented
in the financial statements are so included.’’ SAS No. 31 states ‘‘Assertions about valuation or allocation deal
with whether asset, liability, revenue, and expense components have been included in the financial statements
at appropriate amounts.’’
FIGURE 3
WorldCom Billing Process
Billing #1
Billing #2
SAP R/3
(Revenue & A/R)
• • •
• • •
• • •
Billing #30
Legacy
Telephone Switches Traffic Systems Billing Systems
and billing processes. The complex network and interaction of legacy systems outside the
SAP consolidated instance would have created a significant hindrance to the use of a com-
plete continuous assurance system. As technology evolves to address current limitations
and legacy systems expire, continuous assurance will continue to advance and offer even
more viable opportunities than are currently available for increasing the audit quality and
changing the overall attestation function.
Academic literature has focused primarily on progressing continuous assurance theory
and, therefore, has been hindered by the lack of experimental and empirical research
(Vasarhelyi et al. 2004). Recent research efforts (Alles et al. 2006) provide proofs of concept
by designing and implementing continuous assurance applications into limited, controlled
environments. To gain a better understanding of the overall feasibility and challenges to be
faced, widespread implementation in complex system environments (i.e., legacy systems,
ERP systems, e-commerce applications, etc.) must occur. In these implementation ventures,
opportunities also exist for researchers to explore the integration of neural networks and
other artificial intelligence techniques into the CA analyzer component at the heart of a
continuous assurance system. The successes in earlier research that has used such tech-
niques for fraud detection and bankruptcy detection offer great potential to enhancing the
power and usefulness of continuous assurance applications. Research that integrates these
concepts with continuous assurance models will make significant contributions.
Among the implementation challenges of any continuous assurance technology are
certainly the initial cost of investment in terms of human capital and technological archi-
tecture, the ongoing consumption of system resources, the scalability of continuous assur-
ance applications as technology changes, and the need to constantly refresh industry and
economic trend data required for the accuracy and reliability of the underlying CA Analyzer
rule-set. Academic researchers are well trained and prepared to research these types of
constraints and barriers to continuous assurance, and should assume a leadership role in
addressing these challenges.
Continuous assurance research should also begin to evolve beyond just the technical
aspects of continuous assurance to also developing an understanding of the behavioral
impacts of implementing such systems. From the auditor-auditee relationship aspect, there
are a myriad of organizational and individual level effects that should be considered and
addressed during the implementation of continuous assurance. For instance, how does con-
tinuous assurance affect the trust between auditors and auditees? Does continuous assurance
encourage gaming behavior as auditees look to game the system when facing automated
versus human auditing? From the auditor side of the relationship, there are also a number
of concerns that have to be addressed. Automated continuous assurance does not eliminate
the human component as interpretation of information still exists, but rather shifts the
auditor’s role to a certain degree as the auditor must learn how to sift through audit alerts,
identify alerts that detect real problems, and determine appropriate follow-up procedures
for unusual events detected on a continuous basis. Current prototypes have certainly pro-
vided evidence that there is typically a large volume of alerts generated, and the auditor
must adjust. This suggests the relevance of applying existing theories of information proc-
essing in order to understand the impacts of information overload, information processing
biases, and related information processing challenges that may uniquely exist in a contin-
uous assurance environment. The opportunities for (and need for) research on continuous
assurance hold great promise for researchers who wish to apply their research skills to
moving practice forward and providing normative guidance on the basis by which contin-
uous assurance should evolve and gain traction.
REFERENCES
Alles, M., G. Brennan, A. Kogan, and M. A. Vasarhelyi. 2006. Continuous monitoring of business
process controls: A pilot implementation of a continuous auditing system at siemens. Interna-
tional Journal of Accounting Information Systems 7 (2): 137–161.
Alles, M., A. Kogan, and M. A. Vasarhelyi. 2004. Restoring auditor credibility: tertiary monitoring
and logging of continuous assurance systems. International Journal of Accounting Information
Systems 5 (2): 183–202.
