1 Introduction
The maturation and growth of renewable energy (RE) technologies has opened opportunities for households and other groups to become RE producers as well as consumers. This convergence of production and consumption in space, time and persons is not limited to RE and has been termed ‘prosumerism’ (Ritzer, 2015). When applied to RE, this refers to actors who both produce and consume electricity (Brown et al., 2020). RE prosumers can take a range of forms including individuals, households, communities, companies, and public organisations, and their production should not significantly exceed their own demand (Pieńkowski, 2021). While prosumers may exist within dominant systems of concentrated energy producers separated in time and space from dispersed consumers, they nonetheless represent an alternative basic system of energy provision (Ellsworth-Krebs & Reid, 2016).
Benefits of RE prosumerism over traditional models of energy production, transportation and consumption include the growth of sustainable energy such as wind and solar (and their potential replacement of fossil fuel production and consumption), the localization of financial benefits from energy generation, and techno-economic efficiency by avoiding the need to transport energy long distances through localized production and consumption (Busch et al., 2021). The use of local renewable energy can also help to buffer against price spikes driven by world markets and geopolitical turmoil (Gjorgievski et al., 2022), as well as supporting greater energy security or resilience against natural disasters, physical and cyber attacks, and power system failure cascades (Hirsch et al., 2018). Nonetheless, we should remain aware that prosumerism can still be both captured and driven by capitalist logics and extractive social relations (Ritzer, 2015).
Many renewable energy cooperatives or energy communities (ECs) can be broadly classified as prosumers. In contrast to incumbent energy producers, such communities often adopt a community logic rather than a market logic (Dudka et al., 2023; Wittmayer et al., 2021). Energy communities have been found to address energy poverty (Mey & Diesendorf, 2018; Middlemiss et al., 2019), mobilise financial capital for local energy investment (Dudka et al., 2023; Kooij et al., 2018; Yildiz et al., 2015), facilitate energy democracy and justice (Bianchi & Ginelli, 2018; Hiteva & Sovacool, 2017; Schreuer, 2016; Wahlund & Palm, 2022) and increase public acceptance of renewable energy projects (Hoppe & Warbroek, 2021; Wierling et al., 2018). Since ECs enable citizens to more directly participate in and financially benefit from the ongoing energy transition, and supported by the EU Renewable Energy Directive (RED II), over the last years ECs have increased in scale, scope and number throughout European member states (Blasch et al., 2021). In the European Union, there are more than 10,500 such initiatives (Wierling et al., 2023).
The interaction between prosumers like ECs and the energy system has been primarily governed through net metering policies and subsidies, essentially sending energy back to the grid for guaranteed prices (Londo et al., 2020; Poullikkas et al., 2013). However, problems are emerging with this model. Firstly, by taking customers away from traditional utilities, prosumerism and self-consumption indirectly causes price increases for remaining customers who themselves often cannot afford to invest in their own generation assets (Castaneda et al., 2017; Chesser et al., 2018; Kubli, 2018). This is because utilities attempt to recoup lost revenues from these previous customers turned prosumers/self-consumers by hiking prices for remaining consumers. Secondly, the growth in distributed generation has given rise to grid congestion issues. Natural variability in production (e.g. solar radiation) does not necessarily align with energy usage peaks, leading to problematic imbalances in matching supply and demand (Zhou & Lund, 2023). Utilities are pushing back against prosumer policies (Baker, 2017; Hess, 2016). As a consequence of a variety of these issues, and others, prosumers are nowadays facing a ’post-subsidy era’ as financial support policies for sale of energy to the grid are phased out in many countries (Brown et al., 2019), with low or even negative payback rates being implemented in some countries.
ECs must decide on how best to use their common pool of generated electricity to meet their varied and context-specific goals (Kubli & Puranik, 2023; Van Veelen, 2017). Digitalization and increasing legal possibilities expand the opportunities for forms of community self-consumption (CSC) to achieve specific goals (e.g. autarky, economic benefit) (Watson et al., 2022). Even in CSC arrangements, there is often a surplus of energy remaining which is not used by EC members (Directorate-General for Energy, 2024; Ovaere, 2023). In a study of self-consumption across seven European countries, Ovaere (2023) found that collective self-consumers could use between 39% and 66% of electricity generated. This surplus occurs on many timescales, from hourly to annually.
This provides an opportunity for ECs to exchange with actors or households in the local area, but who are not formally involved in the EC. As a first option for the EC, exchanging energy with neighbours is arguably preferrable to energy storage from a practical standpoint because investment in storage technologies is both costly and also materially wasteful. Particularly if surplus energy can be shared and used immediately without the need for storage. Furthermore, the explicit introduction of storage technologies increases the potential complexity of energy management possibilities. Since the aim of this paper is to compare energy sharing strategies on an abstract level, we chose not to include this. A similar strategy was taken by Mehta and Tiefenbeck (2022, p. 2): “Since storage technologies fundamentally alter physical energy flows and introduce a wide range of energy-management possibilities, we purposely leave them out in order to establish a clear understanding of the solitary effects of each EC design dimension on and the trade-offs between the KPIs”. Finally, while we do not model storage directly, the notion of flexibility introduced later can be considered to include storage capacities of households outside the EC.
In this paper, we focus on this surplus as a common resource of the EC which can be exchanged with neighbour households. However, most research on prosumer energy exchange assumes markets as the main institutional arrangement (Giotitsas et al., 2020; Pires Klein et al., 2021; Singh et al., 2017, 2018). This ignores the potential for other socio-economic relations which might better match EC goals, such as sharing. Several authors have emphasised that non-market energy sharing is also a possibility for exchange of surpluses (Georgarakis et al., 2021; Reuter & Loock, 2017). We therefore study possible (market trading and non-market sharing) mechanisms of collective action leading EC members (several households collectively owning and operating a generation asset) and non-EC neighbour households in the local area to interact to increase local use of energy, thereby reducing grid congestion. Using agent-based simulations, this paper explores alternative governance arrangements for the interaction between ECs, households and the broader energy system. Key performance indicators (and agent objectives) are economic gain, local use of energy, and solidarity. These values have been found, among others, to be key drivers of energy prosumerism and EC motivations by several authors (Brown et al., 2019; Directorate-General for Energy, 2024; Kubli & Puranik, 2023; Montakhabi et al., 2023; Van Veelen, 2017).
The next section reviews the literature on monetary and non-monetary energy exchanges, also defining the key theoretical elements on which our agent-based model is designed. Section 3 presents the model, whose results are then shown in section 4. Section 5 discusses the results and concludes the paper.
