Abstract
Cooking with solid fuels and inefficient cookstoves has adverse consequences for health, environment, and human well-being. Despite the promise of improved cookstoves to reduce these impacts, adoption rates are relatively low. Using a 2-wave sample of 144 households from the baseline and first midline of an ongoing 4-year randomized controlled trial in Rwanda, we analyze the drivers and associations of early adoption of a household energy intervention marketed by a private sector firm. Households sign an annual contract to purchase sustainably produced biomass pellets and lease a fan micro-gasification cookstove with verified emissions reductions in laboratory settings. Using difference-in-differences and fixed effects estimation techniques, we examine the association between take-up of the improved cooking system and household fuel expenditures, health outcomes, and time use for primary cooks. Thirty percent of households adopted the pellet and improved cookstove system. Adopting households had more assets, lower per capita total expenditures and cooking fuel expenditures, and higher per capita hygiene expenditures. Households with married household heads and female cooks were significantly more likely to adopt. Adjusting for confounders, we find significant reduction in primary cooks’ systolic blood pressure, self-reported prevalence of shortness of breath, an indicator of respiratory illness, time spent cooking, and household expenditures on charcoal. Our findings have implications for marketing of future clean fuel and improved cookstove programs in urban settings or where stoves and fuel are purchased. Analysis of follow-up surveys will allow for estimation of long-term impacts of adoption of interventions involving pellets and fan micro-gasification cookstoves.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Approximately three billion people use solid fuels and traditional technologies for cooking and heating (WHO 2016). In Africa and Southeast Asia, over 60% of households are solid fuel users (Bonjour et al. 2013), and take-up of improved cookstoves (ICS) is persistently low (Lewis and Pattanayak 2012). Household air pollution (HAP) resulting from using solid fuels in inefficient cooking technologies accounts for 3.7–4.8 million deaths according to WHO estimates, while the Global Burden of Disease (GBD) Study estimates between 2.2 and 3.6 million deaths from HAP (Landrigan et al. 2018). HAP-related preventable deaths include low birthweight and pneumonia among children, and non-communicable diseases such as stroke, ischemic heart diseases, chronic obstructive pulmonary disease, and lung cancer among adults (Brook et al. 2010; Clark et al. 2012; Alexander et al. 2016; Giorgini et al. 2016; WHO 2016). In low- and middle-income countries, HAP is the largest environmental risk factor for disease burden (Forouzanfar et al. 2015). In addition to health impacts, firewood for cooking is a leading cause of deforestation and environmental degradation (Hofstad et al. 2009; Köhlin et al. 2011), and black carbon from burning solid fuels is a major contributor to regional climate change (Ramanathan and Carmichael 2008).
Though improved cooking technologies and clean fuels designed to reduce air pollution exist, adoption and sustained use in developing countries is a challenge. Systematic reviews focused on the state of knowledge of adoption of improved cookstoves find that higher education, income, household assets, and urban location increase uptake (Lewis and Pattanayak 2012; Rehfuess et al. 2014). On the other hand, socially marginalized status (Lewis and Pattanayak 2012), larger family size, costs associated with high-quality ICS, and requirements for processed or modern fuels to be used with ICS act as barriers to adoption (Rehfuess et al. 2014). Households that purchase instead of collect fuel are more likely to adopt an ICS, as money saving is a tangible benefit to households already paying for fuel. In the context of the enabling environment for ICS adoption, Puzzolo et al. (2013) find that success with early adopters, especially community opinion leaders, and characteristics of the stove and fuel are important determinants of adoption and sustained use, among other factors (e.g., providing loans for businesses producing and promoting ICS; developing an efficient and reliable network of suppliers).
ICS adoption studies are dominated by studies from rural settings. Specific to urban settings in Africa, Gebreegziabher et al. (2012) in urban Ethiopia find that household expenditure, household size, age and education of household head significantly explain household adoption of the electric mitad stove, and Alem et al. (2013) find the price of electricity and firewood and credit access to be significant predictors of electric ICS adoption. Another study from urban Ethiopia finds that ICS price, household income, and wealth (home ownership and separate kitchen) are significant determinants of Mirte and Lakech ICS adoption (Beyene and Koch 2013). In urban Zambia, Tembo et al. (2015) reported that higher income residential area, lower household size, young household head, those with education levels above secondary school, and male-headed households are significantly more likely to use electricity as the sole source of energy. With respect to impacts, most rigorous ICS evaluation studies focus on health impacts (Ezzati and Kammen 2002; Romieu et al. 2009; Clark et al. 2013; Mortimer et al. 2017), failing to consider a broader set of impacts, including socioeconomic and environmental outcomes (Bensch and Peters 2012; Hanna et al. 2016).
