Cooking with Bottled Gas Issues and Challenges in Developing Countries Masami Kojima Cooking with Bottled Gas Issues and Challenges in Developing Countries Masami Kojima December 2022 © 2022 December | International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and should not be attributed in any manner to the World Bank or its affili- ated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. Editor/designer: Nita Congress Cover images (top to bottom): New Africa/Shutterstock (also on p. xii); Kaiskynet Studio/Shutterstock (also on back cover) Contents Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv 8 Comparison of sizes of households abandoning versus stacking polluting fuels . . . 18 Acknowledgments . . . . . . . . . . . . . . . . . . . . . vi 9 Unit subsidy for bottled LPG in Thailand . . . . . . 21 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . vii A.1 Registered FISE beneficiaries and the number of redeemed LPG discount vouchers in Peru . . . . . 32 Executive summary . . . . . . . . . . . . . . . . . . . viii A.2 Beneficiaries of Bonogás Hogares and cash Growing access gap . . . . . . . . . . . . . . . . . . . . 1 disbursed in the Dominican Republic . . . . . . . . . 33 Alternatives to LPG for cooking . . . . . . . . 10 A.3 Beneficiaries and cash disbursed for LPG purchase in El Salvador . . . . . . . . . . . . . . . . . . . . . . 34 Barriers to capturing the full benefits of A.4 Number of DBTL beneficiaries . . . . . . . . . . . . . . . . 35 clean cooking . . . . . . . . . . . . . . . . . . . . . . . . . 15 A.5 Number of refills by PMUY beneficiaries by Lessons from government support for fiscal year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 LPG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Tables Conclusions and implications . . . . . . . . . . 29 1 Break-even price of electricity in $/kWh for substitution of LPG with electricity for cooking . 11 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2 Total monthly consumption of electricity for References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 cooking in kWh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 3 First tariff block size in kWh per month in July Figures 2014 in Sub-Saharan Africa . . . . . . . . . . . . . . . . . . . 12 1 Primary reliance on clean energy for cooking 4 Ratio of the energy charge for the second block and GDP per capita . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 to that for the first block in July 2014 in Sub- 2 Comparison of access to clean energy for Saharan Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12 cooking by urban and rural inhabitants . . . . . . . . 3 5 Maximum combined margin emission factor for 3 Primary energy for cooking . . . . . . . . . . . . . . . . . . . 5 electricity below which cooking with LPG emits 4 Percentage of people living in households citing more CO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 different primary energy sources for cooking 6 Statistics on people who live in households that and annual expenditures per capita . . . . . . . . . . . 6 used LPG and had abandoned polluting fuels . . 19 5 Difference between the top and bottom quintile 7 LPG subsidies and effective end-user prices in the percentage of the population citing paid in June 2022 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 different forms of primary energy for cooking . . 7 8 Effective prices paid in June 2022 and 6 History of LPG prices in the Arab Gulf in 2022 household income needed for cooking with U.S. dollars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 LPG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 7 Breakdown of adoption of LPG: abandonment of polluting fuels, primary use, and all users of LPG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 iii Foreword S tatistics on access to clean cooking are sober- households using LPG—such as 50 percent by 2030. ing. People primarily cooking “clean”—as This study was carried out to identify the conditions defined by the World Health Organization in under which governments should promote LPG as terms of reduced harmful emissions and including an option for attaining universal access to clean clean fuels and technology such as electricity, nat- household energy and how to do so. ural gas, liquefied petroleum gas (LPG), alcohol, and solar cookers—have increased sizably as a percent- The findings confirm those of earlier studies: LPG age of total population, but not in absolute numbers. prices remain high and beyond the reach of many From 2010 to 2020, the percentage of people whose households in developing countries, despite rising primary means of cooking was clean increased from incomes. Virtually no developing country govern- 49 to 69  percent, but faster population growth in ment can overcome the affordability challenge by lower-income countries led to a much less dramatic subsidizing LPG—a strategy that will prove to be fis- corresponding decline in the number of people with- cally unsustainable, even if the subsidy excludes the out access to clean cooking: from 3.0 billion to only better-off. And unlike electricity, which has regis- 2.4 billion. The percentage of Sub-Saharan Africa’s tered much greater progress in access because there rural population with access to clean cooking was are few alternatives to it, there are plenty of cheaper 3 percent in 2010 and just 5 percent by 2020. At this alternatives to LPG, even if they are polluting and rate, it is highly unlikely that the world will meet the inconvenient. But even in situations where electric- 2030 Sustainable Development Goal of universal pri- ity is unreliable and LPG is the cleanest cooking the mary reliance on clean fuels and technology. better-off can afford, many still use solid fuels or ker- osene. It is worth examining the market conditions The greatest barrier to adoption of clean cooking that discourage the better-off from using LPG, such energy is financial. All clean fuels and technology as long delays in cylinder refills. Further, the under- require considerably more cash outlays than unpro- filling of LPG bottles, which raises the effective prices cessed biomass. Low consumption means that paid by consumers, has plagued many markets. Inter- clean energy suppliers cannot take advantage of ventions to stop this practice are especially helpful. economies of scale, leading to low supply and high costs—perpetuating low consumption and making For those who find LPG unaffordable, advanced clean options unaffordable and often unavailable. combustion pellet stoves, which can slash emis- sions when operated correctly, may be a better Since the 1980s, numerous governments have subsi- option. Proper operation will be costlier, because dized LPG to make it more affordable. More recently, pellets must be appropriately sized and purchased, some governments have set targets for the share of while advanced stoves capable of cooking for large iv Foreword v families are much more expensive than tradi- switch to clean energy, governments will be able to tional stoves. As with LPG, economies of scale are create a virtuous cycle of falling prices and rising needed to drive down costs. Because the markets consumption, moving their countries closer to uni- for advanced combustion stoves are still very small, versal access. some initial government support may be helpful. There is one role all governments can play to pro- Demetrios Papathanasiou mote clean household energy without any downside Global Director risks: monitoring and enforcing sensible regulations Energy and Extractives Global Practice and standards to ensure competition. As healthy The World Bank competition drives down costs and more consumers Acknowledgments T his report summarizes the findings and rec- l Emmanuel Skoufias (Angola) ommendations of a study program assessing l Victor Sulla (Botswana, Eswatini, Lesotho, Namibia, the role of liquefied petroleum gas in the con- and South Africa) text of contributing to the attainment of universal l Zakaria Koncobo (Burkina Faso) access to clean household fuel under the United l Martin Mba (Cameroon) Nations’ Sustainable Development Goal (SDG) 7. The l Franck Adoho (Côte d’Ivoire) report is based largely on four background papers: l Tom Bundervoet (Ethiopia) “Subsidizing Bottled Gas: Approaches and Effects l Nadia Belghith (Gabon) on Household Use,” “Primary Household Energy for l Moritz Meyer (The Gambia and Mauritania) Cooking and Heating in 52 Developing Economies,” l Tomomi Tanaka (Ghana) “Comparison of Bottled Gas and Advanced l Utz Pape (Kenya) Combustion Pellet Stoves,” and “Household Use of l Ayago Wambile (Liberia) Bottled Gas for Cooking: Evidence from Sub-Saharan l Habtamu Fuje (Malawi) Africa.” l Marco Ranzani (Mauritius) l Aly Sanoh (Niger) The study has benefited from comments provided l Aibek Uulu (Nigeria) by Robert Bacon, Joeri de Wit, Koffi Ekouevi, Ugo Gen- l Alvin Ndip (Sudan) tilini, Bryan Koo, Sheoli Pargal, Alisha Pinto, Alief Rezza, l Wendy Karamba (Tanzania) Dana Rysankova, Yadviga Semikolenova, and Yabei l Carolina Mejia-Mantilla (Uganda) Zhang, all of the World Bank; Anit Mukherjee of the l Manohar Sharma (Zambia) Center for Global Development; and Harold Anneg- arn of the North-West University, South Africa. Robert Financial and technical support by the Energy Sector Bacon provided a critical review of econometric anal- Management Assistance Program (ESMAP) is grate- ysis in the literature cited; Xin Zhou carried out an fully acknowledged. Through the World Bank Group, analysis of household survey data; and the follow- ESMAP works to accelerate the energy transition ing World Bank staff members provided household required to achieve SDG 7 to ensure access to afford- survey data and aggregated consumption for each able, reliable, sustainable, and modern energy for all. country: vi Abbreviations CO2 carbon dioxide GDP gross domestic product CO2e carbon dioxide equivalent kg kilogram DBTL Direct Benefit Transfer for LPG km kilometer ESMAP Energy Sector Management Assistance kWh kilowatt-hour Program LPG liquefied petroleum gas FEPC Fondo para la Estabilización de Precios de los Combustibles Derivados del Petróleo PMUY Pradhan Mantri Ujjwala Yojana (Petroleum Fuel Price Stabilization Fund) SDG Sustainable Development Goal FISE Fondo de Inclusión Social Energético (Energy Social Inclusion Fund) WHO World Health Organization All dollar amounts are U.S. dollars unless otherwise indicated. vii Executive summary T he richer half of the world cooks almost charcoal being the most commonly observed such exclusively with electricity and gas. An anal- example. ysis of primary cooking energy use patterns in 47 developing economies across the world shows Arguably the greatest challenge to cooking with that either electricity or gas dominates clean cook- LPG is its relatively high price compared to income, ing, but not both; and it is rare to see an even split exacerbated by high price volatility and currency between the two. Polluting cooking fuels are mainly depreciation. Since 1995, the free-on-board prices wood, charcoal, and kerosene. In countries with an of LPG have varied by nearly an order of magnitude; extensive network of natural gas pipelines in urban expressed in 2022 U.S. dollars, the prices ranged areas, households that cook with gas shift from liq- from $0.18 per kilogram (kg) in August 1998 to $1.50 uefied petroleum gas (LPG) to natural gas as income per kg in March 2012. The end users pay much more rises, signaling that the better-off live in urban because the costs of transportation, storage, bottling, areas. Absent natural gas pipelines, richer house- and distribution can be considerable. The net-of-tax holds choose LPG in about a dozen economies and costs downstream of refineries and gas processing electricity in another dozen; charcoal or kerosene is plants in the first half of 2022 varied by more than a favored by the rich in lower-income countries. factor of five, from less than $0.20 per kg in Thailand to more than $1.00 per kg in South Africa. Irrespective National household expenditure surveys in of global LPG prices, the high end of the downstream Sub-Saharan Africa—which lags all other regions in costs makes LPG even more unaffordable. Further, share of population with access to clean household there are large economies of scale in LPG supply, energy—show that clean cooking energy comprises making it costly to supply low-consumption areas, predominantly LPG and electricity, with electricity which tend to be rural or peri-urban with many used mainly in southern Africa. Even in urban areas, lesser-off households. more than half the population uses charcoal as the primary cooking fuel in some low-income countries. The available evidence suggests that LPG cannot be The vast majority of the rural population cites fire- a fuel of the poor in many countries, and lower-price wood as the primary cooking fuel in every country alternatives are needed. Advanced combustion except Mauritius and South Africa. Differences in pri- stoves burning densified wood pellets can slash mary energy use patterns between the rich and the emissions of harmful pollutants if operated properly poor in urban and rural areas show that, in some and can be more affordable. One important differ- countries, fuels abandoned by the rich in urban ence from LPG is that pellets are likely to be sourced areas may be those sought by the rich in rural areas, from local biomass, avoiding exposure to global com- modity price volatility. The emissions performance viii Executive summary ix of pellet stoves, however, can be compromised by To protect health and local air, it is not sufficient for polluting start-up materials and practices, the het- households to adopt clean energy. They must also erogeneity of biomass (size, moisture content, type), abandon polluting fuels and practices. Abandon- excessive emissions during fuel reloading and the ment is a function of affordability first and foremost, burn-out phase, and failing components such as followed by ready availability of clean options. To batteries. Greater efforts are needed to overcome prepare large dishes over many hours—sometimes the identified problems so these stoves will be more with vigorous stirring, which requires a stable cook- widely used and have consistently low emissions in stove—it is much cheaper and more expedient to the field. use traditional stoves. These benefits make it dif- ficult for households to abandon “stove stacking,” Over the long run, concerns about climate change whereby they continue to use polluting fuels and may lead the entire world to shift to only electric- stoves even as they adopt clean fuels and technol- ity for cooking combined with decarbonization of ogy as their primary sources of energy for cooking. power generation. Switching to electricity would also enhance the overall welfare if doing so is no more Using data from national household expendi- expensive and electricity is reliable. The break-even living in ture surveys in Sub-Saharan Africa, the rich ­ price below which cooking with electricity becomes relatively small households were found—not surpris- cheaper than LPG depends on stove efficiency and ingly—far more likely to abandon polluting fuels and the price of LPG. The threshold price increases with to cook exclusively with LPG and electricity. Among increasing LPG price and decreasing LPG stove effi- those living in households using LPG as the primary ciency, and could range from $0.10 to $0.30 per cooking fuel, half or more had abandoned polluting kilowatt-hour. Cooking exclusively with electric- cooking energy in 6 out of 12 countries in which a ity could increase monthly consumption by about sizable share of the population comprised primary 100 kilowatt-hours or more, raising monthly elec- LPG users. Within each per capita expenditure quin- tricity bills markedly and more than proportionately tile, households that had abandoned polluting fuels because most developing countries have electricity were consistently smaller than those that contin- tariffs that rise with consumption. ued to stack stoves (mainly charcoal or kerosene). However, LPG-using but stove-stacking households The impact of fuel combustion on global warming is in the top expenditure quintile were on average complex. In comparing electricity and LPG, there is a smaller than those in the fourth quintile that had grid emission factor above which use of LPG would abandoned polluting fuels—challenging the hypoth- emit less carbon dioxide (CO2). Depending on rela- esis that the smaller the household, the cleaner. tive stove efficiencies, there may be about 140–170 economies out of 232 examined worldwide in which In all cases but one, the share of total expenditures cooking with LPG would emit less CO2 than with spent on LPG was smaller than the 5 percent limit electricity. Even when LPG substitutes for firewood, suggested for affordability in the Energy Sector Man- a potentially renewable fuel, CO2-equivalent emis- agement Assistance Program’s Multi-Tier Framework sions may decrease, in part because combustion of for access—that is, cooking energy is affordable if firewood burned in most stoves releases much more 5 percent or less of household income is spent on black carbon—which has a global warming potential the fuel and the stove. These findings among the two to nearly three orders of magnitude higher than top 40 percent of households point to reasons for that of CO2—than combustion of LPG, which emits stove-stacking that seem to go beyond the question essentially no black carbon. x Cooking with Bottled Gas: Issues and Challenges in Developing Countries of affordability, such as poor LPG market conditions they continued to provide targeted cash transfers. and personal preferences. As a result, the effective average prices paid in June 2022 varied from a low of $0.18 per kg in El Salvador— To help households shift to clean cooking, many with both a universal price subsidy and targeted governments have provided support for household conditional cash transfers—to a high of $2.30 per