1. Introduction
Energy retrofit (ER) of the building stock is a key component in reducing greenhouse gas emissions for cold and temperate climate zones. However, despite various policies, the actual ER rate is far below the required dynamics to face rapid global warming for most countries (BPIE 2022). Since the domain of ER originates in engineering and architecture, most available counselling tools to support decision-makers in politics focus on technological issues to foster ER using optimisation models, economic models or potential analysis models (Shu & Zhao 2023). These approaches often assume that the decisions of humans are purely rational (Homo economicus), only a matter of economic cost–benefit analysis and sufficient knowledge of the building owners to decide for ER (Müller 2013). Although cost–benefits and knowledge are important, the decision-making process of building owners for ER is subject to bounded rationality due to cognitive biases, as, for example, proposed by prospect theory (Ebrahimigharehbaghi et al. 2022). Therefore, policy design must take those biases into account, as the consideration of loss aversion (the human tendency to avoid losses at high costs) requires much stronger policies to meet the target goals (Knobloch et al. 2019). Furthermore, the heterogeneity of individual decision-makers leads to complex multi-actor dynamics and hence the necessity to implement this uncertainty into energy systems modelling to obtain reliable policy assessments (Li 2017). Hence, a recent model of ER captures individual attributes of homeowners, a multi-staged decision process with different influences on each stage and social influence through personal networks to simulate the resulting behaviour for different scenarios (Liu et al. 2024). However, such mathematical modelling comes with the necessity to restrict the number of psychological variables and to make specific assumptions concerning the parameters involved meanwhile suggesting a precision of predictions and recommendations that cannot be fulfilled due to the number of especially soft (social) parameters and the dependency on specific, yet unknown, context conditions of policy implications. To integrate decision-making theories from different disciplines as economics, psychology and sociology, new approaches should be reconsidered to bridge the gap between the disciplines and the different levels of description (Wilson & Dowlatabadi 2007).
Complex systems theory provides such an approach encompassing insights from different disciplines and influences on different levels of description towards a better understanding of human decision processes (Scherbaum et al. 2008). To grasp the social complexity of decision-making, quantitative physical and empirical modelling should be complemented by qualitative system modelling approaches that consider interconnectedness and the side-effects in socio-technical systems (STS) (Schünemann et al. 2024). Accordingly, the existing research gaps could be bridged as (1) the lack of consideration of how the barriers and drivers for ER decisions are interlinked within STS; and (2) the challenges of implementing complexity into a decision-making model. Several studies analyse the drivers and barriers in detail (see Appendix B in the supplemental data online), but these lack systemic consideration of how the complex decision-making process of building owners in the STS is affected (Liu et al. 2024). A systemic perspective of ER as an innovation diffusion process is considered by a few studies, but these do not focus on the decision-making process of building owners (Bobrova et al. 2018; Guo et al. 2019; Müller 2013; Onat et al. 2014).
The present study aims to develop a decision-making model that gathers economic, psychological and sociological factors under the umbrella of a complex systems perspective. This is done by developing a qualitative system model (Causal Loop Diagram—CLD; Sterman 2002) whose core structure is based on a psychological decision-making model. As the underlying decision model, the Rubicon model is selected, a psychological framework that describes the transition between different phases in goal-directed behaviour (Heckhausen & Gollwitzer 1987). It divides the process into motivation, planning, action and evaluation phases emphasising a ‘Rubicon’, or point of no return, when a decision is made, shifting the focus from deliberation to goal pursuit. Combining this individual decision-making perspective with the systemic perspective by including the perceived social norms formed by media and social environment, the system model can demonstrate the added values of this approach. It does this by explaining the low dynamics of ER and identifying crucial leverage points to foster the ER dynamics. The focus is on the decision-making process of single-family homeowners (in the following, ‘homeowners’) for ER in the German context. However, most of the findings can be transferred to other building owners and countries with small adjustments. Based on the systemic perspective, a complexity-informed policy advice approach (Nel & Taeihagh 2024) is applied and general recommendations for policy are derived from the model.
Another aim of this article is to demonstrate how psychological models can be used to develop complex social system models able to represent and understand innovation diffusion processes for sustainable transformations in more detail.
