Using Google Trends To Detect Revenue Misreporting

It’s public and free and could help auditors

The fact that less than 5% of firms that filed financial statements with the Securities & Exchange Commission in 2021 subsequently had to issue restatements suggests overzealous massaging of financial results is the exception.

But that might be a bit optimistic. Research published in 2023 in the Review of Accounting Studies estimates that 41% of companies misrepresent their financial reports. The prospect that most financial reporting that isn’t on the up and up flies under the radar indicates auditors could use some help vetting financial statements. 

UCLA Anderson’s Siew Hong Teoh, Chinese University of Hong Kong Shenzhen’s Peng-Chia Chiu, Chinese University of Hong Kong’s Yinglei Zhang and California State University Long Beach’s Xuan Huang step up with a new metric that helps spot potentially sketchy revenue reporting — which is a major source of financial misstatements, if not outright fraud — in a paper published in Accounting, Organizations and Society.  

How Search Relates to Reported Revenue

Leveraging data from Google Trends, a publicly available real-time tracker of consumer online searches, the researchers compare the change in search volume for a firm’s products with its reported revenue.

The use of Google Trends data for this type of analysis — referred to as nowcasting — has proven to be an effective tool to gain insights for indicators such as unemployment claims and auto sales. The authors’ fraud detection approach is sublimely straightforward. Their MUP metric (shorthand for managing revenue up) is on the prowl for firms whose quarterly change in Google Trends search volume ranked in the bottom quartile in its industry while its reported revenue growth ranked in the top quartile. 

You don’t need an advanced degree in accounting or finance to grasp the head-scratching mismatch of reporting strong sales growth when your products seem to be less popular, as measured by Google search volume. 

Indeed, the researchers found that MUP firms in their study had 165% higher odds of subsequently restating (correcting) their initial reported revenue, compared with firms that didn’t set off a MUP alarm. Moreover, the MUP firms tended to have larger increases in accounts receivables and made lower allowances for bad debts, two accounting moves that are “common channels for upward earnings management.” 

Given that Google search is at heart a consumer tool, the MUP fraud detection metric is more effective when focused on firms dependent on direct consumer sales, rather than the B2B crowd. Still, given that two-thirds of the U.S. economy is consumer-driven, there seems to be plenty of use for the MUP metric.

In Search of Too-Aggressive Financial Reporting

Teoh, Chiu, Zhang and Huang created a database of nearly 1,900 publicly traded firms with products that showed up in Google searches from January 2004 to December 2020, and whose historic financial data were available through Compustat. They calculated the percentage change in a brand’s quarterly Google search volume compared with the same quarter in the previous year. They also did the same quarterly calculation for seasonally adjusted sales growth. They had more than 46,000 observations to work with. 

The average quarterly change in Google search volume is 1%, but for the worst quartile it clocks in at a 14% decrease. The average quarterly sales growth was 6%, but for the lowest quartile it was an average decline of 4%. About 5% of the researchers’ universe fell into a mismatch MUP categorization of strong reported revenue when search volume is lagging. The researchers also stacked the MUP against four other metrics past research has shown are “known fraud detectors” and found MUP remained a strong predictor of potential revenue chicanery. 

Not only is MUP based on easy access to free data, the authors stress its charm as an external metric that is blessedly devoid of management control. That makes it a potentially important new tool for auditors, who under typical auditing practice must triangulate different types of data that in one way or the other rely on management disclosure. 

Teoh, Chiu, Zhang and Huang also serve up a simplified version of their already simple data analysis (that’s a feature, not a bug) to make it even easier for auditors to screen for potential problems: A firm whose reported sales growth is up 15% or more, while their Google search volume is down 15%, “raises flags for potential revenue misstatements that auditors should investigate further,” the authors note.

That said, as they explain in the paper, the bigger win would be for regulators to create or support systems that would automate the collection of data that would make the MUP metric easy for interested parties to calculate. 

“Facilitating the detection of potential misreporting of firm fundamentals would reduce the benefits of misreporting to firm managers and therefore discourage the practice,” they write. “Sunshine may indeed be the best disinfectant for fraudulent reporting of revenues.”