use of LPG. Historically, the most common form of kg in South Africa with no subsidies, or a difference support has been a universal price subsidy. Such a of an order of magnitude. The history of LPG pric- subsidy scheme is regressive because the rich con- ing in the countries studied point to the difficulties sume far more LPG than the poor, takes government of ending price subsidies once started and the high resources away from essential services such as risk of a return to universal price subsidies, despite health and education, and provides powerful finan- their well-documented and widely acknowledged cial incentives for illegal diversion. shortcomings. In response, several governments over the last On balance, cumulative evidence from analyses car- decade have replaced universal price subsidies ried out since oil prices began rising markedly in with cash transfers, most of them conditional on 2004 shows that global LPG prices have spiked suf- LPG purchase. Targeted cash transfers have helped ficiently often to put LPG beyond the reach of the slash diversion, black marketing, and subsidy bur- poor unless downstream costs can be reduced dens. Targeting has been enabled by advances in to those seen in low-cost markets and global LPG mobile and other information technologies, and in prices remain relatively low. turn have enabled the LPG market move closer to market-based pricing, thereby reducing economic For the foregoing reasons, most poor households in distortions. Some governments have subsidized the low- and lower-middle-income countries are unlikely initial adoption of LPG by those previously not using to use LPG unless it is heavily subsidized; among the it by helping to pay for the cylinder deposit fee, LPG rural poor, LPG may need to be virtually free for sus- stove purchase, or both. tained regular use. With the poor having many other pressing needs and being severely cash constrained, Unsurprisingly, evidence points to price subsidies— both LPG and electricity compete with free biomass whether at the point of retail or offered in the form for cooking. Focus group discussions and surveys of cash transfers—boosting overall household use of show that clean cooking ranks lower in spending pri- LPG. However, while start-up subsidies covering the ority than food, clean water, and electricity. There is initial costs of LPG adoption can help poor house- a thin slice of middle class in every country that may holds, without these price subsidies being continued, be persuaded to switch to LPG as their primary cook- many beneficiaries have been found to use LPG ing fuel if a subsidy is provided for the start-up cost, sparingly or even discontinue its use later because but the start-up subsidy for the poor is much less they find LPG unaffordable. helpful for the purpose of promoting sustained reg- ular use of LPG without continuing and large price Despite the subsidy reform policy intended to pave subsidies. the way for market-based pricing of LPG, rising inter- national oil and gas prices in 2021 and 2022, against These findings and observations are not intended a backdrop of rapidly rising food and other prices, to suggest that governments and international saw a wave of policy reversals as many governments institutions should not pursue LPG as a household reverted back to universal price subsidies even as fuel. Universal access metrics do not distinguish Executive summary xi between the rich and the poor, and anything that cannot switch to LPG, if the better-off stop using shifts households—including those that are rich or biomass altogether, it will free up more renewable middle class—to clean household energy will help biomass for the poor to use in efficient—ultimately progress toward universal access. If the rich cannot advanced combustion—stoves. be persuaded to adopt LPG because of market dys- functions and other reasons, there is essentially no The study findings suggest governments may wish chance of the nonrich doing so. to focus more on how to shift households to LPG to create a critical mass of consumers and economies In this context, the first priority is to improve market of scale that can drive down costs and prices—start- conditions so respectable firms are not exiting the ing with the better-off, if they are continuing to use market and the rich do not face high transaction biomass in traditional stoves—than on a target for costs of using LPG. This means strictly enforcing a LPG penetration, especially for rural households, or ban against underfilling (thereby lowering effective financial assistance. The regulatory actions pro- prices paid by consumers) and cylinder safety reg- posed can help lower prices, increase safety, are ulations (although enhanced safety will increase cost-effective for facilitation of LPG adoption and costs); facilitating hospitality arrangements and use, and carry virtually no risk of adverse unintended third-party access to import terminals and storage consequences. tanks to allow different LPG suppliers to access large, costly infrastructure, thereby reducing duplication It takes a long time to develop a market in which of infrastructure and lowering the barrier to entry; safety and other rules are enforced across all sell- and creating healthy and fair competition to drive ers. Taking regulatory, monitoring, and enforcement down prices charged and increase the quality of ser- steps to create such a market is arguably the most vice (notably that for cylinder refills). Even if the poor important and central role of every government. Growing access gap clean (WHO 2014). Of these options, the first three are by far the most widely used, and are in fact virtu- ally the only sources of energy used by households About one-third of the global population, some in high-income countries. 2.4 billion people, still lacked access to clean cooking fuels and technology as of 2020 (IEA et al. 2022), the Electricity is safer and cleaner than gas in terms of most recent year for which data are available. Many emissions at the point of cooking. Until recently, thus continue to collect firewood, walking farther and however, it was felt to be no match for the versa- farther away to find free biomass. Women, children, tility and accuracy of temperature control provided and men in some rural areas spend multiple precious by gas for cooking. Gas stoves respond immediately hours each week collecting biomass—time that could to temperature-setting changes, giving precise con- otherwise be devoted to productive activities, to trol; whereas electric stoves respond more slowly, caring for families, to attending school, or to studying. especially when the temperature is being lowered or In addition to wasting time, poor access to clean fuels the heat is being turned off. Gas, but not electricity, takes a toll on health. The World Health Organization enables charring and flambéing of food. (WHO) estimates that more than 3.8 million people die every year from illnesses caused by household But induction cooking has changed the paradigm. An use of polluting energy (WHO 2022). induction cooktop uses electric currents to heat pots and pans through magnetic induction and adjusts temperature quickly, achieves a more uniform dis- Universal access to clean tribution of heat than any other means of cooking, energy focuses on pollutant keeps the kitchen cooler than electric or gas stoves, emissions and has extremely high efficiency. Induction cook- stoves are generally more expensive and require Recognizing the fundamental importance of access induction-compatible cookware, possibly requir- to clean household energy, the United Nations’ ing new purchase. Cooking with electricity (whether 2030 Agenda for Sustainable Development (UN electric or induction cookstoves) would also require 2015) includes “access to affordable, reliable, sustain- a reliable supply, which is lacking in many devel- able, and modern energy services” as Sustainable oping countries that are prone to frequent power Development Goal (SDG) 7. One indicator for mea- shortages. suring achievement of this goal is the proportion of the population with primary reliance on clean fuels and technology for cooking, heating, and lighting. Of Regional differences are these three access activities, lighting with electricity diverging rather than has seen the most rapid progress; conversely, heat- ing has seen the slowest (Kojima 2021b). narrowing At the current rate of progress—which has been To track progress in access to clean cooking—the slowed further by the COVID-19 pandemic—the 2030 focus of this report—the WHO defines “clean” in SDG target will be missed. Global access to clean terms of fuels and technologies with minimal fine cooking energy has been rising steadily, but not suf- particulate and carbon monoxide emissions. It thus ficiently, and unevenly. The data tracked since 2000 considers electricity, natural gas, liquefied petroleum show that the pace of increase in access in South gas (LPG), biogas, alcohol, and solar cookers to be Asia—the region with the second lowest access 1 2 Cooking with Bottled Gas: Issues and Challenges in Developing Countries rate—has nearly tripled and outstripped comparable scale in supplying electricity and gas, and costs fall progress in the rest of the world; by contrast, though, markedly with rising consumption and concentra- the rate of increase in Sub-Saharan Africa—which tion of consumers. It is thus almost always cheaper has the lowest rate of access—has been very slow, to supply urban consumption centers than rural with not even one-fifth of the population achieving ones, and the highest costs are incurred in supplying access in 2019 (figure 1). sparsely populated areas with low overall consump- tion. Rural population densities are 39 and 309 The differential progress between the two regions is persons per square kilometer (km) in Sub-Saharan in part due to faster income growth in South Asia. Africa and South Asia,2 respectively, making the cost In 2000, valued at purchasing power parity, gross of supply of clean energy potentially much lower in domestic product (GDP) per capita in Sub-Saharan South Asia. This difference is reflected in urban and Africa was marginally higher than in South Asia; by rural access rates, shown as solid and dotted lines, 2020, it was nearly two-fifths lower (figure 1). At respectively, in figure 2. In Sub-Saharan Africa, the market exchange rates, GDP per capita in Sub-Saharan rural access rate has barely moved in the past two Africa in 2000 was 58 percent higher than in South decades. By contrast, rural residents in South Asia Asia, but 9 percent lower by 2020. Another factor began to rely much more on clean energy for cook- accounting for the difference is population density: ing over the last decade, almost catching up with the everything else being equal, the higher the concen- global rural average by 2020. South Asia has outper- tration of consumers, the lower the cost of supply; formed Sub-Saharan Africa in urban areas as well. South Asia has a higher population density than Sub-Saharan Africa. Rising income does not Figure 1 disguises the urban-rural divide. Arguably the most important determinant of whether to use always lead to adoption of clean energy for cooking is its cost relative to income, clean energy and for this reason the “richer half” of the world— For those cooking with clean energy, the first deci- people in higher-income countries—can afford to sion is whether to cook with electricity or gas. If a cook exclusively with electricity and gas. In all coun- household has chosen to cook with gas, natural gas tries, the rich are found more in urban areas than is always preferable to LPG as it is cheaper and more rural;1 in developing countries, this distribution of convenient. Once a household is connected to a wealth results in greater access to clean house- natural gas pipeline, natural gas is always available, hold energy for urban households than their rural whereas an LPG cylinder eventually runs out and counterparts. must be refilled. Most cities in high-income coun- tries have natural gas pipelines connected to homes. Further amplifying the effect of income disparity But rural households in these countries are rarely, is the fact that there are enormous economies of if ever, connected. Absent biogas, rural households around the world wishing to cook with gas must 1 In a study of 24 national household expenditure surveys use LPG. For this reason, cooking gas is synonymous in Sub-Saharan Africa in which 22 surveys enabled com- with LPG for virtually all rural households—and even putation of per capita expenditures in urban and rural areas, the ratio of rural to urban expenditures averaged 0.52—meaning urban expenditures were, on average, double that of rural—ranging from 0.27 in Zambia to 0.77 2 World Bank staff calculations based on land areas and in Sudan (Kojima and Zhou 2022, table A1.1). population available in World Development Indicators. Growing access gap 3 Figure 1  Primary reliance on clean energy for cooking and GDP per capita 80 18,000 70 16,000 14,000 Population with access (%) 60 GDP per capita ($) World 12,000 50 10,000 40 8,000 30 South Asia 6,000 20 Sub-Saharan Africa 4,000 10 2,000 0 0 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 Source: World Bank staff calculations based on data available at World Health Organization Household Energy Database and World Development Indicators for GDP per capita. Note: GDP per capita is at purchasing power parity in constant 2017 international dollars. South Asia is as defined by the World Bank at https://www.worldbank.org/en/region/sar. Figure 2  Comparison of access to clean energy for cooking by urban and rural inhabitants 100 90 80 World urban 70 Population (%) 60 South Asia urban 50 World rural 40 30 Sub-Saharan urban 20 10 South Asia rural Sub-Saharan rural 0 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 Source: World Bank staff calculations based on data available at World Health Organization Household Energy Database. 4 Cooking with Bottled Gas: Issues and Challenges in Developing Countries many urban households, including all urban house- electricity used mainly in South Africa, its immediate holds in Sub-Saharan Africa. neighbors, and Zambia. Even in urban areas, more than half the population used charcoal as the pri- An analysis of primary cooking energy use patterns, mary cooking fuel in Liberia, Malawi, Tanzania, and clean and polluting, in 47 developing economies Uganda; the vast majority of the rural popula- from all regions of the world (figure 3a) shows that tion cited firewood as the primary cooking fuel in either electricity or gas typically dominates clean every country except South Africa. Figure 4a shows cooking, but not both; and it is rare to see an even the results for the countries in the region for which split between the two. Polluting fuels are mainly household expenditure surveys from recent years wood, charcoal, and kerosene. How households including data on LPG are available. Firewood dom- change energy use patterns as income rises can be inated in rural areas even in three countries with seen by looking at the differences in cooking energy the highest rural per capita expenditures:3 Gabon, patterns between the top and bottom wealth quin- Namibia, and Botswana (figure 4b). The largest tiles (figure 3b). shares of clean energy in rural areas were in South Africa, Gabon, and Sudan, in that order. l In economies with an extensive network of nat- ural gas pipelines in urban areas, the shift is from Differences in primary energy use patterns between LPG to natural gas—as in Belarus, Tunisia, Algeria, the rich and the poor in urban and rural areas show Argentina, Georgia, and the Kyrgyz Republic—sig- that in some countries fuels abandoned by the rich naling that the better-off live in urban areas. in urban areas may be those sought by the rich in l In Serbia, North Macedonia, Montenegro, Mongo- rural areas. In Angola, Côte d’Ivoire, Ghana, Kenya, lia, Kosovo, and Zimbabwe, households switched Mauritania, Sudan, and Zambia, urban house- overwhelmingly from solid biomass to electric- holds abandoned charcoal, while rural households ity for cooking as income rose, signaling historical adopted it as income rose (figure 5). By contrast, in availability of affordable and reliable electricity. Burkina Faso, Ethiopia, The Gambia, Liberia, Malawi, l In Turks and Caicos Islands, and to a lesser extent Niger, Tanzania, and Uganda, some households in in Kazakhstan, electricity substituted for LPG with both urban and rural areas adopted charcoal with rising income. rising income. A much larger proportion of the top l Richer households chose LPG in about a dozen quintile than the bottom relied on clean primary economies, whereas charcoal or kerosene was energy in urban Angola, Botswana, Cameroon, Côte favored by the rich at the low end of the income d’Ivoire, Eswatini, Lesotho, Namibia, and Zambia, and spectrum; in these countries—Togo, the Lao in rural Botswana and Eswatini. These differences People’s Democratic Republic, the Democratic arise from the income gap between urban and rural Republic of Congo, The Gambia, Guinea-Bissau, areas, the availability of different fuels, and the rela- Malawi, Mali, Guinea, Madagascar, Sierra Leone, tive magnitudes of fuel price subsidies for LPG. Taking and the Central African Republic—LPG is not yet Angola as an example, LPG was heavily subsidized at affordable or easily available even for the rela- the time of the household survey, making it attrac- tively well-off. tive to switch to LPG rather than charcoal with rising income not only in urban areas but also in rural. National household expenditure surveys from Sub-Saharan Africa—which has no natural gas for household use—show that clean cooking energy 3 South Africa is not considered in this comparison, as its comprises predominantly LPG and electricity, with survey did not collect data to calculate total household expenditures. Growing access gap 5 Figure 3  Primary energy for cooking a. Share of population reliant on b. Difference between top and bottom wealth quintiles in different forms of energy shares of population reliant on different forms of energy Turkmenistan Belarus Tunisia Algeria Turks & Caicos Islands Argentina Iraq West Bank & Gaza Kazakhstan Suriname Georgia Vietnam Mexico Tonga Belize Serbia Thailand North Macedonia Kyrgyz Republic Montenegro Paraguay Mongolia Samoa Honduras Nepal Mauritania Kosovo Zimbabwe Congo, Rep. Côte d’Ivoire Bangladesh Ghana Kiribati Togo Nigeria Lao PDR Chad Congo, Dem. Rep. São Tomé and Príncipe Gambia, The Guinea-Bissau Malawi Mali Guinea Madagascar Sierra Leone Central African Republic 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100 Percentage Percentage Electricity LPG Natural gas Other clean Wood Charcoal/coal Other fuels Source: World Bank staff calculations using data available at https://mics.unicef.org/surveys. Note: The data analyzed are from UNICEF’s Multiple Indicator Cluster Surveys for which wealth quintiles are available starting in 2015. “Other fuels” are dominated by dung in Mongolia; kerosene in Kiribati, Nigeria, and São Tomé and Príncipe, and crop residues in Bangladesh. Where the sum does not total 100 in 3a, the shortfall is due to those who reported no cooking at home or did not reply to the questions on cooking. 