Beyond Adoption: Sustainability and Resilience Dimensions of Household Biogas Systems in West Java, Indonesia

Abstract

This study examines the determinants and impacts of household biogas adoption among dairy-based mixed crop–livestock systems in West Java, Indonesia. Using primary survey data from 201 households, we estimate adoption drivers through logistic regression and assess post-adoption outcomes using propensity score matching combined with doubly robust estimation. The results show that adoption is primarily driven by structural feasibility and institutional exposure, particularly livestock ownership, participation in technical training, perceived time-saving benefits, and fuel-cost pressure, while general socioeconomic variables such as income and education are not statistically significant. Treatment-effect estimates indicate that adoption leads to significant reductions in LPG and firewood consumption, as well as decreased use of chemical fertilizers, reflecting partial substitution of external inputs with locally available resources. However, these benefits are unevenly distributed, with stronger effects observed among households with larger livestock holdings, while training plays a more critical role for smaller-scale farmers. The findings are interpreted through a sustainability–resilience framework, which is used as an analytical lens rather than a causal measurement model. The results highlight the importance of institutional support, service provision, and policy alignment in determining the durability and scalability of biogas adoption. The study contributes to the literature by integrating determinants of adoption with causal impact estimation and situating household-level outcomes within broader socio-technical systems.

1. Introduction

The world is moving towards cleaner energy sources. People are looking for ways to use energy that is not controlled by a single entity. One way is to use biogas systems in people’s homes [1,2]. Biogas systems convert animal waste into energy that people can use. This can help people become more self-sufficient and rely less on traditional energy sources. It can also improve indoor air quality and lower household energy costs.

There are benefits to biogas systems. They can help reduce waste from animals and turn it into something useful. They can also help people have a stable source of energy. However, with all these good things, not many people are using biogas systems. This is especially true in countries where people have little money [3,4].

In Indonesia, the government plans to use renewable energy sources. However, most people still use energy from fuels, like gas. This is because the government helps pay for the gas, making it cheaper for people to use. In areas, there is a lot of animal waste that could be used to make biogas; instead, people are still using gas [5,6].

West Java is a place to look at how biogas systems are working. This area has many animals. It is good for making biogas. Even here, not many people are using biogas systems. Some studies have looked at why this is the case. They found that one reason is that the systems are poorly built. Another reason is that people do not have access to people who can help them fix the systems if they break. It is also hard for people to get the money they need to buy the systems. The gas is still a good option for people because it is cheap.

These studies show that deciding to use biogas systems is not about the money. It is also about how easy the systems are to use and how well they work. People need to know how to use the systems and how to fix them if they break. The government and other organizations can help by providing support and ensuring the systems are of high quality.

Some studies have used methods to see if biogas systems are really helping people. They found that biogas systems can help people save money on energy. These studies only looked at what happened in the short term. They did not look at how the systems changed over time. A different way of thinking about systems can help us understand this better. It is called sustainability and resilience. This way of thinking examines how systems can keep functioning when things get tough. It also looks at how systems can change and adapt over time. This is important for biogas systems because they need to keep working even when conditions are not ideal.

When we apply this way of thinking to biogas systems, we can see that they are not about the technology. They are also about how people use them and how they are supported. For example, if people have access to training and support, they are more likely to use the biogas systems and keep using them over time. Most studies have examined either why people adopt biogas systems or how these systems affect people’s lives. They have not looked at both things together. This study will examine both why people adopt biogas systems and how these systems affect people’s lives. It will use methods to assess how biogas systems are working in West Java. It will also look at how the systems can be supported and made sustainable over time.

Biogas systems are a way to generate energy that is better for the environment. The biogas systems can help people in West Java have a stable source of energy. The government and other organizations can help by providing support and ensuring the biogas systems are of high quality. The biogas systems are not about the technology; they are also about how people use them and how they are supported. The biogas systems can help people save money on energy and improve their lives. The biogas systems are a part of the move towards using energy that is better for the environment.

Indonesia represents a particularly important context for examining these dynamics. As one of the world’s most populous countries and a major greenhouse gas emitter, Indonesia remains heavily dependent on fossil fuels, especially liquefied petroleum gas (LPG), whose long-standing subsidy regime strongly shapes household energy choices. Although national policy frameworks—including the National Energy Policy (KEN), the General Plan for National Energy (RUEN), and the goals of the renewable energy mix, targets—formally promote the expansion of renewable energy, implementation at the household level has been fragmented. In livestock-intensive rural regions, livestock manure constitutes a substantial but underutilized resource for biogas development, indicating a persistent gap between policy ambition and on-the-ground outcomes [7].

Table 1. Household energy access and fuel use indicators in Indonesia compared with Southeast Asia [8].

Indicator	Indonesia	Southeast Asia (Regional Average/ASEAN)	Source
Access to electricity (% of population)	99–100%	90%	World Bank WDI—Access to electricity (% of population) [9]
Access to clean cooking fuels (% of population)	80–90%	60–70%	World Bank WDI—Clean fuels indicator [9]
Household reliance on traditional biomass (solid fuels)	26%	40%	OECD/IEA regional report (2017) [9,10]
Renewable energy share in final energy consumption (includes bioenergy)	12–15%	23%	World Bank WDI; IRENA report [9,10,11]

situates Indonesia within the regional energy context. Although access to electricity exceeds 99%, reliance on traditional biomass remains significant in rural areas, underscoring the persistent challenges of achieving clean household energy transitions. Access to clean cooking is still less common throughout Southeast Asia, with regional averages of about 60% and significant differences between urban and rural areas. According to the World Bank’s World Development Indicators, access to electricity (% of population) reflects the share of people connected to national grids or off-grid services [8]. This variety emphasizes ongoing reliance on biomass and unequal access to clean energy that affects household environmental impacts, exposure to air pollution, and greenhouse emission profiles—a crucial context for comprehending the environmental consequences of biogas adoption in your research.

A growing body of recent research has strengthened the evidence base on biogas adoption by applying quasi-experimental methods, such as propensity score matching, to estimate causal impacts. These studies demonstrate that adopting biogas can generate measurable welfare benefits, particularly through reductions in household energy expenditures. However, their analytical focus has largely remained confined to private household outcomes, offering limited insight into the institutional, governance, and policy conditions that determine whether adoption is sustained, scaled, or translated into broader system-level change [9].

West Java, Indonesia’s leading dairy production region, illustrates this paradox clearly. The province combines high livestock density, established cooperative structures, and favorable agro-ecological conditions that should support biogas diffusion. Nevertheless, adoption remains modest even in this relatively enabling context. Previous studies identify constraints including uneven construction quality, limited maintenance services, financing barriers, inconsistent program design, and strong competition from subsidized LPG [9,10]. At the household level, adoption decisions are shaped not only by livestock ownership and expected economic returns, but also by access to training, perceived convenience, and confidence in the long-term reliability of the system [9,11].