2. Methods
2.1 Systemic decision-making model development process
Figure 1 gives an overview of the main steps conducted to develop the systemic decision-making model discussed in this article. An intensive unstructured review was undertaken on the motivational factors and barriers that influence homeowners’ decision-making regarding ER (more details are given in the following section). Based on this list, the most relevant ones were selected for the implementation in the systemic decision-making model, including feedback from informal exchanges with stakeholders and experts in the field of ER. The factors represented the basic elements for qualitative system model (i.e. CLD) development, whose general structure is inspired by the Rubicon model, an explanatory model from psychology. This structure was identified as best suitable from an unstructured review of existing decision-making and innovation diffusion theories from the fields of psychology, sociology and economic science. Here, the deep expertise of the co-authors was included as well. Technical details of the model creation are explained in Section 2.3.
Figure 1
Main steps in the development process of the systemic decision-making model of homeowners regarding energy retrofit realisation.
A first version of the systemic decision-making model was created. This included several rounds of feedback on the model and revisions.
As the aim of the model is to be explainable, the variables were significantly reduced (about half), which is a critical step because this does not happen without loss of information in the model (discussed in Section 4.3). In addition, the model was visually improved (colouring, arrangement of the variables, highlighting of feedback loops) and essential processes were added so that it is easier to understand. Based on this explainable model, a 27-min video was created and uploaded on YouTube where the systemic decision-making was explained by a stepwise fade-in of considered variables, connections and resulting feedback loops. The video was developed with the purpose of sending it to 31 German experts and stakeholders in the ER domain and asking them to validate the developed systemic decision-making model of homeowners. The feedback of 14 experts was collected mainly by direct dialogue in virtual format or via answers formulated in texts. Section 2.4 gives a more detailed explanation of this expert validation process. The diverse remarks of the experts based on the video presentation were collected and included in several revision iterations in the model to finally derive the systemic decision-making model of the homeowners regarding ER realisation.
This model development process proved to be very efficient in gaining a validated qualitative and visually based system model. At the beginning of the project, the original intention was to conduct several participatory modelling workshops (based on group model building; Hovmand 2014) with stakeholders and experts. Due to the time constraints of the stakeholders and experts, it was decided to use the outlined process of validating the model.
2.2 Drivers and barriers for energy retrofit
The required system variables and their interconnections for the system model are identified by detailed literature reviews. For this unstructured review of the international scientific literature, different search strings (see Appendix A in the supplemental data online) were applied in different literature databases (SCOPUS, Web of Science, Google Scholar). In addition, the German grey literature on barriers and drivers was reviewed using similar research strings (see Appendix A online) in the German language in Google Scholar and in an adapted form in Google. The research was focused on identifying articles and reports discussing the drivers and barriers in homeowners’ decision-making for realising ER. Scientific articles and reports were screened in an unstructured manner, neglecting the large part of the literature discussing technical issues and concentrating on those focusing on homeowners’ decision-making. Using the results of the databases and snowballing effect, 30 of the most relevant literature sources were analysed in detail, which are listed in Appendix B online. The mentioned drivers and barriers from each of those 30 sources were extracted into an Excel table. This created a comprehensive overview of factors: some which were rarely discussed, and others were mentioned frequently in the literature. From this overview, the barriers and driver details were grouped into categories, e.g. the barrier ‘financing difficulties’ to allow an indication of their frequency of mentioning from all 30 sources. Here, the barriers and drivers were divided into intrinsic and extrinsic ones. Table 1 lists the main identified barrier and driver categories that were mentioned more than twice from all 30 sources. The most relevant ones, which were mentioned at least six times, are highlighted in bold font. Those identified drivers and barriers represented variables for the first draft of the CLD model creation process.
Table 1
Major intrinsic and extrinsic drivers and barriers for homeowners’ energy retrofit (ER) decisions.
| DRIVERS FOR ER | BARRIERS FOR ER | ||
|---|---|---|---|
| INTRINSIC | EXTRINSIC | INTRINSIC | EXTRINSIC |
| Preservation/increase of property value | Financial incentives | Efforts and inconveniences of the renovation process | Higher construction costs due to ER |
| Reduction of heating costs | Positive examples and word-of-mouth | Financing difficulties (low income) | Long payback period for investments |
| Personal contribution to climate goals | Lack of awareness of the contribution to climate change | Limited availability of qualified planners and craftsman | |
| Increase in living comfort | Feeling overwhelmed by the complexity of ER | Shortage in the supply of building materials | |
| Lack of knowledge about the energy condition of the property | Negative examples and word-of-mouth | ||
| Lack of know-how | Lack of information to support decision-making | ||
| Fear of unsatisfactory results | Bureaucratic demands | ||
| Inappropriate individual contextual and social circumstances for ER | ER solutions are highly individual | ||
[i] Note: Factors shown in bold are mentioned in six or more of all analysed reports and articles. For details of the sources used, see Appendix B in the supplemental data online.
2.3 Developing the systemic decision-making model
System dynamics approaches, also including qualitative CLD models, are known to be very promising for different issues of complex system modelling, including behavioural modelling (Lane & Rouwette 2023). In the present case, the societal process of ER diffusion within society is assumed to occur as described by the innovation diffusion theory. This posits that innovation starts from innovators, then spreads to early adopters, early majority, late majority and finally laggards (Rogers et al. 1969). For ER, the diffusion intensity depends on the goal-seeking behaviour (Forrester 1971) of individual homeowners to weigh the benefits and the achievability of the goal (including self-efficacy) against the perceived efforts accompanying performing an ER. For the CLD system model development, the psychological Rubicon model serves as a structural basis for the specific case of homeowners’ ER decision, which is explained in more detail in Section 3.2. The CLD visualises the interconnectedness of the social ER diffusion process, focusing on feedback loops that originate from the polarised connectors between the system variables (Sterman 2002). The CLD was created using the tool STELLA Architect (v.2.0.3).
2.4 Expert interviews for model validation
For model validation, a video describing the simplified explanatory model was created and sent to different experts and stakeholders active in the domain of ER. The experts were chosen by their identified significant contributions to solving the problem of low ER dynamics in the German context, gained from the literature review and the existing network of the authors. The chosen experts are active in policy advising (non-governmental organisation or federal ministries), in the practice of advising homeowners (energy efficiency counselling) or in research about human decision-making. The aim of expert involvement was not only to validate the model but also to involve them in the discussion of homeowners decision-making process and to develop joint policy suggestions in the future. This feedback was crucial to build trust in the model and thus enhance the adoption of the findings in policy advice.
Expert and stakeholder feedback was important for validating the CLD model. The model was only based on the system knowledge gained from the literature, the mental model and knowledge of the authors, and exchanges with colleagues.
The video was sent to 31 ER experts and stakeholders in Germany with the request to give direct feedback on the model with a specific focus on completeness, structure and incorrectness. A total of 14 experts and stakeholders responded from very different fields (for more details, including their expertise, see Appendix C in the supplemental data online). The feedback was mainly gained by a direct feedback dialogue (mainly in virtual format) or via email feedback.
A video format for a step-by-step model description is a suitable way to avoid an overload of information when presenting a complex system model (Sterman 2002). However, two side effects of the expert interviews were noticed. It became very obvious that the stakeholders and experts with their different expertise gave very different feedback and had different system knowledge regarding the CLD model. The experts did not consider the complexity and interdependencies of homeowner’s decision-making for ER in all aspects of the systemic decision-making model. This reflects the fragmented nature of knowledge when handling highly complex systems (Forrester 1971; Hovmand 2014). In addition, this approach of starting with an existing system model also seems to limit the innovativeness and open feedback opportunities of the experts and stakeholders. After the diverse feedback, the collected issues were used to revise the CLD model and finally to develop the presented version.1
3. Results
3.1 Major drivers and barriers for energy retrofit
As the basis for developing the system model, the scientific and grey literature was screened to select the most relevant drivers and barriers for homeowners to decide for or against ER. Table 1 lists them divided into intrinsic and extrinsic.2 Economic factors are important for ER diffusion, but also psychological (e.g. cognitive bias towards the status quo, no priority for energy efficiency) and communicative barriers (e.g. explaining the ER options and processes for homeowners) (Koch & Rosenberger 2019). The literature review showed that the individual decision process for ER is highly complex, depending on a broad range of internal and external influences from the political environment, subjective norms and values, perception of ER discussion from the media and by word-of-mouth, as well as behavioural control or emotions.
