Does subsidising cleaner cookstoves backfire? Lessons from Rwanda

Energy efficiency is often sold as a win-win – lower bills and lower emissions – and a ‘low hanging fruit’ for abating carbon (Gerarden et al. 2017). The rebound effect, however, remains a classic concern: when efficiency lowers the effective price of energy, people may consume more, offsetting part of the savings (Gillingham et al. 2016). While ‘micro’ rebounds (i.e. adopters use more) are well studied, market-level spillovers that reach non-adopters through price changes are less understood – this is referred to as the ‘macro’ rebound.

We study a specific yet important type of energy consumption: biomass for cooking in sub-Saharan Africa. The macro rebound matters a lot here as biomass (firewood and charcoal) supplies over 75% of energy across sub-Saharan Africa, mostly for cooking (IEA 2019). It has profound implications for forest degradation and consequently local and global environmental problems. Energy-efficient biomass cookstoves are a commonly used policy strategy to alleviate these pressures on forests. As these cookstoves spread widely, will local biomass markets respond in ways that undo environmental gains? To study this, we develop a theoretical framework generally applicable to local biomass markets, and evaluate a government intervention that provided these cookstoves in Rwanda (Munyehirwe, Ankel-Peters, Sievert, Bulte, and Fiala 2025).

Understanding potential spillovers of energy-efficient biomass cookstoves

We study local macro rebound effects – that is, when the diffusion of an efficient technology shifts village-level biomass demand, triggering adjustments in local prices or resource scarcity that, in turn, change the consumption patterns of non-adopters. In an open-access setting – where no individuals or groups hold exclusive use rights – these adjustments can be subtle and, at times, counterintuitive.

Our model features households that produce and consume crops and biomass for cooking. Energy-efficient biomass cookstoves raise the ‘conversion rate’ of biomass to heat, allowing adopters to use less fuel to prepare the same meals. The novel contribution of our model is to examine how the dissemination of these cookstoves affects equilibrium prices and forest stocks. With open-access harvesting, the biomass supply curve can bend backward: higher biomass prices increase extraction effort and over-harvesting shrinks forest stocks, raising extraction costs and ultimately reducing sustainable yields. The diffusion of cookstoves can therefore generate two distinct responses:

• Incremental local macro rebound effects: modest price declines encourage some additional use among non-adopters, clawing back part of adopters’ savings.
• Transformational local macro rebound effects: if efficiency pushes the system past a tipping point, forest stocks recover, prices fall, and – counterintuitively – harvesting rises alongside healthier forests as regeneration improves.

Evaluating the impact of cookstove subsidies in Rwanda

Empirically, we evaluate a government pilot that randomly varied subsidies for a low-cost, locally produced firewood stove (known as the Canarumwe, with a market price of around €3) across rural villages. In total, 84 villages participated in the intervention of which 63 treatment villages were assigned to one of three price points (market price, medium subsidy, high subsidy). Endline surveys took place roughly two and a half years later.

Three facts anchor this analysis:

1. Adoption responds strongly to price changes. At market price, around 5% of households used the cookstoves on the last cooking day. The medium subsidy raised this to 11%, while the high subsidy yielded 40% usage. Meaningful village-level firewood demand shifts occur primarily in the high-subsidy arm.
2. Most households collect firewood rather than buy it. For many, the ‘price’ of firewood is time spent collecting, not cash outlays.
3. Adopters use less firewood. At the household level, cookstoves substantially reduce firewood use. At the village level, those in the high-subsidy arm show lower total firewood use than controls – suggesting evidence of net savings.

Do market spillovers undo the savings?

We probe local macro rebound effects in three ways:

1. Consumption of non-adopters. Non-adopters in treated villages do not consume more firewood than comparable non-adopters in control villages.
2. Firewood scarcity or ‘time cost’. In treated villages, there is no reduction in the time it takes to collect firewood.
3. Monetary prices. Reported peak-season firewood prices are marginally lower in treated villages, though this effect is not robust across specifications.

In this setting, we find no meaningful local macro rebound effects. Where adoption is high, village-level firewood use falls, non-adopters do not increase use, and scarcity proxies do not ease.

Why might local macro rebound effects be small?

Our model helps diagnose when local macro rebound effects should be large or small. Several features of rural Rwanda limit spillovers:

• Connected village markets can dilute effects. If several villages within a settlement share a firewood market, surplus wood from a high-adoption village may flow to neighbours, stabilising scarcity and prices. In extreme circumstances, prices become exogenous: adopters use less, but non-adopters cannot access more, so no local rebound emerges. With transport costs high and charcoal consumption negligible, firewood rarely moves far; still, some settlement-level integration likely exists, diluting treatment effects and making small local macro rebound effects harder to detect.
• Collection has little marginal cost. Many households gather firewood while doing other tasks (e.g. walking to fields or school). If the marginal time cost of collection is near zero, non-adopters face little incentive to change behaviour even if firewood is more abundantly available – adopters simply burn less and others carry on as before.
• Food, not fuel, is binding. In very poor, land-scarce settings, producing (or buying) more food – not acquiring fuel – constrains number of meals. Cheaper energy does not translate into more cooking when crop production is the bottleneck. We deem this explanation very relevant in the Rwandan context.

A practical caveat: Statistical power

Given their nature, ‘macro’ rebound effects mean that even small spillovers (e.g. a 5% increase among non-adopters) can substantially erode village-level savings when adoption is modest. Detecting such changes requires large samples. In our pilot, minimum detectable effects for key ‘price’ outcomes were large (30–100%), allowing us to rule out only major local macro rebound effects. Understanding policy-relevant spillovers may require higher adoption saturation or bigger samples.

Policy takeaways: Spillovers of technology adoption

Our results point to a wide range of policy implications:

1. Energy-efficient biomass cookstoves deliver real savings. In many rural areas of sub-Saharan Africa, local macro rebound effects appear limited – strengthening the case for these programmes as part of climate and forest strategies (Jeuland et al. 2020).
2. Subsidies move adoption. Demand is price-sensitive and modest subsidies deliver meaningful results. This complements existing evidence that price support can jump-start uptake for low-cost, locally appropriate designs (Bensch and Peters 2020, Berkouwer and Dean 2022).
3. Diagnose your market. Institutions, market integration, and binding constraints (e.g. time costs of collection and land scarcity) determine whether local macro rebound effects are a risk to mitigate or lever to harness. Where charcoal markets are integrated, local price changes are even less likely unless adoption scales across the trading network.

References

Bensch, G, and J Peters (2020), “One-off subsidies and long-run adoption—Experimental evidence on improved cooking stoves in Senegal,” American Journal of Agricultural Economics 102(1): 72–90.

Berkouwer, S B, and J T Dean (2022), “Credit, attention, and externalities in the adoption of energy-efficient technologies,” American Economic Review 112(10): 3291–3330.

Gerarden, T D, R G Newell, and R N Stavins (2017), “Assessing the energy-efficiency gap,” Journal of Economic Literature 55(4): 1486–1525.

Gillingham, K, D Rapson, and G Wagner (2016), “The rebound effect and energy efficiency policy,” Review of Environmental Economics and Policy 10(1): 68–88.

International Energy Association (IEA) (2019), “World energy outlook 2019.”

Jeuland, M A, S K Pattanayak, and J Ankel-Peters (2020), “Do improved cooking stoves inevitably go up in smoke? Evidence from India and Senegal,” VoxDev.

Munyehirwe, A, J Ankel-Peters, M Sievert, E Bulte, and N Fiala (2025), “Energy efficiency and local macro rebound effects: Theory and experimental evidence from Rwanda,” The World Bank Economic Review, lhaf014.