6 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Figure 4  Percentage of people living in households citing different primary energy sources for cooking and annual expenditures per capita a. Urban areas 100 6,000 Annual expenditures per capita ($) 90 5,000 80 Urban population (%) 70 4,000 60 50 3,000 40 2,000 30 20 1,000 10 0 0 Ethiopia Gabon Gambia, The Ghana Lesotho Malawi Mauritania Namibia Botswana South Africa Tanzania Zambia Angola Burkina Faso Cameroon Côte d’Ivoire Eswatini Kenya Liberia Niger Nigeria Sudan Uganda 100 3,000 90 Annual expenditures per capita ($) 2,500 80 Rural population (%) 70 2,000 60 50 1,500 40 1,000 30 20 500 10 0 0 Angola Botswana Cameroon Côte d’Ivoire Eswatini Ethiopia Gabon Gambia, The Ghana Lesotho Malawi Mauritania Namibia Kenya Niger Nigeria South Africa Tanzania Sudan Uganda Zambia Burkina Faso Liberia Clean Wood Charcoal/coal Kerosene Others Expenditures per capita in 2022 dollars Source: Kojima and Zhou 2022. Note: Wood combines firewood and charcoal in Botswana. The South African survey did not ask enough questions to enable calculation of total household expenditures. Where the sum does not add up to 100, the shortfall is due to those who did not cook at home or did not answer the question on cooking. Expenditures are converted to U.S. dollars using market exchange rates. Others = mostly agricultural, other biomass residues, and dung. Growing access gap 7 Figure 5  Difference between the top and bottom quintile in the percentage of the population citing different forms of primary energy for cooking a. Urban areas 100 80 60 40 Urban quintile (%) 20 0 −20 −40 −60 −80 −100 Eswatini Gabon Cameroon Liberia Lesotho Sudan Malawi Mauritania Angola Burkina Faso Kenya Zambia Ghana Tanzania Uganda Niger Ethiopia Namibia Nigeria Gambia, The Côte d’Ivoire Botswana b. Rural areas 100 80 60 40 Rural quintile (%) 20 0 −20 −40 −60 −80 −100 Angola Nigeria Sudan Tanzania Lesotho Botswana Ghana Kenya Malawi Eswatini Liberia Namibia Niger Ethiopia Gabon Uganda Zambia Côte d’Ivoire Gambia, The Burkina Faso Cameroon Mauritania LPG Electricity Biogas Wood Charcoal Kerosene Others Source: Kojima and Zhou 2022. Note: The numbers subtract the share of the population in the bottom quintile from that in the top quintile. Wood combines firewood and charcoal in Botswana. Others = mostly agricultural and other biomass residues and dung. 8 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Evolving market and policy (because the frequency of refills is reduced), but the costlier each refill and the cylinder deposit fee. environments exacerbate Conversely, the smaller the cylinder, the smaller the clean cooking energy the cash outlay for the cylinder deposit fee and access challenge each refill, but the higher the unit cost and the greater the inconvenience due to frequent refilling. HIGH PRICE VOLATILITY AND THE Against this tension between costs and afford- HIGH COST OF LPG DELIVERY ability, sellers in deregulated LPG markets with no subsidies have experimented with cylinders of Arguably the greatest challenge to those wishing to varying sizes, eventually retaining cylinders of 10 cook with LPG is its relatively high price compared to kilograms (kg) or larger, with a smaller size such income, aggravated by high price volatility. In con- as 2 kg for camping and outdoor barbecues (Mat- stant dollar terms, the free-on-board prices of LPG thews and Zeissig 2011). By contrast, in regulated since 1995 have varied by nearly an order of mag- markets in poorer countries such as Indonesia, nitude (figure 6).4 The magnitude of price variation 3 kg cylinders are designated for household use, would be even greater in countries with currency which is costly and inefficient from an economic depreciation. Since 1995, for example, about 80 coun- point of view. tries have seen their currencies halve in value against Whereas kerosene can be purchased in any the dollar, and about 40 have seen them reduced to small quantity without markedly affecting the unit less than one-fifth, exacerbating the price volatility cost of supply, LPG cylinders are refilled only when of LPG and making it much costlier. they are empty, making purchases infrequent and To arrive at retail prices paid by households for LPG, akin to payment of utility bills. It is possible to the costs of transportation, storage, bottling, and meter LPG and pay small amounts at a time, but distribution as well as any taxes and fees levied by doing so increases the unit cost further because the government are added to the free-on-board of the cost of metering and monitoring. In Kenya, prices. How much retail prices exceed the free-on- unit prices of metered, pay-as-you-go LPG are board prices depends on economies of scale, about 15 percent higher than those of LPG sold in distances involved, competition in the market, and 13 kg cylinders.5 At $2 per kg in April 2022 in Kenya, the government’s tax policy. The downstream costs a 15 percent surcharge would amount to $0.30 per are markedly higher for LPG than for other liquid kg—more than 30 percent of the free-on-board fuels because LPG is a gas at room temperature prices in that month. and normal atmospheric pressure, requiring it to l Costs associated uniquely with LPG reduce its be compressed in metal containers. This additional market and further squeeze its ability to attain constraint makes LPG affordability more challenging economies of scale. No other liquid fuel except for several reasons: LPG requires bottling plants and truck transport l Economies of scale and indivisibility are closely of tens of thousands of small cylinders for house- linked. The larger the cylinder, the lower the unit hold use. Because bottling plants have large cost and the more convenient for the household economies of scale, only a handful of such plants are located in a given area, making it impractical 4 Free-on-board prices do not include the cost of shipping, insurance, evaporative losses, and other charges. 5 Personal communication with Paygo Energy in May 2022. Growing access gap 9 Figure 6  History of LPG prices in the Arab Gulf in 2022 U.S. dollars 1.60 1.40 1.20 1.00 $ per kg 0.80 0.60 0.40 0.20 0 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-23 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-22 Source: RIM LPG Intelligence Daily for propane and butane prices and FRED Economic Data for the U.S. GDP deflator. Note: Prices average Saudi Aramco contract prices for propane and butane, adjusted by the U.S. GDP deflator. to consider partial filling to reduce the cash outlay travel 90 km or more to refill their cylinders, as a for each refill. Ghana is an exception in this regard, government audit of an LPG subsidy scheme in making LPG users responsible for cylinder trans- India found (India Comptroller and Auditor Gen- port and refill (including partial refills) at numerous eral 2019). filling stations. However, “horrifying spectacles of victims of explosions involving…[LPG] cylinders… l Short-selling increases effective prices paid by usually leading to fatality” (Kojima 2021c, 34) have LPG consumers. Short-selling is a problem for prompted the government to begin shifting the liquid fuels around the world (Kojima and Bacon market to a centralized system of cylinder man- 2001), but avoiding detection of LPG short-selling agement. Fears about cylinder explosions and is facilitated by its being sold in opaque cylinders. fires deter households from considering LPG Underfilling and filling cylinders with anything (Kojima 2021c). But because proper cylinder main- that adds to the weight are two common ways tenance adds to the cost, many LPG marketers in of illegally increasing profits. Moreover, profi- developing countries avoid proper maintenance, teers have been known to decant properly filled refurbishment, and retirement of old cylinders, all cylinders leaving bottling plants—a practice that at a significant cost to safety. has drawn public anger and attention, especially LPG entails unique transportation require- when decanting has resulted in cylinder explo- ments as well. Trucking cylinders without sions and lives lost (Kojima 2021c). incurring excessive costs requires tarred roads in How much downstream costs add to those in reasonable condition, but many rural areas are figure 6 varies from market to market. Taking Brazil, not served by such roads. The result is that some South Africa, and Thailand as illustrations, total costs LPG purchasers with no home delivery have to 10 Cooking with Bottled Gas: Issues and Challenges in Developing Countries downstream of producer prices—prices at the refin- African countries surveyed, not all expenditures were ery or gas processing plant gates—in March 2022 in cash. Household surveys impute cash values to ranged from less than $0.20 per kg in Thailand to items acquired free of charge, such as homegrown $0.62 per kg in Brazil and slightly more than $1 per food and self-collected firewood—and the share of kg in South Africa before taxes (in Brazil and South freely acquired items is material, especially in rural Africa) and a subsidy (in Thailand) were applied. The areas and among lower-income households. By con- retail prices inclusive of taxes and subsidies were trast, purchasing clean commercial energy, including $0.51 per kg in Thailand, $1.69 per kg in Brazil, and LPG, always requires cash payment, further adding to $2.26 per kg in South Africa. the challenge of cooking with LPG. One measure of affordability is whether total expen- ditures on cooking energy can be limited to less than 5 percent of total household spending. The Energy Alternatives to LPG Sector Management Assistance Program’s (ESMAP’s) for cooking Multi-Tier Framework proposes a maximum of 5 per- cent of household income to be spent on cooking Costs and reliability drive energy—comprising fuel, cookstove, and associated choice between electricity equipment, such as the LPG cylinder, regulator, and hose—as an affordability ceiling (Bhatia and Angelou and LPG 2015). At $1.50 per kg and 10–15 kg a month,6 cylinder The relative prices of electricity and LPG influence refills dominate the cost of cooking with LPG, trans- a household’s selection of energy for cooking. For lating to a requisite minimal household income of example, better-off households in southern Africa $300–450 per month. cook with electricity rather than LPG because, tradi- tionally, it has been cheaper to cook with electricity, But price volatility in some months can raise the and power outages have been extremely rare (at retail price of LPG considerably. In times of high least until the mid-2000s). The break-even price global LPG prices, the retail price may rise to $2.50 below which cooking with electricity (using either per kg, for which minimal monthly household electric or induction stoves) becomes cheaper than income would need to be $500–$750 for monthly other alternatives is a function, among others, of consumption of 10–15 kg. For a family of four, $750 stove efficiency. per month would translate to $2,250 per capita per year—which exceeds average GDP per capita in both Illustrative break-even prices are shown in table 1, Sub-Saharan Africa and South Asia at the current which assumes two efficiencies for LPG stoves. The market exchange rate. Of the 23 Sub-Saharan Afri- first is the minimal efficiency of 68 percent required can countries in figure 4 for which expenditures per by the Bureau of Indian Standards (India Minis- capita could be calculated, only 4 had per capita try of Petroleum and Natural Gas 2011). The second expenditures in excess of $2,250: Botswana, Gabon, is 55  percent, which was the average thermal effi- Mauritius, and Namibia. Further, in the Sub-Saharan ciency of five commercially available LPG stoves obtained in Cameroon, China, Peru, and Uganda that were independently tested by researchers (Shen 6 Even the upper bound of consumption may be insuf- ficient to cover all cooking needs, particularly the preparation of dishes requiring hours of simmering for large families. Alternatives to LPG for cooking 11 Table 1  Break-even price of electricity in $/kWh for substitution of LPG with electricity for cooking Alternative Alternative LPG price LPG stove efficiency stove type stove efficiency $1.00/kg $1.50/kg $2.00/kg $2.50/kg Induction 85% 0.10 0.15 0.20 0.25 68% Electric 74% 0.09 0.13 0.17 0.22 Induction 85% 0.12 0.18 0.24 0.31 55% Electric 74% 0.11 0.16 0.21 0.27 Source: World Bank staff calculations. et al. 2018).7 For induction and electric stoves, effi- straightforward process, because—unlike LPG, for ciencies of 85 percent and 74 percent, respectively, which the unit price is fixed for a given cylinder size— were assumed;8 there are indications, however, that the electricity tariff schedule is complex. There may induction stoves may not be any more efficient than in fact be two or more schedules for residential con- electric stoves when food is being cooked for a long sumers, and within each schedule, there are usually time.9 two or more blocks in which the price almost always rises with consumption. A detailed analysis of elec- Taking an LPG price of $1.50 per kg as an example, tricity tariff schedules in 39 Sub-Saharan African the calculations show it would be cheaper to use countries found that 26 countries had more than an induction stove than an LPG stove with an effi- one schedule for residential consumers. In 31 coun- ciency of 55 percent if the price of electricity is below tries, there were two or more blocks for residential $0.18 per kWh. If the LPG stove meets the minimal consumers—Ethiopia had seven such consumption efficiency standard in India, however, the price of brackets for the energy charge for all residential con- electricity cannot be any higher than $0.15/kWh or sumers and five for the fixed monthly charge for $0.13/kWh for induction and electric stoves, respec- residential consumers with single-phase power—with tively, in these illustrative calculations. unit energy charges increasing with consumption. In countries with more than one block, nine had In most countries, determining the maximum unit volume-differentiated tariffs, whereby a single unit energy cost as shown in table 1 would not be a charge, determined only by total consumption, is applied to the entire consumption, or a combina- tion of increasing block and volume-differentiated 7 All tested LPG stoves were new with the exception of tariffs (Kojima et al. 2016). Although the analysis was a severely corroded one used in Cameroon from 2009 to 2016. Surprisingly, the worn-out stove had the high- of the tariff schedules in effect in July 2014, the pat- est efficiency, 57 percent at high power and 68 percent tern found is likely to be applicable today, suggesting at low, against the average of the other four of 51 percent that switching energy for cooking from a fuel to elec- and 55 percent, respectively. tricity would likely increase the average electricity 8 Induction estimate is from the Energy Star website, cost—and, in countries with volume-differentiated 2021-2022 Residential Induction Cooking Tops; the elec- tariffs, do so substantially. tric estimate is from the Electric Power Research Institute, “Induction Cooking Technology Design and Assessment.” The incremental electricity consumption after 9 Century Life website, Is Induction More Efficient Than switching to electric or induction stoves for cook- Electric Coil or Gas? An Energy Efficiency Comparison Between Stoves. ing influences the impact of rising block and 12 Cooking with Bottled Gas: Issues and Challenges in Developing Countries volume-differentiated tariffs on household expen- had lifeline blocks of 100 kWh a month or smaller ditures on electricity. Table 2 shows equivalent (table 3). consumption when electricity rather than LPG is used for cooking. Taking 12.5 kg a month as repre- Because electricity consumption for cooking is in sentative of the amount of energy needed to do addition to the electricity being used for lighting and most or all cooking with electricity, monthly electric- powering appliances, a household cooking fully with ity consumption may rise by 100 kWh or more. electricity is most likely to exceed the limit of the first block for the energy charge. The ratio of the elec- Although the data are old, the above review of elec- tricity charge in the second block to that in the first tricity tariff schedules in Sub-Saharan Africa indicates was more than 2 in 13 countries (table 4)—poten- representative sizes of the first block in rising block tially significantly increasing the monthly electricity or volume-differentiated tariff structures, often bill when cooking with electricity. Combining LPG referred to as the lifeline block. The average size and electricity for cooking helps moderate electric- of the