Essays in the economics of education and child labor

In Chapter 1, my co-author and I explore the effect that increased access to the financial system by the poor has on child labor prevalence across countries. We use a data set provided by the Consultative Group to Assist the Poor (CGAP) on the number of clients of alternative financial institutions per country. With this data set, we capture the access that the poor has to the financial system, given the special nature of those financial institutions. Our results indicate that countries with mechanisms that facilitate access to the financial system by the poor have a lower prevalence of child labor. Measures of general access to credit (not just by poor households) have no significant impact on child labor.;In Chapter 3, I use data on intended instructional time per school year to adjust the average years of schooling for length of the school year. The literature has looked at differences in school years in different dimensions, for example through outcome variables such as test scores, or differences in resources spent per student. Controlling for the length is potentially important because instructional time varies widely across countries. I find that making years of schooling comparable in length improves the measure of human capital and helps to better explain differences in income per person across countries, especially among OECD countries.;In Chapter 2, my co-authors and I investigate how differential improvements in child mortality may affect the composition of the population and as a result the average education in a country. In particular, less educated groups face higher child modality compared to educated ones, but also benefit relatively more from health interventions. In response to these interventions the less educated increase their representation in the population, which can lower the average years of schooling. Due to this composition effect macroeconomic data may indicate a negative impact of health interventions on education where none exists at the individual level. We find that for a measles vaccination program in Bangladesh the composition effect exists, but is relatively small.