Insights from sustainability and resilience research offer a complementary lens for addressing this limitation. Integrated assessments of farming systems emphasize that technology adoption outcomes are embedded within multi-level socio-technical systems, where household decisions interact with institutional support structures, service provision, and policy regimes [9,12]. Within this literature, resilience is commonly conceptualized through three interrelated attributes: robustness, referring to the capacity to maintain core functions under stress; adaptability, denoting the ability to adjust practices through learning and resource reallocation; and transformability, capturing the potential for structural change when existing systems become unsustainable. These attributes depend not only on household assets, but also on institutional coherence, advisory services, and policy alignment [11,13].

Applied to the adoption of household biogas in West Java, this perspective suggests that the outcomes of adoption cannot be fully understood only by welfare metrics. Robustness relates to feedstock stability and energy security; adaptability reflects access to training, knowledge, and operational confidence; and transformability depends on coordination among cooperatives, extension services, and national energy-policy regimes. For example, fuel-cost pressure—often identified as an adoption driver—cannot be separated from Indonesia’s LPG subsidy structure, while perceived time-saving benefits align with resilience research emphasizing labor constraints as a key determinant of adaptive capacity [14,15].

Previous research on household biogas systems has typically examined either adoption determinants—such as income, herd size, credit access, and training—or post-installation environmental and welfare outcomes, with limited integration between the two. Moreover, claims about improved energy security or livelihood resilience are often made without grounding them in measurable household indicators. This study contributes to the literature in three ways. This study makes three distinct contributions. First, it integrates the analysis of adoption determinants with causal estimation of post-adoption outcomes within a single empirical framework, addressing a common gap in the biogas literature where these dimensions are often examined separately. Second, it provides new empirical evidence from dairy-based mixed crop–livestock systems in West Java, a context characterized by high technical potential for biogas alongside persistent adoption constraints shaped by subsidized LPG markets. Third, rather than treating resilience as a measurable outcome, the study employs a sustainability–resilience framework as an interpretive lens to situate empirically estimated household-level effects within broader institutional and policy contexts.

Biogas Development in Indonesia

Biogas development has received increasing attention in Indonesia, particularly in livestock-intensive regions such as West Java, where environmental pressures from manure management have intensified over time. Dairy production centers in Lembang and cattle-farming districts, including Garut and Tasikmalaya, have experienced sustained methane and nitrous oxide emissions associated with livestock waste, trends documented since the early 1990s and now recognized as a growing component of national greenhouse gas emissions [12,16]. Mixed crop–livestock (MCL) systems dominate smallholder agriculture in West Java, where limited land availability encourages integration of crop and livestock production. These systems enable nutrient recycling through manure use in fodder and vegetable cultivation, but also generate environmental risks when waste is poorly managed. In densely populated upland areas, inadequate manure handling contributes to water and air pollution, creating a need for improved waste-management technologies such as anaerobic digestion [17]. Nationally, approximately 42% of Indonesian households engage in farming, with more than half operating smallholder MCL systems as their primary livelihood strategy. These systems are globally significant, as nearly two-thirds of the rural poor depend on integrated crop–livestock production for subsistence and income generation [12,18,19].

Within dairy clusters such as Lembang, the integration of crop and livestock activities supports nutrient recycling within farm boundaries and ensures forage availability for smallholder cattle herds. As dynamic production systems, MCL farms link agronomic, livestock, and socioeconomic components of household livelihoods [8,20]. However, when manure is inadequately managed, these same systems become environmentally vulnerable. This challenge is particularly acute in West Java’s densely populated upland landscapes, where improper storage and disposal of livestock waste frequently result in localized water and air pollution [20,21].

Anaerobic digestion offers a technically well-established response to these challenges by stabilizing organic waste, reducing odors and pathogens, improving nutrient recovery, and producing biogas as a renewable household energy source [22,23]. For smallholder farmers in West Java, these benefits extend beyond environmental protection to include improved nutrient management in fodder and vegetable production systems, which are common across Lembang, Garut, and Tasikmalaya. At the national level, Indonesia’s energy consumption grew by approximately 3% annually between 2000 and 2011 and is projected to increase by 4.7% per year through 2030, driven by economic growth and rising household demand [8,14,18,20]. Despite relatively modest growth in household energy consumption, fossil fuels continue to dominate the energy mix, supplying nearly 80% of total demand [22]. In this context, household-scale biogas adoption has been shown to reduce expenditures on cooking fuels such as LPG and firewood by at least 40%, highlighting its potential economic relevance for rural households [24]. Biogas technology was introduced in Indonesia in the 1970s and expanded during the 1980s through government programs [9,13,15]. Despite subsequent promotion initiatives, including the BIRU program and support from international organizations, the total number of household digesters remains relatively low compared with other developing countries. Subsequent promotion by public and private actors led to the construction of more than 7000 household digesters nationwide by 2012, supported in part by organizations such as SNV The Netherlands [18]. Nevertheless, Indonesia’s total number of digesters remains low relative to other developing countries. Two main digester types are used: communal systems requiring more than 30 cattle and thus accessible primarily to larger farms [25], and household-scale digesters ranging from 4 to 12 m³, which are better suited to smallholders typically keeping two or three cows [26]. Despite their lower investment requirements and dual production of energy and organic fertilizer [27], diffusion of household-scale biogas systems among MCL farmers has remained slow for more than three decades.

Table 2. Estimated annual greenhouse gas emissions (CO₂-equivalent) from traditional biomass fuels (firewood, cow dung) compared with biogas under typical rural household use. Data adapted from Akter et al. (2021) [28].

Fuel Type	Estimated GHG Emissions (CO2-eq per Year)
Firewood	122.5 t CO2 eq annual (for sample households)
Cow dung	47.3 t CO2 eq annual (for sample households)
Biogas	1.9 t CO2 eq annual (net emission per biogas unit)

below compares estimated annual GHG emissions across common household fuels, illustrating that biogas systems are associated with substantially lower emissions compared to firewood and cow dung combustion [28]. Previous research highlights that perceived benefits—particularly access to reliable cooking fuel and organic fertilizer—play an important role in motivating biogas adoption. These benefits are expected to reduce dependence on firewood and LPGand low er the use of chemical fertilizers in crop production [29]. However, much of the existing literature relies on descriptive comparisons between adopters and non-adopters, making it difficult to distinguish the effects of biogas adoption from pre-existing household characteristics [30,31]. Emission comparisons presented in Table 2 should be interpreted as illustrative estimates derived from published emission factors rather than as direct measurements from surveyed households. Although recent treatment-effect studies have strengthened causal inference by isolating welfare impacts, their analytical focus has largely remained on short-term household outcomes, offering limited insight into the institutional and system-level conditions that shape adoption persistence, performance, and scalability [32,33].

