Publication:

Working Papers:

Abstract: I examine the impact of the Clean India Mission sanitation program on children’s cognition outcomes. Using survey data from the Annual Status of Education Report, and program coverage data from the Indian government, I utilize the time-varying district level roll-out of the program to examine changes in children’s math and literacy test-scores. I find positive and significant improvements in math scores. This finding is robust to two different staggered difference-in-differences strategies. Literacy score effects are mostly positive, but many of these effects are statistically similar to zero. To support parallel trends, I examine changes in differential pre-trends for an unbiased treatment estimate. Multiple pathways, through which the program can affect cognition, are recognized. To address non-random program roll-out, I test if pre-treatment district-level characteristics predict program coverage. Changes in test score measurements and the respective treatment variables lead to statistically similar causal estimates.

Link to paper 

Featured in: Ideas for India


Abstract: We evaluate the impact of a large education reform (Right to Education Act, 2009) in India on children’s health, measured as height-for-age z-scores. Using data from the India Human Development Surveys, we employ a difference-in-differences approach that exploits differential timing in the adoption of the reform across states. We find positive health effects due to the reform with larger impacts in states that adopted it earlier. The benefits do not differ across boys and girls, and low and high caste groups. We test for different mechanisms, at the child- and school-level through which the reform can affect children’s health. Our findings are robust to multiple robustness checks, including a regression discontinuity design that uses children’s birth-year as a source of variation for their potential exposure to the mandates of the Right to Education Act.


Abstract: This work evaluates the impact of the 2009 Right to Education Act on children’s health and cognition in Andhra Pradesh, using data from the Young Lives Study. We exploit the timing of the survey rounds and variations in children’s age, across two cohorts in each round, to adopt a quasi-experimental methodology for studying the relationship. Difference-in-differences (DD) and doubly robust DD strategies report positive impact of RTE on children’s health. Event study estimates support this finding and reveal statistically insignificant human capital differences between the treatment and control cohorts in the pre-treatment period. Triple-differences estimates show that the act has generated some differential impact across different sub-groups. We also provide evidence that our results are not driven by the division of Andhra Pradesh into two different states in 2014, or by the adoption of other simultaneous government policies during the study period.


Abstract: This work evaluates the medium-term association between New Mexico’s Pre-K support and county-level rates of children’s math and reading proficiency rates. We use county-level data and employ a county-fixed effects model for studying the relationship. Results show that lagged per-capita state funding in the Pre-K program is associated with statistically significant improvements in the percentage of children in grade 3 who are proficient in math and reading. We separately study the association between Pre-K funding and math and reading proficiency rates for males and females and estimate whether there are increasing, decreasing, or constant returns to an additional dollar of Pre-K funding per capita. We also show that an increase in the percentage of Hispanic populations in the county is associated with small improvements in the correlation between state Pre-K funding and children’s math and reading proficiency rates.


Abstract: I examine the effect of household debt and women’s labor force participation in India. Using data from the 2011/12 India Human Development Survey, I employ a Two Stage Least Squares (TSLS) strategy to study the relationship. The percentage of neighboring households with debt within the primary sampling unit is used as an instrument to predict changes in household debt. 2nd stage results show that household debt significantly increases the likelihood of women’s employment. This effect is particularly strong among women aged 36 to 45, those with limited or no schooling, and both high and low caste groups. I also find that household debt pushes women into agricultural and non-agricultural labor. Results are supported by several robustness checks including inverse probability weighting, biprobit estimation, and an alternative TSLS specification. Exclusion restriction is addressed by showing that the instrument does not predict changes in several household/primary sampling unit -level variables that can affect women’s employment.

Work in Progress:

Policy Brief: