Narrative-Network Dynamics in Tipping Processes Towards Low-Carbon Energy Futures: The Case of Indonesia

The shift from coal and carbon-intensive regions (CCIRs) is challenging, particularly in poorer but fossil-fuel-rich countries like Indonesia, where coal is abundant. As Indonesia’s primary export is coal, reducing its dependence on this fossil fuel poses a challenge to the economy (S. E. Hosseini & Abdul Wahid, 2014; Sewandono & Munandar, 2021). However, addressing climate change necessitates swift and systemic changes globally, particularly in regions most intensive in the use of fossil fuels. For these adjustments to occur, it is necessary to understand the conditions and processes that may accelerate deliberated structural changes aligned with climate and sustainability goals. Without one-fit-all solutions, this requires implementing multiple interventions adapted and emerging from local contexts and endogenous capacities, leading to positive tipping points towards sustainability (Tàbara et al., 2018).

Recent literature has highlighted the role of deliberate agency in the social tipping processes. In this regard, social networks and their visions can be understood as key to building the preconditions for the emergence of tipping points because of their ability to mobilise collective actions, catalyse transformative change and promote adaptive capacities in response to change (Folke et al., 2005, 2010; Milkoreit et al., 2018; Winkelmann et al., 2022). Research has revealed that agents communicate and transmit their visions of the future within their networks to produce new joint activities or interventions for change (Markard et al., 2012; Olsson et al., 2014; Winkelmann et al., 2022). In particular, transformative visions explore alternate, desirable, and attainable futures to be shared across agents to trigger systemic changes (Pereno & Barbero, 2020; Tàbara et al., 2018). However, the ways social networks communicate their visions in complex social-ecological systems remain under-researched (S. A. H. Hosseini, 2009; Sayles et al., 2019).

Narrative analysis has emerged as a valuable tool in analysing complex social-ecological systems such as the energy sector (Moezzi et al., 2017). Through narratives, researchers can capture the complex dynamics of coupled social-ecological systems and explore the interactions between social networks and visions (Hahn et al., 2008). Hence, the transformation of dominant narratives can be seen as a prerequisite in any deliberate attempt to change social-ecological systems. This is particularly true when dominant narratives can no longer explain existing realities, or alternative narrative framings challenge prevalent narratives or present hitherto unexplored prospects (Moore et al., 2014).

Capturing the complex dynamics in switching energy systems towards renewables is particularly challenging in low-income countries. Additionally, tracking links between changes in energy narratives and actual structural practices on the ground has yet to be addressed in developing economies like those in Asia (Apfel et al., 2021). This is challenging for research because of the intricate and asynchronous interactions between narrative shifts and social changes. Also, narratives are developed in advance of actual policy measures or adjustments. In complex social-ecological systems such as energy systems, with high uncertainty and multiple outcomes that may be spurred by a single intervention (Papageorgiou et al., 2020; Shen et al., 2022), transformative narratives are often built upon practical and shared experiences from agents who communicate their aspirations and learnings through alternative visions (Hinkel et al., 2020). Pioneering agents generate transformative capacities through these visions via multiple learning networks and feedback loops (Ofoegbu & Ifejika Speranza, 2021). Therefore, the analysis of narratives provides evocative accounts that allow interpretation of the complex cultural processes that involve changes in collective systems of meaning (Geertz, 1973). Such studies can thus be enriched by examining the diverse agents’ interactions in their unique cultural and political contexts and explaining transformation within the networks (Emirbayer & Goodwin, 1994).

Such combined analysis of agents’ networks and narratives can be especially useful for understanding the possibilities of systemic shifts in the energy sector (e.g., Hermwille, 2016; Judson et al., 2020; Chailleux, 2020). Given the absence of empirical and integrated analysis in non-Western countries, our research investigates how social-ecological networks and agents’ visions can serve as anticipatory social tipping signals, building the conditions for the transformation of systems. Our reference system comprises the agents influencing the Indonesian energy sector’s policy-making, technological infrastructures, and market innovations. Against this backdrop, we employed a mixed-method empirical approach to examine the transformative visions of key agents and their interactions in two provinces of Indonesia: Banten and Bali.