Against this background, this study examines household adoption of biogas in dairy-based MCL systems in West Java using a treatment-effect framework, extending the interpretation beyond welfare outcomes. Rather than asking only whether biogas adoption generates measurable benefits, the analysis situates adoption within a broader sustainability–resilience perspective that emphasizes household capabilities, institutional support, and policy environments. By integrating causal estimation with system-oriented interpretation, the study contributes new insights into why biogas adoption remains uneven even in resource-rich contexts and under what conditions it can support durable sustainability and resilience outcomes among smallholder farming households [20,34,35]. The bar chart in Figure 1 shows a steady increase in installed biogas capacity over the five-year period. Capacity rises from approximately 95 MW in 2021 to 110 MW in 2022, then continues to grow to 118 MW in 2023, 125 MW in 2024, and reaches 135 MW in 2025. This upward trend indicates consistent expansion in biogas energy development and increasing investment in renewable energy infrastructure during the period.

Despite the long history of biogas promotion in Indonesia and the substantial technical potential associated with livestock-based mixed crop–livestock systems, adoption and sustained use among smallholder households remain limited. Existing studies provide valuable descriptive insights into expected benefits and perceived constraints, yet they often lack robust empirical strategies capable of isolating the causal effects of biogas adoption from underlying household characteristics and contextual factors. As a result, it remains unclear which household and institutional conditions most strongly drive adoption and whether the anticipated energy, environmental, and livelihood benefits are realized in practice [20,37,38].

Figure 2 illustrates the integrated biogas production system and its various applications. Organic feedstocks such as livestock manure, agricultural crops, wastewater, and food waste are collected and fed into an anaerobic digester, where microorganisms break down the organic matter in the absence of oxygen to produce biogas. The generated biogas can be used directly for heat and electricity generation, or upgraded to biomethane. Upgraded biomethane can then be utilized as a transport fuel or injected into the natural gas grid. In addition to energy production, the digestion process produces a by-product known as digestate, which can be applied as fertilizer, soil amendments, or livestock bedding, contributing to nutrient recycling and supporting sustainable agricultural practices [39].

To address these limitations, this study applies a household-level quantitative approach that explicitly distinguishes between the determinants of biogas adoption and the realized impacts of adoption on key household outcomes. By combining regression analysis with treatment-effect estimation and situating the results within a sustainability–resilience framework, the study provides a systematic assessment of how household characteristics, institutional support, and structural conditions jointly shape biogas adoption and its contribution to sustainable and resilient rural livelihood [5,13,14,41]. The following section details the study design, sampling strategy, and analytical methods employed to achieve these objectives. While the econometric approach follows established treatment-effect methods used in biogas adoption research, its purpose here is not limited to estimating welfare impacts but to provide a causal foundation for interpreting adoption outcomes within a sustainability–resilience framework that emphasizes institutional and system-level conditions [42,43,44].

2. Materials and Methods

2.1. Study Design and Data Collection

The analysis is based on primary survey data collected from 201 dairy-farming households in West Java Province. Figure 3 shows the western part of Indonesia, spanning approximately 46,299 km² between 5°50′–7°50′ S and 105°–108°30′ E, and is home to around 47 million inhabitants. Adopter households were initially identified through cooperative installation records, while non-adopters were sampled from neighboring dairy households within the same villages to ensure comparable agro-ecological and market conditions. To account for differential selection probabilities across districts and cooperative affiliations, inverse-probability weights were constructed, and all models were estimated with and without weights to assess robustness. The sustainability–resilience framework is used as an interpretive lens rather than as a directly estimated latent construct, situating observable household outcomes within broader socio-technical systems. Accordingly, the empirical analysis relies primarily on measurable socioeconomic and institutional indicators, including livestock ownership, participation in training, and changes in household energy use [27,36,45,46].

2.2. Variables

2.2.1. Adoption Indicator

A binary variable equals 1 if the household operates a functional biogas digester at the time of survey and 0 otherwise.

We estimate the probability of adoption using a logistic regression:

$$P(Adop{t}_i=1)=\Lambda \left(X_i\beta \right)$$

where $X_i$ includes:

• Herd size: Number of dairy cattle owned by the household, representing manure availability for biogas production.
• Household income: Log of annual household income, capturing the household’s financial capacity.
• Education of household head: Years of formal schooling completed by the household head.
• Access to credit: Binary indicator reflecting whether the household has access to formal or informal credit.
• Participation in training: Binary variable equal to 1 if the household attended at least one biogas-related training session in the past 24 months.
• Training intensity: Number of training sessions attended, used to explore variation in technical exposure where data are available.
• Asset index (PCA): Composite indicator of household wealth constructed using principal component analysis.
• Distance to cooperative: Distance to the nearest cooperative or technical service provider, representing access to institutional support.
• District fixed effects: Dummy variables included to control for unobserved district-level differences.
• Average marginal effects (AMEs): Reported to facilitate interpretation of the logistic regression results.

2.2.2. Structural Variables

• Herd size (number of dairy cattle): Measures livestock ownership and potential manure availability for biogas production.
• Log of annual household income: Log-transformed annual household income used to capture financial capacity while reducing skewness in income distribution.
• Education of household head (years of schooling): Total years of formal education completed by the household head, representing human capital.
• Asset index (constructed via principal component analysis): Composite indicator of household wealth derived from ownership of productive and household assets using PCA.

2.2.3. Institutional Variables

• Participation in technical training (number of sessions attended)—Measured as a binary indicator equal to 1 if the household attended at least one biogas-related training session in the past 24 months. This variable captures exposure to training but does not reflect differences in duration, quality, or technical depth. Estimated effects should therefore be interpreted as general exposure and may be subject to attenuation bias.
• Access to cooperative-based maintenance services—Binary indicator reflecting whether the household has access to maintenance support from a local cooperative.
• Distance to nearest technical service provider—Distance from the household to the nearest biogas technician or service provider, representing accessibility of technical support.
• Economic exposure—Measured as the share of LPG expenditure relative to annual household income, capturing household exposure to cooking-fuel costs.

2.2.4. Outcome Variables

• Monthly LPG consumption (kg): Quantity of LPG used per month, indicating reliance on commercial cooking fuel.
• Monthly firewood use (kg): Amount of firewood consumed per month as a traditional household energy source.
• Annual chemical fertilizer purchases (kg): Quantity of chemical fertilizers purchased annually, reflecting input substitution after biogas adoption.
• Time spent collecting fuel (hours/week): Weekly time devoted to collecting fuelwood or other traditional fuels.
• Self-reported ability to conduct minor maintenance: Household’s reported capability to perform basic biogas digester maintenance tasks.

