Trade-Offs in Objective and Subjective Performance Evaluation: A Case Study Examining the Validity of Agency Theory Predictions

AN EXAMINATION OF THE CONTROL SYSTEM USED BY WEGMANS FOOD MARKETS, INC., TO MEASURE AND EVALUATE THE PERFORMANCE OF ITS MANAGERS LOOKS AT WHETHER THE PRINCIPLES OF AGENCY THEORY CARRY THROUGH IN PRACTICE. THE AUTHORS INTERVIEWED SEVERAL TOP-LEVEL EXECUTIVES TO ASCERTAIN THE USEFULNESS AND FUNCTIONALITY OF THE SUBJECTIVE CONTROL SYSTEM. The control function in organizations includes systems and procedures for establishing performance goals and ensuring that all organization members work toward those goals. (1) An effective control system must (a) establish decision rights (authority) for managers, (b) set performance expectations, and (c) measure outcomes in comparison with expectations. Agency Theory is an economics-based control system design that has been used extensively to model the control system choices available to firms. (2) This article summarizes predictions that arise from the agency model and presents the results of a case study designed to assess the validity of agency model projections. Given that many business school professors continue to propound the relevance and validity of Agency Theory principles, practicing managers need to know if these principles apply in the real world, provide useful guidance, and, thus, have merit. This article addresses these issues head on. CONCEPTUAL FRAMEWORK Our assessment focuses on the use of control systems for specifying and measuring the performance of managers, an area where accounting information plays a critical role. Performance goals for managers are difficult to stipulate and evaluate. In fact, options available to managers and choices made by them might not be clear to superiors, so control of managers' behavior is often based on financial and operational proxies that may be linked to incentives. (3) Therefore, it typically falls to accountants, as the information experts in organizations, to provide managerial performance measures. (4) Decision rights, performance expectations, and performance measures for responsibility area managers are summarized in Table 1. Note that none of the performance expectations can be measured objectively or reliably. Superiors cannot explicitly determine, for example, whether cost center managers have minimized the use of resources to meet output requirements and consequently can never be sure if optimal choices were made. Thus, superiors have to use indirect and more subjective proxies for measuring managerial performance, including financial performance measures such as standard costs and segment earnings and nonfinancial measures such as performance evaluations and job ratings. Given these difficulties, one might ask whether control systems are needed at the managerial level at all. Is it not sufficient to admonish managers to do the right thing? Is it not sufficient to hire hard-working managers? (5) There are two reasons for applying control systems to managers in organizations. First, there is the goal congruence problem. Lack of goal congruence arises among managers, their superiors, and other stakeholders because managers value goals of self-interest, such as pay, promotion, further career opportunities, and leisure time, while owners of for-profit organizations value increases in their financial capital, characterized by rising firm value. (6) Stakeholders in not-for-profit organizations may have more altruistic goals, but a lack of goal congruence with managers still may persist. A second reason for a managerial control system is the private information problem. Managers may have access to private information about their area of responsibility that is unavailable to their superiors. For example, they are often far more familiar with day-to-day operations and thus are better able to detect deviations from the norm, both good and bad. In fact, obtaining and implementing private, decision-relevant information is precisely the role of a subordinate manager. If the superior's decision-relevant information were as complete as the subordinate's, there would be little need to delegate authority to a cadre of middle managers. …