first block in effect in July 2014 was 91 kWh ity consumption while taking advantage of efficient a month, but the range varied widely. The smallest stoves such as electric pressure cookers, as explored was 15 kWh, and 24 of 31 countries in the sample in five case studies by ESMAP (2020). Table 2  Total monthly consumption of electricity for cooking in kWh LPG stove Alternative stove Monthly consumption of LPG efficiency type 5 kg 7.5 kg 10 kg 12.5 kg 15 kg 17.5 kg 20 kg Induction 51 76 100 130 150 180 200 68% Electric 58 87 120 150 170 200 230 Induction 41 61 82 100 120 140 160 55% Electric 47 70 94 120 140 160 190 Source: World Bank staff calculations. Note: The induction and electric stove efficiencies are assumed to be 85% and 74%, respectively as in table 1. Only two significant figures are shown. Table 3  First tariff block size in kWh per month in July 2014 in Sub-Saharan Africa First block size, kWh 15–25 26–50 51–75 76–100 101–150 151–200 201–500 Number of countries 5 11 4 4 3 2 1 Source: Adapted from Kojima et al. 2016. Note: The first block size is the maximal monthly consumption in kWh above which the energy charge for the next block will be applied. Table 4  Ratio of the energy charge for the second block to that for the first block in July 2014 in Sub-Saharan Africa Ratioa 1.04–1.5 1.51–2.0 2.01–3.0 3.1–4 4.1–5 5.1–5.5 Number of countries 11 6 6 4 1 2 Source: Adapted from Kojima et al. 2016. a. The ratio of the effective energy charge, inclusive of ad valorem tax such as value-added tax but exclusive of fixed charges and taxes applied to fixed charges, between the second block (or the second level of service if there are several schedules of a single block each with increasing installed capacity) and the first block. Alternatives to LPG for cooking 13 ENTER ADVANCED COMBUSTION subject to considerable price volatility. One study BIOMASS STOVES of start-up materials found 43 unique ignition materials, with the most commonly cited crite- For many households, and especially those in rural rion for selection being whatever is available and areas, LPG will continue to be unaffordable and out burns quickly—which is unlikely to burn cleanly of reach. For these households, use of domestic bio- (Fedak et al. 2019). mass in the form of densified pellets in advanced l Refueling increases particulate emissions mark- combustion stoves may offer the best chance of edly. Avoiding refueling would require sufficient attaining access to clean energy. Such stoves can fuel in the stove before cooking and most likely slash emissions of harmful pollutants without rules out dishes that take hours to prepare, such subjecting households to global petroleum price as githeri, a Kenyan traditional meal of maize and volatility. Where there is no reliable electricity and beans. road infrastructure is poor, advanced combustion l At the end of cooking, pellet burnout and dis- stoves may be a viable alternative for clean cooking, posal can result in high, visible emissions. Many although these stoves require alternative sources of cooks let the stoves smolder because proper and electricity. immediate disposal of hot pellets and ashes after When operated properly, advanced combus- cooking is difficult and entails physical risk. tion stoves enable full health benefits from l Deterioration in emissions levels with stove emissions reduction to be reaped. One difference age has been found to be high. This deteriora- with electricity and gas, however, is that the emis- tion points to the need not only for regular stove sions performance of advanced combustion stoves maintenance but also for improving stove design can be compromised by start-up materials and prac- to increase performance durability in the field. tices, heterogeneity of biomass used (size, moisture l A lack of access to reliable electricity leads to content, type), and excessive emissions during fuel more equipment breakdowns. If the source of reloading and the burn-out phase (Champion and electricity is not the grid but a system comprising Grieshop 2019; Kojima 2022a): a solar photovoltaic panel backed up by a battery, battery and other equipment failures are worry- l Properly sized pellets are essential. There ingly common and increase emissions further, at are economies of scale in pellet manufacture, times even rendering the stove inoperative (Clark and making such pellets primarily for cook- et al. 2017; Dickinson et al. 2019). ing by households rather than for industrial use has been found to be costly (Carter et al. 2018). Technical solutions to the second and third prob- Ensuring that as much biomass as possible be lems, such as auto-ignition and automated pellet renewable adds to the challenge. loading mechanisms, are not yet commercially avail- l The initial start-up of a pellet stove can cause able for cookstoves and will increase the cost. significant fine particulate emissions, a problem For advanced combustion stoves to play a signifi- that gas and electricity do not pose. Igniting a cant role in fostering universal access to clean energy, pellet stove with anything other than kerosene the challenges outlined above must be addressed increases emissions significantly. Using too much to yield affordability, convenience, and health ben- kerosene results in small plumes of black smoke; efits. Markets for advanced combustion stoves are too little kerosene means a longer ignition time. very small at the moment. Given economies of scale Kerosene can be expensive in rural areas and is in both pellet and stove manufacture, concerted 14 Cooking with Bottled Gas: Issues and Challenges in Developing Countries efforts across governments, the private sector, and assuming 70 percent of displaced wood was renew- international organizations to improve stove design able (that is, carbon neutral). Even the most optimistic and scale up production are likely to be fruitful. scenario of all displaced wood being entirely carbon neutral still reduced net emissions by 3.7  million THE IMPACT ON GREENHOUSE GAS tonnes because of near elimination of black carbon EMISSIONS VARIES FROM COUNTRY emissions. TO COUNTRY Decarbonization of electricity generation, com- The effort to mitigate global warming has added bined with universal use of electricity for cooking, another dimension to the global goal of achieving would ensure clean cooking from both the global universal access. LPG is a fossil fuel, but biomass is and local perspectives. However, electricity gener- potentially renewable, making it intuitively more ation is not yet carbon neutral in most economies. appealing for slowing global warming. However, the Focusing on CO2, the main greenhouse gas emitted impact of fuel combustion on global warming is during the generation of electricity and combustion more complex. Where LPG substitutes for firewood, of LPG, there is a threshold level for the combined carbon dioxide equivalent (CO2e) emissions may margin emission factor for grid electricity below decrease, in part because combustion of firewood which use of electricity rather than LPG for cooking burned in most stoves releases much more black would emit less CO2. Such thresholds as a function carbon than combustion of LPG and has a global of stove efficiency are shown in table 5. Also shown warming potential two to three orders of magnitude is the number of economies in which the combined higher than that of CO2 (Myhre et al. 2013). margin emission factors may be below the threshold level. These data suggest that if electricity is available, Singh, Pachauri, and Zerriffi (2017) took data from not any costlier than LPG (as calculated in table 1), national household expenditure surveys collected and reliable—which is unfortunately not the case in in 2001 and 2011 in India and estimated changes in many developing countries—cooking with electricity net CO2e emissions following substitution of LPG for would reduce emissions of CO2 in the coming years wood. The analysis concluded that replacement of and would also be just as affordable, if not more. wood by LPG by households over the intervening decade reduced wood consumption by 7.2  million These statistics should be treated with caution tonnes and CO2e emissions by 6.7  million tonnes, because the combined margin emission factors used Table 5  Maximum combined margin emission factor for electricity below which cooking with LPG emits more CO2 LPG stove efficiency Alternative stove type Alternative stove efficiency g CO2/kWh Numbera Induction 85% 280 67 68% Electric 74% 250 60 Induction 85% 350 88 55% Electric 74% 310 72 Source: World Bank staff calculations and tabulation based on combined margin emission factors in AHG-001 available at https://unfccc.int/ climate-action/sectoral-engagement/ifis-harmonization-of-standards-for-ghg-accounting/ifi-twg-list-of-methodologies. Note: The induction and electric stove efficiencies are assumed to be 85% and 74%, respectively, as in table 1. Only two significant figures are shown for g CO2/kWh. a. The number of economies with combined margin emission factors smaller than the threshold level expressed in g CO2/kWh out of 232 economies in the database. Barriers to capturing the full benefits of clean cooking 15 in table 5 are derived by taking data from Brazil, China, households would benefit fully from the adoption of India, Indonesia, Mexico, the Russian Federation, and clean household energy. South Africa and then running a simplified regression for all other developing countries, leading to large Over the last two decades, the energy ladder uncertainties in the latter. Country-specific anal- theory has been challenged and replaced with a ysis should be carried out for decision making. The multiple-fuel model in which households “stack” emission factors also do not include scope 3 emis- fuels and stoves. The latter model was first reported sions—upstream emissions in the production of oil, in a 2000 study of rural households in Mexico, which natural gas, and coal and their transportation—which found that households adopted multiple-fuel can be considerable for both electricity and LPG. cooking in a bidirectional process, suggesting that separate, coexisting factors simultaneously pushed households away from biomass and pulled them Barriers to capturing back. Even when households had been using LPG for many years, they rarely abandoned fuelwood the full benefits of use. Thus, fuelwood savings from using LPG ranged clean cooking from 35 percent on average in one village to as little as zero percent in another (Masera, Saatkamp, and Kammen 2000). To protect health and the local environment, it is not enough for households simply to adopt clean energy. They must also abandon polluting fuels, technologies, WHY MULTIPLE FUEL USE PERSISTS and practices. Abandonment is a function of afford- Households retain polluting practices because it ability first and foremost, followed by the ready and costs less to do so. Clean sources of energy with continuous availability of clean options. which it is possible to do all cooking are electric- ity, natural gas, LPG, biogas, alcohol, and densified wood pellets used in advanced combustion stoves. Securing abandonment of In addition to higher stove and fuel costs than tra- polluting practices is elusive ditional biomass stoves, the literature documents several other challenges facing clean energy options. Through the 1990s, many practitioners in the field In developing countries, these stoves tend to be too of household energy believed that households small and unstable for large pots and pans and for would abandon cheaper but more polluting forms vigorous stirring, making it difficult to cook certain of energy and switch to cleaner and more conve- types of dishes and for large families. There are fears nient—albeit costlier—energy as they became richer. about burns and explosions, especially with gaseous Termed the energy ladder theory, this widely held fuels, and LPG stoves come with the added inconve- belief envisaged households moving up the rungs nience of having to wait for LPG cylinder refill delivery of energy sources, abandoning cheaper forms as or take an empty cylinder to the closest retail outlet— they acquired more income. According to this theory, which may be tens of kilometers away—to exchange households would start with three-stone fires and for a full cylinder. In high-income countries, those then abandon traditional use of biomass completely using LPG have two cylinders to make sure they do as their incomes rose, moving to charcoal, kero- not run out, but the initial payment for two cylinders sene, and eventually gas or electricity. In so doing, is unaffordable for many households in developing countries. 16 Cooking with Bottled Gas: Issues and Challenges in Developing Countries LPG stovetops for use in developing countries, Many more better-off including those tested in the above-cited paper by Shen et al. (2018), are relatively small and have households can switch only two burners, meaning that households cannot entirely to clean cooking cook large meals or several dishes in parallel. LPG and advanced combustion stoves are less sturdy EXCLUSIVE VERSUS NON-EXCLUSIVE compared to three-stone fires, potentially caus- USERS OF LPG ing instability if heavy weights are placed on them. Household expenditure surveys can provide infor- Three-stone fire stoves sit on the ground, whereas mation on households that use LPG and have both LPG and advanced combustion stoves are ele- abandoned all forms of polluting energy. This study vated: LPG stoves for safety, because LPG sinks to the used household survey data in Sub-Saharan Africa ground if it leaks; and advanced combustion stoves and confined the analysis to 12 countries in which because the cooktop sits on a combustion chamber. more than one-fifth of the population, either nation- Gitau et al. (2019) found that large Kenyan families ally (nine countries) or in urban areas (three countries), preferred three-stone fires over these cleaner cook- cited LPG as the primary cooking fuel. The share of stoves because they cooked large meals in large pots. LPG users who had stopped using polluting energy varied. In Côte d’Ivoire, Gabon, Mauritania, Mauritius, These problems have been fully addressed in and urban Nigeria, more than three-fifths of primary high-income countries, where large and sturdy LPG users had abandoned polluting fuels, exceed- stoves with multiple burners have been on the ing 90  percent in Gabon and reaching 99  percent market for decades and where almost all house- in Mauritius. At the opposite end of the spectrum, holds meet their cooking and heating needs with the corresponding share was less than one-third in gas and electricity. But such stoves are unaffordable Lesotho, urban Burkina Faso, and urban Kenya. With to many households in the rest of the world, partic- the exception of urban Burkina Faso, more than ularly in developing countries. three-fifths of all households using LPG had adopted it as their primary cooking fuel (figure 7). Cooking certain dishes consumes large quanti- ties of fuel, incurring associated costs. For example, Ochieng et al. (2020) found that it could take as long CHARACTERISTICS OF THE RICH WHO as eight hours to cook githeri, a staple consumed by ABANDON POLLUTING FUELS nearly all the 40 ethnic groups in Kenya and often As expected, the rich were far more likely to abandon cooked in bulk. The amount of fuel and the size of polluting fuels. However, in every country, house- the pot needed led most households to rule out LPG holds that had abandoned polluting fuels were and other modern stoves for making githeri. markedly smaller in size than those using LPG but stacking polluting fuels. In most cases, abandoning households spent less on LPG, suggesting that they consumed less LPG than the stacking households. In the top 20 percent (top expenditure quintile) of the population, the characteristics of these two types of households, both using LPG for cooking, were revealing. Abandoning households were richer (as measured by per capita expenditure) in 11 countries; the only exception being Mauritius, where there Barriers to capturing the full benefits of clean cooking 17 Figure 7  Breakdown of adoption of LPG: abandonment of polluting fuels, primary use, and all users of LPG 100 Other users 90 Primary 80 Abandonment 70 Population (%) 60 50 40 30 20 10 0 Angola Cameroon Côte Gabon Ghana Lesotho Mauritania Mauritius Sudan Burkina Kenya* Nigeria* d’Ivoire Faso* Source: Kojima and Zhou 2022. Note: The analysis is confined to those areas—nationally or urban areas—where more than one-fifth of the population lived in households citing LPG as the primary cooking fuel. For the first nine countries, the entire sample is taken because the one-fifth threshold is met nation- ally. In Burkina Faso, Kenya, and Nigeria, the one-fifth threshold is met only in urban areas. Households abandoning polluting fuels include very few LPG users who did not cite LPG as their primary cooking fuel. However, the sum of abandonment and primary bars in the figure equals the percentage of the total population citing LPG as the primary cooking fuel. Taking Cameroon as an example, the figure shows that 31 percent of the population was using LPG in any capacity, 20 percent of the population cited LPG as the primary cooking fuel, and 7 per- cent used LPG and no other polluting liquid or solid fuels. Abandonment = percentage of people living in households that used LPG and had abandoned all forms of polluting fuels; primary = percentage of people living in households that cited LPG as their primary cooking fuel but had retained the use of polluting fuels; other users = percentage of people who reported nonzero expenditures on LPG but did not cite it as their primary cooking fuel and had retained the use of polluting fuels; * = urban areas only. were too few stacking households to be representa- polluting fuels was larger than the household size of tive. Because abandoning households were smaller, those in the top expenditure quintile that continued in nine countries, their average household expendi- to use polluting fuels. Figure 8 compares household tures were lower than those of stacking households. size for the top two expenditure quintiles in 10 of the This also meant that in half the countries, despite 12 countries shown in figure 7.10 spending less on LPG, the percentage of total house- hold expenditures spent on LPG by those who had The surprising finding about the relationship in abandoned all other fuels was higher. household size between the fourth and top quin- tiles can be seen by comparing the full height The observations about household size may signal of the orange bar in figure 8 with the patterned the ease of abandoning polluting fuels when cooking for small families. However, the average household size of those in the fourth expenditure Mauritius and Burkina Faso are not shown because Mau- 10 quintile—which by definition are poorer than anyone ritius had too few households stacking fuels, and urban in the top expenditure quintile—that had abandoned Burkina Faso had too few households that had aban- doned polluting fuels. 