2.3. Empirical Strategy

Fuel-cost pressure is measured as a perception-based indicator reflecting whether households report cooking fuel costs as financially burdensome. This approach is used because detailed expenditure data are not consistently available in smallholder contexts. While perceptions are relevant for behavioral decision-making, they may diverge from objective expenditure shares. The variable should therefore be interpreted as capturing perceived affordability rather than actual energy poverty. Future research should combine perception-based and objective indicators to improve measurement validity.

Perception-based indicators are therefore commonly used in studies of household energy behavior to capture the behavioral influence of economic pressure. Nevertheless, perceived fuel-cost pressure may not perfectly correspond to objective expenditure levels. The variable should therefore be interpreted as reflecting perceived affordability constraints rather than a precise energy-poverty indicator. Future research could complement perception-based measures with objective indicators, such as fuel expenditure as a share of household income or multidimensional energy-poverty indices, to provide additional validation.

Table 3. Demographic of Respondents and economic characteristics of biogas adopter (N: 201).

Characteristic	Category	Adopters (n = 101)	%	Non-Adopters (n = 100)	%
Gender of household head	Male	78	77.2	82	82.0
	Female	23	22.8	18	18.0
Education level	No schooling/Illiterate	6	5.9	23	23.0
	Primary school	21	20.8	37	37.0
	Secondary education	54	53.5	33	33.0
	Post-secondary/Vocational/University	20	19.8	7	7.0
Household size (persons)	1–3	22	21.8	38	38.0
	4–6	61	60.4	49	49.0
	≥7	18	17.8	13	13.0
Age of household head (years)	25–34	9	8.8	22	22.0
	35–44	41	40.6	28	28.0
	45–54	31	30.7	29	29.0
	55–64	15	14.9	13	13.0
	≥65	5	5.0	8	8.0
Number of cattle owned	1–4	11	10.9	32	32.0
	5–8	59	58.4	53	53.0
	≥9	31	30.7	15	15.0
Landholding size (ha)	<0.25	7	6.9	22	22.0
	0.25–0.50	25	24.8	36	36.0
	0.51–1.00	33	32.7	28	28.0
	1.01–1.50	23	22.8	10	10.0
	>1.50	13	12.8	4	4.0

presents the distribution of key household and farm characteristics among biogas adopters (n = 101) and non-adopters (n = 100). Results indicate that most households in both groups are male-headed, accounting for 77.2% of adopters and 82.0% of non-adopters, while female-headed households represent a smaller proportion. In terms of education, adopters tend to have higher educational attainment, with 53.5% having secondary education and 19.8% having post-secondary or vocational education, compared with 33.0% and 7.0%, respectively, among non-adopters. Conversely, illiteracy is more common among non-adopters (23.0%) than adopters (5.9%).

Regarding household size, most adopter households consist of 4–6 members (60.4%), whereas non-adopters show a higher proportion of smaller households (1–3 members, 38.0%). The age distribution suggests that adoption is most common among household heads aged 35–44 years (40.6%), followed by those aged 45–54 years (30.7%).

Farm resource variables also differ between the two groups. A larger share of adopters own more cattle, with 30.7% owning nine or more, compared to 15.0% among non-adopters, indicating that livestock availability—an important source of feedstock for biogas production—may influence adoption. Similarly, adopters generally possess larger landholdings, with 35.6% owning more than 1 ha, compared to 14.0% among non-adopters. Overall, the data suggest that education level, livestock ownership, and landholding size are important socio-economic factors associated with the adoption of biogas technology. Participation in training is measured as a binary indicator capturing whether the household attended at least one training session within the previous 24 months. This variable reflects exposure to training but does not capture differences in training quality, duration, or content. As a result, the estimated coefficient should be interpreted as an average effect of exposure rather than the effectiveness of specific training designs. This simplification may introduce measurement error and potential attenuation bias, which should be considered when interpreting the results.

Fuel-cost pressure is measured as a perception-based indicator reflecting whether households consider cooking-fuel expenses to be financially burdensome. This approach is used due to the limited availability of reliable expenditure records in smallholder contexts. While perceptions are relevant for decision-making, they may not fully correspond to objective energy expenditure levels. The variable should therefore be interpreted as capturing perceived affordability rather than actual energy poverty.

2.3.1. Adoption Model

A binary logistic regression estimates the probability of adoption as a function of structural and institutional variables. Average marginal effects (AMEs) are reported for interpretability. Standard errors are clustered at the village level. Model diagnostics include Nagelkerke pseudo-R², Hosmer–Lemeshow goodness-of-fit tests, and variance inflation factors (VIF) to assess multicollinearity.

2.3.2. Treatment Effects

To estimate the impact of adoption on outcome variables, propensity score matching (PSM) is implemented. Nearest-neighbor and kernel matching approaches are applied. Doubly robust estimation combines inverse probability weighting with outcome regression adjustment. Balance quality is assessed using standardized mean differences (SMD) before and after matching. Sensitivity to hidden bias is evaluated using Rosenbaum bounds.

2.3.3. Econometric Analysis of Adoption Determinants

Determinants of household biogas adoption were examined using a binary logistic regression model, with adoption status as the dependent variable. Explanatory variables were selected based on innovation–diffusion theory, prior biogas adoption studies, and the institutional context of West Java, and include household demographics, livestock ownership, income, education, electricity access, perceived fuel-cost pressure, perceived time-saving benefits, cooperative engagement, and training participation.

Continuous variables were standardized to facilitate interpretation. Model adequacy was assessed using multicollinearity diagnostics, likelihood ratio tests, goodness-of-fit measures, and receiver operating characteristic (ROC) analysis.

2.3.4. Sustainability–Resilience Interpretation

The empirical results are interpreted using a sustainability–resilience framework that situates household biogas adoption within broader socio-technical systems. Resilience is not estimated as a latent construct but used as an analytical lens linking observed outcomes to three attributes: robustness, adaptability, and transformability. Robustness refers to the stability of household energy supply under fuel-cost pressures. Adaptability reflects households’ operational capabilities and learning processes supported by training and technical knowledge. Transformability concerns institutional coordination and alignment between renewable energy initiatives and prevailing energy-policy regimes. This framework provides a basis for interpreting how adoption outcomes relate to the durability and scalability of decentralized energy systems. The analysis estimates the magnitude of an unobserved factor (Γ) required to alter the statistical significance of the treatment effects. In this framework, robustness reflects the stability of household energy supply and the capacity to maintain energy functions under fuel-cost pressures; adaptability relates to households’ operational capability, learning processes, and access to training and technical knowledge; and transformability refers to the broader institutional and policy conditions, including coordination among actors and alignment between renewable energy initiatives and prevailing energy-policy regimes, that shape long-term system change. This interpretive approach enables the empirical findings to be assessed in terms of their durability, scalability, and long-term relevance for livestock-based rural systems in West Java [40,47,48].

