中文摘要
我们利用带有地理编码的新闻数据，估计了印度高利害考试的举行时间与媒体报道的学生自杀事件之间的时间关联。在举行NEET-UG或JEE-Mains考试的月份，媒体报道的学生自杀事件数量高出17.5%，而在利害程度较低的会考前后，或在成年人中，未观察到类似模式。这种关联集中出现在全国性入学考试与会考时间重合时，提示累积考试负担的重要性。这些模式在女孩中、考试参与率较高的邦以及城市化程度较高的区更为明显。事件研究估计显示，自杀事件在临近考试时及考试期间增加，而成绩公布前后的变化较弱，估计也较不精确。我们没有发现证据表明，气旋或选举月份媒体关注度的整体变化能够解释这一模式，但无法排除考试相关的特定报道偏差。在缺乏心理健康数据的情况下，我们建立了一个替代数据库，以探讨一个政策问题并推动未来研究。
Abstract
We estimate the temporal association between high-stakes exam timing and media-reported student suicides in India using geocoded news data. Student suicides are 17.5 percent higher during months in which NEET-UG or JEE-Mains examinations are held, with no comparable pattern around lower-stakes board exams or among adults. The association is concentrated when national entrance exams coincide with board exams, suggesting the salience of cumulative exam burden. These patterns are stronger among girls, in states with higher exam participation, and in urbanised districts. Event-study estimates show suicides increasing immediately before and during exams, while movements around result announcements are weaker and less precisely estimated. We find no evidence that broad shifts in media attention during cyclone or election months account for the pattern, although exam-specific reporting bias cannot be ruled out. In the absence of mental health data, we generate an alternative database to address a policy question and spur future research.