2 Literature
Most research on prosumer energy exchange assumes (profit-oriented) market trading as the primary coordination mechanism (Giotitsas et al., 2020; Pires Klein et al., 2021; Singh et al., 2017, 2018), and examples of this assumption are in various studies (Hahnel et al., 2020; Long et al., 2017; Morstyn et al., 2018; Pumphrey et al., 2020; Wang et al., 2017; Zhang et al., 2018): “Local markets are likely to be key for managing distributed renewable generation and for coordinating decentralized decision models that satisfy large numbers of self-interested autonomous agents” (Parag & Sovacool, 2016, p. 2). Importantly, this assumed mechanism is even present in community-oriented visions of prosumerism (Capper et al., 2022; Moret & Pinson, 2019; Schwidtal et al., 2023).
However, there is no inherent reason why market trading is the most desirable means of coordination. Some recent authors have pointed towards the (potential) role of non-economic drivers for energy sharing. For example, Singh et al. (2018) found that non-monetary returns were not demanded from prosumers in an off-grid energy system in India, especially from individuals with closer ties to the sender. Georgarakis et al.’s (2021, p. 7) survey of Dutch prosumers found that “60% of respondents would be willing to give surplus electricity for free and 76% would be willing to do so for an indirect financial return. In addition, 77% of the participants who would give away surplus for free and 70% of those who would do so for an indirect monetary return, would give it to a household that cannot afford electricity”. Similarly, in a survey conducted in Switzerland, Germany, Norway and Spain, Reuter and Loock (2017) find that between 45–65% of prosumers would be willing to share excess electricity without a financial return.
Sharing may be preferred over selling to achieve different objectives. In their review of literature on peer-to-peer energy transaction, Jogunola et al. (2017) state that energy can be shared for noneconomic goals such as the accrual of social capital. Similarly, Wilkinson et al. (2020) find that participants in the first real-world trial of P2P energy trading platforms in Australia are highly interested in social equity, objecting to the market-centric setup they were presented with. In their P2P trading experiment conducted in Germany and the UK, Hahnel and Fell found evidence for “altruistic pricing with friends and family, local schools and hospitals, and low-income households” (Hahnel & Fell, 2022, p. 10). In a similar vein in the US context, Gazmararian and Tingley (2024) recently propose a model whereby solar energy owners donate their financial (not energy) returns to community groups or lower socio-economic status actors in their vicinity to develop their own solar projects. The authors find that “a majority of Americans (60%) would be willing to participate, and their donations could raise substantial capital to build out renewable energy for eligible communities” (Gazmararian & Tingley, 2024, p. 12).
ECs might be ideal vehicles for energy sharing since Georgarakis et al. (2021, p. 8) find that energy cooperative members are more “willing to exchange surplus electricity for free (68% vs. 53%) and for indirect financial returns (85% vs. 68%)” than non-members. While altruistic or egalitarian-minded individuals might of course be more inclined to join an energy cooperative in the first place, one should not discount the influential role institutional membership has on attitudes. Similarly, Gazmararian and Tingley’s (2024) survey results show that polycentric governance approaches increase the likelihood of participation and success in solar sharing. Moreover, market-based energy governance arrangements are based on a vision of rational choice, rather than based on empirical evidence from people’s everyday social lives (Singh et al., 2017), and are not aligned with the mutual and public interest goals of many energy communities (Bauwens & Defourny, 2017; Dudka et al., 2023, 2024). Indeed, energy communities are already beginning to experiment with forms of solidaristic energy sharing (Directorate-General for Energy, 2024). For this reason, Adams et al. (2021) suggest that P2P scholarship engage with economic anthropology, cultural economics and economics sociology, as disciplines which deal extensively with non-market forms of exchange.
There are many different classifications of economic exchange in the anthropological literature, with different and overlapping terminologies, including market exchange, commodity exchange, hierarchy, redistribution, reciprocity (including negative, balanced and generalized forms), gift reciprocity, pure gifts, communal shareholding, and communism (Graeber, 2001, 2014; Karatani, 2014; Mauss, 2004; Polanyi, 2001; Sahlins, 2017). Singh et al. (2018) draw on economic anthropology literature to develop a returns continuum for prosumer energy exchanges, ranging from in-cash returns to more intangible returns such as solidarity and social cohesion. They show how expected returns differ depending on the social relation. However, it is important to note that this is not static but is an “intricate sociocultural process” (Singh et al., 2018, p. 195). Depending on the context, those eligible for intangible returns can shrink or grow (Sahlins, 2017). For example, in societies with extensive kinship systems, people may be willing to provide goods or services for intangible returns to a wider group than those living in atomistic capitalistic societies. Sahlins (2017) also mentions that solidarity might shrink (or grow) in times of hardship due to external or social conditions. Reciprocity therefore varies over space and time.
The specific benefits of non-market oriented energy exchanges are primarily to do with socio-economic justice. Prosumerism (including participation in ECs) has remained largely directed towards those with enough financial or other resources to invest in technologies like solar PV – or “resourceful prosumers” (Boekelo & Kloppenburg, 2023). While market trading might make locally generated electricity accessible to lower socio-economic status consumers, the bulk of the benefits derived from the energy transition will ultimately be directed to those who can generate electricity (Darmawan, 2019). Non-market energy sharing instead holds potential to ameliorate (some of) these distributional imbalances. It might also counter the growth and consumption logic institutionally embedded in energy systems (including prosumerism), by reconceiving energy as a shared commons as opposed to a commodity with which to accumulate wealth (Bauwens et al., 2024; Burke, 2021; Ritzel et al., 2022).
To frame our model, we therefore borrow Singh et al.’s (2017) distinction between “mutual energy sharing” and “mutual energy trading”. The former is the action of the EC sharing the surplus to local householders (who are not EC members) for free. This is, following the authors’ definition, “a social and personal energy exchange where an energy-giver and energy-receiver participate for the sake of social relationship between them”. The latter is the EC decision to sell the surplus to local householders for a price, which is defined as “a social and personal energy exchange where an energy-giver and energy-receiver participate in a calculated exchange for the sake of a commensurate material or monetary gain” (Singh et al., 2017, p. 109).
If ECs are to share energy with neighbours, the question arises of who should be shared with. In other words, how should the common pool of surplus energy be governed to reach community goals? Energy exchange is intricately tied up with energy demand response, since recipients might be expected to vary their demand to use surpluses at times they exist. Energy sharing and exchange thus provide an opportunity to coordinate and promote local use of energy for grid balancing (Directorate-General for Energy, 2024). So-called demand ‘flexibility’ is an important factor shaping the ability of actors to use the energy and reap benefits, which therefore also affects the extent to which energy can be consumed locally (Inderberg et al., 2024; Kubli et al., 2018; Libertson, 2022, 2024).