We first adapted the Integrated Sustainability Assessment (ISA; Weaver et al., 2006) to identify how the agents envision and strategise the future with the needed capacities to transform the energy system. In addition, our methodology included a narrative analysis using the alternative pathways framework proposed by Lieu et al. (2020) and a more specific Q-methodology for exploring the content of their transformative visions (Charli-Joseph et al., 2018; Silaen et al., 2019; Nieuwenhuis et al., 2022; González-González et al., 2023). Social Network Analysis (SNA), with the region as the ecological boundary, was also employed to track agent structure dynamics (Devisscher et al., 2017; Froehlich et al., 2020). We hypothesised that the use of qualitative and quantitative SNA could constitute a useful tool to assess and anticipate possible social ‘thresholds’ of transformative change or tipping points and, in this manner, how agents’ interactions may help to anticipate potential cascades of positive change or have telecoupling effects on other networks (Franzke et al., 2022; Froehlich et al., 2020). Although SNA is commonly used to map human nodes, it is limited to tracking social-ecological interactions. Given that we recognise the energy sector as part of social-ecological systems, we enriched the qualitative SNA with the concepts of social-ecological networks and nodes (Janssen et al., 2006).

Indonesia is one of the largest emitters of greenhouse gases. Yet, the country is seemingly moving towards a high-intensity development pathway by relying on fossil fuels (e.g., coal, oil). Such a business-as-usual policy route is particularly visible in the two contrasting provinces of Banten and Bali. Due to their unique energy challenges and renewable energy potential, Banten and Bali provinces make good energy case studies to explore the narrative and agent-network preconditions for structural change (Fig. 1).

A map of Indonesia marks the location pins of Banten and Bali.

Banten Province is known as an economic hub. It is home to about 12.9 million people and thousands of medium and large industries (Rahayuningsih, 2017), such as coal-fired power plants, petrochemicals, and steel production plants. In contrast, Bali Province is recognised as the leading tourist island in Indonesia. As a top tourist destination, the highest energy demand comes from tourist activities (e.g., accommodation, food, transportation). Banten and Bali are connected through an electricity grid in which 50% of the electricity generated in Banten is distributed to the remainder of Java and Bali. Almost half Bali’s electricity demand is supplied from Java through a sea cable. The electricity from Banten usually acts as regional balance¹ in western Java in addition to securing overall Java-Bali’s electricity provision, especially during peak hours.

Indonesia’s energy sector is organised by national and local governments and non-governmental, corporate and intergovernmental entities that encourage sustainable development initiatives. At the national level, the government sets national energy policies, regulates power plant operations, and promotes renewable energy sources. The Ministry of Energy and Mineral Resources (MEMR) oversees the country’s exploration and extraction of natural resources and developing energy infrastructure. PLN (Perusahaan Listrik Negara)—the state-owned electricity company—is responsible for generating, transmitting, and distributing electricity to the country. As such, the company plays a significant role in implementing national energy policies, together with private generating companies.

At the provincial level, the roles of local government in regulating the energy sector can vary from province to province. However, some common responsibilities of local governments are promoting clean energy sources, ensuring the availability of reliable and affordable energy, and regulating the exploration and extraction of natural resources within their jurisdiction. The responsibilities of MEMR in each province are carried out by the local government body, Dinas Energi dan Sumber Daya Mineral (ESDM) or the Agency of Energy and Mineral Resources. For instance, affairs related to the energy sector in Banten are coordinated by the Energy and Mineral Resource Agency (ESDM-Banten) in collaboration with the district down to the local village government. The same pattern applies in Bali (i.e., ESDM-Bali).

