1 Introduction

The United Nations Sustainable Development Goals SDG-5 and SDG-7 emphasize the importance of energy and gender in achieving sustainable development [1]. Unfortunately, almost 3 billion people worldwide lack access to clean energy [2]. Women, especially those in rural areas, bear the burden of energy-related tasks, such as gathering and transporting solid biomass, charcoal, and agricultural waste. This reliance on traditional fuels is the primary source of indoor air pollution, which leads to social and health issues primarily affecting women [3]. Indoor air pollution is a significant cause of death globally, with 2.31 million people losing their lives in 2019 alone [4]. In Ethiopia, indoor air pollution caused by the use of traditional solid fuels is a significant threat to public health, especially for women and girls, leading to respiratory illnesses [5]. The frequent use of biomass in Ethiopia’s rural communities will likely exacerbate this problem.

Research shows that women bear the disproportionate burden of household work and require more energy for cooking, cleaning, and other household activities [6]. However, they have limited access to economic resources and decision-making power to choose the most efficient and sustainable types of energy. As a result, women often rely on traditional and inefficient forms of energy, such as biomass fuels, which have adverse health and environmental impacts [7]. This energy poverty exacerbates gender inequalities and hinders sustainable development goals [8]. Women and girls spent more time than men and boys collecting firewood [9]. A report on gender gaps in labor outcomes stated that, in Ethiopia, the proportion of women collecting water (71%) and firewood (54%) is twice that of men, which is 29% and 28%, respectively [10]. These figures show that energy-related issues are more connected to women and girls in developing countries, especially among rural communities. Collecting and transporting household energy sources is a routine day-to-day task for the household members. In rural Ethiopia, households spend about 11 to 12 h per week collecting energy sources, and the collection is done mainly by female household members. Here, 60% of the total firewood collection is done by female members, and over 90% of the animal dung collection is done by females aged between 18 and 59 [11].

The importance of energy in rural women’s lives, particularly for cooking and lighting, cannot be overstated. Access to clean energy can significantly improve their quality of life by reducing their workload and giving them time for other activities, such as economic gain, employment, education, and working hours. Empowering women and girls through education, income, and awareness is crucial to achieving this goal. This can increase the use of cleaner energy sources, resulting in overall household development [12]. For instance, in India, women empowered in training and financial aspects are more likely to use renewable energy [13]. In Ethiopia, the economic empowerment of respondents has been shown to have a significant and positive impact on the choice of improved energy sources [14].

Empowering women and girls through education, income, and awareness can lead to cleaner energy use, resulting in overall development in the household [13, 15]. Women’s economic empowerment and decision-making could help accelerate the transition to clean and efficient energy use in rural households [16]. Research shows that wage-employed wives in rural households contribute more to decisions to use clean energy than self-employed ones [17, 18]. Women’s educational status and income level are mainly associated with the use of electricity and solar rather than traditional biomass [19, 20]. Therefore, policymakers and concerned bodies need to understand the dimensions of economic empowerment and other factors determining energy choices in rural households to bring real change and help rural women and girls. This article aims to explore these crucial issues.

2 Women’s economic empowerment

Women’s empowerment can be defined as “the process of increasing women’s access to control over the strategic life choices that affect them and access to the opportunities that allow them fully to realize their capacities.” [21]. Women’s empowerment can be expressed in terms of their participation in family and community decision-making, their contribution to household income, their participation in community leadership, their share in controlling household assets, having personal time, and others, depending on their status at home and in the community [22]. A woman’s path to positive personal and societal development is a journey. These changes strengthen her impact on her family, society, and institutions, which can alter her future life [23].

Women’s empowerment is a broad concept encompassing various strategies to enhance women’s social, economic, and political status. Scholars have developed different tools and methodological approaches tailored to specific contexts to measure women’s empowerment. For instance [22], introduced the Women’s Empowerment in Agriculture Index (WEAI) to measure the gender gap in agricultural productivity, becoming the most commonly used survey-based tool. Subsequently, the WEAI underwent modifications in 2015 and 2020 to refine its assessment [24]. Additionally, other indices, such as Kabeer’s Empowerment Index [25, 26] and Composite Women’s Empowerment Index (CWEI) [27], have been developed. Researchers such as [28,29,30,31] have also employed diverse approaches to measuring women’s empowerment in their studies.

