How Do Performance Goals Influence Exploration-Exploitation Choices?

Abstract

Employees in organizations are frequently subject to performance goals such as sales or publication targets. However, often employees do not know what actions will allow them to meet these goals. To perform such tasks effectively, employees need to explore to quickly learn from experience which among the available alternatives offers the higher reward potential, so that they can concentrate subsequent efforts on exploiting it. Prior work models such explore-exploit problems as an adaptive learning process, where employees sequentially sample various options and learn from feedback. However, we currently do not know how performance goals influence this adaptive learning process. We argue that performance goals influence the adaptive learning process by modifying how feedback is perceived. Individuals subject to challenging goals are more likely to interpret feedback from poor alternatives as failures. Therefore, they quickly develop high belief strength that the inferior alternative is worse than the superior alternative, enabling them to reduce “useless exploration,” but also making them slow to adapt to environmental shocks. We test our predictions in a series of laboratory experiments and find that decision makers subject to challenging goals exploit more (relative to those with moderate goals). We also show that such an exploitation focus, while beneficial in stable environments, is detrimental in unstable ones. Our finding that challenging performance goals improve performance in learning tasks stands in contrast to prior findings that such goals inhibit performance in search tasks, an insight that warrants further study to improve our understanding of goal setting in the knowledge economy. History: This paper has been accepted for the Organization Science Special Issue on Experiments in Organizational Theory. Funding: K. Srikanth was supported by SMU Seed Funding Grant for the initial version of this paper. Supplemental Material: The e-companion is available at https://doi.org/10.1287/orsc.2019.13311 .