The Clean Cooking Transition

Measured by people, the cooking transition has two phases. Between 1990 and 2000, gas spread quickly but population grew faster, and the number of people in low- and middle-income countries cooking with polluting fuels rose to a peak of about 3.12 billion around the turn of the century. Since then it has fallen every year, to 2.07 billion in 2023 by WHO's fuel-type estimates, while the number cooking with gas or electricity has risen from 1.73 to 4.51 billion. The clean share rose from 27% in 1990 to 36% in 2000, 56% in 2015 and 69% in 2023.

The fuels did not fall together. Coal for cooking collapsed from 680 million users in 1990 to 59 million, almost all because of China's switch to gas and electricity. Kerosene fell from 294 to 54 million, as Indonesia, India and others replaced it with LPG. Wood and other biomass kept rising until 2008 and has fallen by about 780 million since then. Charcoal rose throughout, from 109 million in 1990 to 252 million in 2023. It is cleaner-burning than wood at the point of use but still a polluting fuel, and it is made inefficiently from wood cut ever further from the cities it supplies.

LPG and natural gas carried the transition: 3.95 billion people now cook mainly with them, against 559 million with electricity. That reflects history and convenience. LPG needs no grid and heats food quickly, and cylinders can reach small towns by truck. Its limits are also clear: it is a fossil fuel, often imported, and its price moves with world markets. Electricity's share is small but growing fastest in relative terms, and as grids reach more homes and electric pressure cookers and induction hobs get cheaper, it is becoming the second lane of the transition.

Tracking the change in the number of people using each fuel between 2000 and 2023 shows three regional pathways. In South Asia, the number cooking with LPG and natural gas rose by about 910 million while the number using wood and biomass fell by 380 million and kerosene by 85 million. India drove this, rising from 22% clean cooking in 2000 to 49% in 2015 and 78% in 2023 through LPG connections for poor households, especially the Pradhan Mantri Ujjwala Yojana. In East Asia, gas rose by 800 million and electricity by 360 million while coal fell by 400 million and biomass by 370 million. China's urbanisation, gas networks and rural electrification did most of this, and electricity is now the main cooking fuel for a quarter of Chinese people. Indonesia moved from 6% to 89% clean in two decades after replacing kerosene with LPG.

The third pathway is Africa's, and it has so far been a ladder with missing rungs. Between 2000 and 2023 sub-Saharan Africa's population grew from 680 million to 1.26 billion. The number of people using clean fuels rose from about 60 to 280 million, but the number using wood and biomass rose by about 220 million and charcoal by 150 million. As people move to towns where wood must be bought, many shift to charcoal, which is lighter to transport and burns with less smoke. In the region's cities, charcoal is the main fuel for 28% of people. Clean access is 43% in cities and 8% in the countryside, a gap wider than any other region's.

A fourth pattern runs through all regions: electric cooking where power is cheap and reliable. South Africa (83% electric), Myanmar (50%), China (25%) and parts of Central Asia show that electricity can be a mass cooking fuel, but these cases rest on low tariffs, hydropower or coal-fired grids, and reliable supply. In Ethiopia, cheap hydropower gave cities a foothold for electric cooking (7% nationally). In Zambia, clean access fell from 15% to 10% as population growth outran grid reach and charcoal filled the gap. Latin America and the Middle East have completed most of an LPG-led transition, and their remaining gaps are rural and in poor countries such as Haiti, Afghanistan and Yemen.

Between 2015 and 2023 the countries that moved fastest gained between two and four percentage points of clean cooking access a year. Cambodia rose from 22% to 53%, India from 49% to 78%, Myanmar from 22% to 44%, Viet Nam from 67% to 90%, Indonesia from 69% to 89%, Nigeria from 5% to 26% and Kenya from 12% to 30%. At those rates, a country moves from minority to near-universal access within about 25 years, roughly the pace that Indonesia, Peru and Viet Nam achieved.

At the other end, 26 countries gained less than half a point a year, and together they hold 585 million people without access. Ethiopia (7%), the DR Congo (4%), Uganda (2%), Mozambique (7%), Madagascar (2%), Mali (1%) and Malawi (2%) have barely moved. Afghanistan, Yemen, Somalia and Haiti are held back by conflict and fragility, and Zambia has gone backwards. At their current pace, many of these countries would not reach near-universal access this century. They share three features: low electricity access, little or no LPG distribution infrastructure, and large rural populations that collect wood rather than buy fuel.