Our study addresses the challenge of HAP in urban settings in sub-Saharan Africa (SSA) through examination of the determinants of early adoption of clean cooking solutions in Rwanda. We use the term “adoption” to describe enrollment in the clean household energy program signified by signing of a contract with the firm marketing the household energy program. We focus on early adoption, meaning that household have relatively recently signed contracts with the private sector firm. Most households that “adopted” (as per our definition), may not have exclusively switched to clean cooking. Throughout this paper, we use our definition of “adoption” and “take-up” (enrollment in the clean household energy program) interchangeably. In addition to explaining determinants of adoption as we have defined it, we explore initial associations with indicators of household health and well-being.
With a population density of 481.7 per square kilometer (United Nations Statistics Division 2017), Rwanda is the most densely populated country in SSA. HAP is the fourth highest risk factor for disease burden in the country (Forouzanfar et al. 2015) and over 95% of the population relies on biomass for cooking (GACC 2016). While the Rwandan government has been supportive of ICS projects, including promoting the locally produced clay stove, Canarumwe (Rayens 2015), and the imported Tier 2 EcoZoom Dura ICS promoted through health centers by DelAgua Health, an environmental health technology firm that collaborates with governments in low income countries, the long-term impacts of these projects on improving health and well-being of household members, reducing particulate matter emissions and reducing biomass use are limited (Rosa et al. 2014; Kirby 2017).
The private sector firm we study, Inyenyeri, is a for-profit fuel and ICS social enterprise, currently marketing renewable biomass fuel pellets and a fan micro-gasification stove to consumers in Gisenyi in western Rwanda. Inyenyeri is unique among household energy interventions in Rwanda, and in the region because (a) it is market-driven and customer service oriented, (b) it combines renewable energy fuel pellets with the currently best available biomass burning stove the Mimi Moto, which has been rated as an International Workshop Agreements (IWA) Tier 4 stove (GACC 2015), (c) it focuses marketing efforts on both the incentives of household decision makers and cooks, and (d) the marketing model has been designed to be scalable and sustainable, with the objective of expansion throughout Rwanda and beyond.
Inyenyeri customers sign an annual contract to make a monthly purchase of renewable biomass pellets, while receiving the Mimi Moto stove at no additional cost on a lease basis. Pellets are produced in a pelletizing factory on the shores of Lake Kivu from sustainably sourced biomass feedstock (e.g., small eucalyptus trees and branches, elephant grass, sawdust). Inyenyeri originally marketed to study participants using the Philips fan micro-gasification stove. For a variety of reasons, they transitioned to the Mimi Moto in early 2016 (Jagger and Das 2018). When our baseline and midline data collections took place the structure of the marketing model was that customers had the option of signing up for 30, 45, or 60 kg of pellets per month, with the recommended quantity dependent on household size. The cost of each pellet package is 6000, 9000, and 12,000 Rwandan Francs (RWF) (7.80, 11.70, and 15.60 USD), respectively. Depending on the fuel-pellet package that households purchase, they received 1 (30 kg), 2 (45 kg), or 3 (60 kg) stoves. The current marketing model as of early 2018 is a fully pay-as-you-go system. As part of their business model, Inyenyeri offers free delivery, training, repairs, and replacement of stoves. For a detailed discussion of Inyenyeri’s pilot activities and lessons learned between 2012 and 2017, as well as an overview of the study timeline see Jagger and Das (2018).
In this paper, we examine the determinants of early adoption of the firm’s improved household energy system among 144 households, and the initial associations of this program with household fuel expenditures, primary cooks’ health and time use,Footnote 1 8 months after intervention. Our hypothesis is that among households that adopt Inyenyeri’s system, household expenditures on cooking fuel are lower, prevalence of health symptoms are reduced, and less time is spent cooking.
Methods
Study Design
Our analysis leverages data from a sub-sample of households from a large and ongoing household-level randomized controlled trial designed to evaluate the exposure, health, and welfare impacts of Inyenyeri’s household energy system for a sample of 1462 urban households in Gisenyi, Rwanda. The impact evaluation consists of a baseline survey (June 2015), two midline [midline 1 (June 2016) and midline 2 (June 2017)] surveys involving a sub-sample of 180 households, and an endline (scheduled for October 2018) survey on the full sample. The study takes place in 22 purposively selected neighborhoods in two cells (Bugoyi and Kivumu) of Gisenyi town in Rubavu District. Our surveys included structured household decision maker and primary cook questionnaires to collect socioeconomic and demographic data; information about all aspects of cooking and fuel use in the household; knowledge and preferences on stoves and fuels; knowledge and mitigation strategies for coping with the varied impacts of HAP, and customer experiences with Inyenyeri.
Sampling
The population of neighborhoods under consideration included Inyenyeri’s existing and potential new urban markets in close proximity to the firm’s two main retail locations. The sampling frame for the study comprised the population of households (N = 2354). Households that were existing customers of the firm were removed from the sampling frame. Using a random number generator, each household head was assigned a random number. The first 1500 households were selected as the study population, and the remaining households in the final sampling frame were designated as replacement households. Two-thirds of the 1500 selected households were randomly assigned to the intervention group (treatment), and the remaining one-third of households were assigned the delayed-entry control group.Footnote 2 A total of 1462 households were surveyed at baseline.