3.2 The qualitative systemic decision-making model
The identified drivers and barriers in Table 1 were taken as system variables for developing a first draft of the decision-making CLD model. Additional variables were implemented from the expert feedback dialogues in a revised version. The model aims to represent visually the interrelated factors relevant to the decision of homeowners to undertake ER. The model includes a large number of soft variables describing the individual decision process with influences from the societal perspective by perceived social norms. Therefore, the system model is a mixture of an individual decision-making process with societal (aggregated) influences by societal aspects (e.g. how ER is debated in society).
3.2.1 Model and system structure
The general structure of the developed system model shown in Figure 2 is based on the four phases of the Rubicon model that describes the social process of goal realisation from contemplation to concrete action. A central concept is the shift or tipping point from motivation to volition, symbolically referred to as the moment of ‘crossing the Rubicon’. Table 2 explains the four phases of the Rubicon model and its implementation into the systemic CLD model. In the narrow sense, the state of the homeowner (uninterested, interested, determined, holding a final ER plan or living in an ER home) is represented by stocks and its transition by flows as it is common for stock-flow diagrams, an extension of CLD (Sterman 2002). This representation is used to enhance the visibility of the stages of homeowners and their development process. The focus of the model explanation will be on the pre-decisional motivation phase (Rubicon phase 1) because it describes the crucial process of uninterested homeowners deciding to proceed with ER. In this pre-decisional phase of the model, the fundamental condition for action is laid—based on cost–benefit analyses of the homeowner and a range of psychological factors. The other three Rubicon phases are important to reflect the circumstances of the homeowners in realising ER and their satisfaction with the ER result which evokes important feedback loops (see the next section). To raise homeowners’ general interest in ER, the major identified factors influencing their attitude towards ER are:
Figure 2
Causal Loop Diagram (CLD) as a qualitative system model representation describing homeowners’ energy retrofit (ER) decision-making process.
Table 2
Phases of the Rubicon framework with their general description and how it is implemented in the system model shown in Figure 2.
| RUBICON PHASE | GENERAL DESCRIPTION | IN THE SYSTEM MODEL |
|---|---|---|
| Phase 1 (pre-decisional) | Motivating for energy retrofit (ER) including weighing pros and cons (Why should I do ER?) | Divided into two steps:
|
| Rubicon | Transition from motivation to volition | Motivated/interested homeowners decide to do ER |
| Phase 2 (post-decisional) | Volition with planning action of ER (How to realise ER?) | The determined homeowner plans the detailed ER realisation with external expertise (variables with orange text in the CLD model) |
| Phase 3 (action) | Realising ER | The homeowner executes the ER of the home (variables with dark red text in the CLD model) |
| Phase 4 (evaluation) | Reflecting if ER results are satisfying (Was it successful?) | The homeowner living in the ER home reflects on whether the effort of the ER process is worth the result (variable with dark red text in the CLD model) |
[i] Note: The four phases of the psychological Rubicon framework are: Phase 1 (motivation) with red and purple variables, overcoming the Rubicon to phase 2 (volition) with orange variables, phase 3 (action) and phase 4 (evaluation) with dark red variables. Variables in grey are external. Reinforcing (R) and balancing (B) feedback loops are numbered according to their description in the text and highlighted in Figures 3 and 4 for better visualisation.
Figure 3
Highlighting the seven identified reinforcing feedback loops of the systemic decision-making Causal Loop Diagram (CLD) model.
Figure 4
Highlighting the three identified balancing feedback loops of the systemic decision-making Causal Loop Diagram (CLD) model.
the perceived self-efficacy (central variable in Albert Bandura’s Social Cognitive Theory; Boone et al. 1977) to master the complex challenge of going through the ER process
the intensity of homeowners’ climate change awareness and thus their understanding of the necessity of climate mitigation measures such as ER
the perceived effectiveness of ER measures (based on external sources such as the media or friends)
the expected complexity of the ER process itself (from external sources such as the media or friends)
the general belief in the success of the national heat transition strategy (group identification according to Rees & Bamberg 2014)
the perceived social norm regarding ER within the associated group (e.g. family, friends, relatives) of the homeowners and
specific external triggers to conduct ER such as the general need of the real estate for refurbishment, change of home ownership or lucrative external offers.