2.3.5. Sustainability–Resilience Framework

Household-level results are interpreted using a sustainability–resilience framework to situate biogas adoption within broader socio-technical systems. Rather than directly measuring resilience, resilience concepts are used analytically to interpret how household capabilities, institutional support, and policy conditions shape adoption trajectories and outcomes [48,49].

The sustainability–resilience framework is used as an interpretive tool rather than a causal measurement framework. Observable indicators such as energy use, time allocation, and maintenance capability are interpreted in relation to resilience attributes (robustness, adaptability, transformability), but no direct causal claims are made regarding resilience outcomes. Instead of treating resilience as a formally measured construct, this study uses the sustainability–resilience framework as a way to interpret how household-level outcomes fit within wider socio-technical systems.

Indicators such as energy use and substitution, time allocation, operational capability, and institutional engagement are therefore discussed in relation to resilience dimensions, but the framework itself is not used to establish causal relationships or quantify resilience outcomes. Within this perspective, robustness refers to a household’s ability to maintain basic energy functions when facing fuel-cost pressures.

Adaptability captures the role of learning, access to training, and the development of operational skills, while transformability relates to how well institutional arrangements and policy environments support longer-term structural change. Taken together, this approach helps assess whether biogas adoption contributes to durable, scalable, and long-term-relevant outcomes for the development of livestock-based rural systems [25,50,51].

Figure 4 above presents a conceptual framework illustrating the key factors influencing the adoption and sustained use of household biogas technology, as well as the resulting sustainability outcomes. The framework highlights a sequence of interrelated components, beginning with the policy and energy regime, which includes factors such as LPG subsidies, renewable energy policies, and governance coherence, that shape the broader enabling environment for renewable energy technologies.

The second component, institutional enablement, refers to the support systems that facilitate adoption, including training and extension services, cooperatives, and maintenance services. These institutional mechanisms help households gain the knowledge, technical support, and organizational assistance required to implement and manage biogas systems effectively.

The third component focuses on household capability, which includes household-level resources and perceptions that influence adoption decisions. These factors include livestock ownership (a key source of feedstock), labor availability, perceptions of time-saving benefits, and fuel-cost pressures that motivate households to seek alternative energy sources.

These enabling conditions influence adoption and performance, which refers to the installation and sustained use of biogas systems at the household level. Finally, the framework links adoption outcomes to broader sustainability outcomes, including system robustness, adaptability, and the potential for long-term transformation of rural energy systems, although the figure indicates that transformability may remain limited depending on contextual constraints.

Overall, the framework emphasizes that biogas adoption is influenced by multi-level factors including policy, institutional support, and household resources, which together determine the effectiveness and sustainability of renewable energy technologies in rural settings [29,39,53,54].

3. Results

3.1. Descriptive Characteristics of Adopters and Non-Adopters

Table 4. Description and Expected Effects of Socioeconomic and Institutional Variables Influencing Household Biogas Adoption.

Variable	Type	Description	Operationalization	Sign	Justification
X1 Age	Continuous (centered, scaled)	Age of household head (years)	(Age − mean)/SD	±	Mixed: younger HH heads more open to innovation, older may have resources but risk-aversion.
X2 Gender	Categorical	Gender of household head (male =1, female =0)	Dummy	+	Male heads more likely to control resources, but female-led HH may value timesaving strongly.
X3 Family size	Continuous (centered, scaled)	Number of household members	Raw count, standardized	+	Larger HH consume more energy → higher incentive to adopt biogas.
X4 Education	Continuous (years of schooling)	Years of formal schooling of household head	Centered/scaled	±	Theory: higher education improves technology uptake; empirical (West Java) showed negative results due to LPG substitution.
X5 Household income	Continuous (annual, million Rupiah)	Annual HH income	Log-transformed	+	Higher income eases upfront investment and maintenance costs.
X6 Electricity access	Categorical	Access to grid electricity (1 = yes, 0 = no)	Dummy	−	Grid electricity/LPG access can substitute biogas → reduces adoption probability.
X7 Fuel-cost pressure	Categorical	Household perceives fuel costs as high	Dummy (1 = yes)	+	Strong predictor: high fuel burden motivates biogas adoption.
X8 Livestock ownership	Continuous (cow equivalents)	Number of cattle owned	Centered/scaled	+	Provides feedstock for biogas, but diminishing returns if herd size too large.
X9 Timesaving	Categorical	Household reports time saved due to biogas	Dummy (1 = yes)	+	Major driver; particularly valued by women (fuelwood collection, cooking).
X10 Training on biogas technology	Categorical	Hands-on training received in last 24 months	Dummy (1 = yes)	+	Increases technical knowledge, reduces digester failure risk, improves adoption odds.

outlines the key socioeconomic and behavioral variables influencing household biogas adoption, detailing their types, measurement methods, and expected effects. It highlights how demographic, economic, and institutional factors, such as age, income, livestock ownership, fuel-cost pressure, and training jointly shape the likelihood and sustainability of biogas adoption and summarizes the socioeconomic and farm characteristics of biogas adopters (n = 101) and non-adopters (n = 100). Several differences are observed, particularly in education, livestock ownership, and landholding size.

Adopter households exhibit higher educational attainment: 53.5% have completed secondary education and 19.8% post-secondary education, compared with 33.0% and 7.0% among non-adopters, respectively. Non-adopters are more concentrated among households with no schooling or only primary education. Household size and age distributions differ modestly, with adopters more frequently located in medium-sized households (4–6 members) and economically active age groups (35–54 years) [46,55].

Livestock ownership shows the most pronounced contrast. Nearly 31% of adopters own nine or more cattle, compared with 15% of non-adopters, while one-third of non-adopters own fewer than five cattle. Landholding size follows a similar pattern, with adopters more likely to operate larger farms. Gender composition is comparable across groups, with male-headed households accounting for approximately four-fifths of both adopters and non-adopters [56,57,58]. These descriptive patterns indicate that adopters generally possess stronger productive asset bases, though causal inference requires multivariate and counterfactual analysis.

3.2. Determinants of Biogas Adoption

Table 5. The Logit Model Results in Determining Biogas Adoption.