A sub-sample of 180 households was randomly selected from the sample of 1462 households for the comprehensive midline 1 and midline 2 Health, Poverty and Cooking (HPC) surveys. The sub-sample was originally split using the same ratio of treatment to control as the larger sample (2:1). After the baseline survey, completed in September 2015, the firm began marketing stoves to all households in the treatment group. Due to concerns about low take-up rates, the 60 control group households in our sub-sample were re-categorized as treatment and marketed to by Inyenyeri prior to the midline 1 data collection.Footnote 3Footnote 4 After midline 1, we were only able to verify that 144 households had contact with Inyenyeri, thus, because we focus on the average treatment effect (ATE), the our sample for this analysis is 144 households.
Household, Poverty and Cooking Survey (HPC)
The midline 1/midline 2 HPC is a shortened version of the full baseline/endline questionnaire. Modules include: household roster; education; assets; housing and cooking infrastructure, facilities and access to services; perceptions of fuels and cooking technologies and their impacts; risk and time preferences and social capital; health of family members and primary cook; time use for household members and primary cook; and cook history (e.g., duration as primary cook and experience with cooking). Modules on adoption and sustained use were added to the midline 1 HPC to learn about respondents’ experiences with the firm and the design of the marketing model.
Attrition from our sample between baseline and midline 1 was a challenge as is common in urban-based longitudinal surveys. At midline 1, we were able to collect data from 115 baseline households. Sixty-five households had to be replaced, because they had either moved outside the study area or were not available to be interviewed.Footnote 5 The 65 replacement households were drawn at random from the 435 baseline delayed-entry control households. The sample for the analysis in this paper includes 144 of the 180 sub-sample households where the firm conducted their door-to-door marketing after our baseline survey was concluded in September 2015. In 91 of the 144 households (63%), we were able to interview the same primary cook at baseline and midline 1. Tracking individuals over time is important for understanding health impacts, but also mitigating confounding that can occur when a household transitions from one primary cook to another.
Analysis
A logit model (Eq. 1) was used to estimate the relationship between socioeconomic determinants of the household and adoption of the improved household energy system in midline 1:
where Yj denotes signing a contract for household j and εj is the error term. Household-level characteristics included household size; stove used in the 30 days prior to baseline survey; number of durable goodsFootnote 6; ownership of land; and log of per capita total expenditures, cooking fuel expenditures, hygiene expenditures (in the 4 weeks prior to baseline survey). We included binary indicators for whether the primary cook was hired and female. For the characteristics of the household head, we considered age, sex, education level, whether s/he thought that some stoves and fuels produce less smoke than others, and whether household head was aware of the environmental health impacts of cooking with biomass.
Second, to assess the average treatment effect (ATE) of Inyenyeri’s household energy system on household expenditures, we used the following model (Eq. 2):
where Yjt denotes cooking fuel expenditures or charcoal expenditures for household j at time t, Contractj is an indicator that equals 1 if household j signed a contract with Inyenyeri, Timet equals 1 if time period is midline 1, and β3 is the ATE estimator, or the effect of signing a contract with Inyenyeri. δi are individual-level controls, αj are household-level confounding variables, and εij is the error term.
Third, to assess the average treatment effect (ATE) of Inyenyeri’s household energy system on primary cooks’ blood pressure,Footnote 7 and time use, we used the following model (Eq 3):
where Hijt denotes blood pressure, or time use for primary cook i in household j at time t, and the remaining variables are the same as in Eq. 2.
Fourth, to assess the ATE of Inyenyeri’s household energy system on binary indicators of self-reported health symptoms of primary cooks, we used the following model (Eq. 4):
where Hijt denotes health symptom for primary cook i in household j at time t, and the remaining variables are the same as in Eq. 2.
In Eqs. 1–3, we use robust standard errors and in Eq. 4, we use bootstrapped standard errors.
The key variable of interest is β3, which provides the difference-in-differences (DiD) in the outcome with respect to signing a contract with the firm. Since contract signing was a choice, the interpretation of β3 in the models above are unlikely to be causal since unobserved (to the researcher) factors could determine both stove take-up and the outcomes (e.g., innate ability or superior information-processing skills of household members). To account for this, we also estimated household fixed effects (FE) models for all outcomes, which eliminate time-invariant unobserved differences across households, which are likely to be the main source of endogeneity in this context.Footnote 8 Estimates from these models are more likely to provide an estimate of the causal effect of signing a contract on the outcome of interest.
Results
Implementation of the Improved Household Energy Program
Inyenyeri’s marketing campaign, launched in September 2015 immediately after the completion of our baseline data collection, consisted of several targeted strategies: (1) marketing using billboards and radio programs with core messages such as “cook fast,” “stay clean,” “life made easy,” and “always the cheapest fuel”; (2) village-level cooking demonstrations; and (3) door-to-door visits from Customer Service Representatives (CSRs) to explain the contract model and conduct in-home cooking demonstrations.
