通过更精准的对象选择提高监管效能：来自美国职业安全与健康管理局（OSHA）的证据

中文摘要

我们研究监管机构如何以最优方式确定检查对象。本文的案例研究对象是美国职业安全与健康管理局（OSHA）的一项随机分配部分检查的项目。平均而言，每次检查使随后五年内的严重伤害减少2.4起（9%）。我们采用新的机器学习方法估计不同对象选择规则的效果。若将检查对象设定为预期可避免伤害数量最高的企业，OSHA原本可以避免两倍之多的伤害；若将检查对象设定为预期伤害水平最高的企业，所避免的伤害数量也几乎同样多。在我们考察的十年间，任一方法都可产生近10亿美元的社会价值。

Abstract

We study how a regulator can best target inspections. Our case study is a US Occupational Safety and Health Administration (OSHA) program that randomly allocated some inspections. On average, each inspection led to 2.4 (9 percent) fewer serious injuries over the next five years. We use new machine learning methods to estimate the effects of alternative targeting rules. OSHA could have averted twice as many injuries by targeting the highest expected averted injuries and nearly as many by targeting the highest expected level of injuries. Either approach would have generated nearly $1 billion in social value over the decade we examine.