Research Papers:

Can Technology Facilitate Scale? Evidence from a Randomized Evaluation of High-Dosage Tutoring (with Monica Bhatt, Jon Guryan, Bhavya Mishra & Michael LaForest)

High-dosage tutoring is an effective way to improve student learning (Nickow et al., 2024; Guryan et al., 2023). Finding ways to deliver high-dosage tutoring at large scale remains a challenge. Two primary challenges to scaling are cost and staffing. One possible solution is to reduce costs by substituting some tutor time with computer-assisted learning (CAL) technology. The question is: Does doing so compromise effectiveness? This paper provides evidence from a randomized controlled trial (RCT) of approximately 4,000 students in two large school districts in 2018-19 and 2019-20. The RCT tested the effectiveness of an in-school math tutoring program where students worked in groups of four, with two students working with an in-person tutor while the other two worked on CAL, alternating every other day. The tutoring model had per-pupil costs approximately 30 percent lower than the 2-to-1 tutoring model studied in Guryan et al. (2023). We find gains in students’ math standardized test scores of 0.23 standard deviations for participating students, which are almost as large as the effect sizes of the 2-to-1 tutoring model reported in Guryan et al. (2023). These findings suggest strategic use of technology may be a way to increase the scalability of HDT.

Learn more about our work by reading NBER working paper or visiting the Education Lab project page.

Media coverage:  U.S. Bureau of Labor Statistics Monthly Labor Review; Evidence Based Policy; CommonWealth Beacon; K-12DIVE

"Unmet Expectations: The Impacts of School Construction on Female Outcomes in Rural Punjab, Pakistan” (Job Market Paper) (2 minute Video Documentary Style) (2 minute Video with Slides) (Blog) (Won Best Student Paper at AGEW 2021)

In this paper, I study the long-term and intergenerational effects of expanding educational opportunities through school construction in a low education setting with significant gender inequality (rural Punjab, Pakistan). Using administrative data on historical school construction that started in the 1960s, I exploit variation across birth cohorts and regions in the timing of school construction to build on a Difference-in-Differences approach that allows for staggered school construction in this context. I find that an additional girls’ school per 1000 children at the district level led to a 4-5 pp (20-25%) higher likelihood of girls completing primary education and increased their years of education by around 0.5 years. I do not find any statistically significant impact of the boys' school construction on their education. These findings indicate that improved access to schooling is driving the results on educational attainment since mobility restrictions, due to social norms, are only relevant for females in this setting.  I also find evidence of intergenerational impacts on educational attainment of children of mothers that are more exposed to the school construction program. However, I do not find corresponding improvements in female labor market or marriage market outcomes as a result of improved educational attainment. These findings contribute to a growing literature that finds that females in developing countries continue to face barriers in terms of later life outcomes despite improvements to their educational attainment especially in settings where social norms and the economic position of males may mediate these effects.  My findings provide valuable insights to policy makers interested in returns to girls education as well as addressing gender inequality in developing countries. 

Watch a quick summary of my JMP in 2 minutes (Documentary Style)

Human Capital Investments of Young Hispanics: The Bright Side of the Great Recession?

In this paper, I analyze the impact of the labor demand shocks induced by the Great Recession on the human capital investments of young Hispanics, a disadvantaged sub-group in United States on the basis of their education. I use a Bartik approach where I construct labor demand shocks separately by race group and ethnicity at the metropolitan area level to analyze their impact on schooling decision of young cohorts. I find that the higher responsiveness of young Hispanics, even conditional on the differential size of the shock experienced by Hispanics relative to other groups, can explain more than half of the observed convergence in High School dropout rates and college attendance rates between Hispanics and other groups following the Great Recession. My results are robust to specification checks for endogenous migration and highlight the impact shocks to opportunity cost can have on educational decisions of a historically disadvantaged group.

(Selected) Research Work in Progress:

Impact of Lady Health Workers on fertility outcomes of rural females: Evidence from Pakistan.

 “Local labor market outcomes and the slowdown in Mexican emigration to the United States” with Darren Lubotsky and Benjamin Feigenberg.

 “Determinants of non-compliance with Covid-19 regulations: the role of right wing groups?” with Hasin Yusuf.

 “Chance for All or Opportunity for Some: The case of Low Fee Private Schools in a low education setting.”