Accounting errors are a commonly acknowledged phenomenon. A 2007 Wall Street Journal report found about 10 per cent of public companies in the US issued financial restatements to address errors, and it is likely many more errors did not get corrected. Yet despite this magnitude, there has been little empirical testing of the effect of errors on firm reporting incentives.

Research by Vivian W Fang, Allen H Huang and Wenyu Wang addresses that gap using real data that shows errors have two effects with very different impacts on incentives. On the one hand, the errors can act as a camouflage for bias, and on the other hand, they can lower the value-relevance of accounting information.

“The potential camouflage effect arises because there is accounting noise that makes fraud detection more difficult, potentially lowering the manager’s costs of manipulating earnings and facilitating opportunistic reporting. But there is an offsetting effect: as noise in the accounting process increases, earnings become less value-relevant to the market, which reduces the manager’s benefits of biasing earnings and dampens her incentive to do so,” they said.

Using misstatements from a broad sample of US firms from 1996 to 2005, the authors show that there is a “hump-shaped” relation between a firm’s probability of engaging in intentional misstatements and the percentage of its industry peers engaging in unintentional misstatements in the same quarter.

The hump shape supports the idea that camouflage and value-reducing effects are counteracting each other, although there was a tilt towards the camouflage effect for the majority of the sampled firms.

The authors also looked at the causes of errors, finding that some were due to transaction complexity or lengthy and complicated accounting rules. The finding that errors were linked to regulation ambiguity in particular was pertinent to concerns that increasingly complex accounting rules may be leading to these errors. “This suggests that accounting regulations could alter a firm’s reporting incentives through their effects on firm error rates,” they said.

Overall, the findings demonstrated that errors were not extraneous and had important implications for earnings quality.

“They have competing effects on firms’ incentives for bias,” the authors said. “In addition, each of the two effects we document carries its own implications: the value relevance-reducing effect suggests that errors affect pricing, and the camouflage effect suggests errors raise the difficulty of regulatory enforcement.”

The results hold at both the firm and industry levels. For instance, firms in industries with a higher prevalence of errors were associated with lower earnings response coefficients. “We found errors decreased the sensitivity of the stock price to unexpected earnings at earnings announcements, consistent with the value relevance-reducing effect. Errors also made it more difficult to detect intentional misstatements,” they said.