Empirical essays on childhood human capital in Ethiopia

Abstract

This thesis contains three empirical papers on childhood human capital accumulation in Ethiopia. The first paper examines the long-term education impacts of exposure to the 1998-2000 Ethiopia-Eritrea conflict. I exploit exogenous variation on regional and birth-year intensity of the conflict, using cross-sectional school survey data. The empirical findings indicate that exposure to the conflict during early childhood decreases student achievement in mathematics and language scores a decade later (mainly for girls). In addition, exposure to the conflict increases the probability of grade repetition (for boys and girls) and school dropout (for boys only). The paper provides first estimates on the long-term effect of exposure to conflict on human capital accumulation. It further contributes to the study of gender differences with regard to exposure to conflict.

The second paper extends the analysis of the first paper by investigating the effect of conflict on childhood health and education using unique child-level panel data from Ethiopia. It also examines to what extent child health operates as a mechanism through which conflict affects childhood education outcomes. Identification is based on a difference-in-difference approach, using two points in time at which older and younger children have the same average age and controlling for observable household and child-level time-variant characteristics. The paper contributes to an empirical literature that relies predominantly on cross-sectional comparisons of child cohorts born before and after the war in war-affected and unaffected regions. The results show that war-exposed children have a one-third of a standard deviation lower height-for-age and higher incidence of stunting. In addition, exposed children are less likely to be enrolled in school, complete fewer grades (given enrollment), and are more likely to exhibit reading problems (given enrollment). Suggestive evidence indicates that the conflict reduces child education directly as well as through its effect on child health. The final paper examines whether student retention improves achievement later and to what extent the former is correlated with school dropout. The relationship between grade repetition and subsequent achievements as well as school dropout remains entirely an empirical question as theoretical predictions are inconclusive. I apply bivariate probit and endogenous treatment regression models to cross-sectional student-level survey data from Ethiopia to examine this relationship. The results indicate that student retention does not improve achievement in mathematics and evidence on the improvement in verbal test scores is relatively weak. In addition, grade repetition is highly correlated with school dropout. The findings have important policy implications because they suggest that grade repetition may be viewed as a waste of resources in the absence of higher test scores, indicating that students in Ethiopia should not be retained and that an investigation of alternative policies is needed.