From our midline 1 survey, we observed that 81.9% of households indicated that they had heard of Inyenyeri. The majority of households had seen Inyenyeri billboards (83%), learned about the firm from friends (81.4%), and had been visited by a customer service representative (72.7%). A smaller percentage attended village cooking demonstrations (49.2%), and/or heard an Inyenyeri radio program (22.9%). Adoption (i.e., households that signed contracts with the firm) was 29.9% at midline 1. Contract-signing households, generally had two ICS (i.e., opted to sign up for the firm’s mid-range pellet package).
To assess the extent to which household’s adoption the household energy system made a partial or total switch to using pellets and the fan micro-gasification stove, we collected detailed information on the share of meals cooked during the past 30 days on various technologies (Fig. 1). Households that adopted the Inyenyeri system appeared to be replacing cooking on both portable and fixed charcoal stoves with the biomass pellet and fan micro-gasification stove combination.Footnote 9 However, even among adopter households, there was evidence of stove stacking; households continued to use pre-existing technologies for a large share of cooking events. For the midline 1 HPC survey, households were also asked to recall what technologies and fuels they used to cook each meal over 3 days immediately prior to the survey. Households that had adopted the Inyenyeri household energy system reported using the fan micro-gasification stove for 3.5 out of nine possible cooking events during the past 3 days. For both of these subjective measures of stove use, fuel use perfectly corresponded to stove use; households using forced-air gasifier stoves reported using them only with biomass pellets.
Motivators and Barriers to Adoption of Inyenyeri’s Improved Household Energy Program
We observed that on average, households had 6 members, 7.4 durable goods/assets, and fixed charcoal stoves were the most used cooking technology (Table 1). Per capita total expenditure in the 4 weeks prior to the baseline survey was approximately 57,797 RWF (75.14 USD), of which cooking fuel expenditure was 2633.7 RWF (3.42 USD), and hygiene expenditureFootnote 10 was 2607.2 RWF (3.39 USD). Over 32% of households hired a cook, 78% cooks were female and 29% households had a female household head. The average age of household head was 48 years, 66% were married, and over 57% were educated at the secondary level and above. Over 65% household heads had heard of the negative impacts of cooking with biomass and more than 87% perceived some fuels and stoves to be less smoke-producing than others. Households that adopted the Inyenyeri system had significantly more durable household goods (8.7) than non-adopters (6.9), higher per capita hygiene expenditures (3678.5 RWF (4.78 USD) compared to 2151.1 RWF (2.80 USD) of non-adopters), and more married household heads (83.7%) than non-adopters (59.4%). Weak statistically significant differences were observed between adopter and non-adopter households on hired primary cooks and female household heads, but on other independent variables, there were no significant differences.
In our first logistic regression model (Table 2), we did not consider perceptions of the household main decision maker on the health, environmental and climate impacts of reliance on biomass for cooking, but included their perceptions about some stoves and fuels being less smoke-producing than others. The second regression model included variables indicative of main decision maker’s level of awareness about health, environmental and climate impacts.
We observed that households with more durable goods, high per capita hygiene expenditures, and female primary cooks were significantly more likely to adopt stoves (p = 0.05).
Per capita total expenditures and per capita fuel expenditures had statistically significant weak negative association with likelihood of adoption of Inyenyeri’s system (p = 0.10). On including variables about awareness of health, forest and climate impacts, we found that where household heads had knowledge of the health impacts for cooks and children from cooking with biomass on traditional stoves, households were more likely to adopt the new household energy system (p = 0.10). However, awareness of the environmental impacts of cooking with charcoal and other unsustainably harvested biomass was associated with significantly lower likelihood of adoption of the Inyenyeri’s system (p = 0.05).
Association Between the Improved Household Energy Program, Health and Well-Being
Health of Primary Cooks
We observed a high prevalence of shortness of breath among primary cooks at baseline (44.1%), followed by burns (18.9%) and night phlegm (9.1%). The average systolic blood pressure among primary cooks was 120.6 mmHg, and average diastolic blood pressure was 78.3 mmHg (Table 1). We restricted our sample to households with the same primary cook at baseline and midline 1 (N = 91). Our first empirical strategy was to investigate the differences in the means between the prevalence of health symptoms at baseline and midline 1. Among self-reported health symptoms where we observed statistically significant differences (reductions in prevalence of burns, night phlegm, shortness of breath and blood pressure), we estimated the ATE of adoption of the household energy system (Table 3).Footnote 11 In the household FE model, we observed a statistically significant decreased prevalence of shortness of breath (p = 0.01) and decrease in systolic blood pressure (p = 0.10) in primary cooks among households that adopted the Inyenyeri household energy system. Diastolic blood pressure also decreased, although this decrease was not statistically significant.
There were no significant differences between the two groups of households on other dependent variables with the exception of time spent in other activities where the mean was significantly higher among non-adopters (1.9 vs. 0.7 h).
