Increase in population cooking with clean fuels and other improved technologies due to policy action
For TS1, which considers policies promoting clean fuel use by all households in these 120 countries where cooking with polluting fuels is currently widespread, we find that the share of households expected to use any clean fuel increases from the current level just below 50% to between 65 and 80% after scale up is achieved, depending on the policy package (Fig. 2). Although aiming for a complete shift of the population, this estimated increase is based on model assumptions included to reflect realistic uptake and use of improved options (note that regarding the latter, the model assumes a 48% use rate of new technologies to account for stove and fuel stacking), as supported by evidence from peer-reviewed evaluations. For TS2 and TS3, the shares of households using any clean fuel following policy intervention range between 59 and 67% (and the increase in other improved technology use ranges between 4 and 10%), and 55 and 60% (increase in other improved: 6–15%), respectively.
We determined the most cost-beneficial intervention in each transition, defined as that with the highest NPV, in each of the 120 countries. For the majority of these LMICs, and under most transitions, the “best” intervention combined a stove subsidy and financing, because this option increases affordability of new stoves but avoids the higher recurring costs of fuel subsidies (which, as noted above, do not change the extent to which a stove is used after acquisition). In a few countries (for example in Niger) and for some transitions, stove subsidy on its own was most cost-beneficial. Importantly, the highest NPV policies are not necessarily those that lead to the highest shares of people adopting clean or improved options in all countries, but stove subsidy plus financing is predicted to raise the share of households using improved or clean solutions in these countries by 25–32 percentage points, which represents more than half of the global access gap. In what follows, we focus on the results pertaining to the most cost-beneficial interventions supporting each transition scenario.
Net benefits of the most cost-beneficial policy options
The highest net benefits occur under the most ambitious transition scenario, TS1, reaching more than US$1.4 trillion over the 2020–2050 period. The NPV is about 30% lower for TS2 (US$943 billion) and 50% lower (US$656 billion) in TS3 (Fig. 3). Net benefits are considerably higher in rural areas than in urban locations, reflecting the relatively lower current penetration of cleaner cooking in rural locations. By construction, TS1 and TS2 yield identical results in urban areas, generating about $196 billion in net benefits; this declines to US$164 billion in TS3.
Populous countries with relatively low access to clean cooking technology have a disproportionate influence on these results. In particular, across rural and urban areas and transition scenarios, China and India contribute the most to overall NPV. In TS1, net benefits in China reach $887 billion (rural) and $117 billion (urban), or 69% of the global NPV; and India contributes $292 billion (rural) and $6 billion (urban), or 21% of the global NPV. Country-specific results appear in Appendix Table D1.
The total net benefits for the most socially beneficial policies thus also vary considerably across regions; we present regional breakdowns – using WHO region definitions – on a per capita basis to provide more comparable figures. This region-wise breakdown for the most socially beneficial policies reveals that implementing the policies with highest NPV for each country yields positive net benefits in all regions. Rural net benefits per capita are highest in the Western Pacific Region (WPR) (Fig. 4). In urban locations, per capita NPV for the most beneficial policies remains highest in WPR, followed closely by Africa (AFR). The Eastern Mediterranean Region (EMR) and SEAR have the lowest per capita NPV in the urban transition scenarios, owing to these regions’ already high use of clean fuels in urban locations.
Annualized costs and benefits and implications for investment
As noted above, the net benefits per capita are positive in all regions for the most beneficial policies. For driving investment at a global scale, it is useful to also consider their annualized costs, which represent the overall burden to governments and donors as well as households. We refer to household costs as private costs, since these accrue to those adopting the improved technology. Mirroring the results in Fig. 3, the total annualized net benefits in urban and rural areas are highest for TS1 (US$28 billion/yr) and lowest for TS3 (US$13 billion/yr) (Fig. 5), and are much higher in rural settings for all scenarios. Importantly, total costs (government and private) are highest in TS1 ($3 billion), which also produces the highest net (and total) benefits. Indeed, substantial investment is required to achieve the most ambitious transition to all clean solutions; this is a major challenge considering the low income of households and governments in the most affected countries.
In all three transition scenarios, across rural and urban areas, WPR contributes the largest share (59–61%) of global benefits, followed by SEAR (22–27%) (Supplementary Figures A1-A3). The Americas (AMR) contribute the least (1–2%) to global benefits. In terms of global costs (and excluding private net fuel costs, which are sometimes negative), SEAR’s contribution is highest (18–39%), followed by AFR (24–37%) and WPR (24–31%). Thus, investment needs are greatest in the regions with the largest number of people living in poverty globally. For TS1, there are significant private fuel savings among urban users in AFR, owing to high dependence of urban Africans on relatively costly and inefficient charcoal fuel.
Categories of costs and benefits
We next disaggregate these annual costs and benefits into their specific components; the benefits of the most socially beneficial policies in rural areas are principally driven by mortality reductions (US$25 billion/yr). Avoided CO2 equivalent emissions (~ US$1.6 billion/yr) provide the second largest contribution to benefits (Fig. 6). Other benefit categories – morbidity reductions, time savings, and forest preservation – represent a smaller fraction of the total benefits. The results for urban areas are proportionally similar, but lower in magnitude.
