产品制造业补贴的效果分析：应对自我申报偏差和重复申报企业

This study analyzes the effects of subsidies for SMEs in Japan (Monozukuri Subsidy), taking into account self-reporting bias and repeat applications. We find that data reported by subsidy recipients to the funding agency are contaminated with errors (due to, e.g., incorrect digits) and that these incorrect, self-reported data significantly distort outcome values. In addition, the estimation results using a fuzzy regression discontinuity design do not robustly find significant differences in the sales and the sales per capita between recipients and non-recipients of subsidies. We also find that firms that are not accepted for the Monozukuri Subsidy in FY2015 are more likely to apply in the subsequent rounds. Neither self-reporting bias nor repeated applications have been addressed in previous analyses of the Monozukuri Subsidy. Further development is needed both theoretically and empirically in terms of reliable data collection and bias correction for repeated applications.