The literature on responsibility accounting offers two broad imperatives. First, a manager is best evaluated based on what he controls. Second, responsibility follows a hierarchical arrangement: the departmental manager is responsible for the (controllable) departmental activities, the plant manager is responsible for the activities in the various departments within the plant, the division manager is responsible for the (aggregate) activities of various plants, and so on. This literature embodies a variety of theories of behavior and organization. Examples include expectancy theory (Ronen and Livingstone [1975] and Brownell and McInnes [1968]), contingency theory (Bruns and Waterhouse [1975] and Merchant [1984]), neoclassical economics (Gordon [1964]), and principal-agent modeling (Baiman [1982]). Principal-agent modeling emphasizes information asymmetries in understanding responsibility accounting. In particular, the link between controllability and informativeness (Holmstrom [1979; 1982], Baiman and Demski [1980], and Antle and Demski [1988]) has been well explicated, and the use of relative performance evaluation (Antle and Smith [1986], Demski and Sappington [1984], and Ma, Moore, and Turnbull [1988]) has received significant attention.1 Far less attention has been afforded the hierarchical aspect of responsibility accounting. This is the focus of our analysis.