Variance Analysis in a Multi-Task Agency Setting

Abstract

Variance investigation has frequently been a subject of management accounting research. Besides technical aspects of variance decomposition, incentive effects have been a main field of research. From an agency perspective, the literature has mainly focussed on the trade-off of risk sharing and incentives. Applying Holmstrom's (1979) informativeness criterion, several results on the use of variance analysis procedures have been derived. In the last decade, however, economic agency research has emphasized that misallocation of effort rather than the above-mentioned trade-off is the central issue in the provision of incentives. Starting with Holmstrom/Milgrom (1991), a rich literature has analyzed the effects of dysfunctional behaviour, i.e. the lack of alignment between the principal's objective and measured performance. This paper investigates the role of variance analysis procedures for aligning the agent's and the principal's objectives. Borrowing a model proposed by Baker (1992), we analyze a risk-neutral agency with linear contracts where the agent receives postcontract, pre-decision information on his productivity with respect to a distorted performance measure. If this signal is informative w.r.t. the agent's marginal product concerning the principal's objective, variance investigation procedures alleviate the problem of dysfunctional behaviour. The results carry over to the multiple-task agency, but interpretation of the variance investigation procedure is less apparent in this context.

JEL Classification: D81, D82, M12, M40, M46