Search

Chapter 1 studies the effect of reducing class-size, by hiring more tenured teachers, on students’ scholastic performance. Using a maimonides’ rule-kind of instrumental variable strategy and a longitudinal panel data from India, I find that a 10% reduction in pupil-teacher ratio (PTR) results in 0.02 improvement in value-added scores. However, a cost effectiveness calculation suggests that reduction in PTR, by hiring more regular teachers, entails cost-benefit ratio of 2.8, meaning that for every INR 100 spent on hiring a regular teacher, the student gains only INR 35 in future income. Thus, despite of positive impact on student learning, it is an economically costly program.Chapter 2 examines the long-term health and educational effects of the Bhopal Gas Disaster (BGD), one of the most serious industrial accidents in history, using geolocated data from India’s National Family Health Survey-4 (NFHS-4) and the 1999 Indian Socio-Economic Survey [NSSO-1999]. A spatial difference-in-differences strategy estimated the relative effect of being in utero near Bhopal relative to other cohorts and to those further from Bhopal. We find that men who were in utero at the time were more likely to have a disability that affected their employment 15 years later, and had higher rates of cancer and lower educational attainment over 30 years later. These results indicate social costs stemming from the BGD that extend far beyond the mortality and morbidity experienced in the immediate aftermath. In Chapter 3, I combine a novel dataset of >1 million patent applications from India with the firm level R&D expenditures to study the impact of weak Intellectual Property Rights (IPR) on innovation activities. With the announcement of TRIPS, the technology classes that were allowed to patent only the process innovations (a weak form of IPR) before, started showing uptick in innovation activities both in terms of patent counts and industry-level R&D expenditures, compared to technol

This dissertation studies the role of human capital in urbanization and labor market in China.Chapter 1 models and quantifies the importance of within-firm skill complementarity in explaining cross-city productivity gaps in China. I argue that skill complementarity is an important driver of skill concentration which augments these productivity gaps with agglomeration economies. I develop a spatial general equilibrium model that captures an economy inhabited by heterogeneous individuals who form production teams through assortative matching and sort across cities in these teams. I structurally estimate the model using firm-level census data.Through counterfactual analysis, I find that within-firm skill complementarity accounts for 18% of cross-city productivity gaps in China. I further examine the general equilibrium effects of place-based policies: subsidizing skilled individuals to reside in second-tier cities. The simulated equilibrium shows local gains from such policies at the expense of other cities, suggesting an equity-efficiency trade-off in a spatial economy.Chapter 2 estimates the income gains from migrating for jobs after graduation using survey data on college graduates. I apply propensity score matching and compare students who have similar propensity to move. I find 12-15% gains in starting salary from this geographic mobility. The effect does not vary significantly across family background and education. Further analysis on mechanisms suggests that the migration premium is mainly attributed to local agglomeration factors at the destination.Chapter 3 turns to one type of human capital and studies the impact of retaking in English test on the labor market. I draw evidence from a national English test and exploit a manipulated regression discontinuity at the passing cutoff for certificates. While there is a positive relationship between scores and wages, I find a 10% jump in starting salary after graduation for those who barely pass the test and bunching

This dissertation consists of three chapters which relate to public officials’ capacity and incentives to improve public services.Chapter 1 examines a setting in which government seeks to augment its capacity by enlisting the private sector. It uses a randomized controlled trial (RCT) to measure the effect of a Liberian school reform which outsourced management of public primary schools to private providers. It finds that outsourced public schools saw learning gains of 0.18 standard deviations in both English and mathematics on average, but the effects varied a lot by private provider. Chapter 2 turns to the question of whether improved public services yield electoral rewards for the officials responsible. It leverages the random variation in learning gains provided by Chapter 1’s randomized school reform in Liberia. On average, voters near treated schools were 3 percentage points less likely to vote for the incumbent party’s candidate than those near control schools. This negative average electoral effect, however, masks important heterogeneity. The negative electoral impact of the reform was concentrated in places where the reform reduced children’s learning. In places where the reform significantly improved test scores, it also produced electoral rewards. Chapter 3 also seeks to measure the electoral gains to public good provision, focusing on the construction of the Interstate Highway System (IHS) in the United States of America. It uses a shift-share estimator to isolate exogenous variation in the timing of IHS construction by county. It finds that a mile of IHS construction in an election year increases county-level vote share for the incumbent governor’s party by 0.6-2.2 percentage points during the period 1950-1972. These essays provide new empirical evidence of democratic accountability for public service provision, even as they illuminate directions for future research into the conditions in which democratic accountability binds.

