Three Essays on Education Outcomes and Institutions

Abstract

This dissertation, in the standard three-essay format, covers three loosely connected topics that focus on education outcomes and the quality of a country's institutions in facilitating access to sanitation in Africa. Chapter 1 attempts to estimate peer effects on student effort. I present a structural model of friendship networks in which I introduce a student grade point average (GPA) as a positive function of the student's effort and their own characteristics. I show that my model is functionally different from the standard model as it captures heterogeneity based on whether students have friends or not. I estimate peer effects using the first wave of the National Longitudinal Study of Adolescent to Adult Health (Add Health) and by applying the generalized method of moments (GMM) approach. I find that on average, a one-point increase in the mean GPA of student's peers induces the student to increase their effort that in turn increase their own GPA by 0.856 points. I also find that the estimated endogenous peer effect coefficient is significantly larger than the estimated coefficient obtained under the standard model. Furthermore, I consider an alternative specification by controlling for network endogeneity. I find that the size of the estimated peer effect does not change much. My results are robust and provide a consistent and efficient measure of peer effects, which can inform the efficiency of network-targeted public policies. Chapter 2 examines whether expansion in institutional quality broadens access to improved sanitation in Sub-Sahara Africa. This is a published paper with two co-authors. This paper employs a dynamic panel-data model and data from 44 Sub-Sahara African countries over the period 2002-2015 to estimate the direct effect of institutional quality on access to sanitation. The estimation techniques control for potential endogeneity of regressors and country-specific effects. The results indicate that institutional quality promotes access to improved sanitation with control of corruption, regulatory quality, and voice and accountability playing the most significant roles. The results also show a dichotomy between rural and urban areas in which aspects of institutions increase access to sanitation. Specifically, in urban areas, the populace's ability to participate in selecting government and expressing freedom through associations and free media drives access to sanitation. In contrast, efficient curbing of corruption, increasing rule of law, and enhancing the capacity of governments to formulate and implement sound policies facilitate access to sanitation in rural areas. This dichotomy generates important policy implications as countries move towards achieving the Sustainable Development goal, universal access to improved sanitation.Finally, Chapter 3 estimates partial correlation of teacher quality and language of instruction on student learning deprivation. I use a unique primary school-level dataset on standardized test scores of Senegalese and Mauritanian grade 4 students and teachers (cross-sectional data). Learning deprivation is a dichotomous variable that takes the value 1 if a student reading test score falls below the minimum reading proficiency level, and 0 if otherwise. An instrumental-variable probit model controls to some extent for the endogeneity of teacher quality due to unobserved school-specific factors correlated with both teacher quality and learning deprivation. After controlling for a range of student, socioeconomic, school, district and regional related variables, I find that a decrease of one in the average teacher test score at the school level (teacher quality) is associated with an increase of the likelihood of a student's being learning deprived by 6.05 percentage points. I also show that the learning deprivation of a student who is taught in French is 98 percentage points higher than that of a student who is taught in a familiar language, (i.e., Arabic). The results suggest that policymakers in developing countries should focus on teachers' subject knowledge in teacher recruitment, training, and compensation policies. They also shed light on the importance of using a familiar language.