Variable	β	OR	SE	z	p	Significance
Constant	−3.212	—	0.88	–3.64	<0.001	<0.001
Livestock ownership (TLU)	0.684	1.98	0.21	3.25	0.001	≤0.01
Training participation (1 = yes)	1.247	3.48	0.34	3.65	<0.001	<0.001
Perceived time-saving benefit	0.593	1.81	0.18	3.29	0.001	≤0.01
Fuel-cost pressure	0.441	1.55	0.15	2.94	0.003	≤0.01
Education (years)	0.067	1.07	0.05	1.34	0.181	>0.05
Household income (IDR million/month)	0.014	1.01	0.02	0.62	0.534	>0.05
Household size	0.112	1.12	0.10	1.12	0.262	>0.05
Landholding (m2)	0.00003	1.00	0.00	1.01	0.311	>0.05
Electricity access (1 = yes)	0.185	1.20	0.39	0.47	0.639	>0.05

Note. β = coefficient; OR = odds ratio; SE = standard error.

reports the binary logistic regression results identifying factors associated with biogas adoption. The model demonstrates satisfactory explanatory power, with the likelihood ratio test rejecting the null model (p < 0.001), a Nagelkerke pseudo-R² of 0.42, and a receiver operating characteristic (ROC) value of 0.82, indicating strong discriminatory performance. Variance inflation factors remain below the commonly accepted threshold of 5, suggesting that multicollinearity is not a major concern [54].

Livestock ownership is a key determinant of adoption (β = 0.684, p < 0.001), with an odds ratio of 1.98, confirming the importance of manure availability. Participation in biogas-related training exhibits the strongest effect, increasing adoption likelihood by more than threefold (OR = 3.48, p < 0.001). Perceived time-saving benefits also significantly raise adoption probability (OR = 1.81, p = 0.001), highlighting the role of labor considerations. Fuel-cost pressure is positively associated with adoption (OR = 1.55, p = 0.003), indicating responsiveness to cooking-energy expenditure constraints. The logistic regression model demonstrates satisfactory explanatory power, with a Nagelkerke pseudo-R² of 0.42 and a receiver operating characteristic (ROC) value of 0.82, indicating strong discriminatory performance. Variance inflation factors remain below the commonly accepted threshold of 5, suggesting that multicollinearity is not a major concern.

In contrast, education, household income, household size, landholding size, and access to grid electricity are not statistically significant once other factors are controlled for. These results suggest that adoption decisions are shaped primarily by feedstock availability, exposure to training, and perceived functional benefits rather than by general socioeconomic status.

3.3. Synthesis of Adoption Patterns and Implications for Causal Loop Analysis

To support the interpretation of the empirical results, this study uses a causal loop diagram (CLD) in Figure 5 as a conceptual tool to illustrate the key feedback mechanisms shaping household biogas adoption and its sustainability–resilience implications. Widely applied in system dynamics and sustainability research, CLDs capture reinforcing and balancing relationships among social, institutional, and policy factors without requiring formal simulation. In this study, the diagram integrates survey findings, field observations, and existing literature to map interactions between household capabilities, institutional support, and energy policy conditions, including the influence of LPG subsidies. It complements the econometric analysis by clarifying how these feedback structures help explain why adoption outcomes may persist, stall, or weaken under different governance and service environments. While core drivers such as livestock ownership, training participation, and fuel-cost pressure are empirically estimated and reflected as reinforcing processes, other relationships—particularly those linked to policy dynamics—remain conceptual and are not formally tested [59,60].

These findings justify the application of propensity score matching and doubly robust estimators in subsequent analysis, as simple comparisons between adopters and non-adopters would confound asset endowments with adoption effects. The prominence of training and perceived functional benefits further underscores the importance of institutional engagement and service provision, providing a basis for interpreting adoption outcomes within a sustainability–resilience framework in the discussion section [61].

Figure 6 shows the key factors influencing household adoption of biogas technology based on estimated odds ratios. Training participation has the strongest positive effect on adoption (≈3.5), followed by livestock ownership (≈2.0) and perceived time-saving benefits (≈1.8). Fuel-cost pressure also positively influences adoption (≈1.5). Other variables, including electricity access, education, household size, and landholding size, show moderate positive effects, while household income has only a marginal influence. The dashed line at odds ratio = 1 indicates the threshold of no effect.

The binary logistic regression model demonstrates strong explanatory power in identifying determinants of biogas adoption among dairy-based mixed crop–livestock households in West Java. The likelihood ratio chi-square test (p < 0.001) confirms that the full model significantly outperforms the null specification, with a Nagelkerke pseudo-R² of approximately 0.42 and a receiver operating characteristic (ROC) value of 0.82, indicating robust discriminatory performance. Results show that livestock ownership, participation in biogas-related training, perceived time-saving benefits, and fuel-cost pressure are significant positive drivers of adoption. Livestock ownership reflects the practical requirement of a stable manure supply for digester operation, while training participation highlights the importance of technical knowledge and user confidence. Perceived time-saving benefits emphasize reduced labor burdens, particularly for households responsible for fuel collection and cooking, and higher fuel costs increase the likelihood of adopting biogas as a cost-buffering strategy. In contrast, education, household income, landholding size, household size, and electricity access do not significantly influence adoption decisions, suggesting that uptake is more strongly shaped by resource availability, institutional engagement, and perceived functional benefits than by general socioeconomic status. Robustness checks support these findings: Rosenbaum bounds indicate that an unobserved factor would need to increase the odds of adoption by approximately 1.5–1.7 times (Γ = 1.5–1.7) to invalidate the statistical significance of the estimated treatment effects, and covariate balance after matching is satisfactory, with standardized mean differences below 0.1 for all covariates. Overall, the results suggest that successful biogas diffusion in West Java depends on both material capacity and institutional enablement, reinforcing the technology’s potential contribution to the sustainability and resilience of dairy-based farming systems [20,48,62].

3.4. Treatment Effects and Matching Diagnostics

To assess the quality of the matching procedure,

Table 6. Covariate balance before and after propensity score matching.

Covariate	SMD Before	SMD After
Livestock ownership	0.34	0.07
Training participation	0.29	0.05
Fuel-cost pressure	0.26	0.06
Time-saving perception	0.24	0.05
Household income (log)	0.21	0.04
Education (years)	0.18	0.03
Household size	0.22	0.05
Age of household head	0.17	0.04
Gender of household head	0.15	0.03
Landholding size	0.28	0.06
Electricity access	0.13	0.02
Access to credit	0.20	0.05
Distance to cooperative	0.23	0.06
Asset index (PCA)	0.27	0.05

reports standardized mean differences (SMDs) for all covariates included in the propensity score model before and after matching. Prior to matching, several variables—particularly livestock ownership, training participation, and landholding size—exhibited notable imbalance between adopter and non-adopter groups. After matching, covariate imbalance is substantially reduced, with all absolute SMDs falling below the conventional threshold of 0.10, indicating satisfactory balance between the two groups.

