Examining the Appropriateness of Using Year-Over-Year Accident Numbers to Evaluation the Effectiveness of Product Safety Administration

抄録

Predictive indicators are crucial for budget planning and outcome-based performance measurement. In Japan’s product safety administration, the year-over-year change in the number of product accidents is often used to evaluate government agency activities. However, predicting product accidents is inherently variable. This study examines the predictive accuracy of using the previous year’s data as forecasts and compares it with the actual target value of a “3% reduction.” The results show that such forecasts have an average prediction error of approximately 8%, which exceeds the target value. This finding suggests the irrationality of past performance evaluation methods. Goal setting often includes reduction rates from predictive values, requiring careful consideration of prediction errors. However, achieving goals significantly different from prediction errors may be unrealistic. In outcome-based evaluations, goals must be distinguishable from prediction errors and attainable. If these criteria are not met, a reassessment of outcome-based evaluation methods may be necessary.