中文摘要
许多人类活动可以根据可预测的自然灾害策略性地安排时间，以减轻这些灾害的影响。我们在一个事关重大的医疗情境——医院手术时间安排——中研究空气污染冲击的缓解问题。利用新近可获得的中国某主要城市的住院手术记录，我们追踪了超过100万名患者的术后生存情况，并发现，在颗粒物污染水平较高的日子接受手术的患者，其院内死亡率显著上升。这一效应具有两个重要特征。首先，超额死亡率主要由手术当天的污染解释，而非入院前、手术前或手术后的污染暴露。其次，一个规模较小的高风险群体——接受呼吸系统或癌症手术的老年患者——承受了污染造成的大部分损害。基于这些实证发现，我们分析了一个医院手术时间安排模型。对于超过三分之一的高风险手术，在三天内存在一个污染水平更低的替代日期，将手术改期至该日期可能在生存率方面实现帕累托改进。
致谢：我们衷心感谢北京大学国家发展研究院Ling Li教授在数据获取方面给予的慷慨帮助。感谢Karen Brandon、Shengmao Cao、Tatyana Deryugina、Qing Gong、Elaine Hill、Wei Huang、Anupam Jena、Shanjun Li、Christopher Worsham、Yao Yao、Peng Zhang、匿名审稿人，以及ASHEcon会议、AERE夏季会议、南京大学、山东大学、环境与能源经济学国际线上研讨会和厦门大学的会议或研讨会参与者提出的有益意见。Huang感谢中国国家社会科学基金重点项目（编号：20ZDA075）的经费支持。Junting Chen提供了出色的研究助理支持。所有错误均由我们承担。
Abstract
Many human activities can be strategically timed around forecastable natural hazards to mute their impacts. We study air pollution shock mitigation in a high-stakes healthcare setting: hospital surgery scheduling. Using newly available inpatient surgery records from a major city in China, we track post-surgery survival for over 1 million patients, and document a significant increase of hospital mortality among those who underwent surgeries on days with high particulate matter pollution. This effect has two important features. First, pollution on the surgery day, rather than exposure prior to hospitalization, before or after the surgery, primarily explains the excess mortality. Second, a small but high-risk group – elderly patients undergoing respiratory or cancer operations – bears a majority of pollution’s damages. Based on these empirical findings, we analyze a model of hospital surgery scheduling. For over a third of the high-risk surgeries, there exists an alternative, lower-pollution day within three days such that moving the surgery may lead to a Pareto improvement in survival. 1 1 We are extremely grateful for Professor Ling Li at National School of Development, Peking University for generous help with data access. We thank Karen Brandon, Shengmao Cao, Tatyana Deryugina, Qing Gong, Elaine Hill, Wei Huang, Anupam Jena, Shanjun Li, Christopher Worsham, Yao Yao, Peng Zhang, anonymous referees, and seminar participants at the ASHEcon conference, AERE Summer Conference, Nanjing University, Shandong University, Virtual International Seminar in Environmental and Energy Economics, and Xiamen University for helpful comments. Huang acknowledges financial support from the Key Program of National Social Science Foundation of China (No. 20ZDA075). Junting Chen provided excellent research assistance. All errors are ours.