Academic

In ?Leverage, Sanctions, and Deterrence of Accounting Fraud? Professor Velikonja proposes an interesting and novel way to structure the civil liability penalties for firms accused of committing accounting fraud.  While reading this thoroughly researched article, I wondered most about the reliability of assumptions about the causes and consequences of accounting fraud upon which Velikonja builds her argument.  I am not convinced that we know as much about accounting fraud as Velikonja?s analysis presumes.  To be fair to Velikonja, many of my concerns would also apply to recommendations made by other scholars who offer prescriptions as to the optimal liability regime to deter accounting fraud.  A more detailed consideration of two of the assumptions made by Velikonja might help to assuage my concerns. 

The first assumption I would recommend exploring further is Velikonja?s claim that ?managers control the information revealing who was involved in accounting fraud and thus can impede external investigations and sanctions.?  Velikonja supports this assertion with citations to Kaytal (2003), Brown (2004), and Buell (2007).  In general, relying on a law review article to support an important empirical claim invites skepticism, unless the cited article is specifically aimed at gathering evidence in support of the claim.  I would argue that Velikonja?s use of these citations as evidence for an important element of her argument is no exception.  My own suspicion is that accounting fraud is ?traceable? in ways that other frauds may not be.  And, indeed, Arlen and Carney hypothesize that ?because Fraud on the Market involves false statements made in a public way by company officials to a market populated by large, well-informed investors, ultimate detection of individual wrongdoers should be relatively certain? (1992, at 701).  Perhaps a more precise discussion by Velikonja regarding what kinds of hidden information would be helpful in detecting the ultimate wrongdoers in an accounting fraud would strengthen her claim. 

The second assumption Velikonja makes which I believe merits further discussion (perhaps even in a separate paper) is the assumption that the decision to commit accounting fraud is a calculated one. This is an assumption made by many of the legal scholars writing about the optimal liability regime to deter accounting fraud.  However, there are a growing number of reasons to be suspicious about the claim that accounting fraud is strongly influenced by economic considerations.  Increasingly, the evidence suggests managers ?stumble into? committing accounting fraud rather than committing these frauds with premeditation. 

One reason to suspect that economic considerations are not paramount in the decision to commit accounting fraud are findings such as those reported in Erickson et al. (2006).  Erickson et al. find no connection between the amount of contingent compensation managers receive and how likely managers are to commit fraud.  If accounting fraud were a product of planning and forethought, then we would expect to find a relationship between the amount of a manager?s contingent compensation and the propensity to commit accounting fraud.  More generally, the evidence connecting economic incentives to the propensity to act unethically is mixed, at best (see Dan Ariely?s research, for example).    

Furthermore, it appears to be unlikely that it is ever reasonable for a manager to commit accounting fraud.  Dechow et al. (2008) (cited by Velikonja) finds that firms accused by the SEC of manipulating their financial performance have had strong performance prior to the manipulations, and surmises that most manipulations are motivated by managers? desire to hide a downturn in financial performance.  Yet it is hard to imagine when the perceived benefits from issuing a fraudulent disclosure in order to delay the disclosure of such a downturn would approach the substantial costs that might be incurred if the fraud is uncovered.  Arlen has concluded upon reviewing the consequences to a firm of having the fraud discovered that committing accounting fraud is rarely in the shareholders best interest (Jennifer Arlen, Public versus Private Enforcement of Securities Fraud, at 27).  I would argue a similar conclusion holds for a manager considering the costs and benefits of committing accounting fraud.  Among the potential costs to a manager of committing accounting fraud are: the almost certain loss of employment, a substantial reduction in the value of the untoward manager?s human capital, and the possibility of ending up in jail (Karpoff et al. (2008)). 

A final example that the propensity to commit accounting fraud may depend on factors other than a manager?s personal cost-benefit analysis comes from an experiment several colleagues and I published titled ?Brandeis? Policeman.?  We structured our experiment to replicate many of the salient features of accounting fraud.  We then manipulated several different variables to see if they would reduce the propensity to commit our experimental proxy for accounting fraud.  We found that a disclosure treatment greatly reduced the extent to which participants in our study were willing to act in an unethical manner.  Yet this disclosure treatment was specifically designed not to affect the cost-benefit analysis of study participants.  Our finding is consistent with other experimental evidence that willingness to commit unlawful or unethical behavior in a context similar to committing accounting fraud is rarely a function of economic considerations.

Despite my concerns about Velikonja?s foundational assumptions, her article is well-written, thoroughly-researched and well-argued.  Moreover, I think Velikonja's idea of reducing civil sanctions on firms accused of committing accounting fraud in exchange for their cooperation with investigators may be useful.  I would encourage Velikonja, and other scholars as well, to focus more of their efforts on exploring the interaction between the law and ?the nuanced social calculus of theft? (Stephen Dubner & Steven Levitt, NY Times, June 6, 2004, M:64), rather than designing liability regimes based on more simplistic behavioral and operational assumptions.