Those with so-called ‘flexibility capital’ are largely actors with pre-existing resources which they can mobilise (Libertson, 2022, 2024; Powells & Fell, 2019). These resources can take various forms, including financial and material assets such as EVs or heat pumps, or socio-temporal resources such as the ability to work from home. Providing cheap or free energy to those with flexibility capital risks exacerbating inequalities by rewarding those with the resources to be flexible (Libertson, 2024). Flexibility capital is therefore tied up with notions of energy or flexibility justice (Powells & Fell, 2019). Therefore, flexibility should be an important consideration of ECs in deciding who to share with and who to trade with. For example, would targeting low flexibility users for energy sharing work against some of these distributional issues? This has implications not only in terms of justice outcomes for recipients but also in terms of potential gains to the EC members, as well as maximising local use of energy. The interaction between these considerations is one of the driving mechanisms of our agent-based model.
The model developed in this paper explores how EC and neighbour household strategies around sharing and demand flexibility might develop under different agent objectives: economic gain, local electricity use, and solidarity. The first objective concerns maximizing economic gain through cost savings for householders and profits for the community. The second refers to optimizing local energy use to enhance system efficiency by reducing grid congestion. The third aims at fostering solidarity by allowing householders and community members to consider each other’s satisfaction. Agents are satisfied to the extent that they satisfy these objectives.
These objectives were chosen based on the existing literature. Economic gain is widely assumed to be a key driver of prosumer energy exchange, particularly among the market-based P2P literature (Montakhabi et al., 2023; Watson et al., 2022). Renewable energy cooperatives also work for members’ individual and mutual interests (Bauwens & Defourny, 2017; Dudka et al., 2024). Local use of energy and goals of autarky and autonomy are widely recurring goals of energy prosumers, both individual and collective (Adams et al., 2021; Montakhabi et al., 2023). This is captured by the so-called literature on transactive energy systems, which usually emphasises grid balancing (Adams et al., 2021; Montakhabi et al., 2023; Watson et al., 2022). Solidarity goals are evidenced in many of the studies mentioned above which show that prosumers, especially EC prosumers, are willing to share energy surplus for free with neighbours (Georgarakis et al., 2021; Hahnel & Fell, 2022; Reuter & Loock, 2017; Wilkinson et al., 2020). Sharing and social relationships are also a key value of many prosumer arrangements (Adams et al., 2021). Renewables cooperatives, while frequently working for members’ mutual interest, may also consider public interest (Bauwens & Defourny, 2017; Dudka et al., 2024).
The governance strategies available to ECs revolve around two key decisions: how much of the surplus will be shared versus sold, and what neighbour households should be targeted for sharing. The latter decision is based on household energy demand flexibility. Since flexibility capital is closely linked to other forms of capital, flexibility can thus serve as a useful proxy for household resources. Furthermore, flexibility is an important prerequisite for increasing local energy use. Therefore, the targeting of neighbour households with low or high flexibility likely has important direct consequences for the objectives listed above. These two key decisions will be described below as ‘policy levers’: willingness to share and flexibility targeting.
3 Model
Most agent-based models of P2P energy trading focus on market exchange (May & Huang, 2023; Monroe et al., 2020, 2023; Paudel et al., 2019; Sanayha & Vateekul, 2022; Zhou et al., 2017, 2018). Lovati et al. (2020, 2021) consider more community-controlled models. Fouladvand (2024) reviews the literature of ABMs applied to ECs and identifies a knowledge gap in terms of different business models and institutional and behavioural settings. In line with earlier studies this research adopts a theoretical exploratory perspective, essentially asking the ‘what if’ question of energy sharing.
ABM was chosen as the methodological approach because it is particularly well-suited to capturing the behavior of agents who are not fully rational and make decisions dynamically based on changing conditions; an essential feature for exploring energy sharing. Unlike multi-criteria decision-making methods or mathematical optimization models, which assume static preferences or require global optimization, ABM allows agents to iteratively adjust their decisions over time, making it highly appropriate for addressing the problem explored in this study; where different prosumer strategies are likely to dynamically evolve in interaction with each other in pursuit of their goals.
Specifically, our model aims to simulate the choices faced by an EC when it generates a surplus of energy. The surplus is considered to be generated by the entire community as a collective entity. As such, the model does not distinguish between community member generators and community member non-generators, emphasising shared responsibility. The community manages a single surplus asset, and we assume that the energy needs of the community have already been met. Neighbouring households who are not part of the community do not generate energy (hence they are not prosumers) and, at the start of the simulation, they only benefit from energy provision from the grid.
We also keep the surplus constant during the simulation run. This choice allows us to focus on the dynamics of energy sharing and flexibility adjustments, rather than the fluctuations of surplus. Modelling a dynamic surplus would require incorporating grid dynamics, which falls outside the scope of this study. By keeping the surplus constant, we isolate and examine the behavioural interactions that are the central element of energy sharing.
When a production surplus occurs, community members have three options to manage it: (i) First, they can sell the surplus back to the grid. The alternative options involve allowing neighbouring households who are not part of the community to use this surplus. Therefore, (ii) the second option for community members is to share the surplus for free, while (iii) the third is to sell the surplus to households. Both the second and third options (respectively ‘mutual energy sharing’ and ‘mutual energy trading’ as in Singh et al. (2017)) depend on the level of householders’ flexibility and the flexibility threshold set by the community, meaning that householders commit part of their demand to be used when the surplus occurs which must satisfy community conditions. The second and third options could potentially increase local use of energy surplus and reduce grid congestion, since we assume that neighbours are within the same local distribution network.
The amount of surplus that the community decides to share and the flexibility of householders dynamically change during the simulation due to continuous interactions between the two parties, as is common in market dynamics. In the model, we model householders with unlimited ‘flexibility capital’, hence they can move from 0 to 100% demand flexibility. This assumption precludes us from focusing on the energy justice issues related to energy exchange, but this is not the primary focus of our study.
Both the community and the householders adapt their decisions iteratively and consider three preference criteria. First, they consider the economic outcomes of their choices, such as how much householders reduce their overall energy costs by using energy from the community and how much the community increases its profit by selling and sharing surplus with householders. Second, both householders and community members consider overall system efficiency (we use this term interchangeably with its proxy: local use of energy surplus), specifically how much surplus is used by householders instead of being sent to the grid, thereby reducing grid congestion. Third, householders consider the satisfaction of community members, and community members consider the satisfaction of householders, aiming to maintain solidarity among a population that is collectively addressing a common challenge.
3.1 Model description
This model allows the exploration of how community-driven energy sharing and selling strategies interact with householders’ flexibility and demand, providing insights into the dynamics of energy distribution, economic outcomes, and local use of energy surplus within a local community context. A detailed model formulation is in Annex II.
In the model, a single community consisting of M members interacts with N noncommunity members, referred to as householders. The community has to manage a single surplus (T) of energy. Each timestep, community members (j ∈ M) decide the portion of their surplus to share, denoted by their willingness to share sj,t. The community’s overall sharing level (st), computed as the average value of each members’ willingness to share, determines the maximum surplus available for free distribution (xmax,t), while the remainder is available for sale (ymax,t), hence T = xmax,t + ymax,t. Community members also establish the minimum flexibility level (fmin,t) required for householders to access the shared surplus, averaging the individual preferences of each member regarding the minimum flexibility level (fj,min,t).
The community also sets a selling price (psell,t) for any surplus energy, averaged from individual members’ proposed prices (pj,sell,t). Community members dynamically adjust their proposed selling price based on householders’ satisfaction with previous transactions, mimicking real-world energy market behaviours, where sellers respond to consumer demand and satisfaction. Exceeding surplus, neither shared nor sold, is sent back to the grid, gaining a payback (psend). This price is treated as an exogenous parameter in the model. However, since grid payback conditions vary substantially due to regulation (sometimes even requiring consumers to pay) we conducted a sensitivity analysis to examine the effects on model outcomes.
Householders (i ∈ N) evaluate their options in each timestep, choosing between using the community’s surplus or purchasing from the grid, based on their fixed energy demand (Di) and their individual flexibility level (fi,t). Householders’ flexibility represents the share of demand that householders commit to be used when the surplus occurs. Households dynamically adjust their flexibility at each timestep, using it as a behavioural lever to achieve better outcomes. They reassess their options in each timestep, considering their individual flexibility levels and the external conditions driven by the decisions of the EC.
A householder’s decision to use community surplus is related to their flexibility level (fi,t) meeting or exceeding the community flexibility threshold (fi,t >= fmin,t) and their flexible demand (Dfi,t = Di · fi,t) being within the community’s available shared surplus (Dfi,t <= xmax,t). If these conditions are met, householders can use the surplus without cost and fulfil their remaining demand from the grid at a fixed price (pgrid).
When a householder’s flexibility level falls below the community flexibility threshold (fi,t < fmin,t) but their flexible demand can still be met by the surplus available for sale (Dfi,t <= ymax,t), they face a cost for purchasing community energy (psell,t) and potentially higher overall costs if the community’s selling price exceeds grid prices. Householders only opt to purchase from the community if this results in a lower total cost. In all other situations, householders must purchase all their energy from the grid at a fixed price (pgrid).
After making their decisions, householders adjust their flexibility levels dynamically based on their satisfaction with the previous timestep’s outcomes, aiming to increase their personal satisfaction, following therefore individual adjustment dynamics and directional learning (Nax, 2015).
Community members similarly adjust their strategies over time, modifying their willingness to share surplus (sj,t hence st), the minimum flexibility requirement (fj,min,t hence fmin,t), and the selling price (pj,sell,t hence psell,t). These adjustments are also influenced by their satisfaction. The community computes its profit at each timestep, derived from selling (yt) surplus and sending (zt) any unused surplus back to the grid.
Both householders and community members track their satisfaction (wi,t and wj,t, respectively), which is influenced by their preferences towards the economic outcome (ai and aj), the local use of energy surplus (or system efficiency) (bi and bj), and others’ level of satisfaction (ci and cj), the latter mimicking the level of solidarity. This iterative process allows agents to respond to each other’s experiences and potentially improve overall satisfaction and local use of energy surplus (the system efficiency ht) which is assessed by the proportion of surplus shared (xt) or sold (yt) to households.
3.2 Model initialisation
The model parameters are initialised as shown in Table 1. The population of householders (N) consists of 100 heterogeneous agents, each with different demand, Di, initialised with values drawn from a uniform distribution between 0 and 10. At t = 0, all householders satisfy their demand by purchasing electricity from the grid at a given price (pgrid). Householders also have different preferences for cost reduction (ai), system efficiency (bi), and solidarity (ci), each initialised with values drawn from a uniform distribution between 0 and 1. At the start, the level of flexibility among householders is also heterogeneous, with values drawn from a uniform distribution between 0 and 0.05. This implies that initially, some householders could potentially commit only a maximum of 5% of their own demand to be flexible in response to the community’s request to use surplus.
Table 1
Model initialisation.
| PARAMETERS | VALUE | |
|---|---|---|
| Number of householders | N | 100 |
| Number of community members | M | 20 |
| Community surplus | T | 200 |
| Price for buying from grid | pgrid | 0.5 |
| Payback for sending to the grid | psend | 0.5 |
| Householders | ||
| Demand | Di | ∈ [0,10] |
| Flexibility | fi | ∈ [0,0.05] |
| Agents’ weight for cost reduction | ai | ∈ [0,1] |
| Agents’ weight for system efficiency | bi | ∈ [0,1] |
| Agents’ weight for solidarity (caring about community members’ satisfaction) | ci | ∈ [0,1] |
| Community members | ||
| Willingness to share | sj | ∈ [0,0.05] |
| Minimum flexibility | fj,min | ∈ [0,1] |
| Selling price | pj,sell | ∈ [0,1] |
| Weight for profit increase | aj | ∈ [0,1] |
| Weight for system efficiency | bj | ∈ [0,1] |
| Weight for solidarity (caring about householders’ satisfaction) | cj | ∈ [0,1] |
| Weight for householders’ positive interactions (for selling price) | ej | ∈ [0,1] |
The population of community members (M) consists of 20 heterogeneous agents, each having different preferences for profit increase (aj), system efficiency (bj), and solidarity (cj), with values initialised from a uniform distribution between 0 and 1. The levels of willingness to share (sj) and the minimum flexibility accepted (fj,min) are also heterogeneous among community members, with sj values drawn from a uniform distribution between 0 and 0.05, and fj,min values drawn from a uniform distribution between 0 and 1. This implies that, initially, community members have low intention to share, as this may imply less profit, and varying levels of minimum flexibility accepted, as some may be more inclined to satisfy householders with even low flexibility levels while others only accept high flexibility. Community members are also heterogeneous concerning the price the community should set to sell the surplus to householders (pj,sell), with values initialised from a uniform distribution between 0 and 1. Lastly, community members have heterogeneous sensitivity to householders’ positive interactions (ej), also drawn from a uniform distribution between 0 and 1.
To explore the influence of policy levers on model outcomes, we conduct a policy mix analysis by varying the three parameters: fj,min, sj and pj,sell. We therefore constrain community behaviour within low/high bounds and assess how the system responds to different policy configurations, providing insights into the interaction between these levers.
The community surplus (T) is set to 200 and this remains constant over all of the simulation runs. The price householders need to pay to buy energy from the grid (pgrid) is also kept constant during the simulations at 0.5, as well as the payback price to the community for sending the energy surplus to the grid (psend = 0.5). The impact of these three model parameters will be assessed via a sensitivity analysis.
