The Effect of Relative Performance Evaluation on Earnings Management: A Game-Theoretic Approach

Posted: 16 Apr 2001

Abstract

Because investors and creditors often compare the financial statements of similar or competing firms when deciding how to allocate their funds, it is likely that a firm's financial well-being depends on how well it performs relative to its rivals. In this paper, we consider the problem of earnings management as a noncooperative game among several firms, in which each firm seeks a comparison advantage through its financial statement numbers. Our model indicates that firms may exaggerate their earnings in a world driven by multi-firm comparisons simply because they expect other firms to do so. Thus, very little may be needed for earnings management to emerge in the Nash equilibrium. Our results hold under the following conditions. First, investors and creditors are not able to unravel the earnings management, thus ensuring that some information asymmetry remains. Second, investors and creditors make interfirm comparisons when assessing firm value. Third, firms care about their own fundamental value as well as the market's perception about firm value. We also show that the equilibrium amount of earnings management depends on the characteristics of the earnings management technique itself and on the proportion of stockholders who are long-term investors in the firm.

Keywords: Earnings management, relative performance, Nash equilibrium

JEL Classification: G14, M41, M43

Suggested Citation

Watts, Susan G. and Bagnoli, Mark E., The Effect of Relative Performance Evaluation on Earnings Management: A Game-Theoretic Approach. Available at SSRN: https://ssrn.com/abstract=253966

Susan G. Watts (Contact Author)

Purdue University ( email )

West Lafayette, IN 47906
United States

Mark E. Bagnoli

Purdue University ( email )

Department of Accounting
West Lafayette, IN 47907-1310
United States
765-494-4484 (Phone)
765-496-1778 (Fax)

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
2,847
PlumX Metrics