Household Fuel Expenditures
The Inyenyeri model is designed to be less expensive, or at least equivalent to cooking with charcoal. On average, households purchased cooking fuel worth 12,480 RWF (16.22 USD) during the 4 weeks prior to the survey, of which 11,431 RWF (14.86 USD) were on charcoal (Table 1). We assessed the association between adoption of the Inyenyeri household energy system and overall fuel expenditures, specifically charcoal expenditures (Table 4). We observed no statistically significant relationship between adoption and overall cooking fuel expenditures. However, we observed a strong statistically significant reduction in expenditures on charcoal (p = 0.01) in the fixed effects model, which accounts for unobserved differences between those who signed a contract and those who did not.Footnote 12 Specifically, the pattern of results suggests that those who signed a contract had higher pre-contract charcoal expenditures than those who did not sign a contract.
Time Use of Primary Cooks
As expected, cooks at baseline spent most of their time in the past 7 days cooking (22.7 h), followed by childcare and cleaning (15.4 h) and non-agricultural activities (4.9 h) (Table 1). In keeping with our hypothesis, we found a statistically significant negative association between adopting the improved household energy system and time spent cooking (p = 0.05) in the FE model (Table 5).Footnote 13 We did not find significant changes in time spent on any other activity.
Discussion
Our study finds an adoption/take-up rate of 30% for the pellet/fan micro-gasification improved household energy system, which is one-third the adoption rate reported by Barstow et al. (2014) for the EcoZoom Dura ICS, in Rwanda’s Western Province. However, unlike the more sustainable Inyenyeri business model, the DelAgua program provided free distribution of ICS and water filters and [unsuccessfully] sought to earn carbon credits from verified reduced use of fuelwood (Barstow et al. 2014).
Our analysis of the determinants of early improved household energy adoption is consistent with findings from Rehfuess et al. (2014), indicating that households that are significantly more likely to adopt are those with higher assets, married household heads, female cooks, and where the household head is aware of the negative impacts of traditional cooking methods on human health. Though knowledge of the negative impacts of biomass on local air quality has no significant association with ICS and clean fuel adoption, the negative sign of the coefficient is perhaps indicative of households’ lower valuation of the environment and climate compared to health. These findings have implications for the target group for future interventions and messaging of ICS programs. Similar to Barstow et al. (2016) study that did not find any association between household size and stove stacking behavior, we observed no significant relationship between number of household members and improved energy adoption. Unlike Gebreegziabher et al. (2012) and Tembo et al. (2015), who undertook ICS adoption studies in urban settings in Ethiopia and Zambia, respectively, we did not find any significant association between overall household expenditure, age, sex, or education of household head and improved energy adoption. Our findings should be framed in the context of the location where the study took place. Cooking needs may vary between sites (e.g., cooking of injera in Ethiopia vs. maize meal in Zambia, or dominance of one staple vs. preferences for multiple staples with different cooking requirements as in Rwanda), influencing take-up of new ICS. We also emphasize that the evidence base is very limited for ICS adoption in urban settings in Africa, making it difficult to generalize.
We note that in the months immediately following intervention roll-out, Inyenyeri had only 1–5 tons of pellets in stock owing to problems with their pelletizing equipment, which limited their ability to market, aggressively to new customers (Jagger and Das 2018), thereby affecting adoption in our study. Thurber et al. (2014) noted similar supply chain concerns (increases in costs of raw materials) with bagasseFootnote 14 feedstock inputs for pellets, which constrained Oorja’s value proposition compared to stove alternatives, particularly LPG, over time.
As our study is powered on an expected take-up of ~ 60%, our ability to detect reductions in health symptoms or economic impacts is limited due to a low ICS and clean fuel take-up rate of approximately 30% in our sample of 144 households, and by our sample size (N = 91) for primary cooks, we are able to track during the first 2 years of our study. The simultaneous use of other cooking technologies is another plausible reason for not observing significant reductions in total cooking fuel expenditures and primary cooks’ time spent cooking. Because our data were collected in the same season in each of the 2 years that we cannot reflect on seasonal variation and health or economic impacts.
Based on Beyene and Koch’s (2013) finding that adoption rates of ICS in urban Ethiopia increased over time, and Barstow et al. (2016) finding that rural households in Rwanda reduced their use of stoves other than the EcoZoom Dura ICS by 20% over the duration of their study, we are hopeful that we may see reductions in stove stacking behavior in subsequent follow-up surveys. Additionally, it is important to compare stove use data from surveys and electronic sensors,Footnote 15 as studies from Rwanda show lower ICS use from sensors compared to surveys (Thomas et al. 2013).
Although adopter households continue to use charcoal stoves alongside newly introduced fan micro-gasification stoves, there were significant reductions in their charcoal expenditures in the 4 weeks prior to the survey. This finding is aligned with the firm’s rationale of pricing the biomass pellets competitively with charcoal, with the aim of replacing it in the long run. Those who signed contracts also showed a significant reduction in time spent cooking, most of which was subsequently devoted to non-agricultural activities although the effect on the latter was not statistically significant.
