Welcome to The CPA Journal Archives

Visit cpajournal.com to read the very latest from The CPA Journal


Applying Technology to Accounting Fraud

Maria L. Murphy, CPA

When I first heard the term “RoboCop” at a recent conference, I thought I must have misheard. But speakers from the SEC and the Center for Audit Quality (CAQ) were talking about a new weapon in the battle against accounting fraud. The new SEC chair, Mary Jo White, has been clear about her intent to strengthen SEC enforcement activities, because “the more successful [the SEC is] at being—and being perceived as—the tough cop that everyone rightfully expects, the more confidence in the markets investors will have” (speech at the Council of Institutional Investors' conference, Sep. 26, 2013).

The Accounting Quality Model

The SEC has introduced a new accounting quality model (AQM, or RoboCop), an automated system that uses registrants' data to detect fraud and other “accounting irregularities” in a faster, more efficient way. This new model uses XBRL tags of financial data included in SEC filings and creates risk scores for SEC staff to use in reviews of company filings and investigation activities. (A further description of the AQM appears on page 68 in “The SEC's Renewed Focus on Accounting Fraud: Insights and Implications for Auditors and Public Companies.”)

In numerous public statements, the SEC has acknowledged that it has had fewer fraud enforcement actions and that there have been fewer restatements of financial statements overall in the past five to seven years; however, some are attributing the statistic to the SEC's focus on other issues related to the financial crisis overall, rather than to better behavior by registrants or improvements in financial reporting as a result of Sarbanes-Oxley reforms. The SEC is devoting additional resources to, among other things, finding out whether there has actually been an overall reduction in fraud. It recently formed a Financial Reporting and Audit (FRAud) task force of accountants and lawyers to combat false and misleading financial statements and audit failures. The SEC intends to develop and use “state-of-the-art methodologies” to identify and investigate financial reporting fraud and to generate new accounting fraud investigations for SEC enforcement staff to pursue; it has categorized this as a new way of “crunching data … to isolate potential red flags” (Andrew Ceresney, speech at the American Law Institute Continuing Legal Education, Sep. 19, 2013).

The SEC is applying modeling, including the Jones Model, to assess risk and identify potential earnings management by estimating “discretionary accruals,” which looks at changes in factors like revenue and fixed assets to predict “nondiscretionary accruals.” The AQM goes further by identifying “outlier discretionary accruals” as indicators of earnings management. I am not an economist but, as a CPA, it seems to me that labeling a discretionary accrual as abnormal, or indicating that outliers are an indicator of fraud based on these types of metrics, could potentially send the wrong messages to everyone—pre-parers, auditors, regulators, and investors.

Is Modeling the Answer?

No one would argue that fraud should be targeted, that investors can suffer real harm due to fraud, that abusive earnings management should be prevented, and that audit failures to detect fraud must be mitigated. But there is definitely debate about whether economic modeling and surveillance tools to detect earnings management, such as the AQM, are the right way to address these issues.

My concern is with the hunt for “disclosure anomalies” and the use of technology to identify company outliers, those whose financial metrics are different from their peer groups. How can anyone agree as to what is an “anomaly,” what standard ratios should be for companies within industry groups, or what the key financial drivers are? Is there a need for “risk indicators” and “risk inducers” and scoring filings across peer groups? For years, auditors and companies have been using analytics and peer group analysis as a tool to assess reasonableness of financial information. But it is a reasonableness test, used in combination with other procedures. Results fluctuate period to period in the normal course, and GAAP permits estimations and ranges and judgments to be made.

There are other more useful approaches to combating fraud, including the CAQ's AntiFraud Collaboration, the PCAOB's Standing Advisory Group, and the SEC FRAud task force's collaboration and education goals. Regulators are focusing on improving internal controls over financial reporting through the education of companies and audit committees and through the oversight and inspection of auditors, which should have a direct impact on the reduced potential for fraud. Whistleblower programs are providing useful information to regulators to combat fraudulent activities and reporting.

Models and metrics might assist SEC staff in reviewing registrant filings and other enforcement activities to combat fraud. My concern is that there is too much data and not enough analysis, and that registrants will have to defend their financial statement ratios in comment letters and SEC investigations. Will this lead to public companies, their audit committees, and their auditors trying to anticipate what metric will be an anomaly and crack the RoboCop code? Will companies follow their peer group's accounting policies and make conservative accrual estimates, rather than apply the best GAAP for their situations? Could this not lead to increased rather than decreased earnings management? Will more boilerplate disclosures result? Hopefully, real improvements in financial reporting and audit quality will prove to have been made after all.

The opinions expressed here are my own and do not reflect those of the NYSSCPA, its management, or its staff.

Maria L. Murphy, CPA. Editor-in-Chief. Assistant Manager of Peer Review, mmurphy@nysscpa.org.

Search for archived articles, authors, and topics below:


Login or create a new account