About this Impact evaluation
Under current trends, household energy use in Africa alone will produce 6.7 billion tons of carbon by 2050. Tragically, the stoves that create this carbon will also create indoor air pollution that will kill millions of children. At the same time, deforestation will harm local ecosystems and biodiversity.
Improved stoves have the potential to reduce all of these problems. Importantly, several mechanisms already exist and new ones are evolving for rich nations to subsidize improved stoves that reduce greenhouse gases.
The research in this proposal will examine the scale-up of access to carbon credits and pioneer a market for health credits to finance further deployment of improved cook stoves across sub-Saharan Africa and South Asia. The implementing partner Plan Ghana is eager to expand the pilot improved stove project to hundreds or even thousands of villages in Ghana if sustainable impact-based financing can be found.
Plan Ghana is also part of a much larger NGO, Plan International, which operates in 60 countries worldwide. The opportunity is vast, as there are hundreds of millions of inefficient biomass stoves in use. Establishing a market for health credits will challenge accepted paradigms for financing global health and can ultimately transform the broader development agenda.
This study uses a randomised controlled trial to evaluate the impacts of improved cook-stoves among villages near Tumu in the Upper West region of Ghana. The use of biomass fuels for cooking poses a serious health hazard, because they emit large quantities of health-damaging air pollutants due to very low combustion efficiency. Improved cook-stoves have the potential to reduce this problem and to decrease deforestation and the release of greenhouse gasses owing to their more efficient fuel use.
A lottery was used to assign villages to treatment or control groups. The study ran from February to May 2009 in 8 of 20 sampled villages, selected for ethno-linguistic and geographic variation. For the sample of villages that were offered cook-stoves, treatment households (N = 402) received them in the first period and control households (N = 366) received them 2 months later. With the aid of Plan Ghana, villagers were trained in building (and using) improved cook-stoves to keep implementation costs low. To reduce attrition and evaluation costs, a baseline was not collected. Instead, the analysis relies on a mean comparison between treatment and control groups.
All participants carried out a cooking test in which they prepared a standardised meal. Treatment households used their new cook-stoves, and control households used their traditional ones. Exposure to carbon monoxide (CO) during the test was monitored with CO tubes. Wood use was calculated, and stove use was measured with stove usage monitors. A household exit survey collected data on socioeconomic characteristics; fuel consumption; wood-gathering time; recent cooking activity; perceptions of the improved stove and incidence of respiratory illnesses, coughing, wheezing and trouble breathing for participants in control and treatment groups. Finally, follow-up stove usage was observed in three villages, 8 months after programme implementation. The authors avoid endogeneity problems resulting from households' decisions to comply by using intent-to-treat estimates. The wood use analyses present treatment-on-the-treated estimates and use lottery assignment as the instrumental variable. Standard errors are clustered at the village level.
Cooking tests indicated that use of the new cook-stoves (that is, the treatment condition) reduced fuel use by 5 percent, which was not statistically significant, irrespective of the estimation method (intent-to-treat or treatment-on-the-treated). There were no statistically distinguishable reductions in time for collecting fuel wood or in CO exposure. Cooking outdoors lowered CO exposure for control households. For treated households, the new stoves were slightly successful in reducing CO exposure indoors ('15 ppm, p = 11%).
Results related to self-reported health seem more promising. The treatment group reported approximately half as many days experiencing irritated eyes, headache, bad cough or a sore throat following cooking. They also reported a smaller number of respiratory symptoms (17 percent versus 34 percent of control households reporting at least one symptom). The self-reported respiratory symptoms did not, however, correlate with self-reported new stove use, or with CO measurement. The positive health impact might be the result of experimenter demand and courtesy bias toward Plan Ghana.
The 8-month follow-up survey showed that half of the improved stoves remained in regular use. Households seemed to employ multiple stoves simultaneously. A probit regression showed that study groups are similar on baseline characteristics, apart from the number of cook-stoves. Substantial attrition, which occurred during the cooking test and follow-up survey, had no statistically significant predictors. High heat destroyed 28 percent of all the stove usage monitors. As a result, data could not be collected for 26 percent of treatment households and 32 percent of control households. This attrition rate was comparable for traditional and improved cook-stoves.
According to the authors, their rigorous evaluation method was successful in reducing the costs of evaluating field-based stove performance, user uptake and exposure to smoke. They advise health-improving stove programmes to focus on regions with indoor cooking, since the expected benefits of the new stoves are higher in that setting than for outdoor cooking. Another recommendation is to create monitoring and maintenance systems that ensure proper functioning of the stove chimneys. When installed incorrectly, leaky chimneys may prevent health benefits that are associated with reduced smoke exposure.