The ISA procedure encompasses the following stages: Scoping, Visioning, Experimenting and Evaluating. As a first step, we adapted the general four-stage participatory integrated sustainability assessment (ISA; Weaver et al., 2006) approach to achieving the research objective. We applied network analysis and Q-methodology to examine agents’ interactions and visions in the ISA phases (Fig. 2). The narrative analysis comprises the alternative pathways framework (Lieu et al., 2020) and a Q-methodology analysis to map out different narratives and explore their content differences. We also used the region as the ecological element, according to Cheruvelil et al. (2017), as a spatial unit that captures regional-scale heterogeneity in ecosystem properties and relationships; in this case, the energy system.

A flowchart includes start, stages of the review of relevant policy documents and others, checkpoint 1 of data and info review, semi-structured interviews and quantitative network analysis with Gephi, F G D 2 of 5 to 10 informants, checkpoint 2 of data and info review, Q-method and integrated analysis, and end.

Regarding the research process and interactions with stakeholders informing this research, we first participated in ten webinars on the potential and ongoing Indonesia’s energy transformations towards clean energy futures. Attending webinars helped to identify key agents constituting the social system the reference before conducting interviews. From May 2020 to September 2021, 20 pertinent national and local agents in the sector participated in semi-structured interviews² and acted as informants. We recorded and transcribed the interviews and the results from webinars while keeping the informants’ profiles anonymous. Desktop research examined relevant policy documents, media, and research articles. The desk-based review supported participants’ data for subsequent phases to reduce bias. The interviewing procedure also followed the stages suggested by Weaver et al. (2006):The next step in the first phase was conducting the first focus group discussion (FGD) to validate the transformative visions and the key agents. A checkpoint was suggested at the end of the first phase. Using a snowball approach, the first checkpoint assessed whether the obtained information and networks were sufficient to answer the research questions. The focus at this stage was to identify information on the transformative visions and to map out all key agents and their interactions in the energy system.

A second phase began with the implementation of additional semi-structured interviews to perform SNA within social-ecological networks or SEN. In this phase, ‘Checkpoint 1’ was used to ensure that valuable information on the agents’ networks was not missing. At this stage, additional interviews were not conducted as the agents’ interactions included diverse interactions from policymakers, private companies, and communities. Since the identified network was not considered sufficiently complex (<150,000 nodes), we conducted SNA with Gephi (Akhtar, 2014). Gephi is a free and open-source programme for visualising and studying massive network graphs such as social networks (Bastian et al., 2009). Before calculating the network characteristics with the degree of centrality, we listed the agents’ capacities according to human-made resources and their regional operation. Subsequently, we deployed the concept of region according to Cheruvelil et al. (2017) to capture regional-scale heterogeneity in energy system dynamics as an ecological boundary. The region component also functions to locate agents’ presence according to their authority (Avelino & Rotmans, 2009). Furthermore, we utilised two characteristics of the social networks of interest that Janssen et al. (2006) suggested can be regarded as tipping elements—they can undergo abrupt and irreversible changes in response to small perturbations—for tipping points: level of connectivity and level of centrality. The level of connectivity has two domains: the links’ density and the networks’ reachability: high density implies high reachability (Janssen et al., 2006). The level of centrality indicates nodes in the network with significantly higher-than-average links. We used Gephi 0.10.1 to examine the agents’ networks (without a dynamic approach; Bastian et al., 2009).

