Accounting Complexity and Misreporting: Manipulation or Mistake?

I explore the effect of accounting complexity on misreporting using a setting of revenue restatements. I measure revenue recognition complexity using a factor score based on the number of words and revenue recognition methods from the revenue recognition disclosure in the 10-K just prior to the restatement announcement. Results are consistent with revenue recognition complexity increasing the probability of revenue restatements, after controlling for other determinants of misreporting revenue. These results are significant both statistically and economically and are robust to a number of different specifications. I also test whether misreporting for complex revenue recognition firms is the result of mistakes or manipulation. My tests provide no evidence consistent with complex revenue recognition being associated with manipulating revenue. However, there is evidence that firms that restate revenue and have more complex revenue recognition are less likely to receive an AAER from the SEC and have less negative restatement announcement returns than firms with less complex revenue recognition, suggesting mistakes are more likely for more complex firms. This paper is based on my dissertation. I appreciate the guidance and advice of my dissertation committee members: Michelle Hanlon (Chair), Russell Lundholm, Ilia Dichev, and Dana Muir. The helpful comments of K. Ramesh, Jeff Wilks, Cathy Shakespeare, Lian Fen Lee, Chad Larson, Judson Caskey, Peter Demerjian and University of Michigan workshop participants are also gratefully acknowledged. The Paton Fund, the Ross School of Business, and Leabo Fellowship provided financial support.