Essays on Leniency Effects and Narrative Feedbacks in Subjective Performance Evaluation

Abstract

This thesis is comprised of two related but independent essays, which examine the incentives of evaluators in subjective performance evaluation. I focus on the process of subjective performance evaluation because we know little about how evaluators give rating and provide narrative feedback, despite practical and academic importance of subjective performance evaluation. The first essay examines whether supervisors leniency in subjective performance evaluation is influenced by the prior performance information of subordinates. While prior studies have used proxies for leniency based on aggregate objective performance level or median of subjective scales, I develop the proxy for leniency at the performance measure level based on prior performance level which is found to be more relevant to the current subjective performance score. I analyze archival performance evaluation data of multiple state-owned enterprises (SOEs) in Korea over multiple time periods which enables me to develop the new definition of leniency. Utilizing this data set, I empirically find that lenient rating persists over time, and that larger amount of leniency is applied to low prior performers and harshly evaluated performers in the previous period and smaller amount of leniency is applied to high prior performers and leniently evaluated performers in the previous period. The results have important implications for understanding the incentives of raters which differ with respect to their previous performance information. The second essay examines the determinants of narrative feedback amounts in subjective performance evaluation. The data of narrative feedbacks in Korean SOEs provides a unique setting to investigate the evaluators motivation for giving narrative feedback in subjective performance evaluation. I investigate the characteristics of performance measures, evaluators, and evaluatees that affect evaluators decision for the voluntary feedback about the evaluatees performance. I find that evaluators present large amount of narrative feedbacks about the highly weighted, long–tenured, and both high scored and low scored performance measures, and SOEs with large number of evaluators and large number of employees. More importantly, I divide narrative feedbacks into those of including suggestions and others, and find that the ratio of narrative suggestions to total narrative comments are associated with performance score, performance measure age and uniqueness, and evaluators experience, gender and evaluator group size. Overall, the evidence is consistent with evaluators giving substantial narrative feedback when evaluators face low information acquisition cost, high confrontation cost and high intrinsic motivation for providing comments.