To complement the tabular results, a Love plot illustrating covariate balance before and after matching is provided in Figure 7. These diagnostics support the validity of the matching approach used to estimate treatment effects. The figure illustrates the reduction in covariate imbalance after propensity score matching. Absolute standardized mean differences below 0.10 indicate acceptable balance. After confirming covariate balance, treatment effects are estimated for key household outcomes. Results indicate that biogas adoption leads to significant reductions in LPG and firewood consumption, as well as lower chemical fertilizer use, suggesting a partial substitution of external inputs with locally available resources. These effects are consistent across matching specifications and remain robust under doubly robust estimation.

3.5. Heterogeneous Treatment Effects

Heterogeneity analysis reveals meaningful variation in treatment effects. Households with larger cattle herds experience stronger reductions in LPG consumption, likely reflecting more consistent biogas production. In contrast, the relative influence of training participation appears stronger among smaller-scale farmers, suggesting that knowledge constraints are particularly binding near technical feasibility thresholds. These findings indicate that uniform promotion strategies may generate uneven outcomes across farm types.

4. Discussion

The results indicate that household biogas adoption in West Java is shaped more by capability-related, institutional, and functional factors than by conventional socioeconomic characteristics. Although adopters tend to possess larger livestock herds and landholdings, multivariate analysis shows that adoption decisions are primarily driven by manure availability, participation in training, perceived time-saving benefits, and fuel-cost pressure, while education, household income, landholding size, and electricity access do not exert independent effects once these factors are controlled for. Interpreted through the sustainability–resilience framework, these findings highlight the joint role of household capabilities and enabling institutional environments in shaping adoption outcomes. Heterogeneity analysis further reveals that households with larger cattle herds experience stronger reductions in LPG consumption due to more stable biogas production, whereas training participation has a stronger influence among smaller-scale farmers, suggesting that knowledge constraints are particularly binding near technical feasibility thresholds. These results also imply that uniform promotion strategies may produce uneven outcomes across farm types, and that expanding training programs alone may be insufficient if institutional design and governance arrangements allow benefits to concentrate among already well-connected households.

4.1. Adoption as a Function of Capability Rather than Wealth

Although training participation emerges as a strong predictor of adoption, the present dataset does not allow differentiation between training formats, duration, or technical depth. In practice, training programs may vary considerably—from short informational meetings to intensive hands-on instruction followed by technical support visits. Such variation may influence both the effectiveness of training and the long-term performance of installed digesters. The estimated training effect should therefore be interpreted as reflecting general exposure to training rather than the effectiveness of specific training models. Future research should examine how training design—including practical demonstrations, follow-up support, and cooperative-based learning—affects adoption outcomes and system sustainability [20,30,33].

From a resilience perspective, these findings align with adaptability, which emphasizes households’ capacity to adjust practices in response to labor constraints and energy-price pressures. The prominence of time-saving benefits highlights that adoption is often motivated by immediate functional improvements rather than long-term economic optimization, particularly in smallholder contexts where labor is scarce and multifunctional.

4.2. Livestock Ownership and the Robustness of Household Energy Systems

The strong association between livestock ownership and adoption reflects the biological dependence of biogas systems on consistent manure supply. This finding highlights that adoption is embedded within mixed crop–livestock production structures rather than determined solely by household socioeconomic status [20,62,63].

In resilience terms, livestock ownership contributes to robustness by enabling households to sustain basic energy functions under conditions of fuel-price volatility or biomass scarcity. However, robustness derived from asset endowments also implies unequal entry conditions, as households with fewer animals face higher barriers to adoption. This highlights a potential trade-off between technical efficiency and inclusiveness in household biogas programs [64,65]. Heterogeneity analysis reveals meaningful variation in treatment effects. Households with larger cattle herds experience stronger reductions in LPG consumption, likely reflecting more consistent biogas production. In contrast, the relative influence of training participation appears stronger among smaller-scale farmers, suggesting that knowledge constraints are particularly binding near technical feasibility thresholds. These findings indicate that uniform promotion strategies may generate uneven outcomes across farm types.

4.3. Fuel-Cost Pressure and the Limits of Economic Incentives

The positive association between perceived fuel-cost pressure and adoption suggests that households respond to rising cooking-energy costs by seeking alternative energy solutions. At the same time, the variable captures perceived rather than objectively measured expenditure burdens, reflecting how households interpret their own financial constraints. In rural contexts where income flows are irregular and household budgeting is informal, perceived economic pressure may shape technology adoption decisions more directly than recorded expenditure shares. Nevertheless, future research would benefit from combining perception-based indicators with objective measures of energy affordability to better understand the relationship between perceived and actual energy poverty [24,65].

While participation in training is strongly associated with higher adoption probability, this finding does not imply that simply expanding training programs will automatically generate inclusive adoption outcomes. The present analysis estimates average effects and does not directly observe how training opportunities are distributed among farmers. In practice, access to training may depend on factors such as cooperative membership, social networks, or proximity to extension services. If training resources are disproportionately captured by better-connected households, expanding training without careful targeting could reinforce existing inequalities. For this reason, training initiatives may be most effective when combined with inclusive delivery mechanisms, such as cooperative-based outreach, targeted support for smaller farms, or follow-up technical assistance for households operating near feasibility thresholds.

4.4. Implications for Sustainability and Resilience Outcomes

Taken together, the results suggest that household biogas adoption in West Java contributes to sustainability and resilience primarily through incremental improvements rather than systemic transformation. Adoption enhances labor efficiency, improves manure management, and reduces dependence on traditional biomass, generating environmental and social benefits. However, the durability and scalability of these gains depend critically on institutional conditions that support operation, learning, and long-term system maintenance.

This interpretation aligns with sustainability–resilience research emphasizing that resilience emerges from interactions between household capabilities and enabling environments rather than from technology deployment alone. In the absence of coherent institutional support and policy alignment, biogas adoption risks becoming a fragile intervention—effective for some households but insufficient to drive broader rural energy transitions [66,67].

4.5. Positioning Within the Biogas Impact Literature

A growing body of empirical research uses treatment-effect approaches to assess the welfare impacts of household biogas adoption, particularly reductions in energy expenditure. These studies provide robust evidence that biogas can deliver private economic benefits when selection bias is addressed. However, their analytical focus is largely confined to welfare outcomes.

The present study builds on this causal impact literature while extending it by embedding adoption analysis within a sustainability–resilience framework. Rather than treating adoption as a binary outcome, the analysis emphasizes institutional support, operational capability, and policy–regime interactions that shape the durability of adoption. Training participation and perceived time-saving benefits emerge as central mechanisms, highlighting the importance of learning and service provision—factors that are often peripheral in welfare-focused studies.

Moreover, the Indonesian context introduces system-level constraints that differ from biomass-dominated settings, particularly the influence of subsidized LPG on household energy choices. By explicitly accounting for this regime-level interaction, the study demonstrates that biogas adoption outcomes depend not only on household characteristics but also on alignment between renewable energy initiatives and national energy policies. In this sense, the study complements existing impact evaluations by shifting attention from short-term welfare gains to the conditions under which biogas adoption contributes to long-term sustainability and resilience. While training participation is associated with higher adoption probability, these findings do not imply that expanding training programs alone will guarantee improved outcomes. Institutional design and governance arrangements may influence how training resources are distributed, and poorly targeted programs could disproportionately benefit already well-connected farmers [68].