The relationship between energy use and women’s economic empowerment is a significant area of research that different researchers significantly discuss. According to [32], energy access can catalyze women’s economic empowerment. Similarly [33], found a positive effect of the use of improved cookstoves on women’s participation in side businesses in rural India, demonstrating a direct link between energy use and economic empowerment. Furthermore, the International Center for Research on Women [34] emphasizes the importance of women’s access to modern energy services for improving women’s living standards in the developing world.

In this section, our focus is limited to women’s economic empowerment. Economic empowerment can be defined as the ability of individuals to engage in economic activities and earn a livelihood, as well as the capacity to make decisions about their economic well-being and to access resources and opportunities to improve it [34]. When we talk about empowering women’s economic potential, we mean women’s capacity to engage in the competitive market on an equal footing. Finding a decent profession, utilizing one’s time for personal objectives, enhancing and managing wealth, and having a voice, leadership, and meaningful involvement in economic decision-making at all levels, from the family to international organizations, all fall under this category [35].

The measurement of women’s economic empowerment is complex and challenging, requiring the use of indirect indicators and indices. Researchers have highlighted the need to consider multiple dimensions of economic empowerment, including access to resources, decision-making power, and skill development. Ferrant and Thim [36] argue that economic empowerment should be measured through composite indicators that capture both the breadth and depth of women’s economic participation. Laszlo and Grantham [37] suggest that a gender-equitable economic empowerment index should include indicators such as access to credit, property rights, and participation in decision-making processes. Furthermore, [38] emphasized the importance of considering contextual factors that influence women’s economic empowerment, such as cultural norms and legal frameworks. Therefore, a multidimensional and context-specific approach is necessary to accurately measure women’s economic empowerment.

This study aims to evaluate women’s economic empowerment in rural households. To achieve this, we have drawn upon the research findings of various scholars [22, 28, 39, 40] and employed three dimensions of equal weight: skills, resources, and decision-making ability.

The first dimension, skills, is essential for human development and overall growth. It includes indicators such as educational status, access to information, personal networks, and skills gained from training programs. The second dimension, resources, considers the possession of assets such as paid employment, house or land ownership, credit or bank accounts, leisure time, leadership roles, and control over household income. The third dimension, decision-making ability, measures a woman’s influence over household matters, such as large purchases, visiting family or relatives, control over earnings, and healthcare decisions for herself and her children (see Table 1).

Table 1 Given weight for each indicator and items
Full size table

Each indicator within these dimensions is measured using a binary variable approach, where a positive response is coded as one and a negative response as zero. This approach simplifies the data and facilitates analysis. However, it is important to note that while this method effectively captures binary outcomes, it may not fully capture the nuances or degrees of economic empowerment [41].

For each dimension, independent indices were constructed with a value range of 0 to 1. This range was based on a weighted average of the indicators within each dimension. The value of 0 represents an individual's lowest possible score, indicating a lack of empowerment in that dimension. Conversely, a value of 1 represents the highest possible score, indicating full empowerment in that dimension.

The methodology aligns with the International Center for Research on Women’s (ICRW) emphasis on a comprehensive measurement framework [34]. It also aligns with the Global Development Research Center’s (GDRC) call for clear definitions and transformations of variables in women’s economic empowerment measures [42].

However, the effectiveness of this methodology depends on the accuracy and comprehensiveness of the chosen indicators. It is crucial to ensure that the chosen indicators accurately represent the dimensions of economic empowerment they are meant to measure. This is particularly important in the context of rural households, where a variety of unique factors may influence women’s economic empowerment.

$$EI_{i} = \mathop \sum \limits_{j}^{n} \frac{{k_{ij} }}{n}, \,j = 1,2, \ldots , n$$
(1)

where EI is the empowerment index for each dimension, and kij is the jth indicator in the ith dimension.

