Management Control Using Non-Binding Budgetary Announcements

We use an experiment to investigate the efficacy of a non-binding budgetary announcement made by an owner in order to mitigate a management control problem induced by asymmetric information. The owner's announcement indicates how much funding she will provide for each possible cost report by the manager regarding an investment opportunity. The manager has private knowledge of the cost, incentive to overstate it, and the ability to do so undetected by the owner. The experiment consists of three treatments: (1) the owner fully commits to honor her announcement regarding how she will use the manager's cost report, (2) the owner makes no announcement at all, and (3) the owner makes a non-binding announcement regarding how she will use the manager's cost report. The first two treatments establish empirical benchmarks to gauge the effectiveness of the non-binding announcement. There are three main results. First, owners in the non-binding announcement treatment significantly outperform those in the no announcement treatment throughout the experiment. Second, owners appear to use the non-binding announcement as a bluff in an attempt to convince managers that they will reject a profitable project more often than they intend. This strategy appears to be particularly effective for the owners in the first half of the experiment. Third, the difference in owner welfare between the non-binding announcement and binding announcement treatments is much less than the prediction made from standard game-theoretic assumptions. The third result suggests that, to the extent that commitment is costly, an optimal control system might not employ commitment