Childhood development and human capital accumulation: evidence from Indonesia

Abstract

Abstract

This thesis consists of three research papers. The three papers are independent yet all are connected by the major theme of the importance of early childhood development for the long-term accumulation of human capital later in life. The first paper is about the long-term effects of early childhood malnutrition on adults' educational and health outcomes. Using Indonesia's 1960s famine as a source of exogenous variation and employing the difference-in-differences approach, I find that children who were born in affected regions went on to complete fewer years of schooling. On average, persons who were in utero in the hunger-striken regions accumulated 0,55 fewer years of schooling. They also tend to have lower cognitive ability when they are adults. There is, however, no strong direct evidence of long-term effects of childhood malnutrition on chronic diseases. Only after comparing the offspring of hunger survivors with that of the control group, the symptoms of stunting can be identified. The children of the survivors tend to be taller than their peers, suggesting that their parents were able to survive because they are genetically strong and would have had above-average height had they not experienced stunting. In the second paper, I examine the long-term impact of early-life exposure to the air pollution that resulted from the 1997 forest fires in Indonesia on human capital accumulation. The fires are considered as the worst in Indonesian history in terms of the size of the area affected and the duration. The 1997 wildfires occurred in the second semester during the prolonged dry season in some provinces in Sumatera and Kalimantan, and blanketed the affected areas and the surroundings for 3 months or more. I find that children who were exposed to the smoke when they were in the first 3 years of life went on to experience a shortfall in accumulated years of schooling by around 0.23 years. However, I find limited evidence on health effects and other schooling dimensions. Further, I investigate and find some indications that following the wildfires average households in the hotspot provinces changed their preferences and behaviour towards fewer children, adopting birth control, and increased expenditure allocation for health and education. These adaptive responses may contribute to moderating the true detrimental, long-term effects of the smoke. In the last paper, I examine the effects of exposure to ethnic diversity during the school-age period on the human capital outcome, measured mainly as years of schooling achievement. Combining three different data sets from Indonesia in 2000 and 2010, I link children's neighbourhood-ethnic diversity and the cohorts' average educational outcome at the local level 10 years later. The diversity is measured by the Herfindahl-based index constructed at the sub-district level from the full count 2000 Population Census. In addition, to net out the effect of diversity, I control for neighbourhood characteristics that may be correlated with population diversity and density. My Ordinary Least Square (OLS) estimates tend to support the popular idea that diversity is harmful. More diverse communities are associated with lower mean years of schooling. However, when I employ instrumental variables to mitigate a potential endogeneity issue in the neighbourhood ethnic mixture, the diversity effects become significantly positive. More diverse neighbourhoods tend to provide a favourable educational environment for children, so they are able to accumulate more years of schooling. The mechanism behind this is likely through exposure to greater pools of talents in ethnically diversed communities.