The Cumulative Women’s Economic Empowerment Index (CWEEI) is computed for each woman by adding the empowerment indices across all dimensions—skills, resources, and decision-making ability—and then dividing by the total number of these dimensions. This results in an average score that encapsulates the overall degree of a woman’s economic empowerment across all dimensions. The CWEEI is a continuous measure ranging from 0, which signifies a lack of empowerment, to 1, which signifies the highest possible level of empowerment.

CWEEI can be expressed as:

$$CWEEI_{i} = \mathop \sum \limits_{i}^{m} \frac{{EI_{i} }}{m} , \quad i = 1, 2, 3$$
(2)

where CWEEI is the cumulative women’s economic empowerment index, and EIi is the empowerment index for each dimension.

3 Materials and methods

Primary data collection was conducted between January and July 2022, targeting the rural households in Ethiopia’s Southern Nations, Nationalities, and Peoples’ Region (SNNPR). A multi-stage sampling technique was employed for the selection of the respondent households. The entire rural woredas (districts) in the region were initially grouped into three clusters based on their geographical classifications, such as Highland (Duga), Middle altitude (Weyinadega), and Lowland (Kolla), as illustrated in Fig. 1. Six woredas, two in each cluster (Ezha, Yem, Cheha, Arbaminch Zuria, Endegagn, and Duna), were then randomly selected. Two kebeles (the lowest administrative unit) were selected from each selected woreda for the final household selection. Primary data were collected from 569 rural women (each representing a separate household) using a pre-tested structured interview schedule.

Fig. 1
figure 1

Map of the study areas SNNPR, Ethiopia

Full size image

This paper explores the relationship between women’s economic empowerment and the use of clean energy in rural households. The study focuses on the type of fuel typically used for cooking, categorized according to the theory of the energy ladder [43], such as traditional fuel (e.g., firewood, dung, and agricultural waste), transitional fuel (charcoal), and modern fuel (electricity). This results in a nominal dependent variable with three categories. To assess the impact of women’s economic empowerment on fuel choice, the study employs three indices constructed to measure women’s skills, resource possession, and decision-making power. Additionally, the study includes independent variables such as the household head’s sex, marital status, respondent’s age, geographic location, family size, and income (see Appendix D). The statistical software Stata (Version 14) performed the data analyses.

A multinomial logit model was used to see the relationship between the dependent and other exploratory variables. The multinomial logit model is used when multiple categories or outcomes exist for a dependent variable. The model estimates the probability of each category, given a set of independent variables. The probabilities are estimated using a logit function, which is a transformation of the linear combination of the independent variables. The model assumes that the errors are independently and identically distributed and that the probability of each category is mutually exclusive. The multinomial logit model is an extension to binary logistic regression when the dependent variable has more than two nominal categories. The type of independent variables in the model can be continuous, nominal, or ordinal scales.

Let \(Y = ({Y}_{1}, . . , {Y}_{k})\) be a vector of k integers, denoting the numbers of outcomes of n trials of a random experiment, with k outcomes. The probability of success of each outcome is \({\pi }_{i}\). For n independent observations, the multinomial probability that n1 falls in Category 1, n2 falls in Category 2, …, \({n}_{k}\) falls in Category k, where \(\sum_{j=1}^{n}{y}_{j}=n\). The probability mass function may be expressed as:

$$f\left( {y_{1} , . . , y_{k} ,n,\pi_{1} , \ldots ,\pi_{k} } \right) = P\left( {{\text{Y}}_{1} = y_{1} , \ldots ,Y_{k} = y_{k} } \right) = \left( {\frac{n!}{{y_{1} ! \ldots y_{k} !}}} \right)\pi_{1}^{{y_{1} }} \ldots \pi_{k}^{{y_{k} }}$$
(3)

Multinomial logit regression uses maximum likelihood estimation to evaluate the probability of categorical membership [44]. The model is used when there is no natural order among response categories.