18 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Figure 8  Comparison of sizes of households abandoning versus stacking polluting fuels 7 Q4 abandoning Q4 stacking Q5 abandoning Q5 stacking 6 5 Household size 4 3 2 1 0 Angola Cameroon Côte Gabon Ghana Lesotho Mauritania Sudan Kenya* Nigeria* d’Ivoire Source: World Bank staff calculations using household survey data. Note: The solid portions are incremental household sizes for those staking fuels. For example, in the top quintile in Cameroon, the average size of households abandoning polluting fuels was 2.1, and that of the households stacking fuels was 3.9. * = urban households only; Q4 = quintile 4; Q5 = quintile 5; abandoning = LPG-using households that had abandoned polluting fuels; stacking = households citing LPG as their primary cooking fuel and were using polluting fuels in parallel. (nonsolid) portion of the blue bar. The latter group have stable, large wood-burning stoves to make of households was cooking exclusively with LPG and these traditional dishes in large quantities. electricity despite being poorer in every case and larger on average than those in the top quintile that In addition, a lack of sufficient awareness about the continued to use other fuels. Possible explanations benefits of using clean energy—and conversely, the include differences in market conditions and per- adverse effects of using polluting fuels—may con- sonal preferences: tribute to continued fuel stacking (Kojima, Bacon, and Zhou 2011). l With respect to market conditions, some house- holds may live closer to LPG retail outlets than Barring households in the fourth quintile in Lesotho others or near LPG retailers that respond immedi- and urban Kenya, those that had abandoned all pol- ately to requests for refills, or have two cylinders luting fuels did not spend any more than 3 percent of so the time it takes to refill is not a pressing con- total household expenditures on LPG (table 6). With cern, thereby making it easier to abandon other the exception of Sudan, all households spent more fuels. on electricity than LPG; this is consistent with obser- l For differences in personal preferences, some vations in the literature that households consider families may like dishes that are difficult or costly spending on electricity a higher priority than cooking to cook with LPG, such githeri in Kenya or tuo energy (Kojima 2021c). Those households that con- zaafi—a popular goat stew prepared in north- tinued to use other fuels alongside LPG used mainly ern Ghana, which is cooked for many hours and charcoal or kerosene—even when doing so was not requires vigorous stirring. These families may Lessons from government support for LPG 19 Table 6  Statistics on people who live in households that used LPG and had abandoned polluting fuels Côte Item Angola Cameroon d’Ivoire Gabon Ghana Lesotho Mauritania Sudan Kenyaa Nigeriaa Quintile 4 % of quintile-4 36 7 22 85 10 10 40 22 5 18 population % LPG expenditure 1.4 2.3 1.0 0.7 1.1 3.3 1.9 1.6 4.6 2.7 share % Electricity 2.1 2.5 2.2 2.8 5.3 3.9 7.9 1.4 2.1 1.8 expenditure share Quintile 5 % of quintile-5 43 26 45 84 21 14 45 30 15 40 population % LPG expenditure 1.1 1.7 0.9 0.7 1.0 2.1 1.7 1.4 3.3 2.2 share % Electricity 2.0 1.8 2.0 3.2 4.3 2.7 8.7 1.2 2.2 1.9 expenditure share Source: Kojima and Zhou 2022. a. Urban households only. any cheaper than cooking with LPG alone, as in using deforestation in certain pantropical regions. Accord- kerosene and LPG together in urban Nigeria. ing to one estimate, 27–34 percent of extraction of woodfuel (fuelwood and charcoal) in such regions is These findings suggest considerable scope for unsustainable, and approximately 275 million people abandonment of polluting fuels and the need to in South Asia and East Africa live in areas with deplet- understand why those who have the financial ing woodfuel resources (FAO 2022). means to be exclusive users of clean energy do not do so, such as a lack of reliable LPG supply or difficul- Since the 2000s, the policy focus—and support— ties with cylinder refill. for household energy has been about controlling damage to public health from household air pollu- tion and enhancing gender equality while reducing Lessons from consumption of fuelwood and charcoal to stem government support deforestation. The policy question, however, cannot be reduced merely to that of how to shift house- for LPG holds away from woodfuel, in no small measure because of the significant dependence of livelihoods Government support for LPG in the 1980s and 1990s on it. In Nigeria alone, one-fifth of the population is was motivated by concerns about diminishing forest engaged directly in fuelwood collection and char- cover due to household use of fuelwood and char- coal production, providing an estimated 530,000 coal. Even though the most recent data confirm that full-time-equivalent direct jobs; an additional agricultural expansion is driving nearly nine-tenths 200,000 people—mostly on a full-time basis—pro- of global deforestation, fuelwood and charcoal vide transport services for retail and wholesale trade production remains a significant contributor to (FAO 2022). 20 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Historically, the most common form of government lower-middle-income countries (Kojima 2011). Uni- support for LPG has been a universal price subsidy. versal price subsidies benefit rich and poor alike, but These subsidies continue in many countries, includ- the rich capture a significantly greater proportion ing Angola, Indonesia, Jordan, Morocco, and Sudan, of the total LPG subsidy because they can afford to and have recently been resumed in other countries consume more. An extreme case is the use of subsi- that had previously abandoned them. Price subsidies dized LPG by the rich to heat their swimming pools should be fully accounted for in the national budget and saunas in Brazil (Lucon, Coelho, and Goldemberg and scrutinized annually by the parliament, but 2004). often they are not. Especially in countries with sig- nificant upstream oil and gas production, petroleum Universal price subsidies are offered because they producers—typically national oil companies—may are administratively easy to introduce and politi- be officially or informally required to cross-subsidize cally expedient (Victor 2010). They are all the more LPG using their upstream oil and gas revenue. The attractive in countries where the government has net result is lower upstream fiscal revenue in the a poor track record of delivering social services. form of lower taxes, reduced dividend transfers, or Given their regressive nature and high costs, univer- both. Nonfinancial government policies to promote sal price subsidies would not be the policy choice of LPG are preferable to providing price subsidies, argu- many governments if they could implement more ably the most important of which is to improve the targeted interventions. They persist—at least in regulatory environment and create a market with part—because of weak institutional capacity. Where fair and healthy competition with adequate safe- politically vocal and powerful constituents, who tend guards for consumers. These issues are explored in to be better-off, benefit significantly from a universal this section. price subsidy for LPG, its regressive nature is, ironi- cally, what increases general political support for the administration in power. Universal price subsidies increase LPG use but at an Subsidy reforms are unsustainable cost frequently reversed Universal price subsidies predictably increase Many governments have announced the phaseout household consumption of LPG. Everything else of universal price subsidies for fuels. Some have sub- being equal, the lower the ratio of the cost of LPG to stituted targeted cash transfers for untargeted LPG income, the greater the household use of LPG; and price subsidies, as discussed in the next subsection, lower prices increase both the percentage of house- only to reintroduce universal price subsidies later holds using LPG and the quantities of LPG consumed and implement them side by side with targeted sub- by those households that have chosen LPG. Since sidies; this is discussed further later in this section. January 2008, when Indonesia froze the highly sub- sidized price of LPG sold in 3 kg cylinders, the share In some countries, a fuel moves in and out of a uni- of households using LPG as the primary cooking fuel versal price subsidy regime. An example is bottled has soared more than fourfold (Kojima 2021c). LPG in Thailand. An oil fund in the downstream petro- leum sector has been established to cross-subsidize Because of the relatively high costs of LPG com- fuels. There is a levy on each fuel for the fund, and pared to other cooking fuels, universal price certain fuels are subsidized with negative levies. subsidies tend to be highly regressive in low- and Lessons from government support for LPG 21 Figure 9  Unit subsidy for bottled LPG in Thailand 0.60 Subsidy Collection 0.50 Subsidy or levy paid ($/kg) 0.40 0.30 0.20 0.10 0.00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-22 Jan-23 Source: World Bank staff calculations using data at Thailand Energy Policy and Planning Office, Petroleum Prices. Gasoline blended with 85 percent ethanol and petro- in Morocco, which has maintained the same price for leum diesel blended with 20 percent biodiesel have LPG sold in 3 kg, 6 kg, and 12 kg cylinders since the been consistently cross-subsidized in this way since 1990s (Bladi.net 2022). they were first introduced in the market. Accord- ing to government data, since January 2001, bottled LPG has been subsidized three-fifths of the time and Commercial malpractice the cumulative unit subsidy provided has exceeded and stifling of competition the fund levy charged to LPG more than fourfold, are unintended resulting in a significant net subsidy (figure 9). One consequences outcome is the freezing of the LPG price for two years from April 2020—at the time of the collapse of Subsidized prices lead to illegal diversion. One exam- the global LPG prices in the wake of the COVID-19 ple is the unauthorized and unsafe use of bottled pandemic—to March 2022, when global LPG prices LPG in vehicles designed to run on gasoline with- were much higher. out proper conversion in countries where gasoline costs more per km driven. Fuel shortages caused When subsidized LPG is used extensively by house- by inadequate reimbursement of the subsidy to holds, any pricing reform becomes highly politicized LPG suppliers create financial incentives for black and intractable. The more subsidized the fuel, the marketing of subsidized LPG. While long-distance more widely it is used and the greater the resistance transportation of bottled LPG is not as easy as that to raising its price. In the face of strong opposition of other liquid fuels, if prices in neighboring countries and politicization, retail LPG prices in some countries are sufficiently higher, smuggling could also occur, have been frozen for months on end, as in Sudan; for especially along the borders. Government attempts years as in Angola (since December 2015), Indonesia to limit the subsidy outlay by subsidizing only LPG (since January 2008), Jordan (since January 2016), and sold to households cannot stop illegal diversion. Any Senegal (since February 2016); and even decades, as 22 Cooking with Bottled Gas: Issues and Challenges in Developing Countries price differentiation between two products that are first prerequisite of which is fair and healthy compe- identical in quality provides strong financial incen- tition in the market. tives for illegal diversion of the lower-priced good to the higher-priced market. Advances in digital An example is differential pricing of bottled versus technologies have enabled bulk LPG in Peru. Until 2012, LPG pricing was governed wider adoption of targeted by the rules for the Fondo para la Estabilización de Precios de los Combustibles Derivados del Petróleo subsidies (FEPC, Petroleum Fuel Price Stabilization Fund), first In response to the regressive nature of universal created in 2004 and intended to last only 120 days price subsidies, and exacerbated by the significant but which continues to exist to this day. Prices are economic distortions and inefficiencies associ- allowed to move only within a narrow band, out- ated with them, some governments have begun to side of which the deficit or surplus would be drawn implement more targeted and less distortionary from or transferred to the fund. This mechanism is subsidy delivery mechanisms (Kojima 2021c). These intuitively appealing. In theory, prices are smoothed steps have been aided by advances in digital tech- without needing budgetary support. However, these nologies, especially mobile banking. An important price stabilization funds are self-financing only if departure from the past is a move to market-based there is frequent reversion of global prices to the pricing accompanied by cash compensation to eligi- mean. Global oil price movements have not shown ble LPG purchasers (or sellers that sell LPG to eligible such constant reversion to the mean for decades, LPG purchasers). This approach requires drafting eli- making price stabilization funds seldom, if ever, gibility criteria for cash compensation; creating a self-financing. In Peru, the government was forced list of eligible beneficiaries; determining an appro- to transfer $2.4  billion to the fund between 2005 priate amount and frequency of cash transfers; and and the end of 2011. In 2012, LPG sold in bulk was deciding whether the supplier receives the cash removed from the FEPC, resulting in a material differ- transfer for the LPG purchase made, or if the supplier ence in the unit prices of bottled and bulk LPG, with is bypassed and the cash is delivered directly to the bottled LPG almost always being cheaper. The Peru- purchaser. vian petroleum regulator has presented evidence of diversion of cheaper LPG to the higher-priced market HOW MUCH TARGETING by comparing consumption of each LPG type as a function of its relative unit price (Osinergmin 2019). Selling LPG at market-based prices and transfer- ring cash—either to eligible LPG purchasers or to Because governments set prices when they are LPG suppliers for registered eligible purchases—is subsidized, price subsidies stifle competition in the a marked improvement over price subsidies. Cash market. In deregulated markets, firms compete on transfers conditional on LPG purchase do not elim- price and quality of service; removing the suppliers’ inate economic distortions entirely: LPG suppliers ability to set prices eliminates the ability to compete know and take into account the fact that the cash on price. In some countries, such as India and Indo- transfers reduce the prices purchasers ultimately nesia, fuel subsidies are channeled entirely through pay—although the sharper the targeting, the smaller national oil companies, cementing their market the distortions. power and making it more difficult to prepare the downstream petroleum sector for deregulation—the Lessons from government support for LPG 23 Striking the right balance between over- and under- reform—cash transfers would not be considered a coverage of beneficiaries is one policy challenge. If form of fuel subsidy (Kojima 2017). targeting criteria are so strict cash transfers cover a negligibly small fraction of total LPG consumption, CASH TRANSFER EXPERIENCE AND the economic distortions would be correspond- OTHER OPTIONS ingly small—but so would the impact on household All four countries with cash transfers cited above LPG use, unless the vast majority of the popula- use digital technologies. Among them, Peru has tion can already afford clean modern energy and the sharpest targeting. The percentage of the total the cash transfers are intended to help the remain- population covered as beneficiaries in April 2022 ing few. At the opposite end of the spectrum, if cash was 2.6 percent in Peru, 6.4 percent under the cyl- transfers are generous, provided to a large swath of inder deposit fee program in India, 11 percent in the the population, and cover a sizable fraction of the Dominican Republic, and 18 percent in El Salvador. LPG purchase cost, many more households would The total population covered would be a function use LPG; but low prices widely available to so many of the average household size of the beneficiaries. If, would increase the fiscal burden and likely dimin- for example, the beneficiaries in Peru had an aver- ish the incentives to improve the efficiency of the age household size of five, the cash transfer program supply chain as well as of LPG stoves. would cover 13 percent of the total population. To the extent that the poor generally have larger families CASH DELIVERY MECHANISM than the rest of the population, coverage would be Cash transfers to suppliers may be administra- correspondingly greater than that estimated on the tively easier because there are far fewer suppliers basis