Forensics Journal - Stevenson University 2015 | Page 63
FORENSICS JOURNAL
information, you can calculate a company’s `fraud score’ or `F-Score’,
which can provide a pretty good indication of whether or not the
people inside the company might be manipulating their accounting”
(A tool for average investors to detect public company accounting
fraud, 2012).
To avoid rushing to judgment with respect to the integrity of
management’s financial reporting, the forensic analyst must consider
the totality of circumstances when evaluating changes in the results
of key indexes, ratios or models. When assessing the probability of
fraud, company specific information included in financial reports
should not be evaluated in isolation. External information, such as
overall industry growth, should be incorporated in the analysis
to help distinguish industry anomalies from company anomalies.
Accounting Quality Model
In July 2013, the SEC announced plans to develop an automated
data analysis tool to assist in the detection of financial reporting fraud
i.e. the Accounting Quality Model (AQM), but it was nicknamed
RoboCop by a reporter in a news article printed in The Financial
Times newspaper. The AQM tool combines computer automation
and data analysis to scan through corporate financial reports filed with
the SEC. Since all public companies are now required to submit their
financial reports using the eXtensible Business Reporting Language
(XBRL) filings format it makes corporate data easier to work with
and analyze using the AQM (XBRL.SEC.gov, n.d.).
USING COMPUTER TECHNOLOGY AND AUTOMATION
There are dozens of tools available to assist individuals in performing
data analysis. Some of these tools range from basic software programs
such as Microsoft Access and Excel, to more advanced programs
such as Audit Command Language (ACL), IDEA, and Monarch.
Regardless of the program selected, introducing automation into a
data analysis routine can enhance accuracy, efficiency, timeliness and
completeness. Many auditors use technology to review one hundred
percent of a transaction file. This allows conclusions to be reached
quickly and with greater confidence. Computer technology also
allows forensic accountants to create scripts, a set of predefined
automated procedures. The scripts are then applied in a consistent
manner to data files, when analyzing quarterly and annual financial
reporting data. Ideally, scripts should be designed to assist in the
periodic review of key metrics that point to anomalies in financial
reports. These exception reports are designed to alert the forensic
accountant or data analyst of potential problems which may require
a more focused review.
The AQM tool searches the SEC’s Electronic Data Gathering
Retrieval (EDGAR) database to locate financial statements filed by
companies. It then analyzes the financial statements for relationship
anomalies among the data, either within the company itself, or as
compared to its peer group, and flags the company for closer scrutiny
by an SEC Examiner. To help determine which companies receive
closer attention, and which do not, the AQM tool assigns risk scores
to companies based, in part, on its comparison of discretionary and
nondiscretionary expense data that appear in the financial reports
(A look inside the SEC’s accounting quality model, 2014).
Computer technology can also be used to analyze non-financial data
such as the text in financial statement footnotes or management
discussions and analysis. Some fraud experts believe that analyzing
words can be an effective way of identifying deception on the part
of management. In fact, on April 25, 2011, the SEC posted a request
for information (RFI) on its Federal Business Opportunities website,
seeking information about text mining software that it could use to
potentially predict financial statement fraud. A partial excerpt of the
request reads:
FALSE POSITIVES
The preceding indexes, ratios and models offer focused guidance but
no guarantees that observed anomalies are in fact synonymous with
fraudulent financial reporting. Identifying the presence of material
fraud in financial reporting is more of an art than a science and there
is no sure pathway to prevent or detect this type of fraud. The reason
for this is because the environment in which companies operate is
both volatile and dynamic. While this creates legitimate deviations
from normal business operating results, it also provides opportunities
to manipulate financial reporting. The peaks and troughs of the U.S.
economy distorts business operations which might impact some
industries more than others. The adoption of new accounting rules
impacts year-over-year changes in financial reporting. SEC regulations
change over time, especially in response to major corporate scandals,
and these changes impact the quality and quantity of information
reported to the public. Finally, there are certain areas in accounting
which are subject to management estimates, interpretation and
future projected which are subject to either legitimate change or
false manipulation. The challenge for the forensic analysis is to
differentiate truth from fiction.
“The U.S. Securities & Exchange Commission (SEC) seeks
information about data and text mining software applications
that assess the probability of financial statement fraud
occurring at a given public company. Such application should
utilize textual analysis to identify such things as word patterns
and frequency of usage that may be correlated with financial
fraud...” (SEC seeks information on data and text mining
software applications to analyze financial statements, 2011).
Although text mining is still relatively new and evolving, computer
programs such as RapidMiner and StatSoft allow text mining to be
performed with relative ease.
61