‘Checkpoint 2’ involved assessing specific and shared transformative visions among the agents regarding sustainable futures in the Indonesian energy sector. If semi-structured interviews and FGDs are sufficient to capture the shared visions among the agents qualitatively, Q-methodology can be neglected. Otherwise, Q-methodology is the choice—a statistical method for analysing conflicting discourses around certain topics based on Z-score and eigenvalues. During the Q-methodology exercise, participants virtually sorted statements (see Table 1) on cards into columns ranging from −3 (high disagreement) to +3 (strong agreement): see Fig. 3. Before performing discourse analysis with Q-methodology, we selected several experts and pertinent agents to reduce redundancies and prioritise the relevant statements. Ken-Q—a software tool for conducting Q-methodology analysis online—was used to perform the analysis (Yang & Xu, 2021). A degree of agreement and disagreement for each statement was determined using the Z-score, which allows comparison of the scores of different statements across different visions. The unit of Z-scores is the standard deviation. For example, a statement with a z-score of −2.0 is two standard deviations below the distribution’s mean. Finally, eigenvalues determine vision pluralism among agents: a higher eigenvalue indicates that a perspective accounts for more data variance and offers a distinct view. A factor or perspective with eigenvalues <1 is usually ignored (McParland et al., 2011).

A Q-sort response grid has negative and positive values for disagree, neutral, and agree.

The Alternative Pathways Framework approach (APF; Lieu et al., 2020) was first used to position agents’ general narratives of current and future energy systems collected from semi-structured interviews and FGDs. The Atlas.ti software was used to classify the collected insights to place narratives based on categories: types of human agents, ecological nodes, transformative visions, and tipping interventions (Table 2). Subsequently, the features of future narratives were specified with Q-methodology. APF looks at the mainstream and alternative narratives, where the latter can be characterised as on-stream, off-stream and transformative. The mainstream narrative depicts the dominant energy system, while the alternative narratives operate with different natures from the mainstream system. According to Lieu et al. (2020), the on-stream narrative occupies a niche within the mainstream but does not challenge the regime. Off-stream narratives represent niche clean energy technologies. Transformative narratives encompass radical technological innovations and fundamentally new ways of reorganising the socio-energy system. In Q-methodology, the features of transformative visions were explored beyond technological change. For example, key agents are initiating new political structures and modes of social-ecological interactions aligned with sustainability (a positive tipping point).

We summarise the different perspectives identified from agents’ interactions and desktop research into mainstream and alternative narratives as follows:

Our study looked at the role of narratives and network configuration in creating the cognitive policy and social conditions for tipping systems toward desirable, sustainable futures. We conducted discourse analyses with SNA (with an ecological element) to check to what extent agents’ actions and visions could serve as components of anticipatory social tipping signals for positive tipping points. The Banten and Bali case studies indicated that narratives take time to become actions and create the conditions for transforming social-ecological structures, whilst some narratives may fail to influence reality. Social-ecological network analysis also revealed that some significant agents in the network were central to shifting narrative discourses and energy policies. These results offered two visions that, by making them explicit and allowing them to be discussed, could help mutual learning among different networks.

Most interviewees agreed that renewable diversification is feasible in Indonesia in the future, yet transformative visions about sustainable futures take time to ensure systemic changes. As noticed by other studies (such as Lenton et al., 2022; Pereno & Barbero, 2020; Tàbara et al., 2018), large systemic changes occur due to the interplay of actions at different scales. Understanding tipping dynamics at lower scales, such as communities, may serve as social anticipatory signals of tipping points, either positive or negative, that may occur in larger social-ecological configurations (Gangwal et al., 2020; Lenton et al., 2022; Tàbara et al., 2018). For instance, NGOs like the Indonesian Renewable Energy Society (IRES) and Greenpeace bridge government and communities according to their Vision 1 to influence the governments’ narrative position towards off-stream technologies. Therefore, mapping how different agents learn and interact across scales is crucial to understanding positive tipping points’ potential and complex dynamics.

