APPLICATION OF CAUSAL INFERENCE APPROACHES TO HEALTH AND ENVIRONMENTAL ISSUES

Qingwei Qiao

doi:10.7273/000007449

This dissertation explores the application of causal inference methods in addressing major health and environmental issues. Chapter 1 examines the association between healthy food accessibility and public health outcomes, emphasizing the intensive margin—the sales volume of healthy food retailers—rather than the extensive margin. Using double/debiased machine learning (DDML), the study identifies a strong association between increased sales at healthy food retailers and improved health indicators, particularly an 8.69% reduction in obesity. Interestingly, our analysis also reveals a nonlinear relationship between exposure to healthy food and health outcomes. While low-access areas benefit the most, high-access regions also experience notable improvements, particularly in mental health. These insights contribute to policies aimed at improving food environments in underserved communities. Chapter 2 explores the economic and environmental impacts of China’s COVID-19 lock-downs. Analyzing data from 18 cities between 2019 and 2022, the study finds that lock-downs lead to a 19.3% reduction in nightlights and a 20.7% decrease in NO2 emissions. The stronger decline in NO2 emissions relative to economic activity suggests that industrial output and transportation were the primary drivers. While lockdowns improved short-term environmental conditions, the economic costs often outweighed these benefits, particularly in regions prioritizing economic recovery. The findings further indicate that China remains on the upward-sloping segment of the environmental kuznets curve (EKC), where economic growth continues to contribute to environmental degradation. Chapter 3 utilizes data from a cluster randomized controlled trial conducted in the Zinder region of Niger to investigate the role of social networks in amplifying the impact of public health interventions on child nutrition and growth outcomes. The trial included over 2,000 households across 105 villages, implementing two key interventions: the Care Group (CG) model, which delivers maternal health and nutrition education, and Community-Led Total Nutrition (CLTN), which mobilizes communities to take collective action on child nutrition. Based on reported friends survey data, we identify substantial spillover effects, where households that did not directly receive the intervention still exhibited improvements in child-feeding practices and nutrition outcomes when they had social connections with treated households. Specifically, having one additional treated friend led to an increase of approximately 1.18% in minimum dietary diversity (MDD) and 2.48% in minimum meal frequency (MMF). Our findings highlight knowledge diffusion as the primary mechanism behind these effects, and heterogeneous analysis further indicates a strong association between network structure and the magnitude of these spillovers.

APPLICATION OF CAUSAL INFERENCE APPROACHES TO HEALTH AND ENVIRONMENTAL ISSUES

Files and links (1)

Abstract

Metrics

Details