Accounting Conservatism and Incentives Contracting

Abstract

This article aims to clarify the consequences of accounting conservatism from a stewardship (principal-agent) point of view. Kwon et al. (2001) argue that the limited liability of the agent results in a demand for conservatism. Specifically, they obtain this result in the case of choosing a threshold for a binary classification. I show that such an optimal conservative bias is not a pervasive phenomenon when other measurement issues are analyzed: the principal might be better off with fair (neutral) valuation or even with a higher of rule (liberal bias). The optimal accounting rule depends crucially on two parameters. First, on the risk aversion of the agent, with more risk aversion being in favor of more liberalism. Second, on the ranking of what the prior literature has labeled the Overall Information Content (OIC) of the accounting systems. I further show that the OIC of a conservative or aggressive bias depends on the informational content and bias of the underlying possible estimates. Overall the results suggest that, from a stewardship point of view, the choice between fair value and a possible conservative or liberal bias might be better set at the standard level on a case by case basis, depending on the specific measurement issue, rather than uniformly at a conceptual level for all standards. The paper contributes to the literature by clarifying the link between accounting conservatism and agency theory. It also has direct consequences for standard setters.