This dissertation is composed of three unrelated chapters. Chapter 1 estimates the causal effect of the increase and decrease in ambient light relative to the clock on homicides in Mexico. It does so by using the discontinuous nature of the start and end of daylight saving time (DST) in a regression discontinuity design. It finds no effect on homicides when DST starts in Spring, but finds that transitions out of DST in Fall cause homicides to increase by nearly 7.8% which equates to almost 4 homicides per year with a yearly social cost of 145.2 million Mexican pesos. Neither Spring, nor Fall DST transitions influence homicides involving organized crime. To learn more about the long-term consequences of displacement on women, Chapter 2 examines the marriage market outcomes of forcibly displaced women. It uses data from representative surveys in 7 countries to document that women who are adolescents at the time of displacement are more likely to be married with no such pattern for displaced adolescent men. It provides more robust evidence by using the specific instance of displacement due to the partition of India in 1947. It employs difference-in-differences to find that women who were adolescents when they were displaced by partition were significantly more likely to marry earlier (in line with the cross country evidence), were less likely to continue their education, had more children overall, and do not appear to have married spouses with worse observable characteristics. Finally, chapter 3 tries to estimate the effect of a large scale anti-poverty cash transfer program in Pakistan, Benazir Income Support Program (BISP), on voter turnout and electoral choice. Program eligibility is determined by a threshold on a proxy means test (PMT) score. However, manipulation of the PMT score around the threshold makes the use of traditional regression discontinuity design invalid. In the absence of conclusive causal evidence through bounding techniques, the chapter provides b

Too often governments fail to provide access to quality public services to the poor. My work focuses on this issue and the bottlenecks that impede high-quality government provision of education in sub-Saharan African countries. Chapter 1 studies whether outsourcing public services to private entities improves service delivery in fragile states. It provides experimental evidence from the Partnership Schools for Liberia (PSL) program, which delegated management of 93 public schools to eight different private organizations. Within one academic year, outsourcing increased students scores in English and math by .18$\sigma$, relative to control schools. While the highest-performing providers generated increases in learning of \input{tables/alto_effect}\unskip$\sigma$, the lowest-performing providers had no impact on learning. Consistent with the rules of provider contracts, we find no evidence that providers engaged in student selection. However, providers were allowed to shift pupils from oversubscribed schools and underperforming teachers to other government schools. These results suggest that leveraging the private sector to improve service delivery in fragile states is promising, but they also highlight the importance of procurement rules and contracting details to aligning public and private interests.Chapter 2 studies cross-age tutoring --- in which older students tutor younger students --- as an inexpensive alternative for providing personalized instruction. Tutoring in math has a small positive effect on math test scores. The effect is concentrated among middle-ability students, suggesting that tutors are not able to help advanced learners and those lagging behind grade-level competencies.Chapter 3 studies complementarities across policies in education. While the idea that complementarities across policies can lead to increasing returns has a long tradition in economics, there is limited evidence that clearly identifies such complementarities. It presents evidence

This dissertation describes three field experiments that aim to increase productivity and human capital through incentives, training and access to capital. The first chapter presents the results of a randomized evaluation of the Turkish government's job training programs for the unemployed. The second chapter presents the results of a field experiment in India that relies on a novel design to test effort incentives against output incentives in an education setting. The third chapter presents the design for a field experiment in Uganda that increases small firms' access to networks, incentives, and training.