Essays in development economics, human capital and external validity

Thesis

This thesis comprises three papers under the overarching theme of external validity – a question of how the effectiveness of interventions or policies translate across contexts. I focus on interventions that aim to improve human capital, specifically health and education, in developing countries. To address the topic of external validity, we need comparable interventions and outcomes as well as identification of causal mechanisms that drive intervention effectiveness across contexts. My three papers contribute to each aspect. Two of my papers focus on constructing and comparing outcomes to measure education and human capital globally. In a third paper, I conduct a large-scale randomized controlled trial of a sex education intervention in Botswana in a third of the country. This trial builds on a prior trial showing a similar intervention reduced teenage pregnancy by 28% in Kenya ten years prior. The study in Botswana provides replicable evidence across contexts, showing reductions in pregnancy of up to 40%. The trial also deepens understanding of the underlying mechanism driving behavior by testing novel dimensions, such as the degree to which the messenger delivering the sex education message matters. I find the messenger makes-or-breaks the intervention: while near-peer educators reduce pregnancy, government teachers have a null effect and appear to induce a teenage rebellion response. This contrast is striking, suggesting that an underexplored dimension on which external validity turns – the messenger – might matter more than dimensions which traditionally get more attention, such as time and geography.

Authors: Angrist, N

• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English