Empirical studies in health and development economics

Zichen Deng

Research output: PhD ThesisPhD-Thesis - Research and graduation internal

197 Downloads (Pure)


This thesis consists of three empirical studies that center around development, environmental, and health economics. Chapter 2 leverages newly collected individual-level hunger recall information of the China Family Panel Survey to estimate the causal effect of undernourishment on later-life health. We develop a Two-Sample Instrumental Variable (TSIV) estimator that can deal with heterogeneous samples. The new estimator includes a first step which preprocesses the data by methods such as the nearest-neighbor matching. Using the new estimator, we find a nonlinear relationship between the widely used indicator of famine intensity and personal exposure (measured by hunger recall among famine survivors). The non-linearity in famine exposure may explain the variation in the famine's effect on later life health found in previous studies, as studies explore different parts of this non-linear relation. We furthermore find that exposure to famine induced hunger early in life leads to worse health among females fifty years later. This effect is more significant than the reduced-form effect of previous studies. For males, we find no impact. In our study, the information on personal exposure helps in the justification of instruments used in this literature, as well as providing insight into the proper specification of the reduced-form regressions used in the extensive famine literature. Chapter 3 examines the effects of a massive salt iodization program on the human capital formation of school-aged children in China. To identify the salt iodizing policy's long-term benefits, we use the national salt iodizing program as a quasi-experiment and exploit geographic variation in goiter prevalence before the intervention. Our estimates show that the salt iodization policy has strong and significant effects on cognition for girls. We find robust positive effects of the program for girls. A one standard deviation decrease (12%) in the pre-intervention goiter rate is associated with math and vocabulary scores increasing by roughly 15%. We also see substantial increases in the educational attainment of females. Yet, we do not find any effects for boys. We show in a simple model of parental investment that gender preferences can explain our findings. We consider the role of gender preferences and how this may affect large-scale public programs' effectiveness. Analyses exploiting within the province, village-level variation in gender attitudes find that the gains in cognition are most significant for girls born in regions with the strongest son preferences. The policy's heterogeneous impact confirms the importance of parental gender preferences. Chapter 4 turns the research focus to the implementation of environmental regulations in developing countries. Despite ambitious environmental laws in many countries worldwide, the enforcement of these regulations is often weak. Holding government officials accountable for this lack of enforcement is, in turn, often marred by inadequate information about environmental quality. In this chapter, we study whether better environmental monitoring can solve this issue and improve the policy's effectiveness. We focus on air pollution in China and investigate the impact of a nationwide monitoring program. Using identification strategies that exploit strict assignment criteria set up by the central government, we show that an additional air pollution monitor reduced satellite-based pollution measures by 2-3% and increased enforcement of air-pollution regulations by 20%. To clarify how this pollution reduction came about, we construct a novel data set of city governments' air pollution enforcement activities from 2010 to 2017. We geo-reference enforcement activities and show that the increase in enforcement is driven by firms located close to those monitors, whereas firms faraway face fewer enforcement changes. These effects are caused by local officials that face strong incentives to reduce pollution and are stronger when there is limited scope for data manipulation.
Original languageEnglish
Awarding Institution
  • Vrije Universiteit Amsterdam
  • Lindeboom, Maarten, Supervisor
  • Portrait, France Rose Marie, Co-supervisor
Award date15 Jan 2021
Publication statusPublished - 15 Jan 2021

Cite this