（2007）Internal Controls and Collusion.pdfVIP

Internal Controls and Collusion Kirill Novoselov∗ The University of Texas at Austin February 5, 2007 ∗Department of Accounting, B6400, McCombs School of Business, 1 University Station, Austin, TX 78712-0211. Email: kirill.novoselov@. I would like to thank the members of my dissertation committee Paul Newman (Chair), Andres Almazan, Shane Dikolli, Steve Kachelmeier, and Maxwell Stinchcombe for helpful comments and suggestions. I also thank Jenny Brown, Carlos Corona, Keith Houghton, Volker Laux, Lil Mills, Neil Schreiber, Michael Williamson, and workshop participants at The University of Texas at Austin for valuable insights. All remaining errors are mine. Abstract This paper investigates the use of productive interdependency, or segregation of du- ties, as internal control in a hidden information setting with two productive agents and zero implementation costs. I show that, when the agents can collude and collu- sion is not too costly, implementing internal control reduces agency welfare even as it improves productive efficiency. When this is the case, the principal under certain conditions optimally chooses to use internal control as a threat instead of imple- menting it. I also show that lowering the accuracy of the accounting information system may increase the principal’s expected payoff. 1 Introduction Recent years have seen renewed interest in internal control as a means of improving corporate governance in publicly traded companies.1 This interest is, at least in part, fueled by complaints about the burdens imposed on companies by Section 404 of Sarbanes-Oxley Act of 2002 (SOX) that requires managers to report on, and external auditors to attest to, the adequacy of internal controls over financial re