Conclusion
The analysis of the first two rounds of data collected for a large household energy randomized controlled trial aims to shed light on the determinants of early adoption and associations between adoption and indicators of health and well-being of an improved household energy system. Our study is set in western Rwanda, in sub-Saharan Africa, a region with limited ICS intervention studies, particularly in urban settings. We found considerable influence of sociodemographic variables on household adoption of the improved energy system, and evidence of improvements in health and time use for primary cooks, and significant reduction in monthly charcoal expenditures for households. We acknowledge that these results from our ongoing study have limitations. First, we use a sub sample of 144 households from our full sample of 1462 households where we collect repeated measurements between baseline and endline, to detect impacts of a private sector initiative promoting sustainability produced biomass pellets in tandem with fan micro-gasification stoves. Second, the time lag between the intervention roll-out and follow-up survey (8 months) is insufficient to see high uptake and substantial impacts on many of the variables we are collecting data on. Third, the potential effects of the stove if used exclusively could not be assessed in this study given widespread concurrent use of charcoal in the adopting households.
We will continue to track these 144 households in subsequent surveys prior to endline, in order to ascertain the determinants and impacts of late adopters compared with early adopters. Data from our endline survey with the full sample of 1462 households will enable us to better understand cost and time savings, and potential health improvements from adoption of the Inyenyeri household cooking energy system over a two-year timeframe, and allow us to make stronger causal claims about the effect of stove use on these outcomes.
Notes
Time spent on various activities (domestic chores, agriculture/livestock/fisheries and other) over the 7 days prior to the survey.
A household was excluded from the sample if (a) the main respondent and/or the primary cook refused; (b) the primary cook was less than 15 years old; (c) no cooking was done in the household; (d) the household was currently using or had previously used an improved stove such as the Philips gasifying stove; (e) a sampled household was in the same plot as another that has already been surveyed; and (f) the village chief, key people in the village (such as community health worker) or neighbors had no information about the household sampled. If there was a new household residing in the same plot as the sampled household, and the original inhabitants listed in the sampling frame had moved to a different location, the survey was conducted with the new household.
Households were contacted by arranging appointments and making home visits to demonstrate the ICS and biomass pellets combination.
Personal and area exposure monitoring of carbon monoxide, particulate matter 2.5, and polycyclic aromatic hydrocarbons was conducted at baseline, midline 1, and midline 2, and is planned for endline for the sub-sample of 180 households.
Attritor households are renters and households with ties to other regions of Rwanda or the Democratic Republic of Congo (i.e., more likely to move out of the study area), and households with several people engaged in daytime employment (i.e., more likely to be difficult to locate or schedule an interview with).
The following were the durable goods/assets considered in this analysis: living room suite, refrigerator, freezer, radio, TV set, satellite dish, cooker, video/DVD player, computer and accessories, music system, electric fan, air-conditioner, sewing machine, bed, cupboard/bookcase, table-chair, car, motorcycle (for home use only) and bicycle (for home use only). Our asset variable is a simple count of assets each household owns.
We took three consecutive measurements of primary cooks’ blood pressure after the cook had been at rest for at least 20 min. We used the Omron 5 Series blood pressure monitor which has been validated for measurement of blood pressure according to European standards (Topouchian et al. 2011). Enumerators received focused training on how to use electronic blood pressure measurement instruments. For our analysis, we use the average of the three measurements.
We conducted the Breusch-Pagan and Hausman tests to test for differences between the pooled ordinary least squares (OLS), fixed effects (FE) and random effects (RE) model, and report results where coefficients are significantly different.
Anecdotal evidence gathered from the firm’s customer service representatives suggest that the decision to change from the Philips stove to the Mimi Moto affected take-up of the household energy system.
These expenditure items included broom/brush, sponge, shoe brush and polish, disinfectant and cleaners, laundry services, rubbish collection services, and wages for household domestic help.
The Breusch-Pagan and Hausman tests between pooled OLS and RE, and pooled OLS and FE, respectively, for blood pressure indicate that the pooled OLS model is inappropriate. The Hausman test indicates significant differences between FE and RE estimates only for night phlegm and shortness of breath, suggesting unobserved heterogeneity.
While there were significant differences between the pooled OLS, and FE and RE estimates, the FE and RE estimates were not significantly different for charcoal expenditures.
The Breusch-Pagan and Hausman tests between pooled OLS and RE, and pooled OLS and FE, respectively, for time spent in all activities except labor and other activities, show significant differences. Among all activities, the Hausman test indicates significant differences between FE and RE estimates for time spent in cooking and childcare activities.
Bagasse is the dry pulpy residue left after the extraction of juice from sugar cane, it is commonly used as fuel for electricity generators.
Our study also collected objective data on stove use using temperature loggers (stove use monitoring system-SUMS). Analysis from the same will be published separately.
References
Alem Y, Hassen S, Köhlin G. (2013). The dynamics of electric cookstove adoption: panel data evidence from Ethiopia. Working Papers in Economics No. 557 at University of Gothenburg.