of the average size of all households. The pro- than purchasers. However, if there are credible con- grams studied are detailed further in the appendix. cerns about timely and full reimbursement by the government for the subsidy, cash transfers to LPG Instead of cash transfers, Ghana distributed free 6 kg purchasers may exert more pressure on the govern- cylinders and cookstoves in the Rural LPG Program it ment for reimbursement, because the government’s launched in 2013. In 2022, it expanded the initiative failure to pay on time and in full would become more to the entire country under the LPG for Development widely and rapidly known. In addition, if LPG suppli- Program—consisting of household and commercial ers are not involved in the administration of subsidy modules—with the goal of distributing 203,160 cook- reimbursement, it may be easier for the government stoves by 2025 (Ghana Ministry of Energy 2022). It to focus its attention on the supply chain for creat- does not appear that Ghana’s program is using digi- ing a competitive market and ensuring that suppliers tal technology for enrollment. operate strictly on a commercial basis. Unconditional cash transfers take the LPG supply Choice between conditional chain out of the subsidy delivery mechanism and and unconditional cash have the least distorting effects. They also almost transfers depends on policy always exclude the better-off. It is worth noting objectives and resources that if cash transfers are not conditional on LPG purchase—even if the transfer program is first intro- available duced for the explicit purpose of compensating Conditional cash transfers can cover LPG purchases households for higher LPG prices following a subsidy or the initial costs associated with adoption of LPG 24 Cooking with Bottled Gas: Issues and Challenges in Developing Countries (an LPG stove, the cylinder deposit fee, or both). India, Universal price subsidies El Salvador, and Peru have cash transfers conditional on LPG purchase; India further has cash transfers for have returned on and off, the poor to cover the cost of the cylinder deposit and fully in 2022 fee for first-time buyers. The Dominican Republic Ghana is one of the few countries eliminating uni- also introduced cash transfers when the LPG price versal price subsidies for LPG that has maintained subsidy was phased out in 2008; the transfers have market-based pricing with no distinction in the been unconditional. treatment of LPG depending on end use. Although In offering conditional cash transfers, govern- the stated policy objective of governments remov- ments are picking winners rather than being fuel ing the subsidies was to replace a universal price and technology neutral. Unconditional cash trans- subsidy with cash transfers, the Dominican Repub- fers, in comparison, put decision making entirely in lic, El Salvador, and Peru all reinstated price subsidies the hands of the cash recipients, who alone decide after several years. In India, LPG for household use is their spending priorities. Because unconditional exempt from duties, taxes, and other levies, making cash transfers do not specify how to spend the cash, it essentially subsidized and much cheaper than LPG many lesser-off households choose items other than for other users11 (Kojima 2017). a clean cooking fuel. For example, a study of uncon- The history of LPG pricing in these countries points ditional cash transfers implemented in 2005–06 in to the difficulties of ending price subsidies once Indonesia to mitigate the impact of higher kerosene started and the high risk of policy reversal. It is fre- prices on the poor and near-poor found that the top quently argued that times of low international fuel two expenditure items selected were rice and debt prices are ideally suited for subsidy removal because repayment (Widjaja 2009). In the extreme, the cash the upward price adjustments can be limited or may may be used to purchase polluting fuels. In some even be unnecessary. However, if a period of low circumstances, therefore, it may be appropriate to prices is followed by a prolonged period of rising consider LPG a merit good and provide targeted prices on the world market, political pressure to conditional cash transfers. reverse the subsidy reform can quickly mount. The fiscal burden of unconditional cash transfers Such a period of rising prices was observed from would be greater than that of conditional cash trans- 2017 to 2019, and again from mid-2020. While Ghana fers with the same degree of targeting. Not every and Mexico—both of which used to have large LPG potentially eligible household chooses to participate price subsidies—have maintained their deregulation in a conditional cash transfer program because the policy, the Dominican Republic, El Salvador, and Sen- amount of cash transferred may not be large enough egal, among others, have resorted to freezing prices. to make LPG affordable or spending on LPG may not El Salvador reintroduced a universal price subsidy be a priority for the poor. By contrast, the only deter- rence to receiving unconditional cash transfers is the initial transaction cost of enrollment. 11 From February 2014 to December 2022, the price differ- ence between “unsubsidized” LPG sold in 14.2 kg cylinders and that sold in 19 kg cylinders ranged from 10 percent to 46 percent and averaged 32 percent in 2022. Source: World Bank staff calculations based on retail prices posted on the Indian Oil Corporation’s Indane Cooking Gas webpage. Lessons from government support for LPG 25 in October 2021; while the Dominican Republic has bottled LPG between 2012 and 2018 far exceeded the frozen and subsidized LPG on and off, providing a contributions paid into the fund (Osinergmin 2019). universal price subsidy—which its Bonogás program In April 2020, the government finally removed bot- was intended to end definitively—for many months tled LPG from the fund to end the universal price at a time. subsidy once and for all. But against a backdrop of rising international prices and growing political pres- The history of LPG pricing in Peru is informative, not sure to resubsidize LPG, the government resumed only for the reversal of the subsidy reform policy the price subsidy for bottled LPG in September 2021 but in terms of how the subsidy for LPG intended with the issuance of Supreme Decree No. 023-2021- for household use affected pricing of LPG for busi- EM, which brought bottled LPG back under the FEPC. nesses. As mentioned earlier, the government in In March 2022, Supreme Decree No. 002-2022-EM 2012 removed LPG sold in bulk from the FEPC but brought bulk LPG under the FEPC. Consequently, not bottled LPG, resulting in a large price difference both residential and commercial LPG users are now between bottled and bulk LPG and the diversion of subsidized. bottled LPG to bulk customers. The regulator in 2019 published a report showing that the price subsidy for Table 7  LPG subsidies and effective end-user prices paid in June 2022 Unsubsidized Universal price Cash transfer Price net of Cylinder size price subsidy Retail price to the poor transfer Unit Kg $/kg $/kg $/kg $/kg $/kg Dominican Republic 11.4a n.a. n.a. 1.33 0.75a 0.59 El Salvador 11.4b 1.31 0.33 0.98 0.71c 0.27 Ghana — 1.50 0.00 1.50 0.00 1.50 India 14.2d 1.50 0.59 0.90 0.18 0.72 Peru 10.0e n.a. n.a. 1.43f 0.67 0.76 South Africa — 2.33 0.00 2.33g 0.00 2.33 Thailand — 1.01h 0.39i 0.62j 0.00 0.62 Source: World Bank staff calculations based on data available from Dominican Republic, Supérate; El Salvador, Dirección de Hidrocarburos y Minas; National Petroleum Authority, Indicative Prices; Indian Oil Corporation’s Indane Cooking Gas; Peru, Osinergmin; South Africa Depart- ment of Mineral Resources and Energy, Maximum Retail Price for Liquefied Petroleum Gas May 27, 2022; Thailand Energy Policy and Planning Office, Petroleum Prices; and exchange rates from the World Development Indicators. Note: The effective prices paid are for refilling the cylinder of the designated size. — = not applicable; n.a. = not available. a. Although the Bonogás program in the Dominican Republic takes the form of an unconditional cash transfer, it is intended to help house- holds refill LPG cylinders, typically 25 pounds (11.4 kg), and hence for illustrative purposes the cash transfer is applied to refilling a 11.4 kg cylinder. b. The cash transfer in El Salvador is independent of how much LPG is purchased but government publications use a 25-pound cylinder refill to show the effective end-user price paid. The effective price net of the cash transfer is zero for a 10-pound cylinder and only $0.10 per kg for a 20-pound cylinder. c. The cash transfer recipients are not confined to the poor. d. The unsubsidized price in India is the unit price of LPG sold in 19 kg cylinders, and prices are those in New Delhi. e. Peru limits the use of discount vouchers to cylinders no larger than 10 kg. f. The retail price in Peru is the average of all surveyed prices throughout the country. g. The South African retail price is the average of price ceilings per kg across the country in 54 regions. h. The unsubsidized price is back-calculated from the addition of the retail price in Bangkok and the universal price subsidy. i. This corresponds to the negative oil fund levy. j. The retail price is that in Bangkok for households. 26 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Effective prices paid for LPG Thailand. The difference of $0.99 per kg between the refinery and retail prices exclusive of a 15 percent in 2022 varied by an order value added tax signals very high costs for bottling, of magnitude across the transportation, and retailing in South Africa. countries studied Among the countries with cash transfers, the effec- Various forms of government support for household tive end-user prices paid in India and Peru were use of LPG lowered the effective end-user prices virtually identical and those in the Dominican Repub- paid for refilling LPG cylinders for the sizes shown to lic and Thailand were also broadly comparable; while as little as $0.27 per kg in June 2022—markedly lower that in El Salvador was much lower due to the com- than the benchmark free-on-board price of $0.75 per bination of a relatively large universal price subsidy kg in figure 6 before costs of transportation, stor- and cash transfer. There is a significant difference, age, bottling, and retail and taxes and fees are added however, between Thailand on the one hand and (table 7). Thailand had by far the lowest retail price, the Dominican Republic, El Salvador, and Peru on the on account of a high unit subsidy combined with other. In Thailand, the amount of LPG that can be exceptionally low costs of supply. While Thailand’s purchased at $0.61 per kg is unlimited, whereas in the ex-refinery LPG prices are in line with global prices, latter three countries, the monthly transfer amount the costs of bottling, transportation, and retail are is fixed: hence, the average unit price paid is a func- about $0.20 per kg, far below the costs in all other tion of how much LPG is purchased. countries. In South Africa, which has never subsi- dized LPG and historically has had relatively high The effective prices paid by households in various retail prices, the ex-refinery price of LPG in June 2022 countries as a function of monthly consumption was $1.04 per kg (South Africa Department of Energy of LPG are shown in table 8. Because monthly cash 2022)—noticeably higher than the $0.83 per kg in transfers are limited in absolute terms, the average Table 8  Effective prices paid in June 2022 and household income needed for cooking with LPG Unit price ($/kg) Minimum annual household income ($) Monthly consumption 10 kg 15 kg 20 kg 10 kg 15 kg 20 kg Dominican Republic 0.48 0.77 0.91 1,200 2,800 4,400 El Salvador 0.18 0.44 0.57 440 1,600 2,700 Ghana 1.50 1.50 1.50 3,600 5,400 7,200 India 0.72 0.73 0.78 1,700 2,600 3,700 Peru 0.76 0.98 1.10 1,800 3,500 5,300 South Africa 2.33 2.33 2.33 5,600 8,400 11,200 Thailand 0.62 0.62 0.62 1,500 2,200 3,000 Ratio of highest to lowest 13 5.3 4.1 13 5.3 4.1 Source: World Bank staff calculations based on data available from Dominican Republic, Supérate; El Salvador, Dirección de Hidrocarburos y Minas; National Petroleum Authority, Indicative Prices; Indian Oil Corporation’s Indane Cooking Gas; Peru, Osinergmin; South Africa Depart- ment of Mineral Resources and Energy, Maximum Retail Price for Liquefied Petroleum Gas May 27, 2022; Thailand Energy Policy and Planning Office, Petroleum Prices; and exchange rates from the World Development Indicators. Note: The calculations for the Dominican Republic assume that the monthly unconditional cash transfers are used to purchase LPG. The calculations for El Salvador are based on different size cylinders that deliver the lowest unit cost for the given monthly consumption. Mini- mum annual household income is rounded off to retain only two significant figures unless it exceeds $10,000 and is based on affordability limiting cooking energy costs to 5% of total household income. If the household is using electricity or other sources of energy for cooking in parallel, the minimum household income would be higher. Lessons from government support for LPG 27 price paid is a strong function of monthly LPG con- than on cooking energy. Households consuming sumption in the Dominican Republic, El Salvador, and only 10 kg of LPG a month in particular are likely to Peru. be using other forms of energy for cooking. These observations show that unless LPG is heavily subsi- The average prices paid by households in June dized, it is difficult for it to become the cooking fuel 2022 in the seven countries analyzed exhibited dif- of choice in many low- and lower-middle-income ferences of nearly an order of magnitude at low countries, either alone or together with electricity. consumption between the highest and lowest, making LPG affordable in some countries and a fuel Promotion of LPG among the poor is particularly for the better-off in others. If there are monthly cash challenging. In Ghana, one study found that only transfers that are not tied to the quantities of LPG 8 percent of households in the Rural LPG Promotion purchased—the Dominican Republic offers uncon- program that gave away free LPG stoves and cylin- ditional transfers and El Salvador provides $8.04 per ders were still using LPG after 18 months of receipt. month for any purchase of LPG without requiring a In India, one-sixth of rural households in six states minimum quantity for eligibility—the less LPG con- that had been registered for LPG for more than a sumed, the lower the effective price. In five of the year had never refilled cylinders. These patterns seven countries, the average prices paid by those are consistent with household spending priorities; consuming 10 kg a month were about the same as, specifically, household responses to surveys asking and in some cases markedly lower than, the inter- about spending priorities in India and Peru did not national free-on-board prices of LPG, indicating a mention using LPG for cooking among their top substantial unit subsidy.12 Households consuming priorities (Kojima 2021c). More recent data in India 20 kg a month to do all their cooking with LPG faced obtained by a right-to-information application filed prices exceeding free-on-board prices in all coun- against the three national oil companies adminis- tries except El Salvador and Thailand, confirming tering the PMUY scheme show that during the fiscal their large unit subsidies. year ending on March 31, 2021, 9 million of 80 mil- lion PMUY beneficiaries did not refill at all despite Depending on the degree of subsidy and how much the availability of three free refills during the first LPG is used, the minimum monthly income required nine months of the fiscal year; more than 10 million to make cooking with LPG affordable varied from $54 refilled only once. The average refill rate rose from for buying 10 kg to $930 for buying 20 kg in South 3.21 at the end of 2018 to 3.66 in March 2022; the Africa. The affordability threshold of less than 5 per- latter translates to monthly consumption of 4.3 kg cent spent on cooking needs to be interpreted with (India Ministry of Petroleum and Natural Gas 2022b; caution. If more than one type of energy is used—and Insights on India 2022). many households cooking with LPG use microwave ovens, electric kettles, or electric rice cookers—the amount spent on LPG needs to be less than 5 per- More can be done on the cent, especially given that household survey analysis regulatory and policy front shows households tend to spend more on electricity The preceding section points to how easily gov- ernments can fall into cycles of reform and policy Note that unconditional transfers are not subsidies by 12 reversals. Instead of directing efforts to artifi- definition; thus, Bonogás payments in the Dominican cially keeping end-user prices low, governments Republic do not constitute a subsidy but are captured in the calculated average unit prices. can achieve considerable long-term gains for both 28 Cooking with Bottled Gas: Issues and Challenges in Developing Countries consumers and the economy if they devote more LPG retailer to have a properly calibrated weigh- efforts to improving market conditions through reg- ing instrument. ulatory actions. Regulatory actions can help lower l Promote other measures that can reduce supply prices, increase safety, be cost-effective in facilitat- costs. Governments can do more to make the ing LPG adoption and use, and carry virtually no risk market as efficient as possible and promote of adverse unintended consequences. They can be competition so efficiency gains are passed on to especially helpful in the abandonment of polluting consumers in the form of lower prices. There is fuels by the better-off who have already adopted tangible evidence in Ghana that competition has LPG as their primary cooking fuel and who should increased after deregulation, as measured by a not be beneficiaries of government subsidies for decrease in the