———. 2002. Feasibility and economics of continuous assurance. Auditing: A Journal of Practice &
Theory 21 (1): 125–138 (March).
American Institute of Certified Public Accountants (AICPA). 1980. Evidential Matter. Statement of
Auditing Standards (SAS) No. 31 August. New York, NY: AICPA.
Bell, T. B., J. C. Bedard, K. M. Johnstone, and E. F. Smith. 2002. KRiskSM: A computerized decision
aid for client acceptance and continuous risk assessments. Auditing: A Journal of Practice &
Theory 21 (2): 97–113.
Canadian Institute of Chartered Accountants and American Institute of Certified Public Accountants
(CICA / AICPA). 1999. Continuous Auditing. Research report. Toronto, Canada: CICA.
Eichenwald, K. 2002. For WorldCom, acquisitions were behind its rise and fall. The New York Times
(August 8).
Eining, M. M., J. K. Loebbecke, and J. J. Willingham. 1990. Expert Systems: Issues for Decision
Making in an Auditing Setting. Proceedings of the Third International Symposium on Expert
Systems in Business, Finance, and Accounting, Marina Del Ray, CA.
———, D. R. Jones, and J. K. Loebbecke. 1997. Reliance on decision aids: An examination of
management fraud. Auditing: A Journal of Practice & Theory 16 (Fall): 1–19.
Feder, B. J., and S. Schiesel. 2002. WorldCom finds $3.3 billion more in irregularities. The New York
Times (August 9).
Graham, L., and J. Bedard. 2003. Fraud risk and audit planning. International Journal of Auditing 7
(1): 55–70.
Groomer, S. M., and U. S. Murthy. 1989. Continuous auditing of database applications: An embedded
audit module approach. Journal of Information Systems 3 (2): 53–69.
———, and ———. 2003. Monitoring high volume on-line transaction processing systems using a
continuous sampling approach. International Journal of Auditing 7: 3–19.
Hackenbrack, K. 1993. The effects of experience with different sized clients on auditor evaluations
of fraudulent financial reporting indicators. Auditing: A Journal of Practice & Theory 12
(Spring): 99–110.
Healy, P. 1985. Effect of bonus schemes on accounting decisions. Journal of Accounting and Eco-
nomics (April): 85–107.
Kaplan, R., and D. Kiron. 2004. Accounting fraud at WorldCom. Harvard Business School (July 26).
Kogan, A., E. F. Sudit, and M. A. Vasarhelyi. 1999. Continuous online auditing: A program of
research. Journal of Information Systems 13 (2): 87–103.
Pincus, K. V. 1989. The efficacy of red flags questionnaire for assessing the possibility of fraud.
Accounting, Organization, and Society 14: 153–163.
Schultz, B., and P. Halper. 1998. Legacy lessons. NetworkWorldFusion News (October).
Sender, H. 2002. Accounting issues at WorldCom speak volumes about disclosures. The Wall Street
Journal (August 21).
Vasarhelyi, M. A. 2005a. Would continuous audit have stopped the Enron mess? Working paper,
Rutgers University.
———. 2005b. Concepts in continuous assurance. Working paper, Rutgers University.
———, M. Alles, and A. Kogan. 2004. Principles of analytic monitoring for continuous assurance.
Journal of Emerging Technologies in Accounting 1: 1–21.
———, and F. B. Halper. 1991. The continuous audit of online systems. Auditing: A Journal of
Practice & Theory 10 (1): 110–125.
Winograd, B. N., J. S. Gerson, and B. L. Berlin. 2000. Audit practices of PricewaterhouseCoopers.
Auditing: A Journal of Practice & Theory 19 (2): 175–182.
Woodroof, J., and D. Searcy. 2001. Continuous audit: Model development and implementation within
a debt covenant compliance domain. International Journal of Accounting Information Systems
2 (3): 169–191.