Alexander D, Northcross A, Wilson N, Dutta A, Pandya R, Ibigbami T, Adu D, Olamijulo J, Morhason-Bello O, Karrison T, Ojengbede O, Olopade CO (2016) Randomized controlled ethanol cookstove intervention and blood pressure in pregnant Nigerian women. American Journal of Respiratory and Critical Care Medicine 195(12):1629–1639 (https://doi.org/10.1164/rccm.201606-1177oc).
Barstow CK, Ngabo F, Rosa G, Majorin F, Boisson S, Clasen T, Thomas EA (2014) Designing and piloting a program to provide water filters and improved cookstoves in Rwanda. PLoS ONE 9(3):e92403.
Barstow CK, Nagel CL, Clasen TF, Thomas EA (2016). Process evaluation and assessment of use of a large scale water filter and cookstove program in Rwanda. BMC Public Health 16(1):584.
Bensch G, Peters J (2012) A recipe for success? Randomized free distribution of improved cooking stoves in Senegal.” Ruhr Economic Paper 325.
Beyene AD, Koch, SF (2013) Clean fuel-saving technology adoption in urban Ethiopia. Energy Economics 36:605–613.
Bonjour S, Adair-Rohani H, Wolf, J, Bruce NG, Mehta S, Prüss-Ustün A, Lahiff, M, Rehfuess, EA, Mishra, V,Smith KR (2013) Solid fuel use for household cooking: country and regional estimates for 1980–2010. Environmental Health Perspectives 121(7):784–790 (https://doi.org/10.1289/ehp.1205987).
Brook RD, Rajagopalan S, Pope CA 3rd, Brook JR, Bhatnagar A, Diez-Roux AV, Holguin F, Hong Y, Luepker RV, Mittleman MA, Peters A, Siscovick D, Smith SC Jr, Whitsel L, Kaufman JD (2010) Particulate matter air pollution and cardiovascular disease: an update to the scientific statement from the American Heart Association. American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism 121(21):2331–78 (https://doi.org/10.1161/cir.0b013e3181dbece1).
Clark ML, Bachand AM, Heiderscheidt JM, Yoder SA, Luna B, Volckens J, Koehler KA, Conway S, Reynolds SJ, Peel JL (2012) Impact of a cleaner-burning cookstove intervention on blood pressure in Nicaraguan women. Indoor Air 23(2):105–14 (https://doi.org/10.1111/ina.12003).
Clark ML, Bachand AM, Heiderscheidt JM, Yoder SA, Luna B, Volckens J, Peel JL (2013) Impact of a cleaner-burning cookstove intervention on blood pressure in Nicaraguan women. Indoor Air 23(2):105–114.
Ezzati M, Kammen DM (2002) The health impacts of exposure to indoor air pollution from solid fuels in developing countries: knowledge, gaps, and data needs. Environmental Health Perspectives 110(11):1057.
Forouzanfar MH, Alexander L, Anderson HR, Bachman VF, Biryukov S, Brauer M, Burnett R, Casey D, Coates MM, Cohen A, Delwiche K, et al (2015) Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. The Lancet 386(10010):2287–2323.
Gebreegziabher Z, Mekonnen A, Kassie M, Köhlin G (2012) Urban energy transition and technology adoption: the case of Tigrai, northern Ethiopia. Energy Economics 34(2):410–418.
Giorgini P, Di Giosia P, Grassi D, Rubenfire M, Brook RD, Ferri C (2016) Air pollution exposure and blood pressure: an updated review of the literature. Current Pharmaceutical Design 22(1):28–51. Review.
Global Alliance for Clean Cookstoves (2015) Partner directory-Mimi Moto. Available at: http://cleancookstoves.org/partners/item/999/2205 [Accessed on January 15, 2017].
Global Alliance for Clean Cookstoves (2016) Country profiles. Available at: http://cleancookstoves.org/country-profiles/32-rwanda.html [Accessed on February 7, 2017].
Hanna R, Duflo E, Greenstone M (2016) Up in smoke: the influence of household behavior on the long-run impact of improved cooking stoves. American Economic Journal: Economic Policy 8(1):80–114.
Hofstad O, Köhlin G, Namaalway F (2009) How can emissions from woodfuel be reduced? Realising REDD + : National Strategy and Policy Options (Angelsen A, Brockhaus M, Kanninen M, Sills E, Sunderlin WD, Wertz-Kanounnikoff S). Bogor, Indonesia: Center for International Forestry Research, pp 237–248.
Jagger P, Das I (2018) Implementation and scale-up a biomass pellet and improved cookstove enterprise in Rwanda. Energy for Sustainable Development 46:32–41.
Kirby M (2017). Assessing use, exposure, and health impacts of a water filter and improved cookstove distribution programme in Rwanda (Doctoral dissertation, London School of Hygiene & Tropical Medicine).
Köhlin G, Sills EO, Pattanayak SK, Wilfong C (2011) Energy, gender and development. Policy Research Working Paper, No. WPS 5800. Washington, DC: Social Dimensions of Climate Change Division, World Bank.