As in Self-Determination Theory, the system model distinguishes between intrinsic and extrinsic motivation (Ryan & Deci 2000). The latter is represented by the perceived social norm that plays a key role in the CLD model and depends on two other important roles which are the proportion of positive media reports about ER and the proportion of positive word-of-mouth in society and the associated group. As Constantino et al. (2022: 51) stated:
humans are social animals, and their behaviour is influenced by their pro-social preferences, social networks, social identities, and social norms.
Thus, social norm aspects are of major relevance in describing motivation processes for ER in the present CLD model. However, because of the claim that the CLD should be manageable to read by remaining at a high aggregation level, a deeper analysis of injunctive and descriptive norms as well as group identity processes is not part of the present model version but is highly recommended for deeper analysis (see Section 4.3).
In the CLD model, the second step of the motivation process is the concrete feasibility check where the homeowner weighs their perceived benefits and efforts to decide on ER. If this evaluation is positive, the homeowner overcomes the critical Rubicon from motivation to volition and the detailed planning phase (including qualified planners) starts. After that, the ER can be realised if the capacities of qualified skilled workers, planners and energy consultants as well as the required ER materials are available. The last phase of the Rubicon model represents the evaluation of the homeowner if the ER result and process are satisfying. In the developed CLD, this has a large impact on whether the future word-of-mouth in society and thus social norms are more positive or negative.
3.2.2 Feedback loops
Psychological models such as the Rubicon model describing the phases of decision-making describe linear process steps. Expanding this model systemically by including the societal impacts on decision-making leads to a decision-making system model with a large number of interrelations leading to non-linear dynamics in the described STS. Thus, the Rubicon model is extended by introducing the word-of-mouth feedback from homeowners living in ER homes to society. In the present CLD model, several reinforcing and balancing feedback loops exist. The identified feedback loops are labelled in Figure 2, with R for reinforcing and B for balancing. For improved capture, each feedback loop is highlighted separately in Figure 3 (reinforcing) and Figure 4 (balancing). These individual feedback loops are discussed below:
R1
An important reinforcing feedback loop is obtained between the proportion of positive word-of-mouth and the proportion of positive media coverage about ER in society. This feedback loop is highly relevant because most identified feedback loops from society to individual decisions include word-of-mouth. The reinforcing nature claims that the larger the positive word-of-mouth the more positive media coverage is reached going back to an enhancement in word-of-mouth. However, its reinforcing nature also applies in reverse.
R2–R3
These reinforcing feedback loops state that the higher the number of homeowners interested in or determined to undertake ER and the lower the number of uninterested homeowners, the larger the positive word-of-mouth will be. This leads to an enhancement of perceived social norms regarding ER and thus to an increase in extrinsic motivation to become interested in ER. The opposite can be the case here as well if uninterested homeowners prevail.
R4
The larger the share of homeowners in ER homes—especially in their neighbourhood—the more positive the perceived social norm for ER. This is a vital reinforcing feedback loop because the social environment can create significant social pressure to become (extrinsically) motivated to undertake ER as well (Constantino et al. 2022).
R5
This reinforcing feedback loop describes the dynamics of the social identity process in associated groups because the attitude of the associated social group towards ER has a significant impact on the personal attitude, but this also has an effect on the attitude of the group.
R6–R7
These include more than two reinforcing feedback loops because similar feedback loop paths are summarised. They include elements of R1–R3 with positive word-of-mouth and media-reporting which depends on the share of interested homeowners for ER. Different to these feedback loops, R6 and R7 state a reduction in perceived complexity and uncertainty if media and societal debates on ER are more positive and less contradicting. This has a positive impact on reducing the own effort for ER and the perceived self-efficacy to achieve the aim of ER.
B1
This balancing feedback loop is of major relevance because it describes the limitations of ER dynamics in the STS. In short, it includes the fact that the more homeowners become motivated and decide on ER, the more construction activity in Germany will increase, leading to the lower availability of craftsmen, planners, energy consultants and probably ER materials. This induces a rise in ER costs, experienced complexity and reduction of the quality of the ER process/results, and thus more negative word-of-mouth by more unsatisfied homeowners in ER homes, and with it a lower extrinsic motivation of homeowners in non-ER homes to go through the ER process.
