Taichi Kimura: Evaluation, Bias, and Their Resolution

My specialization is management accounting, particularly performance evaluation. In recent years, I have been focusing on the issues surrounding subjective performance evaluation.

The basis of performance evaluation is what is known as objective performance evaluation. This involves evaluating employee performance using objective indicators such as sales, costs, and profits. The intention is to motivate employees by linking objective performance to incentives. However, several disadvantages of objective performance evaluation have been pointed out. For example, if a sales department employee's performance is evaluated solely on their sales quota achievement rate, that employee might engage in aggressive sales tactics just to meet the quota. Even if the quota is temporarily met, the reputation of a company that engages in such sales activities will likely decline.

Therefore, subjective performance evaluation has gained attention as a way to compensate for the flaws of evaluation based on objective indicators. By having supervisors subjectively judge and evaluate their subordinates' work, aspects that tend to be overlooked in evaluations based on objective indicators can be incorporated. However, this has the disadvantage of allowing bias to enter the evaluation results. Since the evaluation is determined by the evaluator's subjectivity, evaluations may consciously or unconsciously become too lenient (leniency bias) or result in similar ratings for everyone (central tendency bias).

To eliminate bias, a practice called "calibration" has been attracting attention in recent years. In this process, after the immediate supervisor performs the primary evaluation, a calibration committee composed of multiple managers compares and reviews the various evaluation results to re-examine whether each evaluation is appropriate. Calibration is conducted with an awareness of "aligning perspectives," "reconciliation," and "adjustment" of evaluations, and the results of the primary evaluation are rewritten if necessary.

It is unclear whether calibration solves the problems of subjective performance evaluation. In fact, empirical research published in recent years points out that while the introduction of calibration mitigates leniency bias, it worsens central tendency bias because extreme evaluations are adjusted. Regarding performance evaluation, we are still in the midst of trial and error, aiming for better system design.