Working Papers
The Consequences of Repeat Exams and Score Manipulation on Graduation Success: Evidence from Nepal.Abstract: High school graduation increases human capital and has a signaling value. Yet, completion rates conditional on enrollment remain low, especially in developing countries. In this paper, I study the effect of "repeat exams" policy, in which poorly performing students are promoted to the next grade instead of retaining them in the same grade, but are required to repeat the exams in the subjects in which they failed, on their future exam performance and on high school completion. Because this policy seems to incentivize graders to manipulate scores, I also study the direct effect of manipulating exam scores to let students pass if they score just below the passing threshold on their future performance. To investigate these questions, I employ Nepal's unique high school setting, and administrative exam-level data covering the universe of more than 600,000 students taking these exams. I show that barely failing a grade 11 exam and thus repeating this failed exam during the next grade, lowers grade 12 exam performance and high school completion. Because students have to allocate time and effort to learning for currently failed exams they might be unprepared for future exams. I also find that manipulating exam scores lowers students' future performance relative to the counterfactual in which exam scores are left unmanipulated, possibly because it sends an incorrect signal to students regarding the level of study effort required to pass. These results suggest that repeat exams as a stand-alone policy is ineffective, and graders ought to be discouraged from manipulating scores to mitigate low completion rates.
Second Revise and Resubmit Requested, the Economic Journal.
Conference Presentation: GDE 2022
Abstract: We study the resilience of community-based institutions when they are under pressure. Specifically, we investigate the functioning of forest user groups in Nepal in the aftermath of the 2015 earthquake. Disasters, such as earthquakes, can lead to pressure on these groups through their effect on the demand for forest resources as well as through possible negative effects on the enforceability of rules. Our study combines remote sensing data with detailed administrative data on forest user groups and exploits spatial variation in the intensity of the earthquake. Using a triple difference strategy, we show that the earthquake-induced increase in deforestation is significantly lower in locations with more forest user groups. Our result demonstrates that these institutions lead to more sustainable use of forests even under pressure. More generally, our findings show that decentralization and shifting governance of local resources to community-based institutions can be an effective policy in times of increased pressure on resources and local institutions.
Abstract: We study the role of discrimination and in-group biases in the allocation of public funds by community-based committees. Specifically, we investigate the effect of caste hierarchies on the amount of emergency aid given to households after the 2015 earthquake in Nepal. Local committees allocated aid immediately after the earthquake, and –officially– the amount of aid was a function of the magnitude of the housing damage. To identify discrimination and in-group favoritism in this context, we utilize rich data on house characteristics and housing damage and exploit exogenous variation in earthquake intensity. We first provide evidence for caste-based discrimination by these committees: Upper caste households received more aid than lower caste households. Second, we find in-group favoritism among upper caste households: Upper caste households received more aid if individuals from their own (upper) caste were involved in the allocation of aid. In contrast, lower caste households did not benefit from the presence of their own (lower) caste representatives in aid allocation committees. The results highlight the importance of social hierarchies in the study of favoritism and contribute to a better understanding of social structure for the implementation of development policy.
Work in Progress
Kinship-network spillovers in Nepal on socio-economic outcomes and political participation with Simon Heß and Matthias Schündeln.We exploit detailed voter registration data to construct family networks with the goal of understanding the role of networks in two domains. In one project, we combine network data with data on election outcomes, and in another, we combine it with socio-economic data and earthquake data. We have prepared the data for both projects and are currently getting the first set of results.
The role of earthquake-related disruptions on schooling outcomes with Matthias Schündeln.
In this early-stage research, we use administrative exam-level data covering all high school students in Nepal taking these exams from 2014 to 2016 and exploit the variation in the intensity of the 2015 earthquake and the timing of high-school exams, which were postponed in some parts and not in others, to distinguish the effect of earthquake-related disruptions, which may for example induce stress, from the reduced opportunities for classroom learning.
The kinship-network spillovers of being a victim of Maoist Insurgency on future political outcomes with Simon Heß.