Social Networks and Hiring: Experimental Evidence from Pakistan
(with Christina Brown)
In South Asia, three-quarters of ultra-poor households rely on casual labor, with urban workers often seeking short-term construction jobs through social ties or “labour stands”—informal spot markets where workers wait for daily hire. While spot markets like this are generally thought of as the free-market ideal, these markets appear to be rife with failures, including information asymmetries, wage rigidity, and large search costs for employees and employers. The presence of these frictions often increases dependence on social relationships. This project investigates why employers hire through social connections, the mechanisms driving these choices, and how workers adjust their investments in social capital versus productivity in response. To test these questions, we partner with a large construction firm, exogenously varying contractors’ incentives and the information available about worker productivity.
Completed Draft: here
AER registry here; blog post: here
Nudging at Scale: Combining Random and Quasi-Random Variation to Evaluate the Scale-Up of Incentives for Immunization in Pakistan
(with Subhash Chandir, Rachel Glennerster, Edward Jee, Mubarak Taighoon Shah and Danya Arif Siddiqi)
When large government programs are introduced, programmatic imperatives often prevent the ideal randomized evaluation. We demonstrate how leveraging detailed administrative data and combining random and quasi-random variation can generate a plausible measure of impact despite programmatic constraints. We study the scale-up of an incentive for an immunisation program in Sindh Province, Pakistan, leveraging a randomised rollout across seven districts and quasi-random variation to construct program inclusion probabilities. Using administrative data and weighted Callaway-Sant’Anna estimation, we calculate the probability a district is included in the program by permuting across alternative selection criteria for the included districts. Immunization rates for Pentavalent-1 and Measles-1 both rise by almost 15%. Individual-level analysis reveals that the program increased marginal child enrollment by 16%, although persistence was limited; however, enrolled children were more likely to complete their schedules.
Completed Draft: here
AER registry here
Using Previous Health Behavior to Tailor Incentives in Pakistan
(with Christina Brown, Subhash Chandir, Rachel Glennerster, Mubarak, Taighoon Shah, and Danya Arif Siddiqi)
The study leverages an at-scale incentive for immunization programs in Pakistan to understand whether it is possible to use prior behavior or self-selection to identify who will benefit from commitment devices. Households received baseline payments and were randomly assigned to additional incentives tied to strict, moderate, or no deadlines. The design tests whether tailoring incentive structures could potentially have large benefits for timely vaccination.
Progress: Baseline data collection (completed), program implementation and follow-up data collection (in progress)
AER registry here
The Effects of Management Training and Credit Access on the Scale and Quality of Low-Cost Private Schools in Kenya
(with Guthrie Gray-Lobe and Michael Kremer)
The low-cost private school sector accounts for a substantial share of primary enrollment in low- and middle-income countries and is dominated by small, family-run “mom-and-pop” schools operating a single campus. While evidence highlights the role of standardized procedures and modern management practices in improving education quality, it remains unclear whether such small-scale operators can benefit from these practices or whether scale barriers limit their effectiveness. To address these questions, this study employs a randomised evaluation that offers management training and financial access to assess the scale and quality of low-cost private schools in Nairobi and Machakos, Kenya.
Progress: Baseline data collection (completed), program implementation (in progress), and follow-up data collection planned (May 2026).
Last Mile Education Technology: An Evaluation of Building Education Foundations through Innovation & Technology
(with Guthrie Gray-Lobe, Emse Kadzamira and Todd Pugatch)
The study evaluates the impact of scaling up onebillion’s onecourse software in primary schools across Malawi. While the software’s effectiveness in improving foundational learning skills has been demonstrated in nine small-scale trials, we examine its impact on learning outcomes and cost-effectiveness at scale. This will be the first large-scale evaluation of a personalised, adaptive learning program in primary schools in Africa.
Progress: Baseline data collection (completed), program implementation (in progress), and follow-up data collection planned (May 2026).