Herfindahl-Hirschman Index from cooking. Examples of useful regulatory action in this 4,008 in the year of deregulation to 1,143 five years regard follow: later (Kojima 2021c).13 To increase competition, governments can encourage hospitality arrange- l Set standards for thermal efficiency of LPG ments and third-party access to import terminals stoves and monitor and enforce them. Mexico and storage tanks, thereby reducing duplication tightened efficiency standards for LPG-based of infrastructure and lowering the barrier to entry. water heaters in 2011, and higher efficiency may Improving roads can reduce transport costs and account for the reduction in household use of enable more areas to be reached, while improv- LPG in subsequent years. In testing the efficiency ing ports and customs clearance could reduce of LPG stoves from different countries, Shen et congestion and demurrage charges. Better port al. (2018) found that efficiency varied with a dif- infrastructure may also facilitate LPG imports in ference of approximately 10 percent between the larger parcels, again lowering costs. Fair competi- most and the least efficient stoves. tion—essential for increasing efficiency—requires l Monitor and enforce rules against short-selling. establishing a modern regulatory framework, The weight of LPG contained in a cylinder is which may include formal adoption of interna- opaque in many markets, leading to the suspi- tional standards, and effective monitoring and cion that cylinders are underfilled. Short-selling of enforcement to curb commercial malpractice and LPG is difficult to detect at the point of purchase, ensure safety. Where institutional capacity is still but effective monitoring and enforcement can being developed for monitoring and enforcement, minimize this commercial malpractice. In many one option is to establish a system of certified markets, the tare weight of the cylinder may not installers and private inspectors under govern- be clearly marked—and in any event, unless the ment supervision. customer can weigh the filled cylinder before pur- l Consider requiring publication of prices in real chase, the amount of LPG in the cylinder is difficult time. Chile requires such a system for automotive to verify. Even if the cylinder weighs exactly as fuels to promote price competition. Resolution required, it may be filled partially with substances No. 60 issued in January 2012 requires each filling other than LPG (such as sand). It is no coinci- dence that Brazil has a law on LPG devoted solely to enhancing transparency in weighing cylinders, The Herfindahl-Hirschman Index is a commonly accepted 13 starting with the requirement that the cylinder metric for market concentration. Broadly, a market with tare weight be clearly and visibly marked. Kenya an index of less than 1,500 is considered competitive, an index between 1,500 and 2,500 is considered moderately in 2009 issued LPG regulations requiring every concentrated, and an index greater than 2,500 is consid- ered highly concentrated. Conclusions and implications 29 station to enter price information in a national providing basic training on safety features of LPG— online price database, which is published on the especially among women—might be effective in regulator’s website and is accessible on mobile shifting households away from solid fuels to LPG. devices. The website displays minimum and max- imum prices, the address of each filling station, prices, and the date and time of the last price Conclusions and implications change; retailers are required to enter new prices no earlier than 15 minutes before price changes are implemented. This study has reviewed numerous challenges as l Monitor and enforce safety measures. Fears of well as opportunities for clean cooking with a focus fires and explosions are frequently cited as one on LPG. Although both electricity and clean cooking of the main reasons households are reluctant to energy are covered by the goal for universal access, use LPG. Ensuring safety calls for a clear defini- there are some important differences. There are no tion of cylinder ownership; assignment of legal good alternatives to electricity—kerosene lamps responsibility for cylinder maintenance, repair, deliver far fewer lumens than electric light bulbs, and replacement; effective enforcement of the and kerosene cannot be used to charge cell phones ban on cross-filling where such a ban exists; or power television sets. By contrast, several differ- proper training of operators throughout the ent fuels, even those that are considered polluting, supply chain; extensive education campaigns for can cook a meal in a way that is indistinguishable end users; and penalizing companies that refill for those eating it. Many would even argue that for unsafe cylinders. Turkey requires training of all certain dishes, those made over wood taste better, personnel involved in supplying LPG and educat- as witnessed by restaurants in high-income coun- ing consumers about proper handling of LPG. The tries advertising pizzas and bread baked in wood government sets strict rules about the conditions ovens as a selling point. For these and other reasons, under which cylinders can be refilled, and even households do not have as much incentive to pay authorizes charging of small fees to marketing more for clean cooking energy, and surveys indeed companies to finance monitoring and enforce- show that households prioritize electricity markedly ment (Matthews and Zeissig 2011). above cooking energy. l Work with industry to raise awareness and address the public’s concerns. An often over- The budgetary allocation needed to overcome looked factor in the use of LPG is the education the greatest barrier to cooking exclusively or even level of household members. A study using a mostly with gas and electricity—the high cost of two-stage Heckman model to examine the fac- doing so relative to income—would be unaffordable tors influencing adoption and consumption of to all governments in developing countries. Targeted LPG found that, aside from income, the decision cash transfers in a market with deregulated pricing or, to use LPG increased with the highest level of absent adequate competition in the market, prices education attained by female and male house- set at cost recovery levels under efficient opera- hold members (Kojima, Bacon, and Zhou 2011). tions seem to be the only sustainable path over the Education is likely to be a proxy for the level of long run. The global fossil fuel price movements in awareness about the benefits and costs of LPG. the last two decades have shown that international In persuading households to start using LPG, rais- market prices can rise and fall unexpectedly and ing awareness about the benefits of its use and price shocks can last over prolonged periods of time. 30 Cooking with Bottled Gas: Issues and Challenges in Developing Countries However expedient administratively, governments that 90 million poor households have newly gained need to resist the temptation to bring back univer- access to LPG thanks to the PMUY, and cylinder refill sal price subsidies whenever prices keep on rising for rates have been slowly but steadily rising rather than months or even years on end. The temptation to do stagnating or falling as earlier feared. That many of so is strong, especially in the face of other crises, such these households did not take advantage of three as high inflation causing prices of other basic goods free refills in 2020 raises questions about what other and services to rise steeply, a recession, or a natural barriers there may be. Possible explanations include disaster. The proper response to deal with multiple difficulties in obtaining refills and high transaction or acute crises is to expand social safety nets and costs of doing so (such as having to travel long dis- focus on increasing income support to households, tances to get to the closest LPG retail outlet), a lack not engage in interventions compensating sector by of awareness about free refills, late or inadequate sector for higher fuel, electricity, transport, food, and cash transfers, and the inability to come up with other costs. Sectoral interventions are inefficient cash for repayment—linked to each refill—of loans and carry higher risks of economic distortions. The offered by the national oil companies and taken out pressure to keep fuel prices low affects both univer- to buy LPG stoves. sal price subsidies and targeted cash transfers, and further points to the political difficulties of phasing Difficulties with refills are frequently encountered out fuel-specific subsidies. because of the time it takes to turn the cylinders around. Consideration could be given to enabling One way of slowly phasing out sectoral interven- lower-income households to have two cylinders tions is to fix financial support in nominal terms and at a time so when one is emptied, the household allow inflation to erode the value. Peru had fixed the can continue to cook with the second one until the value of the LPG discount voucher until January 2021, first can be refilled. The implicit risk in this plan is when the government increased the voucher value the household’s transferring the second cylinder to three times in succession. a third party for cash. Maintaining an accurate and up-to-date database—which is important under all Another means of limiting sectoral interventions is circumstances—would help minimize such abuse. to shift to unconditional targeted cash transfers. The Dominican Republic’s Bonogás program, although It takes a long time to develop a market with healthy closely tied to LPG, is not a sectoral intervention competition in which safety and other rules are because it is unconditional. Such cash transfers can enforced across all sellers. Taking regulatory, mon- be scaled up or down depending on the nature and itoring, and enforcement measures to create such magnitude of the economic shock, irrespective of a market is arguably the most important and cen- the sector in which it occurs, and cover the entire tral role of any government. Punishing those who vulnerable population. underfill cylinders and setting and enforcing stove efficiency standards can arguably help lower the Targeted cash transfers are mostly aimed at helping effective prices paid by consumers more than any the poor and near-poor, and are unlikely to realize other steps. All governments can—and should— abandonment of polluting fuels. That said, the shift redouble their efforts in these areas to the benefit to clean household energy starts with its adoption, of society at large. however limited the initial use. Data from India show Appendix Implementation of cash transfers T his appendix provides details on cash trans- websites available for any fuel subsidy program in fer programs associated with LPG in four the world.14 Using interactive data visualization soft- countries. ware, the website enables the viewer to pull up any data by type, region, and date. The database pro- vides comprehensive spatial and temporal coverage, Peru enabling public assessment of the overall program since inception. Figure A.1 shows some of the data In July 2012, Peru launched a tightly targeted con- available on the FISE website. The total cost to the ditional cash transfer program under the Fondo de government from the inception of FISE to October Inclusión Social Energético (FISE, Energy Social Inclu- 2022 was S/. 1.36 billion (about $405 million). sion Fund). The program’s eligibility criteria are very restrictive: In addition to being classified as poor The eligibility criteria for FISE preclude most house- or extremely poor by the government, households holds. In July 2021, the government expanded program must live in an area where natural gas is not available coverage by ending the requirement to register in the but LPG is commercially available, and—if connected Household Targeting System. Nevertheless, the ceil- to the electricity grid, which most households in ing on electricity consumption is very limited by any Peru are—its electricity consumption in the preced- standards and particularly disadvantages multifam- ing 12 months could not exceed 360 kWh annually ily households and those living in areas requiring (or 30 kWh a month on average) until July 2021, when greater heating. If the ceiling is breached, the bene- the annual limit was increased to 504 kWh (42 kWh a ficiary is no longer eligible to receive the LPG price month). The government issues LPG discount vouch- discount voucher until the limit on electricity con- ers, now digitized, and LPG sellers are reimbursed by sumption is met again. Figure A.1 shows that the cash transfers to their bank accounts. Vouchers are number of vouchers redeemed has been consistently valid for 60 days, and the amount has been increased lower than the number of active (eligible) beneficia- four times—although when converted to constant ries with the exception of May 2020; this can happen U.S. dollars, the increases have not compensated presumably because vouchers are valid for 60 days for currency depreciation and inflation. The fourth and some redeem two vouchers at a time. After the increase in April 2022 was to last six months, but had relaxation of eligibility criteria in July 2021, there was been extended to last through the calendar year. a noticeable increase in the number of active benefi- ciaries in the following months. The government of Peru has been highly transpar- ent in making data on cash transfers available to the public, maintaining one of the most comprehensive FISE website, Gas doméstico (GLP). 14 31 32 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Figure A.1  Registered FISE beneficiaries and the number of redeemed LPG discount vouchers in Peru 2.0 Beneficiaries and vouchers (millions) 1.5 1.0 0.5 0.0 Jul-12 Jan-13 Jul-13 Jan-14 Jul-14 Jan-15 Jul-15 Jan-16 Jul-16 Jan-17 Jul-17 Jan-18 Jul-18 Jan-19 Jul-19 Jan-20 Jul-20 Jan-21 Jul-21 Jan-22 Jul-22 Active beneficiaries Excluded Suspended Redeemed vouchers Source: Adapted from FISE website, Gas doméstico (GLP). Note: Excluded beneficiaries are presumably those who failed to satisfy all the eligibility criteria, such as the ceiling on the average monthly electricity consumption. Government documents refer to suspension for causes such as transferring the discount vouchers to third parties. Dominican Republic electricity, school, and food programs particularly targeting those classified as extremely poor. The The Dominican Republic was one of the first coun- Bonogás Hogares program reached some 730,000 tries in the world to end a costly universal price people in the first year, rising to 940,000 people by subsidy for LPG and replace it with cash transfers, the beginning of 2022. launching a targeted unconditional cash transfer program in September 2008 as a component under Although the name of the program refers to gas and its umbrella Solidarity Program. The Bonogás Hoga- the goal is to enable the poor to cook with LPG, the res (household LPG token or certificate) program Bonogás program provides a fixed amount of cash provides a fixed amount of cash to eligible house- to eligible households without requiring them to buy holds in extreme and moderate poverty without LPG. The amount of cash transferred was fixed at requiring them to buy LPG. Before launching Bonogás RD$228 (about $6.60 in 2008) for each eligible house- Hogares, the government considered, but eventually hold from the inception of the program to March abandoned as impractical, a proposal to deliver an 2022, when it was more than doubled to RD$470 LPG price subsidy only to the poor. ($8.56). In the same month, the government added about 260,000 more beneficiaries. A transparency The government issues two types of cards for poor website posts data about the execution of Bonogás families identified in the Census of the Unique Hogares and other government programs. The gov- System of Beneficiaries. Ten social assistance pro- ernment transferred a total of about $760  million grams are implemented through the two cards, one under the Bonogás program through October 2022 of which covers the Bonogás program along with (figure A.2). Appendix 33 Figure A.2  Beneficiaries of Bonogás Hogares and cash disbursed in the Dominican Republic 6.0 90 80 Transferred amount ($ millions) 5.0 70 Beneficiaries (millions), amount (RD$ billions) 4.0 60 50 3.0 40 2.0 30 20 1.0 10 0.0 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Sources: Dominican Republic, Supérate Portal de Transparencia website, Subsidio Aliméntate y Bonogás a miembros de hogares, and previ- ously posted government data. Note: 2022 is up to October. El Salvador The political economy of the LPG subsidy reform in 2011 initially defied expectations. The reform El Salvador eliminated its LPG price subsidy in 2011 by made the assistance substantially less regressive launching an unconditional cash transfer program, and was expected to improve the welfare of about which it then made conditional in 2014. The maxi- three-quarters of the population. The poor, some of mum monthly electricity consumption allowed at whom had not been purchasing LPG, received $8.50 the time of the launch was 99 kWh for the full pay- a month at first, rising to $9.10 per month and main- ment and 199 kWh for half payment (Calvo-Gonzalez, tained at that level for 30 months from June 2011 to Cunha, and Trezzi 2015; Ríos Quiñónez 2012). The eli- December 2013. Yet these “winners” of the subsidy gibility criteria are generous for households and reform initially opposed it. In February 2011, the Arch- cover the nonpoor; the scheme also provides for bishop of San Salvador voiced his concern that the cash transfers to institutions and businesses using poor might be left out. A survey in January 2011 about LPG for cooking. Unlike the Dominican Republic, the the proposed subsidy reform found an approval rate government in earlier years varied the monthly cash of only about 30 percent, including from among the transfer amount almost every month to keep the bottom 40 percent of the income distribution. Once cost of refilling a 25-pound (11.4 kg) cylinder constant people began receiving cash transfers, however, the net of the cash transfer. More recently, the govern- satisfaction rate rose steadily, increasing to 66 per- ment has frozen the monthly cash transfer amount cent by August 2012. for many months at a time. The number of beneficia- ries has not varied significantly, remaining at a little A detailed analysis of the changes in the satisfaction more than 1 million (see figure A.3). rate with the targeted unconditional cash trans- fers found that an individual’s knowledge of the reform, trust in the government’s ability to deliver, 34 Cooking with Bottled Gas: Issues and Challenges in Developing Countries Figure A.3  Beneficiaries and cash disbursed for LPG purchase in El Salvador 14 1.4 Disbursement ($, millions) 12 1.2 Beneficiaries (millions) 10 1.0 8 0.8 6 0.6 4 0.4 2 0.2 0 0.0 Jan-14 Jul-14 Jan-15 Jul-15 Jan-16 Jul-16 Jan-17 Jul-17 Jan-18 Jul-18 Jan-19 Jul-19 Jan-20 Jul-20 Jan-21 Jul-21 Jan-22 Jul-22 Total disbursement Beneficiaries Source: Adapted from El Salvador Ministry of the Economy, Subsidios e incentivos fiscales. Note: The number of beneficiaries is not consistently available at the government’s transparency website. and political views explained on average about India 70 percent of the survey responses. The level of sat- isfaction rose primarily because the government The rollout of the Direct Benefit Transfer for LPG had delivered as promised (Calvo-Gonzalez, Cunha, (DBTL; later renamed PAHAL) in India arguably rep- and Trezzi 2015). resented the most ambitious attempt to promote household use of LPG using cash transfers; it soon The cash transfers were modified and made condi- became the largest cash transfer program in the tional beginning in 2013, and the conditional cash world. At its launch in June 2013, the DBTL was poten- transfer program became fully operational in 2014. tially open to all households; they had only to link The subsidy was delivered through the Solidarity their bank accounts and Aadhaar (unique biometric Card to eligible beneficiaries, who would receive a identification) numbers to LPG consumer identifica- price discount equal to the monthly subsidy amount tions. Throughout implementation, there was a limit per household. The Solidarity Card enabled consol- on the number of DBTL-eligible refills a year, initially idation of several benefits in a single instrument. 