Introduction

Climate change mitigation technologies often require significant initial investments followed by sustained usage to generate social benefits. While market mechanisms typically select high-benefit agents first, this creates an adverse selection problem for adoption subsidies. Economists have traditionally favoured marginal-cost (usage) subsidies to align private and social benefits. However, this research demonstrates that in contexts with pervasive demand distortions—such as capital market failures and behavioural biases—upfront purchase price subsidies can be more effective. The study evaluates induction cookstove adoption in Kenya, a setting with severe credit constraints where buyers often face an annualised percentage rate (APR) of 224%.

Theory

The researchers developed a model evaluating green technology adoption under two types of distortions: allocation distortions, which only affect decision-making (such as inattention and present bias), and resource distortions, which also incur real resource costs (such as transaction costs and credit constraints). The theory identifies two channels through which distortions increase subsidy efficacy. Firstly, the selection channel: distortions move the marginal adopter toward higher private and social benefit types. Secondly, the elasticity channel: distortions can increase demand elasticity, reducing the subsidy expenditure required per marginal adoption. In the presence of these distortions, purchase price subsidies can dominate usage subsidies, particularly when energy demand is inelastic.

Study setting and experimental design

The empirical evaluation took place in Nakuru County, Kenya, involving 2,134 charcoal-cooking households. The study focused on the ECOA induction stove, which retails at US$84 for a bundle including three pieces of compatible cookware. The researchers implemented a randomised controlled trial cross-randomising purchase price subsidies (13% or 76% off), usage subsidies (0%, 25%, or 75% reductions in electricity costs for six months), and access to interest-free, four-month loans. Willingness-to-pay (WTP) was elicited using the incentive-compatible Becker, DeGroot, and Marschak mechanism. The researchers used high-frequency temperature sensors and electricity meters to accurately track energy usage and cooking duration.

Private and social impacts of electric stoves

Adopting an induction stove generates substantial private and social benefits. Temperature sensors recorded a reduction in daily charcoal cooking of 30 minutes, and self-reported monthly charcoal spending decreased by US$9.7 (a 64% reduction). While electricity spending rose by US$2.5, aggregate monthly energy expenditure fell by US$8.1 (32%). Over the stove’s estimated 2.2-year lifespan, these savings represent a return on investment of 101% per year at the market price. Socially, each stove abates approximately 2.4 tCO2e per year, primarily by displacing charcoal production and combustion.

Green subsidy efficacy

The study found that usage subsidies had a precisely estimated zero impact on both WTP and actual usage. In contrast, purchase price subsidies generated significant welfare gains. Each subsidy dollar spent on upfront costs generated US$3.4 in fuel savings and US$10 in environmental benefits, assuming a social cost of carbon (SCC) of US$120 per ton. The subsidy cost of abatement was calculated at US$12 per tCO2e, which is roughly one-tenth of the SCC. The observable welfare gain per subsidy dollar was estimated between US$10 and US$12, after factoring in manufacturing costs and the resource costs generated by distortions.

Demand distortions and subsidy efficacy

A primary finding is that demand distortions significantly increase subsidy efficacy. In the control group, average WTP was only 10% of total discounted fuel savings; providing access to an interest-free loan increased this to 19%. This distortion decreased the subsidy cost of abatement by 41%, from US$20 to US$12 per tCO2e. It operated through the two theoretical channels: credit constraints raised the marginal adopter’s positive externality by 19% and fuel savings by 30%, while shifting demand elasticity from -0.9 to -2.3, which lowered the subsidy cost per marginal adoption by 29%.

Model estimation and counterfactuals

Using the experimental variation, the researchers estimated the model to simulate outcomes in undistorted markets. In a counterfactual scenario where WTP equals private benefit (no distortions), the abatement cost rises more than tenfold to US$137 per tCO2e. The results suggest that low-income contexts with large demand distortions offer many of the most cost-effective opportunities for global climate change mitigation financing. While U.S. electric vehicle subsidies cost over US$1,200 per tCO2e abated, the Kenyan context demonstrates that addressing market failures can make green subsidies exceptionally efficient.

Conclusion

The researchers conclude that demand distortions like credit market frictions can increase the social returns to green subsidies by targeting high-impact users and increasing price sensitivity. A widespread transition to electric cooking for the 31 million African households currently using charcoal could abate more than 80 million tons of CO2e annually. Directing global mitigation financing towards these distorted markets may maximise the impact of each marginal dollar spent on climate goals.