The logit of the model is given regarding other categories. For example, if we consider the first category as a reference category, the logit of the other categories may be given by:

$$logit\left( {\pi_{j} } \right) = \log \left( {\frac{{\pi_{j} }}{{\pi_{1} }}} \right) = x_{j}^{T} \beta_{j } \,j = 1, \ldots , J$$
(4)

where (J-1) logit equations are used simultaneously to estimate \({\beta }_{j }.\) Once parameter \({\widehat{\beta }}_{j}\) is calculated then the probability of each category can be found as:

$$\hat{\pi }_{j} = \frac{{{\text{exp}}\left( {x_{j}^{T} \hat{\beta }_{j} } \right)}}{{1 + \mathop \sum \nolimits_{j = 2}^{J} {\text{exp}}\left( { x_{j}^{T} \hat{\beta }_{j} } \right)}} \quad for\;j = 2, \ldots ,J$$
(5)

In the multinomial logit model, the effect that an independent variable has on the dependent variable is reflected by a relative risk ratio (RRR), which is a relative odds ratio that estimates the relative risk of ending up in a specific category of the dependent outcome relative to the base outcome [44]. The assumptions of the multinomial model are essentially similar to those of the binary logit model, with the addition of independence of irrelevant alternatives (IIA). It means that if we add more categories, the relative odds of being in one category compared to the base category will remain the same [45]. In order to interpret the results of the model outputs, it is mandatory to check those assumptions.

4 Results and discussion

A total of 577 respondents from six different districts were interviewed, and 569 completed interview schedules were considered in the final study, indicating a 98.6% response rate. The respondents’ average age was 40 years; the youngest was 19 years old, and the oldest was 70 years old. The research area had an average household size of 4.6 individuals. In terms of annual total household income, an average of ETB 60,691.44 (USD 1188.03) was recorded (see Table 2).

Table 2 Descriptive statistics for continuous variables
Full size table

In the study area, males were the household heads of more than three-quarters of the total households. The majority of the respondents who participated in the study were married, with almost 80% being married. However, around 15% were widowed, and 5% were either single or divorced (see Table 3).

Table 3 Descriptive statistics for categorical variables
Full size table

Women’s economic empowerment may be measured using different techniques, with one common technique combining many indicators into a composite index. In this case, three indicators were used to assess women’s economic empowerment, and a Women’s Economic Empowerment Index was computed as a result. To construct the cumulative Women’s Economic Empowerment Index (CWEEI), each indicator was assigned a weight based on its relevance in measuring women’s economic empowerment. The individual indicator scores were combined to form a single score (see Table 4 and Appendix A). It was found that women in the area had moderate economic empowerment on average, with a CWEEI score of 0.583. The women’s decision-making index (WED) had the highest value of 0.636, indicating that women in the survey area had moderate involvement in household decision-making. The empowerment in skills had a score of 0.602, suggesting that women had a moderate skill level in economic activities. However, the resource possession index score was relatively low at 0.512, indicating that women were empowered less in the resource dimension than in the other indices [46]. In other words, women had limited control or access to resources such as land, finances, or paid jobs.

Table 4 Descriptive statistics for indexes of individual indicators
Full size table

Table 5 shows the average energy preference categories about the cumulative index of women’s economic empowerment. The majority of households, over 66%, opted for traditional energy sources, while more than 23% chose transitional sources, and only 10% used modern energy sources. The result reveals that women with a higher CWEEI score of 0.766 tend to select modern energy sources over those with lower CWEEI scores of 0.537 for traditional and 0.634 for transitional sources. The difference in mean CWEEI values is also significant, as shown in Appendices B1 and B2.

