Scott A. Emett

Compensation or Feedback: Motivating Performance in Multidimensional Tasks

Employees often perform tasks with multiple dimensions. In this study, we examine how employees’ performance on multidimensional tasks differs under different control structures. We conduct two experiments in which we manipulate the presence of compensation controls and the presence of feedback controls on multiple task dimensions. Our findings suggest that when employees are compensated on multiple dimensions they commit to multiple goals and divide their attention among those task dimensions. However, when feedback controls are implemented on one task dimension with compensation controls on another dimension, employees can improve performance on individual dimensions as well as their overall task performance. As a result, we find that employee performance on a multidimensional task can be higher when firms compensate employees on one task dimension and provide feedback on the other task dimension than when firms compensate on both task dimensions. This study highlights the benefits of complementing compensation-based controls (i.e., incentive pay) with non-compensation based controls (e.g., feedback), and provides a theoretical basis to help explain the prevalence of this approach in practice.

Controls and the Asymmetric Stickiness of Norms

This study investigates how formal control systems and the behavior of peers influence behavior in accounting settings. We manipulate formal controls and peer behavior (social norms) in a laboratory experiment, allowing us to precisely investigate the interactive effect of these two factors on behavior. We provide evidence that weak controls lead to more socially-interested behavior, while strong controls lead to more self-interested behavior. We also provide evidence that individuals conform more to social norms that conflict with the behavior that formal controls induce. Finally, we find that individuals preferentially attend and conform to the self-interested actions of peers (as opposed to the socially-interested actions of their peers), causing self-interested norms to be “stickier�? than socially-interested norms for behavior. Our results suggest that the interaction of formal controls and normative influence will lead to a gradual movement toward noncompliance with management expectations or regulatory requirements in accounting contexts.

PCAOB Guidance and Audits of Fair Values for Level 2 Investments

We report an experiment that examines auditors’ adjustment decisions and management-bias assessments related to portfolios of investment securities. We first examine whether current PCAOB guidance leads auditors to correct different amounts of error in investment portfolios that have the same aggregate error, depending on how the error is distributed across securities within portfolios. We hypothesize and find that experienced financial-services auditors correct larger amounts of overstatement for some distributions than for others, holding constant aggregate error. We also find that auditors modify adjustments from what is indicated by standards in the direction that corrects aggregate error in a portfolio, but that they do so incompletely and asymmetrically. Based on prior research in psychology, we also predict and find that auditors identify certain patterns of error as indicative of management bias, but not others. However, management-bias assessments do not affect auditors’ correction decisions, even when auditors are prompted to consider management bias. These results suggest a tendency for auditors to focus on required numerical comparisons without incorporating management-bias assessments into correction decisions. They also highlight a potential deficiency in current auditing standards that managers could exploit by strategically locating errors within particular securities within a portfolio.