Study IDs best ‘red flags’ auditors can use to spot financial fraud

A new analysis of the benchmarks that auditors use to identify financial statement fraud risk finds that the most commonly used benchmarks are less effective at identifying fraud than benchmarks that are less commonly used.

The paper, “Auditor Use of Benchmarks to Assess Fraud Risk: The Case for Industry Data,” is published in the Journal of Forensic Accounting Research.

“We wanted to look not only at the benchmarks that standards say auditors can use, but the benchmarks that auditors are actually using in practice, to see which of these benchmarks is most effective,” says Joe Brazel, co-author of a paper on the work and Jenkins Distinguished Professor of Accounting in North Carolina State University’s Poole College of Management.

“And our findings suggest that auditors may want to increase their use of a benchmark that compares a company’s revenue growth to the revenue growth of their industry sector,” Brazel says. “That benchmark is not one of the most commonly used, but we found that it is most indicative of fraud risk.”

At issue are benchmarks, which are tools that financial statement auditors use as indicators that fraud may have taken place. There are two organizations that set professional standards governing which benchmarks could be used when auditing public companies and privately held companies.

For this study, the researchers began by conducting an in-depth survey of 30 auditing professionals to get an idea of how often auditors were using the benchmarks put forward in standards.

The two most common benchmarks were both trend analyses that looked only at a company’s financial statements. For example, looking at a company’s revenue in the current year and seeing whether it departed significantly from its revenue the previous year. However, the survey found that auditors also used four other benchmarks to lesser degrees.

To assess the extent to which those benchmarks were correlated with actual fraud, the researchers drew on a dataset of 429 confirmed cases of fraud that were documented by the U.S. Securities and Exchange Commission between 1994 and 2014. The researchers also looked at data from the publicly traded companies that had not committed fraud between 1994 and 2014.

“Essentially, we wanted to see which of the benchmarks that auditors are using are highly correlated with companies that committed fraud,” Brazel says. “And we looked at the companies that didn’t commit fraud to make sure benchmarks that were good indicators of fraud weren’t also targeting a high number of fraud-free companies.”

The researchers found the two most commonly used benchmarks—which relied solely on a company’s financial statements—weren’t very useful.

“Neither of those benchmarks was a good indicator,” Brazel says. “What’s more, their utility fluctuated over the 20 years we looked at—sometimes they didn’t work at all.”

By far the most useful benchmark involved evaluating the difference between a company’s revenue growth and the revenue growth of its industry.

“Basically, if a company is reporting revenue that seems too good to be true, auditors should increase their fraud risk assessment and proceed with skepticism,” Brazel says.

The study also identified a related trend that upends conventional wisdom about fraud.

“It’s long been thought that companies commit fraud in order to keep up with their competitors,” Brazel says. “But our research finds that is not the case.

“Instead, we found that when revenues were down across an entire industry, some companies committed fraud in order to keep their reported revenues artificially high. In essence, their revenues declined along with everyone else’s, but they cooked their books to avoid admitting it.”

“We think this particular paper has very practical utility for auditors in the field,” Brazel says. “And we’ve incorporated guidance at the end of the paper that is the sort of thing I would have wanted when I was a practicing auditor.”

Corresponding author of the paper is Keith Jones of the University of Kansas. The paper was co-authored by Qiyang Lian of the University of Missouri-Kansas City.