The Idle Margin: Why Everyday Conservation Goes Undone

Vehicle idling—leaving the engine running while stopped—is a conservation decision in which switching off saves fuel and reduces emissions. Why, then, do drivers persist in idling? Using 10-second GPS data from 6,621 private vehicles in Pakistan, I find that the average vehicle idles for 27 minutes and document misperceptions about idling. A randomised field experiment varies private information, social information, payments, and an in-car reminder. Private information reduces idling by 1.8 minutes, or 9% of the control mean, with effects persisting after treatment ends; estimated benefits are 3.7 times programme costs, about half through drivers' fuel savings. Social information and payments add no detectable reduction. By contrast, the reminder increases idling by 1.0 minute, concentrated in long spells—a pattern consistent with procrastination. Idling is a malleable behavioural margin, while interventions can backfire.

Subsidies and Debt under Nonlinear Electricity Pricing

With Robin Burgess, Tim Dobermann, Michael Greenstone, Faraz Hayat

This paper studies how households respond to electricity subsidies under nonlinear pricing. In a randomised experiment with 1,731 households in rural Pakistan, we vary which blocks of an increasing-block tariff are subsidised, generating distinct changes in effective average and marginal prices. Subsidies do not increase electricity consumption, including when they lower the marginal price of the next unit. Instead, they increase utility debt repayment, with larger effects among households entering treatment with payment stress or existing arrears. The results show that, in this setting, electricity subsidies operate primarily through household liquidity rather than energy demand.

Preaching Against Power Theft

With Robin Burgess, Tim Dobermann, Michael Greenstone, Faraz Hayat

We study whether moral and financial incentives can reduce electricity theft and non-payment in rural Pakistan. We randomise anti-theft messages delivered through Friday sermons across 42 high-loss electricity feeders and embed an electricity-bill subsidy experiment with 1,219 households. Sermons produce little detectable improvement in feeder-level losses or collections. Household subsidies reduce amounts payable and arrears, but do not meaningfully increase electricity consumption or improve collection rates. The results suggest that moral appeals alone may be insufficient where formal enforcement is weak, while financial relief primarily improves household balance sheets.

Hardening the Grid: Technology versus Enforcement

With Robin Burgess, Tim Dobermann, Michael Greenstone, Faraz Hayat

We study whether theft-resistant infrastructure or stronger enforcement is more effective at reducing electricity theft in Pakistan. Building on pilot evidence that aerial bundled cables substantially reduce losses, we partner with LESCO and the Ministry of Energy to randomise 120 high-loss transformers to theft-resistant cables, enhanced inspections, or status-quo operations. We combine transformer metering with billing and payment records to measure losses, collections, arrears, and legal connections. The experiment compares the effectiveness, persistence, and cost of technological and enforcement approaches to improving utility performance.

Technology Subsidies and Crop Residue Burning

With B. Kelsey Jack, Ben Moscona, Michael Greenstone

This project studies whether subsidising cleaner agricultural technology reduces crop residue burning. We leverage lotteries for large government subsidies for residue-management machinery in Punjab, Pakistan, and combine administrative data with ground-based and satellite measures of burning. The project also studies the measurement challenges involved in detecting short-lived, small-scale agricultural fires and whether complementary policies may be needed to translate technology adoption into reductions in burning.

Small Farms, Coordination, and Air Pollution

With Steve Cicala, Kyle Emerick

This project studies whether the environmental costs of small-scale farming arise partly from coordination failures. Using a diagnostic survey of rice farmers in Punjab, Pakistan, we study timing pressure, machinery access, residue value, and demand for coordinated baling services. We use this evidence to design a field pilot testing whether coordinating neighbouring smallholders can spread the fixed costs of residue collection and make cleaner residue management economically viable.

Industrial Pollution and Emissions Disclosure

With Michael Greenstone

This project studies whether making industrial pollution visible can improve environmental compliance. In collaboration with the Punjab Environment Protection Department, we use third-party emissions monitoring to generate plant-level ratings for industrial firms in Lahore. We test whether disclosing these ratings to regulators and civil society increases enforcement, induces firms to reduce emissions, and encourages investment in pollution-control technology.