9 but raised eventually to 12 by April 2014. Because Although the eligibility criteria did not change, making the Aadhaar penetration rate was low, the DBTL was cash transfers conditional reduced the percentage suspended in March 2014; it resumed in January 2015, of households claiming benefits by 12  percentage waiving the requirement to link beneficiaries to Aad- points. Since 2014, the number of beneficiaries has haar until December 2016. remained between about 1.1 million and 1.2 million (figure A.3). The monthly amount was last changed In January 2016, households whose beneficiary and in February 2021 from $5.00 to $8.04. From the pro- spouse had a combined taxable income of more gram’s inception to October 2022, the government than Rs. 1 million (nearly $15,000 at the time) in the had spent $1.0 billion on the cash transfers. previous year were deemed ineligible. Although obtaining accurate information about taxable income in India has been a challenge, this was the only targeting mechanism in the DBTL until 2022. To Appendix 35 reduce the fiscal burden, the government in March Figure A.4  Number of DBTL beneficiaries 2015 launched the Give-It-Up campaign in which 300 well-off households were encouraged to opt out of the DBTL, with the option of reapplying after a hiatus 250 of a year. By March 2020, more than 10 million house- 200 Millions holds had voluntarily given up the subsidy, against 150 263 million households in the DBTL and 1.3 trillion rupees (about $20  billion) transferred.15 The cam- 100 paign notwithstanding, the number of households 50 enrolled in the DBTL steadily rose until 2022, albeit 0 at a declining rate (figure A.4). When the global oil April April April April April 2018 2019 2020 2021 2022 price collapsed in 2020, the conditional cash trans- Source: India Ministry of Petroleum and Natural Gas 2022a. fers under the DBTL were suspended. They were resumed in 2022 only for beneficiaries of Pradhan Mantri Ujjwala Yojana (PMUY), a conditional cash Figure A.5  Number of refills by PMUY beneficiaries by fiscal year transfer program launched in May 2016 to cover the cylinder deposit fee and restricted to poor women; 80 there were 95.4 million beneficiaries in October 2022 FY 2020–21 FY 2021–22 60 (PPAC 2022). Millions In response to the financial hardships caused by the 40 COVID-19 pandemic, the government authorized 20 up to three free refills for PMUY beneficiaries from April 1 to December 31, 2020, resulting in 141.7 million 0 free refills, or an average of 15.7 million free refills a 1 refill 2 refills 3 or more refills month (Insights on India 2022). The free refills appear Source: Moneylife 2022. to have increased LPG consumption; conversely, Note: FY = fiscal year, running from April 1 to March 31. their withdrawal at the end of 2020 appeared to have decreased consumption. Compared to fiscal For the DBTL, the Comptroller and Auditor General year 2020–21 (ending on March 31, 2021), the number collected data from January to October 2015 (India of PMUY beneficiaries refilling three times or more in Comptroller and Auditor General 2016). Published in fiscal year 2021–22 declined by 10.4 million, and those March 2016, the report found multiple registrations refilling only once increased by 9.5 million (figure A.5). by the same beneficiaries in various combinations; cash transfers exceeding 12 times a year, aided by The government of India proactively identified and missing or clearly incorrect key information in the tackled teething problems in rolling out the DBTL database; inadequate documentation for block- and PMUY by commissioning the Comptroller and ing and unblocking beneficiaries, raising the specter Auditor General of India to carry out two in-depth of ineligible customers being unblocked or eligible audits. beneficiaries continuing to be blocked; more than 10 million LPG customers being deterred from regis- tering for the DBTL for a lack of knowledge and the Government of India Ministry of Petroleum and Natural 15 lengthy enrollment process; failure of cash transfers Gas website, PAHAL and Give It Up webpages. 36 Cooking with Bottled Gas: Issues and Challenges in Developing Countries to some beneficiaries because of inaccurate data not being met. It found numerous cases of cylin- entries by the LPG distributors; and failure by the gov- der deposit fees released with inadequate or invalid ernment to reimburse the oil marketing companies documentation or to underage women or to men; on time for the cash transfers. The government had exceptionally long periods between enrollment and set up a complaint mechanism offering a toll-free LPG cylinder delivery, far in excess of the seven days number and a web-based portal, which was deemed permitted; many customers not living within the to be working well. The audit recommended taking stipulated limit of 15 km from distributors and some more steps to improve the accuracy of the cus- as far away as 92 km with no home delivery; delays tomer database, make the process of blocking and of more than 10 days and as long as 664 days for a unblocking more transparent, and inform the public refill against the limit of 7 days; a declining refill rate to increase participation in the DBTL. The national oil from 2015 to 2019 among the beneficiaries, with companies worked to centralize and clean up data- one-third of those who had had LPG for at least a bases for greater accuracy. year having refilled only three times or less; about 200,000 beneficiaries refilling more than 12 times To raise awareness, the government in 2017 launched a year; and two-thirds of those who had taken out a series of community meetings that brought interest-free loans to buy LPG stoves and pay for the together about 100 LPG customers and served as first refill struggling to pay back the loans, even when a platform for them to interact, promote mutual spread over six refills. The performance review rec- learning, and share experiences; the three national ommended full and accurate documentation, safety oil companies had organized 105,123 such meetings campaigns, and monitoring of high-consumption across India by the end of March 2019 (India Comp- households (India Comptroller and Auditor General troller and Auditor General 2019). 2019). The number of LPG distributors increased by more than 40 percent between 2016 and October The audit on the PMUY covered the period from May 2022, presumably helping to ease delays in fulfilling 2016 to December 2018 and found ineligible bene- refill orders (PPAC 2022). fits being conferred and service delivery standards References Bhatia, Mikul, and Niki Angelou. 2015. “Beyond ESMAP (Energy Sector Management Assistance Pro- Connections: Energy Access Redefined.” ESMAP gram). 2020. Cooking with Electricity: A Cost Technical Report 008/15. World Bank, Washing- Perspective. Washington, DC: World Bank. ton, DC. FAO (Food and Agriculture Organization of the Bladi.net. 2022. “Maroc : le prix du gaz butane reste United Nations). 2022. The State of the World’s inchangé depuis plus de 30 ans.” February 18. Forests 2022. Forest Pathways for Green Recovery and Building Inclusive, Resilient and Sustainable Calvo-Gonzalez, Oscar, Barbara Cunha, and Riccardo Economies. Rome: FAO. Trezzi. 2015. “When Winners Feel Like Losers: Evidence from an Energy Subsidy Reform.” Policy Fedak, Kristen M., Nicholas Good, Ethan Walker, Research Working Paper 7265. World Bank, Maggie L. Clark, Christian L’Orange, John Volck- Washington, DC. ens, and Jennifer L. Peel. 2019. “An Expert Survey on the Material Types Used to Start Cookstoves.” Carter, Ellison Milne, Ming Lei Shan, Yuan Zhong, Wei- Energy for Sustainable Development 48: 59–66. meng Ding, Yichen Zhang, Jill Baumgartner, and Xudong Yang. 2018. “Development of Renew- Ghana Ministry of Energy. 2022. “Medium Term able, Densified Biomass for Household Energy in Expenditure Framework for 2022 - 2025.” Accra. China.” Energy for Sustainable Development 46: 42–52. Gitau, K.J., J. Mutune, C. Sundberg, R. Mendum, and M. Njenga. 2019. “Factors Influencing the Adoption Champion, Wyatt M., and Andrew P Grieshop. 2019. of Biochar-Producing Gasifier Cookstoves by “Pellet-fed Gasifier Stoves Approach Gas-Stove Households in Rural Kenya.” Energy for Sustain- Like Performance during In-Home Use in able Development 52: 63–71. Rwanda.” Environ. Sci. Technol. 53: 6570–79. IEA, IRENA, UNSD, World Bank, and WHO (Interna- Clark, S., E. Carter, M. Shan, K. Ni, H. Niu, J.T.W. Tseng, tional Energy Agency, International Renewable S.K. Pattanayak, M. Jeuland, J.J. Schauer, M. Ezzati, Energy Agency, United Nations Statistics Division, C. Wiedinmyer, X. Yang, and J. Baumgartner. 2017. World Bank, and World Health Organization). “Adoption and Use of a Semi-Gasifier Cooking 2022. Tracking SDG 7: The Energy Progress Report. and Water Heating Stove and Fuel Intervention Washington, DC: World Bank. in the Tibetan Plateau, China.” Environmental Research Letters 12 (7): 075004. India Comptroller and Auditor General. 2016. “Report of the Comptroller and Auditor General of India Dickinson, Katherine L., Ricardo Piedrahita, Evan R. on Implementation of PAHAL (DBTL) Scheme Coffey, Ernest Kanyomse, Rex Alirigia, Timothy (Pratyaksh Hanstantrit Labh Yojana) for the Molnar, Yolanda Hagar, Michael P. Hannigan, Abra- period ended 31 March 2016.” Union Government ham Rexford Oduro, and Christine Wiedinmyer. (Commercial) Ministry of Petroleum and Natural 2019. “Adoption of Improved Biomass Stoves and Gas Report No. 25 of 2016 (Compliance Audit). Stove/Fuel Stacking in the REACCTING Interven- tion Study in Northern Ghana.” Energy Policy 130: 361–74. 37 38 Cooking with Bottled Gas: Issues and Challenges in Developing Countries India Comptroller and Auditor General. 2019. “Report Kojima, Masami, and Xin Zhou. 2022. “Household of the Comptroller and Auditor General of India Use of Bottled Gas for Cooking: Evidence from on Pradhan Mantri Ujjwala Yojana.” Union Gov- Sub-Saharan Africa.” Policy Research Working ernment (Commercial) Ministry of Petroleum Paper 10089. World Bank, Washington, DC. and Natural Gas Report No. 14 of 2019 (Perfor- mance Audit). Kojima, Masami, Xin Zhou, Jace Jeesun Han, Joeri de Wit, Robert Bacon, and Chris Trimble. 2016. “Who India Ministry of Petroleum and Natural Gas. 2011. Uses Electricity in Sub-Saharan Africa? Findings “LPG Gas Stoves.” Press release. December 1. from Household Surveys.” Policy Research Work- ing Paper 7789. World Bank, Washington, DC. India Ministry of Petroleum and Natural Gas. 2022a. “The Government Continues to Modulate the Lucon, Oswaldo, Suani Teixeira Coelho, and José Effective Price to Consumer for Domestic LPG.” Goldemberg. 2004. “LPG in Brazil: Lessons and Press release. August 22. Challenges.” Energy for Sustainable Development 8 (3): 82–90. India Ministry of Petroleum and Natural Gas. 2022b. “LPG Connections under PMUY.” Press release. Masera, O.R., B.D. Saatkamp, and D.M. Kammen. 2000. July 21. “From Linear Fuel Switching to Multiple Cook- ing Strategies: A Critique and Alternative to the Insights on India. 2022. “Insights into Editorial: Energy Ladder Model.” World Development 28 Ujjwala Scheme | 90 Lakh Beneficiaries Do Not (12): 2083–2103. Get Refills.” Matthews, William G., and Hilmar R. Zeissig. 2011. Kojima, Masami. 2011. “The Role of Liquefied Residential Market for LPG: A Review of Experience Petroleum Gas in Reducing Energy Poverty.” of 20 Developing Countries. Washington, DC: Extractive Industries for Development Series World Bank. No. 25. World Bank, Washington, DC. Moneylife. 2022. “LPG: 2.11 Crore Customers Did Not Kojima, Masami. 2017. “Energy Subsidies: Identifying Buy a Single Refill in FY22, Says Govt.” News arti- and Quantifying Energy Subsidies.” Energy cle. July 25. Subsidies Good Practice Note 1. World Bank, Washington, DC. Myhre, G., D. Shindell, F.-M. Bréon, W. Collins, J. Fuglest- vedt, J. Huang, D. Koch, J.-F. Lamarque, D. Lee, B. Kojima, Masami. 2021a. “Comparison of Bottled Gas Mendoza, T. Nakajima, A. Robock, G. Stephens, T. and Advanced Combustion Pellet Stoves.” Tech- Takemura, and H. Zhang. 2013. “Anthropogenic nical paper. World Bank, Washington, DC. and Natural Radiative Forcing.” In Climate Change 2013: The Physical Science Basis, edited Kojima, Masami. 2021b. “Primary Household Energy by T.F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S.K. for Cooking and Heating in 52 Developing Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and Economies.” Technical paper. World Bank, Wash- P.M. Midgley. Contribution of Working Group I ington, DC. to the Fifth Assessment Report of the Intergov- Kojima, Masami. 2021c. “Subsidizing Bottled Gas: ernmental Panel on Climate Change. New York: Approaches and Effects on Household Use.” Cambridge University Press. Technical paper. World Bank, Washington, DC. Ochieng, Caroline A., Yabei Zhang, John Kennedy Kojima, Masami, and Robert Bacon. 2001. “Abuses in Nyabwa, Don Ivan Otieno, and Charles Spillane. Fuel Markets: How to Protect Consumers and 2020. “Household Perspectives on Cookstove Public Health.” Viewpoint 237. World Bank, Wash- and Fuel Stacking: A Qualitative Study in Urban ington, DC. and Rural Kenya.” Energy for Sustainable Devel- opment 59: 151–159. Kojima, Masami, Robert Bacon, and Xin Zhou. 2011. “Who Uses Bottled Gas? Evidence from Households in Developing Countries.” Policy Research Working Paper 5731. World Bank, Wash- ington, DC. References 39 Osinergmin (Organismo Supervisor de la Inversión en South Africa Department of Energy. 2022. “Breakdown Energía y Minería). 2019. “Fondo de Estabilización of Petrol, Diesel and Paraffin Prices as on 01 June de Precios de los Combustibles derivados del 2022.” Petroleum Sources website, Department Petróleo: el caso del GLP envasado.” Reporte de of Mineral Resources and Energy. Análisis Económico Sectorial Sector Hidrocarbu- ros Líquidos Año 8 – Nº 12 – Abril 2019, Gerencia UN (United Nations). 2015. Transforming Our World: de Políticas y Análisis Económico. The 2030 Agenda for Sustainable Development. New York: UN. PPAC (Petroleum Planning and Analysis Cell). 2022. “Quarterly LPG Profile Report.” India Ministry of Victor, David. 2010. The Politics of Fossil Fuel Subsidies. Petroleum and Natural Gas. GSI series, “Untold Billions: Fossil-Fuel Subsidies, Their Impacts and the Path to Reform.” Geneva: Ríos Quiñónez, Eva María. 2012. “Gobernabilidad Global Subsidies Initiative, International Insti- Democratica en el Salvador Durante las tute for Sustainable Development. Gestiones Gubernamentales del Periodo 2004-2012 a Traves de la Cooperacion WHO (World Health Organization). 2014. WHO Internacional Multilateral desde un enfoque de Guidelines for Indoor Air Quality: Household Fuel Desarrollo Humano. Perspectivas.” Thesis for a Combustion. Geneva: WHO. bachelor’s degree in international relations, Uni- WHO (World Health Organization). 2022. “Household versidad de El Salvador, San Salvador. air pollution and health.” Shen, Guofeng, Michael D. Hays, Kirk R. Smith, Craig Widjaja, Muliadi. 2009. “An Economic and Social Williams, Jerroll W. Faircloth, and James J. Jetter. Review on Indonesian Direct Cash Trans- 2018. “Evaluating the Performance of Household fer Program to Poor Families Year 2005.” Paper Liquefied Petroleum Gas Cookstoves.” Environ- presented at Association for Public Policy Anal- mental Science & Technology 52 (2): 904–15. ysis and Management International Conference, Singh, D., S. Pachauri, and H. Zerriffi. 2017. “Environmental “Asian Social Protection in Comparative Perspec- Payoffs of LPG Cooking in India.” Environ. Res. Lett. tive,” Singapore, January 7–9, 2009. 12: 115003.