Indian Labor Regulations and the Cost of Corruption: Evidence from the Firm Size Distribution 
(with Amrit Amirapu)
Review of Economics and Statistics, March 2020.
Final draft, VoxDev article, Code and data

Working papers

Generalizing the Results from Social Experiments: Theory and Evidence from India
Conditionally accepted at the Journal of Business and Economic Statistics.

Combining Experimental and Observational Studies in Meta-Analysis: A Debiasing Approach
(with Rachael Meager), revise and resubmit at the American Economic Journal: Applied Economics.

Spatial Spillovers from High-Rise Developments: Evidence from the Mumbai Mills
(with Nick Tsivanidis).

Evaluating Ex Ante Counterfactual Predictions Using Ex Post Causal Inference
(with Cyrus Samii, Rajeev Dehejia, and Kiki Pop-Eleches).

Learning Practical Policies for Populations with Implementation Costs
(with Junlong Aaron Zhou and Cyrus Samii).

Multi-article projects in progress

Modeling Global Sanitation Interventions in Meta-Analysis
Funding: Y-RISE

We classify common interventions intended to decrease open defecation in rural parts of the developing world according to their underlying theoretical mechanisms. We build a structural model nesting all theoretical mechanisms and estimate its parameters in different experimental datasets. We evaluate the model’s performance in predicting intervention effects out of context and compare it to atheoretic statistical alternatives.

Firm Relocation as Environmental Policy: Impacts on Agglomeration and the Environment
Funding: NSF, PEDL, STEG, IGC, Weiss Fund, NYU FAI

We study a policy which randomly relocated 20,000+ firms operating in Delhi to industrial areas outside the city between 2000-2016. We study how the presence (or absence) of industrial activity affects air quality, the location choice of workers, as well as firms with linkages to the relocated firms and therefore agglomeration. Given the scarcity of air quality monitoring in India and other developing countries we develop deep learning methods using satellite imagery to create pollution measures at a fine spatial and temporal resolution.

  • Firm Presence, Environmental Quality, and Economic Activity: Evidence from Randomized Relocation
    (with Namrata Kala).
  • Inferring PM2.5 Levels using Satellite Imagery: A Deep Learning Approach
    (with Namrata Kala and Minas Sifakis).

Predicting the Impacts of Mobile Money: Experimental Site Selection and Ex-Ante Evaluation
Funding: Gates Foundation, Mastercard Impact Fund

We choose six migration corridors from thousands of candidates across South Asia in which to experimentally evaluate a program teaching migrants to send remittances via mobile money, with the goal of deriving policy recommendations for all the candidate sites. We adopt a quasi-Bayesian approach by developing a prior for the joint distribution of site level average treatment effects based on a microeconometric structural mode, and build in robustness to other sources of heterogeneity across sites. We run experiments in our selected sites and evaluate the performance of our methodology in determining site-level relative informativeness. We compare predictions based on the model to alternatives such as predictions made by past and future participants, and local and global experts.

Single-article projects in progress

Measuring Slums from Space
Funding: IGC, Weiss Fund
(with Minas Sifakis, Nicholas Swanson, and Nick Tsivanidis).

We apply frontier machine vision techniques to identify slums from high-resolution daytime satellite imagery in cities across India and Africa over the past 20 years. Our model is built for external validity and scalability, addressing the scarcity of high quality slum maps and making best use of the ones we have.

The Impacts of Refugee Influxes: Evidence from Administrative and Cellphone Data in Jordan
Funding: EBRD, TaiwanBusiness, IGC, Weiss Fund
(with Konhee Chang, Nick Tsivanidis, and Nathaniel Young).

What does a migrant influx imply for the welfare of both urban incumbents and the migrants themselves, and what policies are most effective for relieving the associated problem of congestion? Using call detail records (CDR) for the universe of cellphone transactions made on one of Jordan’s largest operator’s network from mid-2015 until the present, we document patterns of migration into Amman at high spatial and temporal frequency. We use administrative data available between the mid-2000s to mid-2010s to examine the effects of the migrant influx on city structure.