The arithmetic for 2030 is unforgiving. Projecting each country's 2015–2023 pace forward with its recent population growth leaves about 1.57 billion people without clean cooking in 2030, or 1.34 billion if population were held at its 2023 level. The IEA's stated-policies outlook gives a similar 1.7 billion. Universal access by 2030 would require about 375 million people a year to gain clean cooking, two and a half times the 2015–2023 rate of about 150 million a year. The simulator in section 9 shows how much later target years ease that pace.

Electricity access has run far ahead of clean cooking. Comparing the share of each country's population with electricity with the share cooking cleanly gives a lower bound on the number of people who are electrified but still cook with polluting fuels: at least 1.37 billion, or two-thirds of everyone without clean cooking. India accounts for about 310 million, China 155 million, Bangladesh 123 million, Pakistan 113 million, Nigeria 80 million and Ethiopia 62 million. Sub-Saharan Africa has about 390 million, 40% of its people without clean cooking. In Bangladesh, Sri Lanka, Laos, Nepal, Guatemala and Timor-Leste, at least 94% of people have electricity, but fewer than half cook cleanly.

Not everyone with electricity can cook with it. Supply may be unreliable, limited by a small solar home system or too expensive. Still, the economics of efficient electric cooking have changed. Research under the Modern Energy Cooking Services programme found that cooking typical meals in an electric pressure cooker costs roughly a third as much as LPG and less than a fifth as much as charcoal across its study countries, because sealed, insulated, automatically controlled pots use far less energy than open pots. In Ghana an electric pressure cooker cost about a third as much as LPG per pot of jollof rice, though it took longer. In Zambia, electric pressure cookers were the cheapest and quickest option for every dish tested.

Electric cooking has limits. In Ethiopia, cheap resistance stoves spread when tariffs were below about 5 US cents per kWh. Higher tariffs, outages and the evening peak can quickly erase the advantage. Utilities need to plan for cooking load, appliances need upfront finance, and many households will keep several stoves. Cooking with electricity is clean at the point of use, whatever the generating mix, and becomes climate-friendly as the grid decarbonises. For the 1.37 billion who are already connected, it is the fastest lane that does not require new fuel import chains.

WHO attributes about 2.9 million deaths in 2021 in these 106 countries to household air pollution from cooking with polluting fuels. They come from stroke, ischaemic heart disease, chronic lung disease, lung cancer and pneumonia. WHO's fact sheet puts the figure at 3.2 million in 2020, including 237,000 children under five. The burden is split almost equally between East Asia (920,000), South Asia (910,000) and sub-Saharan Africa (758,000), but rates differ greatly. Adjusted for age, sub-Saharan Africa loses about 121 people per 100,000 a year, South Asia 70, East Asia 38 and Latin America 16. The highest rates, above 170 per 100,000, are in the Central African Republic, Somalia, Sierra Leone, Guinea, the DR Congo, Togo, Haiti and Madagascar.

Across countries, the age-standardised death rate rises by about 1.6 deaths per 100,000 for every percentage point of the population cooking with polluting fuels (r = 0.94). That relationship partly reflects how the estimates are built, because WHO derives deaths from fuel use through exposure–response functions. It should be read as a picture of the modelled burden, not as independent evidence. The independent evidence comes from trials and cohort studies, and it carries an important caution.

In the HAPIN trial in Guatemala, India, Peru and Rwanda, which gave households free LPG stoves and fuel for 18 months, the median personal exposure of young children fell from 66 to 24 µg/m³ of PM2.5. Yet there was no significant effect on birth weight or severe infant pneumonia. Exposure stayed well above WHO's guideline probably because of neighbours' smoke, outdoor pollution and occasional traditional cooking. The lesson is not that clean cooking fails but that health gains require near-exclusive clean use across whole communities, together with cleaner outdoor air. Programmes should therefore measure sustained use and the retirement of the old stove, not stoves distributed.

ESMAP's State of Access to Modern Energy Cooking Services estimated that the lack of modern cooking costs the world about $2.4 trillion a year. Of this, $1.4 trillion comes from health, $0.8 trillion from women's lost time and productivity, and $0.2 trillion from climate damage. The estimate refers to the broader population that lacks modern cooking services, about four billion people, and it dwarfs the $150 billion a year ESMAP estimated would be needed for universal modern cooking by 2030. Most of the cost never appears in a household budget or a ministry's accounts, so it rarely appears in investment decisions.