Moreover, for TS1 in rural areas, while yearly mortality benefits and yearly time savings are highest in WPR ($16 billion and $0.5 billion, respectively), SEAR provides the highest value of yearly CO2 equivalent emissions avoided ($0.7 billion) (Supplementary Figure A4). Region-wise results for TS1 in urban locations are slightly different. Yearly mortality benefits and yearly time savings are still highest in WPR ($2 billion and $161 million, respectively), but AFR contributes the highest value of yearly CO2 equivalent emissions avoided ($147 million). Similar regional patterns are found for TS2 and TS3, with relative shares growing slightly in AFR (Figures A6 and A7).
The regional shares of global costs and benefits are partly dependent on the populations of the included countries; decision-makers seeking to compare such numbers with those pertaining to other potential investments may also want to understand them in per capita terms. These per capita net benefits (ranging from $0.5/yr to $22.3/yr for TS1, and somewhat less for TS2 and TS3) – calculated over the entire population in these 120 LMICs and in each global region – are presented in Supplementary Figures A7-A10. The differences across regions are somewhat reduced when accounting for their differing populations.
Furthermore, the specific valuations of benefits – particularly health and time savings – that are included impose weights on non-monetary indicators of benefits, which are directly related to economic development. For example, the value of a statistical life is strongly dependent on income16, and is used to translate avoided deaths into a value of reduced mortality, while time savings are valued at a fraction of local wage rates.17 Thus, the higher benefits projected in the WPR region, and for rural mortality overall, are largely driven by the higher incomes in that region, relative to SEAR or AFR. Figure 7 illustrates how several non-monetized outcomes vary across regions (additional results for forest loss avoided – a relatively minor category of benefits – appear in Supplementary Figure A11). Compared to the corresponding monetized outcomes, the shares of non-monetized benefits are relatively higher in SEAR and AFR and lower in WPR.
For TS1, the most socially beneficial policies in rural areas avoid over 4 million cases of illness and nearly 120,000 deaths per year (for a total of 2.1 million DALYs avoided per year), with an additional 418 thousand cases and 15 thousand deaths per year avoided in urban locations (0.3 million DALYs per year) (Fig. 7). About 21 billion hours are saved every year (17 and 4 billion, in rural and urban settings), and tons of CO2 equivalent emissions avoided reach 380 million per year (310 and 70 million, in rural and urban areas). Finally, annual unsustainable forest loss avoided is about 80 billion kilograms of wood harvest (Figure A11).
Relative to the most socially beneficial policies under TS1, rural areas for TS2 and TS3 would avoid somewhat fewer deaths and illnesses per year (2.4 and 1.4 million cases, and 71 and 43 thousand deaths, respectively). Yearly cases and deaths avoided in urban areas would similarly decline. In TS2 and TS3, about 15 billion and 14 billion hours, respectively, would be saved every year in rural settings (~ 4 billion in urban settings), and tons of CO2 equivalents avoided per year would reach 248 million and 208 million, respectively. Finally, in TS2 and TS3, annual forest loss avoided overall would represent about 64 billion kgs and 54 billion kgs of net loss of forest stock, respectively.
Turning to the costs disaggregation, the major cost categories for the most beneficial policies in rural areas for TS1 are private fuel costs ($1.5 billion/yr); private stove, learning and maintenance costs ($680 million/yr); government costs including administrative and program costs ($598 million/yr); and the costs of stove subsidies ($445 million/yr) (Fig. 8). In urban areas, the highest costs in TS1 are government costs ($263 million/yr) and private stove, learning and maintenance costs ($176 million/yr). In urban areas, private fuel savings to households are an important net benefit from using cleaner fuels since inefficient use of charcoal and kerosene remains costly ($251 million/yr).
The highest cost components vary somewhat by region (Supplementary Figures A12-A14). For example, in TS1 and rural areas, private stove, learning and maintenance costs are highest in AFR ($160 million/yr) and WPR ($211 million/yr), but government administrative and program costs are highest in other regions, especially SEAR ($249 million/yr). In TS2 and TS3, net fuel costs to households decline substantially, owing to efficiency gains of other improved biomass technology (in contrast to TS1 where net fuel costs increase owing to higher cost of clean fuel). In urban locations, the highest yearly cost category in all regions is government administrative and program cost. Urban private fuel savings are realized in four regions (AFR, AMR, EMR and WPR), with urban AFR experiencing the highest yearly gains ($363 million/yr).
Costs per capita, overall and by region, are shown in Figure A15-A18 (these range from $0.07/capita-yr to $1.06/capita-yr across urban and rural areas in transitions 1–3). As with benefits, the differences across regions are somewhat reduced when accounting for their different population sizes, because many more people live in the WPR region which has the highest aggregate benefits and costs. Importantly, these full economic costs do not directly reflect affordability of clean and improved stoves, since they include both pecuniary (e.g., purchased fuel) and non-pecuniary aspects (time costs of fuel collection, and learning costs). Moreover, to ease comparisons across scenarios and relative to other potential non-cooking interventions, they are calculated on a total country population basis, rather than scaled proportionately to the specific sub-population transitioning from polluting to cleaner fuels in each scenario.