Table 5 Summary statistics for energy choice with CWEEI
Full size table

The multinomial logit model investigated the relationship between cooking fuel choice, women’s economic empowerment indices, and other independent variables. However, checking the model diagnosis test before interpreting the model output is better. These included a multicollinearity test, an IIA (Independence of Irrelevant Alternatives) test, and a goodness-of-fit test. The multicollinearity test indicated that all predictor variables had a VIF value less than 10, indicating no severe correlation among the predictor variables. The IIA test showed that the assumption was not violated, indicating the independence of the alternatives. The goodness-of-fit test demonstrated that it was superior to an intercept-only model. The predictive margins were also analyzed, and the model perfectly predicted the dependent variable’s category (see Appendices C1, C2, C3, and C4). As a result, the model is suitable for interpretation.

According to the multinomial logit result (Table 6), respondents’ marital status contributes to choosing energy sources for cooking. The variable’s coefficient indicates that single or divorced respondents favored transitional energy over traditional energy sources. Assuming all other variables are constant, the relative risk of choosing transitional energy over traditional energy for single or divorced respondents is expected to rise by a factor of 11.123. This may be because a larger proportion of single or divorced respondents live in rented houses than married respondents. It is found that more than 13% of single or divorced respondents live in rented houses, while only about 5% of married people live in rented houses. Most rented houses do not have traditional kitchens that allow the use of traditional energy sources, and the owners only allow electricity for lighting. So, renters may be forced to cook with charcoal.

Table 6 Estimation of multinomial model result
Full size table

The coefficient for household size indicates a negative relationship with the household’s energy choice. When family size increases, modern fuel is less likely to be used than traditional solid fuel. Increasing family size by one person will decrease household choice of transitional energy over traditional energy by a factor of 0.827 and modern energy over traditional energy by a factor of 0.711. Larger rural households may demand more energy to satisfy their needs, especially for cooking. This might lead to a higher reliance on traditional energy sources like firewood, animal dung, or agricultural waste. Bigger families may also have more trouble buying or gaining access to modern energy sources like electricity or clean cooking technology, which can be costly or unavailable in some rural locations. Several studies have found that a household with large family members is more likely to use traditional solid energy sources rather than clean and modern energy [18, 47,48,49,50,51].

A household's income plays a significant role in deciding whether to switch to a more advanced energy source. As household income increases, the likelihood of switching to a higher level of energy source also increases. Specifically, if the total logarithmic income of a household increases by one unit, the relative risk of selecting transitional energy sources over traditional sources would increase by a factor of 4.572, assuming all other factors are constant. In the same vein, if everything else stays constant, the relative risk of choosing modern energy over traditional sources would increase by a factor of 12.185. Several researchers have indicated that income is crucial in determining the transition to cleaner energy sources [12, 18, 51,52,53]. For example, a study conducted in China showed that household income positively influences household use of clean energy. As household income increased, the adoption of clean energy technologies also increased [54]. Generally, income is an important factor in energy transition and choice, and higher income levels are associated with a greater likelihood of adopting cleaner energy sources and technologies. Households with lower incomes in rural areas may have limited energy access and may rely on traditional energy sources. Lower-income households may struggle to obtain modern energy sources due to high upfront costs, lack of infrastructure, or limited availability. Additionally, the affordability of traditional biomass in rural areas gives households access to and utilize traditional biomass fuels such as firewood, agricultural residues, and animal dung for cooking purposes at a reasonable cost. Maybe traditional biomass is often the primary energy source due to its availability and accessibility in the study area.

The skill empowerment index was one of the indicators used to assess women’s economic empowerment. Which assesses women’s empowerment in terms of access to training and education. Results show a significant and positive relationship between women’s skill empowerment index and energy choice for cooking. If the respondents’ skill empowerment score increases by one unit, the relative risk of choosing transitional energy sources over traditional sources rises by a factor of 8.785. Similarly, selecting modern energy over traditional sources would raise the relative risk by a factor of 35.134, provided all other variables remained constant. This means women’s empowerment is crucial in adopting clean energy in the study area. Similar research results were also found in Nepal, which states that lower education levels are associated with kerosene and firewood usage [55]. Similarly, a study [56] in rural Bangladesh found that women’s education and decision-making power were positively associated with adopting improved cookstoves and clean energy technologies that reduce indoor air pollution.