Much of that cost is time, and collecting wood and tending smoky fires falls mostly on women and girls. Few studies measure the time saved. A systematic review found two, both in rural India: in one, households using biogas saved 1.2 hours a day collecting fuel and 0.7 hours cooking, about 29 days a year, and in the other they saved 1.5 hours a day collecting fuel. Those hours can go to paid work, schooling or rest, and fewer trips for wood mean less exposure to injury and violence. Gender benefits also depend on design. LPG refills paid in lump sums, or electric appliances bought on credit, can leave women dependent on household decision-makers. Programmes that put connections, accounts and subsidies in women's names, as Ujjwala does, give them more control over the switch.

For the climate, the IEA estimates that universal access by 2030 would cut greenhouse gas emissions by up to 1.5 Gt of CO₂-equivalent, 900 Mt of it in sub-Saharan Africa. The reductions come mostly from less unsustainable wood harvesting and lower methane and black carbon from incomplete combustion, which more than offsets the added LPG emissions. Carbon finance has tried to monetise this but has often overstated it. A 2024 study in Nature Sustainability found that a sample of cookstove offset projects was over-credited about 9.2 times, mainly because of exaggerated adoption and use, underestimated continued use of the old stove and overstated deforestation. A metered Gold Standard methodology came closest, at about 1.5 times. Credits can help close the affordability gap, but only with metered use and conservative baselines.

Each country's chart shows the shares of its people cooking mainly with each fuel from 2000 to 2023, with polluting fuels at the bottom and clean fuels on top. The pathways vary widely. Indonesia shows a near-vertical switch from kerosene to LPG after 2007. India shows LPG steadily displacing wood, with a jump after 2016. China shows coal disappearing and electricity rising. Kenya shows LPG growing from a low base in cities, Ghana shows charcoal and LPG growing side by side, and Ethiopia shows how little has changed outside a small urban electric-cooking market.

Below the chart, the country's urban and rural fuel mixes appear side by side. In most of Africa they are two different countries: urban households buy charcoal and LPG, while rural households gather wood. In much of Asia LPG has reached villages, and rural households often keep wood for some dishes, for heating water or for when a cylinder runs out. WHO counts each household's main fuel. Stacking, where clean and polluting stoves are used in the same home, means the clean shares here somewhat overstate exclusive clean use.

The figures under the chart give clean access in 2023, the 2015–2023 pace, the share of people with electricity, the number of people without clean cooking, the number electrified but cooking with polluting fuels, and the modelled death rate from household air pollution. The same countries can be compared in the sortable table in section 8 and projected forward in the simulator in section 9.

The access layer shows the clean-cooking divide. Near-universal access stretches across Latin America, the Middle East, Central Asia and much of East Asia. Most of sub-Saharan Africa is below 30%, and a second group of low-access countries runs from Afghanistan through Myanmar, Laos and Papua New Guinea to Haiti. The pace layer shows where that picture is changing. The fastest gains are in India, Cambodia, Viet Nam, Myanmar, Nigeria, Côte d'Ivoire and Kenya, while Central and East Africa barely move.

The main-fuel layer shows which fuel feeds each country's kitchens: gas across most of the world, electricity in South Africa, Myanmar, Mongolia, Kyrgyzstan and Cuba, charcoal in Haiti, Liberia and Somalia, and wood across most of the Sahel, the Horn and Central Africa. The eCooking layer shows the share of people who are electrified but cook with polluting fuels, highest in Laos, Timor-Leste, Bangladesh, Sri Lanka and the Gambia. The health layer maps WHO's age-standardised death rates, and the charcoal layer shows where urban demand presses on surrounding forests.

Hover over or tap a country for its card, which shows its 2023 fuel mix, access in 2000, 2015 and 2023, urban and rural access, electricity access, the eCooking gap, deaths and its trajectory class. Clicking a country also selects it in the country explorer above. High-income countries and countries with fewer than one million people are left blank.

The simulator starts from each country's 2023 access and grows its population at its 2015–2023 rate. It compares two futures. In the first, every country continues at its 2015–2023 pace. In the second, every country reaches universal access in the chosen target year along a straight line. The difference between the two gives the number of additional people who must switch each year and the person-years of polluting-fuel cooking that would be avoided to 2050.