The model initialization is parametric. This approach allows us to focus on the behavioural dynamics of energy sharing and to reduce complexity related to physical energy flows. For example, household demand is set at about 5% of the total community surplus, a realistic scenario for communities with high surplus at given periods, an issue often challenging for grids and substations with limited capacity. Excluding units like kW or kWh avoids the need to model additional mechanisms, such as household demand and consumption patterns, which are beyond the scope of this study.
We run the model for 500 timesteps. Each timestep represents an abstract decision-making moment rather than a fixed unit of time. We therefore focus on how agents adjust their decisions under static external conditions such as surplus, demand, and prices (pgrid and psend). By abstracting time, the model emphasizes the long-term evolution of energy sharing behaviours and governance mechanisms, rather than short-term fluctuations in energy system dynamics. This allows us to assess the stable outcomes that emerge from repeated interactions over time.
To control for the effect of random elements in the model on the final results, we run 100 Monte Carlo simulations with different random seeds and present the averages. The model is implemented using the LSD platform (Laboratory for Simulation Development: see www.labsimdev.org). The LSD platform is a powerful tool for developing agent-based models in economics and social sciences. It simplifies model creation by requiring users to define only the computational content, with the platform generating efficient logic automatically. Its scalability supports large agent populations, while compiled C++ code ensures high efficiency and fast execution.
4 Results
4.1 Comparing scenarios
This section presents the results of three scenarios. As discussed in Section 2, the choice of who to share or sell with has implications for various potential goals that ECs or households might want to pursue including their own economic benefit, the local use of energy (or system efficiency), and the welfare of others. The first scenario consists of both householders and community members considering only cost reduction (ai = 1, bi = 0, ci = 0) and profit increase (aj = 1, bj = 0, cj = 0), respectively, when they assess their level of satisfaction (See Annex II for model formulation details). The second scenario adds to the previous one by incorporating concern for system efficiency for both householders (ai + bi = 1, ci = 0) and community members (aj + bj = 1, cj = 0). In the third scenario, both householders and community members also consider solidarity in their satisfaction level (ai + bi + ci = 1 and aj + bj + cj = 1).
Starting with purely economic considerations, then adding concern for system efficiency, and finally incorporating solidarity, this progression reflects a logical and realistic evolution of priorities from basic material wellbeing to broader, idealistic goals. Furthermore, by aligning the criteria for both groups, the results can inform policies that address all stakeholders uniformly, without requiring group-specific strategies. Nevertheless, additional simulations were conducted to explore all possible combinations of the three factors and are summarised in Annex I.
Figure 1 shows the outcome of the three scenarios in three columns. For each scenario, three summary plots are presented: in the first row, the trends over the simulation timesteps are shown for the following model parameters: selling price (psell), system efficiency (h), and the average level of satisfaction for householders (wi) and community members (wj). The second row of charts plots trends regarding the share of the community surplus (T), whether it is shared (x), sold (y), or sent back to the grid (z). The bottom row of charts plots trends related to the average value of householder flexibility (fi), the minimum level of flexibility householders should have to receive surplus for free (fmin), and the value of community sharing (s). Table 2 summarises the values of the model outcome, averaging the last 200 timesteps.
Table 2
Summary of scenarios’ outcomes.
| SCENARIO 1 | SCENARIO 2 | SCENARIO 3 | |
|---|---|---|---|
| psell | 0.99 | 0.76 | 0.41 |
| h | 0.81 | 0.88 | 0.98 |
| wi | 0.35 | 0.19 | 0.03 |
| wj | –0.81 | –0.44 | –0.31 |
| fi | 0.79 | 0.98 | 0.98 |
| fmin | 0.91 | 0.93 | 0.96 |
| s | 0.88 | 0.88 | 0.94 |
| x | 0.81 | 0.88 | 0.94 |
| y | 0 | 0 | 0.04 |
| z | 0.19 | 0.12 | 0.02 |
Figure 1
Outcomes of the three scenarios: trends of the simulation runs. Note: The horizontal axis in all plots represents simulation time, while the vertical axis shows values ranging from 0 to 1. For some model outcomes, such as householders’ flexibility or community sharing, these values represent percentages, whereas for others, like psell, they correspond to actual parameter values.
In the first scenario, the community sets the selling price very high, almost at its maximum value. The community rapidly increases the level of sharing, deciding that up to 88% of the community surplus should be shared for free with householders. Additionally, the minimum level of flexibility required increases rapidly, reaching 91%, meaning the community is willing to share for free only with householders that have very high flexibility levels. However, householders on average choose to be flexible for only about 79% of their demand, which means not all of them are able to enjoy the free energy shared by the community.
Furthermore, those who do not receive the surplus for free do not buy it from the community, as the selling price set by the community is too high, making it less attractive compared to buying energy directly from the grid. Under these conditions, where some householders receive free energy and others buy directly from the grid, community members experience very high negative satisfaction (on average around –0.81), while householders experience positive satisfaction (on average 0.35). Householders’ positive experience leads to an attempt by community members to increase the selling price, which however does not result in any profit as the percentage of surplus sold to householders is zero.
Instead, about 81% of the surplus is shared for free (less than the actual sharing level set by the community) and 19% of the surplus is sent back to the grid, which is not a fully positive outcome as sending energy back to the grid does not help in reducing grid congestion. Nevertheless, in this scenario, system efficiency reaches about 81%.
In the second scenario, system efficiency (hence local use of energy surplus) improves from 81% to 88%, and the selling price set by the community reduces to 0.76. When community members include concerns about system efficiency in their satisfaction assessments, the level of community sharing remains the same (88%), while the minimum flexibility accepted slightly increases (from 91% to 93%). In this scenario, driven by concerns for system efficiency together with economic performance, householders increase their flexibility level, rising on average from 79% to 98% in scenario 1.
These changed attitudes lead to a reduction in the average satisfaction of householders, which decreases to 0.19, while also improving community member satisfaction which, despite still being negative, improves from –0.81 to –0.44. Despite the reduction in the selling price, scenario 2 also does not result in the surplus being sold to householders. However, it leads to an overall increase in surplus sharing (88%) and a lower share of surplus sent back to the grid (12% compared to 19% in scenario 1), which is a better outcome when the goal is to reduce grid congestion.
In the third scenario, where both householders and community members relate satisfaction to economic performance, system efficiency, and solidarity, the result is a very low level of energy surplus sent back to the grid (only 2%) and a local use of surplus of 98%. Furthermore, in this scenario, the community can sell part of its surplus to householders (around 4%) mostly because community members decide to lower the selling price to 0.41. Average satisfaction levels of householders reduce and become almost zero while community members still improve their satisfaction, which is now –0.31 on average.
This analysis shows that solely focusing on economic performance does not lead to optimal system outcomes. In the first scenario, where only cost reduction and profit increase are considered, the system efficiency is 81%, and the selling price is set high, resulting in significant negative satisfaction for community members and positive but limited satisfaction for householders. Introducing concerns about local use of energy surplus in the second scenario raises system efficiency to 88% and slightly reduces the selling price which, while not enabling surplus sales, does improve overall satisfaction for community members and encourages higher household flexibility. The third scenario, which incorporates solidarity alongside economic performance and system efficiency, achieves the highest system efficiency at 98% and significantly lowers the selling price, facilitating some surplus sales to householders. This holistic approach results in near-zero satisfaction for householders but continues to improve community member satisfaction.
4.2 The effect of prices (psend and pgrid)
This section analyses the impact of prices on these dynamics. This is based on the third scenario, where both householders and community members consider system efficiency and solidarity along with the economic effects of sharing or selling the surplus. A full batch sensitivity analysis is conducted on the payback price that the community receives for sending surplus back to the grid (psend) and on the price householders need to pay for buying energy directly from the grid (pgrid). Both values vary in steps of 0.2 within the range of 0.1 to 0.9, resulting in 25 different model configurations. Other parameters assume the values as listed in Table 1. For each configuration, the results presented in Figure 2 are computed as the average of 50 simulation runs over the final 200 time-steps.
Figure 2
Summary tables of price sensitivity analysis. Cells in red show the lowest values while cells in green the highest.
Reducing the payback price charged to the community for sending surplus back to the grid (psend), thereby drastically reducing such incentives, while simultaneously increasing the price for buying from the grid (pgrid), leads to the highest satisfaction levels for both householders and community members. This approach results in the lowest level of surplus free sharing but also the highest level of selling the surplus to householders, even if the selling price set by the community reaches its maximum. Under these conditions, the system efficiency reaches approximately 97%, which, although high, is not the highest achievable value.
The key to these results lies in the behavioural incentives created by adjusting psend and pgrid. By lowering psend, community members are discouraged from sending surplus back to the grid, making it more advantageous to share or sell the surplus within the community. Simultaneously, increasing pgrid makes buying energy from the grid more expensive for householders, encouraging them to buy surplus energy from the community instead. This dynamic leads to higher satisfaction among all parties involved and more efficient use of the community’s energy surplus.
However, the highest system efficiency is not achieved under these conditions but rather with slightly higher value of psend and with slightly lower value of pgrid. This suggests that while extreme pricing adjustments can drive desirable behaviour changes, there is an optimal balance where system efficiency is maximised without necessarily pushing psend and pgrid to their extremes. Higher psend makes it moderately more attractive to send surplus to the grid when conditions for energy sharing are less convenient, while a lower pgrid reduces the financial burden on householders when grid energy is needed.
4.3 The effect of community surplus (T)
Based on Scenario 3 and the initialisation of model parameters in Table 1, this section explores the role of community energy surplus (T). Specifically, it examines the effects of varying levels of community surplus on sharing and selling dynamics, and consequently, on the grid. A sensitivity analysis is run on the amount of surplus the community manages. The parameter T is varied in steps of 50 within the range of 50 to 1000, resulting in 20 different model configurations. As before, for each configuration the results presented in Figure 3 are computed as the average of 50 simulation runs over the final 200 time-steps.
Figure 3
Model outcomes based on varying level of community surplus (T).
Plot (a) shows that as the surplus the community has to manage increases, system efficiency (h) decreases and the price the community sets for selling the surplus to householders (psell) increases. However, both householders’ and community members’ satisfaction gradually increases with higher surplus (wi and wj, respectively). Plot (b) indicates that when T reaches 500, there is a continuous drop in the amount of surplus the community shares for free with householders, accompanied by a simultaneous increase in the surplus sent back to the grid. Interestingly, only at very low levels of surplus (T ≤ 250) is the share of surplus sold to householders above zero. Householders’ flexibility (fi), community members’ minimum flexibility level (fmin), and their willingness to share (s) are consistently above 90% (plot(c)).
To interpret such results, it is important to remember that householders’ demand is initialised with values drawn from a uniform distribution between 0 and 10 (Di ∈ [0,10]). This implies an average demand value of 5 per householder, meaning that if all householders decide to have fully flexible demand (fi = 1), the total maximum flexible demand for 100 householders would be 500. This value is critical as it is the values at which Figure 3 shows a noticeable change in the trend of many model outcomes.
This implies that as long as the community generates a surplus that covers the maximum flexibility level of householders, then free sharing remains the preferred option. In this case, only a small percentage of the surplus is sent back to the grid, resulting in higher system efficiency and a lower community contribution to grid congestion. However, when the community surplus exceeds this value, a larger percentage of the surplus is sent back to the grid, thereby lowering system efficiency. In other words, when householders are unable to absorb the community surplus, then grid congestion can potentially worsen.
This result is influenced by the fact that the payback for sending the surplus back to the grid is still positive (psend = 0.5, as in Table 1). If this community incentive were reduced or eliminated, we could expect that the dynamics would shift further. Specifically, without the incentive to send surplus back to the grid, the community would need to find alternative ways to manage excess energy, possibly leading to innovative storage or usage solutions to maintain high system efficiency and reduce grid congestion.
4.4 Policy mixes analysis
As noted above, community members adjust their strategies over time in response to interactions with householders and the resultant satisfaction in each previous timestep. The three policy levers at community members’ disposal to influence the next round of interactions are their level of sharing (determined by the average of the community members’ willingness to share (s)), the level of flexibility householders should have to receive shared surplus for free (fmin), and the price the community sets for selling the surplus to householders (psell).
A policy-oriented analysis was therefore conducted in which these three policy levers were varied high or low, giving in total eight policy mixes (P1–P8 in Table 3 below). Low level ranges from 0.01 to 0.25, while high level ranges from 0.75 to 0.99. These upper and lower bounds constrain the community’s behaviour within defined limits, reflecting potential government or EC policy guiding community energy sharing. This means that each individual community member can adjust their own choices (sj, fj,min, pj,sell) subject to these boundaries. All model parameters are set as in Table 1, and we base these policy mixes on Scenario 3 where both householders and community members relate satisfaction to economic performance, system efficiency, and solidarity.
