【1155期】 6月19日计量经济学学术讲座：Seeding efficient large-scale public health interventions in diverse spatial-social networks （韩晓祎，长聘副教授，厦门大学）

【主题】Seeding efficient large-scale public health interventions in diverse spatial-social networks 

【报告人】韩晓祎（长聘副教授，厦门大学王亚南经济研究院、经济学院）

【时间】2023年6月19日星期一10:00-11:30

【地点】经济学院701会议室

【语言】英文

【摘要】The selection of target locations for large-scale public health interventions is complex when the take-up of such interventions has peer effects through social networks and health outcomes have spillover effects through spatial networks. To address this issue, we develop a threshold spatial dynamic panel data (TSDPD) model with endogenized human-virus interactions. The model features spillovers through two layers of spatial-social networks, an endogenous region intervention threshold determined by a structural break in within-region transmissibility, and endogenous cross-and within- state mobility network. We apply the model to study the initial COVID-19 vaccination rollout in the United States from February 5 to April 15, 2021. We find strong peer effects in state vaccination rates arising from the friendship network (coeff.=0.911, s.e.=0.023) and strong COVID-19 virus transmission through the mobility network (cross-state transmission coeff.=0.247, s.e.=0.086). Vaccination decreased infections mainly through reduced transmissibility upon passing a full vaccination rate threshold of 2.827% (s.e.=0.045%) for big states and 2.649% (s.e.=0.124%) for small states by population; whereas its impact on the mobility network slightly increased infections. Cumulative infections would have been 1.19 million or 24.85% higher if vaccines were not available. Targeting the six most populated US states or most spatially connected is the most clinically effective in reducing cumulative infections by more than 345,000, and targeting the six most socially influential states is the most clinically effective in increasing the national vaccination rate by 3.7 ppts to 18.73% (95% CI:18.31–19.27%). Targeting the six most socially influential states is more than 20 times as cost-effective as targeting the most populated or most spatially connected ones.