5. Recommendation

5.1. Policy Implications

The results suggest that biogas adoption in dairy-based rural systems is shaped more by structural feasibility and institutional exposure than by general socioeconomic characteristics. In particular, herd size plays a central role, reflecting the need for a reliable manure supply to sustain digester operation. Access to technical training also significantly increases the likelihood of adoption, mainly by reducing uncertainty and improving user confidence. Beyond adoption itself, the findings show that biogas systems can deliver tangible operational benefits. After accounting for observable differences through propensity score matching and doubly robust estimation, adopter households demonstrate lower reliance on LPG and firewood, along with reduced use of chemical fertilizers. These changes indicate a partial shift from external inputs toward locally available resources, contributing to greater energy stability and improved nutrient cycling within mixed crop–livestock systems.

At the same time, these benefits are not distributed evenly across households. Farmers with larger herds tend to experience stronger energy substitution effects, likely because they can sustain more consistent gas production. In contrast, training appears particularly important for smaller-scale farmers, for whom knowledge constraints may be more binding. This pattern suggests that uniform installation strategies risk favoring households that are already better positioned to adopt. A more inclusive approach would therefore combine technical support with targeted outreach, especially for farmers operating near the margins of technical feasibility.

The relationship between perceived fuel-cost pressure and adoption further highlights the role of household decision-making under economic constraints. While households appear responsive to rising energy costs, this variable reflects subjective perceptions rather than measured expenditure burdens. In contexts where incomes are irregular and financial records are limited, perceived affordability may shape behavior more directly than objective indicators. However, this also introduces potential measurement limitations, as perceptions do not always align with actual energy poverty conditions.

These considerations are important when interpreting the policy implications. Although training is consistently associated with higher adoption rates, this should not be taken as sufficient justification for simply expanding training programs. The analysis captures average effects but does not observe how access to training is distributed in practice. Participation may depend on factors such as cooperative membership, proximity to extension services, or existing social networks. Without careful targeting, there is a risk that additional resources could disproportionately benefit already well-connected households. For this reason, policy efforts should focus not only on increasing training availability but also on ensuring equitable access through targeted delivery mechanisms, cooperative-based outreach, and appropriate institutional safeguards.

Finally, some caution is warranted due to measurement constraints. Training participation is recorded as a binary indicator and does not capture differences in quality, duration, or technical depth, which may attenuate estimated effects. Similarly, fuel-cost pressure is based on self-reported perceptions rather than objective expenditure data. While both variables are relevant for understanding behavior, they may introduce bias if interpreted too narrowly. Future research would benefit from incorporating more detailed measures of training characteristics and combining subjective indicators with objective energy expenditure data to strengthen measurement validity.

5.2. Institutional Implications

The sustainability of decentralized energy systems depends not only on initial installation but also on the reliability of institutional support systems. Indicators collected in this study, including maintenance response times, availability of spare parts, and frequency of follow-up visits, reveal gaps in service provision. In several cases, maintenance delays exceed two weeks, and spare parts are inconsistently available in local markets. These service constraints may reduce system functionality and discourage continued use.

The findings therefore suggest that installation-focused approaches alone are insufficient to ensure long-term system performance. Strengthening maintenance ecosystems, technical service networks, and cooperative-based support mechanisms is essential for sustaining biogas systems after installation. Without reliable service provision, decentralized energy technologies risk becoming short-lived interventions rather than durable components of rural energy systems.

National energy policy also influences adoption incentives. The persistence of LPG subsidies reduces the relative economic attractiveness of biogas systems, even when households experience fuel-cost pressures. However, subsidy removal alone would not necessarily guarantee wider adoption, as LPG programs also serve energy-access and social protection objectives. Instead, improved coordination between renewable energy initiatives and existing energy policies may create more consistent incentives for decentralized energy transitions. Institutional capacity is evaluated using independent service indicators reported by respondents, including average maintenance response times, availability of spare parts, and frequency of follow-up visits. These operational indicators provide a basis for assessing service reliability without inferring institutional performance directly from adoption rates [69].

5.3. Research Implications

Several limitations highlight opportunities for future research. First, the cross-sectional design limits the ability to examine long-term adoption trajectories and system performance over time. Longitudinal studies or panel datasets would provide stronger evidence on the durability of adoption and the persistence of energy and input-use adjustments. Second, some outcome measures rely on recall-based information on household energy consumption, which may introduce measurement error.

Although sensitivity analysis indicates that substantial unobserved bias would be required to overturn the estimated treatment effects, unmeasured household characteristics—such as entrepreneurial orientation or management ability—may still influence both adoption and outcomes. Future research could address these limitations through quasi-experimental designs, program rollout analysis, or administrative maintenance records.

Another limitation concerns the measurement of training participation and fuel-cost pressure, both of which rely on simplified survey indicators that may not fully capture underlying variation in economic conditions or institutional support structures.

Finally, while this study focuses on West Java, similar structural conditions—including livestock-based rural economies, fossil fuel subsidies, and fragmented service infrastructures—exist in many low- and middle-income countries. Comparative studies across regions would help clarify how institutional ecosystems shape the long-term viability of decentralized renewable energy systems.

6. Conclusions

These results confirm that household biogas uptake depends more on operational capability and institutional engagement than on general socioeconomic status, consistent with evidence from other developing-country contexts [1,8,28,69,70,71]. Interpreted through a sustainability–resilience lens, biogas adoption contributes to rural systems mainly through incremental gains in robustness and adaptability, including improved manure management, reduced reliance on traditional biomass, and enhanced labor efficiency. However, the persistence of subsidized liquefied petroleum gas limits the extent to which these gains translate into broader energy-system transformation. This misalignment between renewable energy initiatives and prevailing energy-pricing regimes constrains the long-term scalability and transformative potential of household biogas systems in Indonesia.

The results further underscore the central role of institutional support. Without consistent service provision and policy coordination, biogas installations risk becoming short-lived interventions rather than durable components of rural energy systems, a challenge documented in previous Indonesian programs. From a policy perspective, effective biogas scaling requires a shift from installation-focused approaches to integrated strategies that strengthen technical support, align energy and agricultural policies, and gradually address distortive fossil-fuel subsidies. This study is subject to limitations, including reliance on cross-sectional data and indirect assessment of environmental impacts. Future research should employ longitudinal designs, incorporate biophysical indicators, and examine how alternative institutional and service-delivery models affect the durability and resilience of household biogas systems.