Table 3
Policy mix scenarios. Summary of outcomes for eight policy mixes (P1 to P8) combining low and high levels of: (i) the minimum flexibility required for householders to access free surplus (fmin), (ii) the community’s willingness to share surplus (s), and (iii) the price set by the community for selling surplus to householders (psell). Low levels range from 0.01 to 0.25, and high levels range from 0.75 to 0.99.
| P1 | P2 | P3 | P4 | P5 | P6 | P7 | P8 | |
|---|---|---|---|---|---|---|---|---|
| Level of minimum flexibility (fj,min) | Low | Low | Low | Low | High | High | High | High |
| Level of community sharing (sj) | Low | Low | High | High | Low | Low | High | High |
| Level of price for selling surplus (pj,sell) | Low | High | Low | High | Low | High | Low | High |
| Selling price (psell) | 0.14 | 0.91 | 0.12 | 0.86 | 0.12 | 0.91 | 0.12 | 0.86 |
| System efficiency (h) | 0.09 | 0.09 | 0.97 | 0.97 | 0.96 | 0.09 | 0.99 | 0.97 |
| Householders (w– i) | 0.03 | 0.03 | 0.03 | 0.03 | 0.04 | 0.03 | 0.04 | 0.03 |
| Community members (w– j) | –0.01 | –0.01 | –0.32 | –0.32 | –0.24 | –0.01 | –0.33 | –0.32 |
| Householders flexibility (f–i) | 0.92 | 0.92 | 0.98 | 0.98 | 0.97 | 0.92 | 0.98 | 0.98 |
| Minimum flexibility accepted (fmin) | 0.09 | 0.09 | 0.11 | 0.11 | 0.98 | 0.97 | 0.97 | 0.97 |
| Community sharing (s) | 0.09 | 0.09 | 0.97 | 0.97 | 0.1 | 0.09 | 0.97 | 0.97 |
| Shared (x) | 0.09 | 0.09 | 0.97 | 0.97 | 0.1 | 0.09 | 0.97 | 0.97 |
| Sold (y) | 0 | 0 | 0 | 0 | 0.86 | 0 | 0.02 | 0 |
| Sent (z) | 0.91 | 0.91 | 0.03 | 0.03 | 0.04 | 0.91 | 0.01 | 0.03 |
Of particular interest from this analysis is the relative importance of sharing when compared with selling price in driving outcomes across the policy mixes. From Table 3, it is evident that the selling price (psell) alone does not significantly impact the outcomes, as there is no clear relationship between low or high selling price levels and other model results. In contrast, the level of sharing (s) emerges as a key driver in these policy mixes. Specifically, low sharing levels result in low system efficiency (averaging around 9%) and the highest proportion of surplus sent back to the grid (approximately 91%). However, an interesting interaction occurs in P5, where low sharing is combined with a low selling price. In this scenario, system efficiency increases to 96%, and the surplus sent back to the grid reduces substantially. This indicates that while the selling price alone has minimal effect, its interaction with low sharing levels significantly influences outcomes. When the community reduces its willingness to share (s) but simultaneously lowers the selling price (psell), householders are incentivised to purchase from the community instead of the grid. This is because the community’s selling price is much lower than the grid’s purchasing price (pgrid = 0.5, as in Table 1). As a result, this scenario leads to the highest level of community surplus sold to householders (approximately 86%), highlighting its importance in shifting energy consumption from the central grid to the community surplus.
Furthermore, we can identify three targeted policy mixes, each addressing a specific EC goal. Firstly, the economic profit policy mix (P6) is characterised by demand for high flexibility, a low level of sharing, and a high selling price to households. Secondly, the local use policy mix (P7) is characterised by demand for high flexibility, a high level of sharing, and a low selling price to households. Thirdly, the solidarity policy mix (P3) is characterised by low flexibility requirements, a high level of sharing, and a low selling price to households.
Comparing these three policy mixes specifically, household satisfaction remains stable across each of these (and all) policy mixes. System efficiency is low in P6 (economic profit policy mix), although community members are on average more satisfied. This is likely because they benefit financially from selling more energy back to the grid. The solidarity policy mix (P3) and the local use policy mix (P7) have quite similar outcomes in terms of system efficiency, and both householder and community member satisfaction. It is interesting that the policy feature (i.e. minimum flexibility required) distinguishing these mixes (minimum flexibility threshold) does not seem to affect outcomes on average. This potentially points to the feasibility of energy sharing strategies which target low flexibility households (often those with fewer resources), without necessarily sacrificing local use.
5 Discussion
In this paper we have developed a model to investigate the benefits and potential limitations of energy sharing in local prosumer interactions. Our findings have generally strengthened the case for the facilitation and implementation of non-market energy sharing. Key findings reveal that solely focusing on economic performance does not lead to optimal use of energy surplus. Instead, incorporating solidarity and concerns over local energy use achieves better outcomes in terms of system efficiency, as well as prosumer satisfaction. Furthermore, the solidarity policy mix which targets low-flexibility households (often those with lower resources) does not compromise the system efficiency.
5.1 Solidarity and the energy commons
These findings have relevance for recent contributions which advocate for more collectively oriented energy prosumer relations. Several authors posit the commons as a central organising idea towards which these ‘co-prosumers’ are oriented (Bauwens et al., 2024; Burke, 2021; Byrne et al., 2009; Giotitsas et al., 2015, 2020, 2022; Hall et al., 2019; Marzban et al., 2023; Melville et al., 2017; Ritzel et al., 2022; Van Zyl-Bulitta, 2019; Wolsink, 2020). They argue that energy and electricity has potential to be organised in a way that incentivises collective benefit and sustainability rather than individual profit-maximisation, and that commons provide an ideal institutional foundation for this. Collective ownership and control arrangements, such as those in ECs, can be considered examples of energy commons (Gollwitzer et al., 2018). However, there is a danger that commons become insular and exclusive when based purely on direct reciprocity and personal relations (Bauwens et al., 2024; Euler, 2019). An important question is therefore how to avoid energy commons becoming exclusive. Relatedly, if ECs treat their surplus energy as a commodity, there is a risk that they simply replicate capitalistic growth logics. Our model has shown how solidaristic energy sharing is a viable alternative to market-based surplus trading, facilitating an “upwards” movement from pursuit of mutual interest of EC members to public interest (Bauwens & Defourny, 2017; Dudka et al., 2024). Furthermore, if beneficiary households can use their savings to invest in their own generation assets, as proposed by Gazmararian and Tingley (2024), this could provide a pathway to achieve system-wide and more equitable prosumerism (Ruotsalainen et al., 2017).
Under scenario 3 in our simulations, in which all agents considered both the local use of energy surplus and the satisfaction of others (solidarity), superior outcomes for system efficiency are observed. There is also a convergence in community and household satisfaction, with community members being considerably less unsatisfied and householders being less, but still positively, satisfied. This could be due to a ‘solidarity effect’ since agents’ satisfaction depends on the satisfaction of others. This is an interesting finding which stems from our assumptions about agent behavior contrasting with the usually assumed independent, self-interested homo economicus often seen in such models (Singh et al., 2017).
5.2 The sharing economy
Another related concept which has been applied to P2P energy exchange is the sharing economy (Schneiders et al., 2022). P2P energy exchange has been framed as sharing because it involves the shared use of underutilized assets, and the transfer of energy as a service rather than an exchange of goods. Since ownership of the asset is not transferred, it is not a direct trade of commodities. However, this notion of sharing is a weak one and might be better understood as assetization (Birch & Muniesa, 2020; Birch & Ward, 2022) or a ‘neo-capitalist approach’ (Martin, 2016; Pasimeni, 2021). As Schneiders et al. (2022) recognise, this does not form a break with commodification of energy. Montakhabi et al. (2020) also discuss the possibility for an energy sharing economy which is essentially rental assets and still commodification. Drawing on economic anthropology literature, we want to distinguish between market-based exchange and sharing-based exchange, with the former being focused on a calculation of commensurate material gain, whereas the latter is focused on achieving stronger social ties. Our model explicitly differentiates between these two forms of socio-economic relations.
