This dissertation consists of three empirical essays focused on understanding how informational and behavioral barriers influence human capital investments in developing countries. In particular, I study two key actors in education production – teachers and parents, and how their beliefs shape investments in child human capital with implications for student learning and inequality.
Chapter 1 uses a field experiment with teachers in India to investigate the role and malleability of teachers’ beliefs. In many developing countries, teachers often perceive only a weak mapping between their effort and what students learn. I conduct an experimental evaluation of a psychosocial intervention in India that targets teachers’ beliefs about perceived control – self-beliefs about one’s ability to influence outcomes. I study the extent to which this intervention affects teachers’ beliefs, their effort in class, and their students’ academic performance. I devise a novel experimental task to elicit beliefs through revealed preference, about the relationship between their teaching effort and the performance of students in their classroom. I find that the intervention induced a 14% increase in teachers’ beliefs about their ability to increase learning, as measured by the revealed preference task. Treated teachers exert greater effort at the intensive margin, scoring higher on an index of classroom effort. They also spend more time grading student work and provide more detailed feedback to students. Finally, students taught by teachers in the treatment group learn more, scoring 0.09 SD higher in the end-of-year exams. These findings suggest that teacher beliefs can serve as a powerful lever for changing teaching practice and raising learning levels in developing countries.
Chapter 2 studies the relationship between parental perceptions about children’s performance and parental investment in children’s human capital, and how this relationship evolves over the course of schooling. Using rich longitudinal data on investments, test scores, and parental assessments, I implement alternative specifications for the parental investment function that allow investment to depend on the entire history of lagged investment and inputs, account for past parental beliefs to circumvent reverse causality, and use household fixed effects to account for fixed characteristics at the household level. I find that compared to children with poor perceived performance, children with better perceived performance are up to 16 percentage points more likely to be enrolled in private as opposed to public schools, and receive up to 40% higher investment in schooling. This relationship intensifies as children progress from primary to secondary school. Results are robust across specifications, with evidence of complementarity between perceived ability and schooling. Within a household, parents’ behavior is reinforcing, with more spent on the child believed to be the better performer. These findings inform our understanding of parental investment response and intra-household allocation of human capital investment decisions.
Chapter 3 (co-authored with Daniel Chen, Sultan Mehmood, and Shaheen Naseer) uses a field experiment to evaluate the impact of providing information about teacher value-added to public school teachers in Pakistan. We show that growth mindset training shifts teachers’ beliefs about the malleability of intelligence, and reduces stereotypes against first-generation learners and students from disadvantaged backgrounds. In contrast, exposure to narrative or empirical evidence about teacher value-added did not have statistically significant effects. We document patterns of teachers’beliefs in a resource-constrained setting and show that perceived returns to effort are increasing in parental education and past performance of students, indicating that teachers view these as complementary inputs for teaching.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/qdzg-xc47 |
| Date | January 2024 |
| Creators | Kaur, Jalnidh |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0027 seconds