The eCooking slider divides the required new access between electric cooking and LPG or other clean fuels. For context, the simulator also reports how many people in the chosen area already have electricity but cook with polluting fuels. If the electric share of new access exceeds that pool, grid or off-grid electrification must also expand. For sub-Saharan Africa, it compares the pace with the IEA's estimate that about $4 billion a year would close the region's gap.

The model is deliberately simple. It ignores stacking, assumes recent population growth continues and treats the pace of the last eight years as the counterfactual. It does not estimate deaths avoided, because health gains depend on exclusive use and outdoor air quality, as the HAPIN trial showed, and scaling WHO's deaths with fuel shares would overstate precision. Its purpose is to show the size of the effort a target implies.

The first lane is LPG at scale, and its record is the strongest. Indonesia's conversion programme from 2007 distributed more than 57 million starter kits, each with a 3 kg cylinder, first fill, burner and regulator, and replaced subsidised kerosene in about two-thirds of households within five years. By 2025, 90% of households cooked mainly with LPG. India's Ujjwala scheme has given about 103 million poor households a connection, but the average beneficiary still buys only about 4.5 refills a year (up from 3 in 2019–20), well short of full use. Its targeted subsidy has been renewed at ₹300 per 14.2 kg cylinder. The lessons are to subsidise the connection and the first refills, keep refills affordable and available in small sizes, and plan for supply security. Indonesia imports more than 80% of its LPG and India about 60%, mostly from the Middle East, so price shocks reach kitchens directly.

The second lane is electric cooking for the 1.37 billion who are already connected. It needs efficient appliances (electric pressure cookers and induction hobs) financed on the electricity bill or through pay-as-you-go, cooking-friendly tariffs and reliable supply, and utilities that plan for cooking load. Mini-grid and solar-home-system providers can add cooking where batteries and panels are sized for it. The enablers apply to both lanes. The first is finance: the IEA estimates that about $4 billion a year would close Africa's gap, against $770 million invested in 2024 and $2.2 billion pledged at the 2024 Paris summit, of which about $740 million had been disbursed by mid-2026, nearly half to LPG and 19% to Kenya. The second is carbon and results-based finance that pays only for metered, verified use. The third is an honest plan for charcoal, the fuel of a quarter of urban Africa, through efficient kilns, legal and taxed supply chains and sustainably managed woodlots during the transition.

One rule cuts across everything: count sustained, near-exclusive use. The HAPIN results, the over-crediting of offsets and the low refill rates under Ujjwala show that programmes measured by stoves distributed can succeed on paper and fail in the kitchen. Survey modules on stacking, smart meters on LPG cylinders and electric cookers, and health surveillance can show what changes. The countries that moved fastest — Indonesia, Viet Nam, Peru, India and Cambodia — combined a clear national target, a subsidised entry point, a dense distribution network and sustained political attention. With those ingredients, a 25-year transition is possible anywhere.

Coverage: 106 low- and middle-income countries (World Bank 2024 income groups) with at least one million people and WHO household energy estimates. High-income countries are excluded; nearly all have universal access. Clean access is the proportion of people relying primarily on clean fuels and technologies (SDG 7.1.2; WHO Global Health Observatory), and fuel shares are WHO's modelled proportions of people by main cooking fuel (biomass, charcoal, coal, kerosene, gas, electricity) for total, urban and rural areas, 2000–2023. The global series uses WHO's aggregate number of people by fuel, 1990–2023, for all countries in its model. People counts multiply shares by World Bank population (WDI), and urban and rural splits use the WDI urban share.

Pace is the average annual change in clean access, 2015–2023. Trajectory classes are near-universal (≥ 95%), fast (≥ 2 pp a year), steady (1–2), slow (0.5–1) and stalled (< 0.5). "95% by" extrapolates the pace linearly. The eCooking gap is the population share with electricity (WDI, 2023) minus the share cooking cleanly, a lower bound on people who are electrified but cook with polluting fuels, since some clean cooks lack electricity. Health: WHO deaths attributable to household air pollution and age-standardised death rates, 2021. The simulator projects population at its 2015–2023 growth rate and compares the 2015–2023 pace with a linear path to 100% in the target year.

Caveats: WHO's estimates are modelled from surveys and are smoothed, rounded to whole percentages and revised between releases. "Main fuel" ignores stacking, so clean shares overstate exclusive clean use. Deaths are derived from fuel use through exposure–response functions, so their correlation with fuel shares is partly by construction. The cost-of-inaction, investment and emissions figures are quoted from ESMAP and the IEA, with different scopes and years. All projections are extrapolations, not forecasts.