Landrigan PJ, Fuller BE, Acosta NRJ, Odeyi O, Arnold R, Basu N, Balde AB, Bertollini R, Bose-O'Reilly S, Boufford JI, et al (2018) The Lancet Commission on pollution and health. The Lancet 391(10119):462–512. https://doi.org/10.1016/S0140-6736(17)32345-0
Lewis JJ, Pattanayak SK (2012) Who adopts improved fuels and cookstoves? A systematic review. Environmental Health Perspectives 120(5):637.
Mortimer K, Ndamala CB, Naunje AW, Malava J, Katundu C, Weston W, Wang D (2017) A cleaner burning biomass-fuelled cookstove intervention to prevent pneumonia in children under 5 years old in rural Malawi (the Cooking and Pneumonia Study): a cluster randomised controlled trial. The Lancet 389(10065):167–175.
Puzzolo E, Stanistreet D, Pope D, Bruce N, Rehfuess E (2013) Factors influencing the large-scale uptake by households of cleaner and more efficient household energy technologies. London: EPPI Centre.
Ramanathan V, Carmichael G (2008) Global and regional climate changes due to black carbon. Nature Geoscience 1(4):221–227.
Rayens KM (2015) The use of improved technology and market-based incentives to increase forest resource and biodiversity conservation in Rwanda. Theses and Dissertations—Forestry. Paper 26. http://uknowledge.uky.edu/forestry_etds/26/.
Rehfuess EA, Puzzolo E, Stanistreet D, Pope D, Bruce NG (2014) Enablers and barriers to large-scale uptake of improved solid fuel stoves: a systematic review. Environmental Health Perspectives (Online) 122(2):120.
Romieu, I, Riojas-Rodríguez H, Marrón-Mares AT, Schilmann A, Perez-Padilla R, Masera O (2009) Improved biomass stove intervention in rural Mexico: impact on the respiratory health of women. American Journal of Respiratory and Critical Care Medicine 180(7):649–656.
Rosa G, Majorin F, Boisson S, Barstow C, Johnson M, Kirby M, Ngabo F, Thomas E, Clasen T (2014) Assessing the impact of water filters and improved cook stoves on drinking water quality and household air pollution: a randomised controlled trial in Rwanda. PLoS ONE 9(3):e91011.
Tembo ST, Mulenga BP, Sitko NJ (2015) Cooking fuel choice in urban Zambia: implications on forest cover (No. 202883). Michigan State University, Department of Agricultural, Food, and Resource Economics.
Thomas EA, Barstow CK, Rosa G, Majorin F, Clasen T (2013) Use of remotely reporting electronic sensors for assessing use of water filters and cookstoves in Rwanda. Environmental Science & Technology 47(23):13602–13610.
Thurber MC, Phadke H, Nagavarapu S, Shrimali G, Zerriffi H (2014) ‘Oorja’in India: assessing a large-scale commercial distribution of advanced biomass stoves to households. Energy for Sustainable Development 19:138–150.
Topouchian J, Agnoletti D, Blacher J, Youseef A, Ibanez I, Khabouth J, Khawaja S, Beaino L, Asmar R (2011) Validation of four automatic devices for self-measurement of blood pressure according to the international protocol of the European Society of Hypertension. Vascular Health and Risk Management 7:709–717.
United Nations Statistics Division (2017) Available at: http://data.un.org/CountryProfile.aspx?crName=RWANDA [Accessed on February 7, 2017].
World Health Organization (2016) Household air pollution Fact sheet. Available at: http://www.who.int/mediacentre/factsheets/fs292/en/ [Accessed on February 13, 2017].
Acknowledgements
This impact evaluation is led and executed by The University of North Carolina at Chapel Hill, United States (UNC-CH) in collaboration with The Access Project (Baseline) and Laterite Africa (Midline 1 and onward). The first midline survey was funded by the United Nations Foundation/Global Alliance for Clean Cookstoves (UNF-15- 708: Evaluating Sustained Adoption of Inyenyeri’s Improved Stove and Fuel Initiative in Rwanda). The project was funded as part of a joint effort by the Global Alliance for Clean Cookstoves and the USAID Translating Research into Action Project (TRAction). The baseline, endline, and all other midline surveys are funded by the National Institutes of Health/National Institute of Environmental Health Sciences (R01ES023861: The Health and Poverty Effects of a Large-scale Cookstove Initiative in Rwanda). We are grateful to the Carolina Population Center (P2C HD050924) at UNC-CH for general support. This research has been reviewed and approved by the Institutional Review Board at UNC-CH, the Rwanda National Ethics Committee, and the National Institute of Statistics Rwanda.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author(s) declare that they have no competing interests.
Rights and permissions
About this article
Cite this article
Jagger, P., Das, I., Handa, S. et al. Early Adoption of an Improved Household Energy System in Urban Rwanda. EcoHealth 16, 7–20 (2019). https://doi.org/10.1007/s10393-018-1391-9
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10393-018-1391-9