Women’s empowerment on the resource possession index (WEP) also contributes to the choice of energy sources in rural areas. Table 6 shows that a woman with a better (WEP) index has a better chance of adopting clean energy sources. If a woman’s WEP index score increases by one unit, the relative risk of choosing modern energy sources over traditional sources rises by a factor of 11.126, keeping other variables constant. Choudhuri and Desai [19] also found that women’s access to salaried work and control over household expenditure decisions were positively associated with using clean fuels such as liquefied petroleum gas (LPG) and electricity. Women who own assets in rural households, such as land and livestock, may feel more secure and stable, which might encourage and provide economic freedom to adopt clean energy sources [57].

The ability to decide on household matters is also a crucial factor in the women’s empowerment index, significantly impacting the choice of energy source for cooking. The results show that when a woman’s empowerment to participate in household decision-making scores increases by one%, the relative risk of choosing transitional energy sources over traditional sources rises by a factor of 3.416, while selecting modern energy over traditional sources would raise the relative risk by a factor of 39.140, all other variables held constant. Thus, it is clear that women’s empowerment, particularly their decision-making power in the household, plays a crucial role in determining the choice of energy source for cooking. This relationship between women’s empowerment and the adoption of clean cooking technologies has been the subject of numerous studies, which have consistently found a positive association. For instance, a study in India found that women’s education and decision-making power in the household were important determinants of adopting clean cooking technologies [58]. Another study conducted across 31 African countries shows a positive association between women’s decision-making power in the household and the use of clean cooking fuels [59]. The findings of these studies are consistent with the results of the current study, providing compelling evidence that women’s empowerment, particularly their involvement in household decision-making, plays a crucial role in promoting the adoption of clean cooking fuels.

5 Conclusions

The study was based on primary data collected from 569 rural households in Southern Ethiopia. The results show that the majority of households, over 66%, opted for traditional energy sources, more than 23% chose to use transitional, and only 10% reported using a modern energy source. The average age of the respondents is 40 years, and the research area had an average household size of 4.6 individuals. Most households had male heads (75.4%), and most respondents were married (79.96%).

Women in the study area had moderate economic empowerment, with a CWEEI score of 0.583. Women showed moderate skill in economic activities (0.602) and moderate involvement in household decision-making (0.636). Nonetheless, compared to other indices in the research, the ownership of resources score was comparatively low (0.512), showing that women have lower ownership resources scores than the other indices. As a result, policies and initiatives should attempt to boost women’s resource ownership and involvement in economic activity. Furthermore, the study discovered that women with a higher CWEEI prefer modern energy sources over traditional or transitional ones. This shows that increasing women’s economic empowerment may lead to more sustainable energy choices.

The multinomial logit model found a significant relationship between household energy choice and marital status, household size, total household income, skill empowerment index, resource possession index, and women’s decision-making index. Households with larger family sizes were more likely to use a traditional energy source, possibly due to the need for more significant amounts of energy to accommodate the household’s energy needs. Policymakers and program implementers should consider household size when designing energy programs and interventions, and family planning programs should get serious attention in the area. Total household income was also a significant predictor, suggesting that higher-income households may have more access to and preference for modern energy sources. Promoting income-generating activities and exploring subsidized options for low-income households may be beneficial for encouraging the adoption of clean cooking.

Women’s economic empowerment is critical in enhancing the adoption and use of clean energy in rural households. Empowering women through education, decision-making, and income-generating activities can enhance their ability to understand and use clean energy technologies effectively. This, in turn, can improve rural households’ quality of life and livelihoods while mitigating environmental degradation. Therefore, policy interventions promoting clean energy use in rural areas should prioritize women’s economic empowerment as a key strategy.

The research only focused on women’s economic empowerment and did not consider a broader analysis of women’s empowerment. To improve future research, it would be beneficial to investigate how different aspects of women’s empowerment, such as health, employment, political participation, access to information, and education, are related to each other and to the development outcomes. Additionally, research should consider a broader range of the population, more than one region, in the country when examining the choice of energy sources to identify potential opportunities for intervention and areas for improvement.