5.3 Model extensions and limitations
This model is among the first attempts to simulate energy exchange without relying solely on traditional market dynamics. We have therefore opted to illustrate the main outcomes of the model using simple dynamics and without making strong assumptions. Nonetheless, the agent-based model presented in this paper allows for the easy implementation of additional features.
For instance, we have simulated the condition of a community with a given, stable surplus. This is a limitation of our model as community surpluses vary substantially with time. As a model extension, this surplus could be made dynamic (e.g. to match natural variability in sunlight or wind). Similarly, prices (psend and pgrid) are currently held constant but could be made dynamic in the model to follow price changes. For example, as local use increases, so does grid price. This would tie in with utility death spiral literature (Athawale & Felder, 2022; Castaneda et al., 2017; Laws et al., 2017).
Extending the model to incorporate temporal dynamics could address the limitation of using an abstract timestep. Hourly or real-time intervals would enable the exploration of interactions between governance mechanisms and physical energy system behaviors.
Currently the model simulates one community with fixed size. An interesting model extension would be to allow the community to grow or for more households to be brought into the exchange or, in other words, the extension of residual control rights or rights to residual surplus (Bauwens & Defourny, 2017).
Furthermore, in the current simulation, community members have the option to send the surplus back to the grid or to sell it to non-community members. In the future, we plan to add a condition that allows community members to buy extra devices, such as batteries to store surplus (thereby increasing their flexibility), or even to increase their own demand. This could involve a kind of resource coordination game among community members with different technologies.
We assume flexibility can be freely changed by all agents. The notion of flexibility capital emphasises that this is not the case in reality since people are constrained by their resources and lifestyles (Libertson, 2022, 2024; Powells & Fell, 2019). Incorporating these limitations into the model and analysing their distributional effects is an interesting avenue for further study. Household purchase of energy from the community should perhaps also require flexibility.
The abovementioned ‘solidarity effect’ is an interesting result. However, if the starting point of an EC’s energy sharing is to improve the material economic condition of poorer neighbour households, then a more ‘objective’ indicator might be preferrable for solidarity. This could capture the economic effects on households independently of their satisfaction caused by community member benefits and local use of energy. Similarly, the price-setting mechanism between communities and households could focus only on economic benefit of others, rather than their full satisfaction.
The model can be extended to two-way P2P type of exchanges between households, also among community members. This could also incorporate investment decisions by prosumer households.
The proposed model has some limitations and relies on certain assumptions that should be acknowledged. For instance, both householders and community members adjust their behaviour (fi, fj,min, sj) depending on their satisfaction, following a fractional power growth. Although this mirrors potential behaviour, we do not have empirical evidence to test such a trend. However, it is the best way to simulate adaptive behaviour in dynamically changing conditions.
We assume institutional barriers to EC energy sharing can be overcome. These institutional barriers include legal rights for different actors including households and ECs to engage in energy exchange, grid management and energy market arrangements which favour larger, commercial actors, lack of targeted policy support for emergent local energy exchange business models, lack of guiding rules and standards for the organisation of such initiatives, and incumbent utility resistance (Mourik et al., 2020; Van Summeren et al., 2020). While some promising opportunities are emerging to overcome such barriers in some contexts (for example EU and its member states recent legislation), the resilience and stasis of these institutional barriers should not be overlooked. We assume technological feasibility in terms of both the hardware (e.g. renewable energy generation assets) and software (e.g. energy management systems) required to perform energy exchange. While the proposed dynamics and relations are technologically feasible in theory, they have not been sufficiently proven in practice (Giotitsas et al., 2022). We assume very simplistic democratic decision-making within the community based on the average of member preferences.
This research, like most research in this area is speculative in nature, such as hypothetical surveys (Georgarakis et al., 2021), simulation models or scenario-building workshops and interviews (Marzban et al., 2023; Montakhabi et al., 2023; Wilkinson et al., 2020). We call for more real-world experiments of both market and non-market energy sharing, especially within energy communities, such as Wilkinson et al. (2020).
5.4 Implications of findings for EC and government policy
What are the main dynamics we observe taking place in the model, and what might they mean for EC and state actors in this space?
We find that, as long as the community generates a surplus that covers the maximum flexibility level of householders, sharing is the dominant option, even if agents are solely focussed on their own economic benefit. If agents include solidarity in their objectives, this leads to marginally higher sharing interactions, and also slightly higher flexibility among householders. Results also show that integrating local use of electricity and solidarity into agent objectives can increase overall outcomes (highest local use of electricity and improved community satisfaction, albeit lower but still positive household satisfaction) resulting in lower selling prices with some surplus sold to householders. ECs could therefore use their local influence to promote energy solidarity values among both community members and local neighbour households.
Government policies that reduce the financial incentives for sending surplus energy back to the grid while increasing the cost of buying energy from the grid can promote local energy sharing and sales. This approach increases the use of locally generated energy surplus, enhancing community satisfaction and reducing reliance for householders on external energy sources. This basically describes the situation of a utility death spiral, where utilities push back against net metering and also raise costs for normal consumers. If this model replicates real-world dynamics in this sense, then the implication is that this can drive further local energy exchange (rather than curtailing local generation), but potentially at the risk of causing economic distress to consumers who are essentially forced to switch providers to local ECs. This is likely an undesirable scenario as it creates local dependency relations.
Our policy mix analysis also reveals some relevant findings for both government policymakers but also ECs and their networks, who might be unsure how best to interact with their neighbours in complex socio-technical environments. While these findings are of course not conclusive, our simplified policy mix warns that ECs who target economic profit may hinder system efficiency by discouraging local use from sharing or selling electricity. In contrast, the local use and solidarity policy mixes attain higher levels of local use, albeit at the expense of EC member profits. Of further interest is the result that the solidarity policy mix which targets low-flexibility households (often those with lower resources) does not compromise system efficiency.
To conclude, researchers and policymakers should be more open to the possibilities for non-market mediated forms of energy exchange and should create conditions which allow for these to flourish. Implicit assumptions about intrinsic profit-orientation of prosumers is not only empirically and theoretically unfounded, but also potentially counterproductive in resolving grid congestion problems.
Additional File
The additional file for this article can be found as follows:
Acknowledgements
The authors would like to thank the anonymous reviewers for their time and valuable feedback.
Competing Interests
The authors have no competing interests to declare.