Worker productivity often depends on privately owned, portable inputs that workers finance themselves. When liquidity constraints bind, investment in these inputs may be inefficiently low, depressing both worker earnings and firm profits. I study whether firm subsidies that enable acquisition of worker-owned capital can raise productivity while remaining privately profitable for the firm. I partner with a large home-services platform firm in India where workers travel to customer homes to complete service jobs. In this setting, efficient last-mile connectivity is a key input for worker productivity, making a scooter essential worker-owned capital. I implement a randomized experiment in which workers receive subsidies toward scooter loan down payments. The subsidy increases scooter adoption by 32 percentage points. Administrative data show large increases in jobs completed and earnings. Treated workers travel farther, accept more lucrative early-morning and late-evening jobs, and deliver quality improvements. The firm recovers the subsidy within 18-29 weeks through increased revenue share and reduced attrition, subsequently scaling the program to additional cities. A complementary discrete-choice experiment shows demand for this worker-owned capital is more responsive to reducing upfront payment requirements than to lowering overall prices, consistent with liquidity constraints as the primary barrier. The results demonstrate that enabling investment in worker-owned capital can be both productivity-enhancing and privately optimal for firms while benefiting workers.

World Bank Blog Post

Funding: Gender Growth and Labor Markets IZA Bonn, Center for Effective Global Action

Cash-back subsidy programs require households to pay full price for the subsidized good and be reimbursed later, thus creating liquidity hurdles. We exploit India’s LPG subsidy program, where the net-of-subsidy price is fixed while the over-the-counter price varies quasi-exogenously, to identify the causal effect of liquidity constraints. For low-asset households, a 1% increase in the over-the-counter price (with net price unchanged) reduces LPG purchases by 1.5% of the mean. Survey evidence shows substitution towards polluting solid fuels and worse child health. Transfer programs that neglect seemingly minor liquidity frictions may reduce essential consumption and under-deliver on welfare goals.

Submitted.

Abstract: Cities in developing countries are critical engines of growth but face heightened vulnerability to climate change, particularly from increasingly frequent extreme rainfall events. Rapid urban expansion has often outpaced infrastructure investment, raising concerns that extreme weather events could exacerbate existing inequalities. We study how urban workers adapt to extreme rainfall using high-frequency labor market data from a location-based platform in Delhi.  Using over four million daily-level job data, we find that earnings fall by 4 percent on rain shock days, driven by declines in labor supply rather than customer demand. Workers partially smooth income losses through intertemporal substitution but exhibit no spatial adaptation. The effects are highly unequal: workers in peripheral, poorer neighborhoods experience income losses three times larger than those in wealthier areas. Infrastructure quality, particularly drainage and metro connectivity, is positively associated with neighborhood resilience. Our findings suggest that without deliberate adaptation, climate change will reinforce existing spatial inequalities and limit the economic potential of developing cities.

Draft available on request.

Small firms in developing economies often under-adopt proven, productivity-enhancing tools. This project asks whether owners’ prior beliefs - about costs, usability, reliability, and payoffs - systematically deter experimentation and adoption, and whether belief updating can shift technology choices and firm outcomes. We study a dense cluster of garment firms in Mathura, India, where most producers still use outdated “umbrella” machines despite the widespread availability of industrial Juki machines that raise speed and quality and increase energy efficiency. We elicit detailed prior beliefs and willingness to pay for a short trial, then run a randomized evaluation that offers time-limited access to the new machine alongside a standard price offer. We track adoption, product upgrading, productivity, profits, labor reorganization, and energy use from baseline through endline. The design isolates the role of pessimistic or inaccurate priors in suppressing experimentation, and tests whether low-cost belief correction - through hands-on trials - causally increases take-up and improves firm performance. A central innovation is our mapped firm-to-firm network: we collect social network data to measure peer learning, spillovers, and the diffusion of technology within production clusters over time. 

[Endline ongoing.]

External Collaborators: Small Industries Development Bank of India (SIDBI), Foundation for MSME Clusters (FMC)

Funding: The Weiss Fund, Center for Effective Global Action, Michigan State University, SurveyCTO

AEA Registry 

Analysis stage.