The work performed for the project, throughout the term of the Fellowship, consists of the following steps. First, a wide variety of data was gathered for the purpose of the analysis. This data includes (a) data on the educational performance of students, school characteristics, teacher characteristics, (b) data on various measures of conflict including physical barriers and measures of conflict-related violence, (c) data on various geographic level characteristics that might be correlated with conflict intensity. Second, this data was cleaned and harmonized to create a merged dataset. This merged dataset allows the researcher to observe student performance in all localities in the West Bank and measures of conflict and mobility restrictions in that locality. Third, using this merged dataset, various econometric models were estimated to analyze the impact of conflict on educational performance. This includes binary choice models for binary education outcomes (e.g. whether the student passed the exam) and high-dimensional fixed effect models where the fixed effects are for the locality of residence and the locality of the school. Fourth, to ensure that the estimates capture the effect of conflict, various robustness and placebo checks were conducted. This includes checking whether the sample of students who sit for exams is affected by exposure to conflict and whether students move endogenously in response to exposure to conflict. Fifth, to examine the mechanisms driving the baseline results, additional data sources were used to investigate the impact of conflict on the school learning environment and the psychological wellbeing of children.
The baseline results indicate that mobility restrictions in the form of checkpoints have adverse impacts on educational performance. Each additional checkpoint 5–10 km away from the school vicinity reduces the probability that a student passes the exam by over 2.5 percentage points (pp) and reduces the overall exam score (out of 100%) by 1.36 pp (Table 1). The results also suggest that the impact of mobility restrictions is fairly similar across genders, although the estimated impact on the probability of passing the TGE is more negative for female students compared to male students (Table 2). Examining how the results differ across subjects, the results indicate large variation across subjects. Mathematics scores are most adversely impacted, especially among female students (Table 3). Importantly, the impacts of mobility restrictions operate through a distinct channel to conflict-related violence which has been the focus of the previous literature. Evidence of the three mechanisms at play was found: First, each minute increased travel time due to delays at checkpoints reduces the probability of passing by 0.6 pp (Table 4). Second, mobility restrictions significantly increase the probability of students suffering from a lack of concentration (Table 5). Third, mobility restrictions impact the labour and capital supply at schools, potentially worsening the school learning environment (Table 6).