Whilst transformative visions alone may not be sufficient to trigger and consolidate change, we argue that the different configurations of network structures play a crucial and differentiated role in enacting systemic change. Targeting specific nodes or subgroups in the networks that contribute simultaneously to narrative building and policymaking can be extremely effective in knowing where to target actionable tipping interventions. For example, most agents driven by Vision 1 aim to interact with the government in amending the government’s narrative position from mainstream to off-stream technologies. Other international organisations are also increasingly playing a role in the socio-energy network configuration, as is the case of the assistance from the United Kingdom for constructing a dam and hydropower plant in the Buleleng Regency, Bali (Investor Daily, 2019). Another example is that the proportion of clean energy in Banten’s electricity rose to 5.9% according to Local Regulation of Banten No. 7/2022 when PLN moved towards clean energy deployment, given its access to human-made resources and direct control over the region (Banten Local Government, personal communication, 2021).

At the same time, other weak-tie agents in the social energy network, such as civil society organisations and small businesses, are also important, albeit performing different functions in the energy transformation processes. Some private companies with Vision 2 in both provinces are beginning to deploy clean energies despite the absence of local directives. For instance, one company in Banten disclosed that their green initiatives were voluntary and contributing to national GHG reduction pledges. Ultimately, SNA with ecological elements helped identify a centrality approach over a concrete region and its power dynamics to trigger or hinder change.

This study also demonstrates that the analysis of narrative-network dynamics can constitute an appropriate methodology to identify anticipatory social tipping points (ASTS) that create the previous conditions of systemic change, new opportunities for mutual learning and joint transformative strategy building among diverse social networks. The energy sector is projected to change rapidly and systemically if agents with different visions can collaborate (Pereira et al., 2015). Understanding these complex socio-economic and political processes is paramount in non-Western countries with vast amounts of fossil fuels like Indonesia. For instance, agents that aim to influence the government (Vision 1) could utilise best practices such as private sector installation of clean energy sources to influence the mainstream narrative of governmental agencies. A shift from the mainstream to the off-stream narrative can signal positive or negative tipping points before interventions, such as policies favouring renewable diversification (off-stream narrative) through carbon pricing or subsidies.

Using participatory discourse analyses and social-ecological network analysis, our study looked at the role of narratives and network configuration in creating the conditions for tipping systems toward desirable sustainable futures. Discourse and network analyses are used to check to what extent the actions and visions of agents in networks can serve as components of anticipatory social tipping signals for positive tipping points. The Banten and Bali case studies revealed that transformative visions of sustainability can inspire positive tipping points. Still, time and effort are required for narratives to become actions and change social-ecological structures. Furthermore, not all narratives successfully influence the conditions leading to transformative changes. Through network analysis, it was possible to discover some key agents within the networks that could be identified as having a centrality position in changing narrative discourses and energy policies. These results also offered two kinds of visions that, if combined or reconciled, may help mutual learning among different networks.

Most interviewees agreed that renewable diversification is feasible in Indonesia, but transformative visions of sustainability do not ensure instant systemic changes. Precedent studies showed that large systemic changes result from the interplay of actions at different scales. Tipping dynamics at lower scales, such as in communities, can indicate potential tipping points in larger social-ecological systems, either positive or negative. For example, local actions like Indonesia’s micro-hydro and geothermal energy can lead to rapid and systemic changes in the national energy sector if combined with other small-scale actions. Therefore, it is important to map the role of multiple learning networks and feedback loops by all relevant agents at different scales to understand the possible emergence of positive tipping points. Whilst transformative visions alone may not be sufficient to trigger and consolidate change, we argue that the different configurations of network structures affect in various and, in some cases, complementary ways the building of conditions for systemic change. Targeting specific nodes or subgroups in the networks that contribute to narrative building and policy-making simultaneously can be an effective approach to knowing where to target actionable tipping interventions. In this case study, social network analysis with ecological elements helped identify a centrality approach over a concrete region and the power dynamics to trigger- or hinder-change.

Nevertheless, this research also acknowledges two important limitations: first, the social networks were observed at a certain time, so their long-term effects on broader structural dynamics cannot be fully assessed. Second, this study only focused on one ecological element, the energy flows and resources of the selected regions. Empirical research may need to explore other biophysical components and interaction features to obtain a more comprehensive view of social-ecological agents and tipping dynamics.