Data and Code for: Does Getting Forecasts Earlier Matter? Evidence from Winter Advisories and Vehicle Crashes

Abstract: Significant investments are directed toward improving the accuracy and early availability of forecasts. However, the value of longer lead times on forecasts is unclear. Using data on winter weather advisories and vehicle crashes in the US, I show that advisories with longer lead times reduce crashes, even when they are less accurate than advisories with shorter lead times. Further, marginal benefits do not decrease with lead time. The benefits come from individual and institutional responses. When advisories arrive earlier, people visit fewer places, and snowplow crews intensify the road maintenance operations. These results have policy implications for providing effective forecasts.<br><br>This package has data and code to replicate the study.

摘要：当前已有大量资源投入到提升气象预报精准度与发布时效的工作中。然而，预报更长预见期（lead time）的价值仍不明确。本研究依托美国冬季气象预警与机动车交通事故数据开展分析，结果表明：即便长预见期预警的精准度不及短预见期预警，其仍可有效减少交通事故的发生量。进一步研究发现，预警的边际收益并不会随预见期的增加而递减。此类收益源于个体与机构的响应行为：当预警发布更早时，民众会减少外出活动，除雪作业团队也会强化道路养护作业力度。本研究结果对于构建高效的气象预报服务体系具有重要的政策参考价值。<br><br>本配套数据包包含可复现本研究的全部数据与代码。