B2–B3
These state that the lower the availability of skilled workers, planners, energy consultants and probably ER materials, the longer the ER duration (including waiting time) and the lower the amount of qualified planning which limits the number of parallel ERs as it does in B1.
More reinforcing feedback loops than balancing ones exist in the CLD. While the latter keeps the STS stable and implies a limit in ER dynamics, all reinforcing feedback loops are found in Rubicon phase 1 (motivating homeowners). As mentioned, the reinforcing feedback loop can be interpreted in two directions for ER:
negative societal debates and a high number of uninterested homeowners in ER can lead to an even lower extrinsic motivation of homeowners to become interested in ER making the problem even worse
if the societal debates in society are more positive, then a different feedback loop leads to a higher intrinsic motivation for ER and thus to much higher ER dynamics.
From the CLD model, several leverage points to enhance the ER dynamics can be drawn for policy development.
4. Discussion
To identify potential leverage points addressing the complex transformative problem of too-low ER dynamics, a complexity-informed systems approach to policy advice is applied (Nel & Taeihagh 2024). A systemic decision-making model covering the psychological insights of the bounded rationality of human decision-making (Ebrahimigharehbaghi et al. 2022) and the multiple interdependencies and non-intended side-effects policies might evoke in the STS (Forrester 1971) is presented. Although such an explorative systemic model cannot predict future development, it can provide answers to the impact of policies by applying ‘what-if’ scenarios to tackle transformative policies with very unknown future paths (Schünemann et al. 2024). After suggesting policies, the applied interdisciplinary systemic approach and its limitations are discussed in the following.
4.1 Policy advice
Three main leverages to enhance ER dynamics can be derived from the developed systemic decision-making model, especially from the identified feedback loops (Figures 3 and 4). These are listed in Table 3 together with the summarised suggested policies:
Table 3
Suggested governmental policies to enhance the energy retrofit (ER) dynamics.
| OVERALL APPROACHES TO ENHANCE ER DYNAMICS BASED ON THE CLD | CORRESPONDING RUBICON PHASE IN THE CLD | KEY VARIABLES FROM THE CLD MODEL | SUGGESTIONS FOR GOVERNMENTAL POLICIES/INTERVENTIONS |
|---|---|---|---|
| Raising enthusiasm and trust for ER in society | Phase 1 (pre-decisional) |
| Informational policies with positive storytelling to build trust and identification with ER in society and reduce uncertainty |
| Shifting the perception of ‘own benefits’ relative to ‘own efforts’ for homeowners to be more advantageous | Phases 1–2 (pre- and post-decisional) |
| Lowering the effort side by reducing the perceived complexity of ER (e.g. one-stop-shop concept, reducing the time for ER to a few weeks), by designing socially just and user-friendly subsidies constant in their availability |
| Reducing the bottleneck of insufficient capacity of craftsmen, planners, energy consultants and ER material | Phases 3–4 (action and evaluation) |
| Support the education of ER specialists and the increase in producing ER material (delays in capacity-building crucial) |
[i] Note: These are derived from the systemic Causal Loop Diagram (CLD) decision-making model to understand homeowners’ motivations and concerns.
Raising enthusiasm and trust for ER in society
ER is currently not popular in German society (Grimm & Groß 2023). How people can be motivated for sustainability transformations, such as ER, is at the very core of psychology in understanding how existing unsustainable social norms and values can be changed. Concepts such as social norms and social intervention (Constantino et al. 2022), group identity and collective climate action (Rees & Bamberg 2014) are very promising approaches to analyse this in-depth. Generating trust in ER (in the process, actors and measures) plays an essential role (De Wilde 2019). In this context, the German debate about mandatory climate-friendly heating in the new Building Energy Act was unhelpful and enhanced the uncertainty and negative word-of-mouth within society and the media significantly (Grimm & Groß 2023). The developed CLD model shows that informational policies are necessary, and mainstream positive storytelling aims to demonstrate that ER is manageable and important to achieve a common desired future. Rees & Bamberg (2014: 467) stated that the:
collective action approach is based on the assumption that group identification is a powerful driver of social change.
If ER becomes a relevant issue in the associated group of homeowners in need of ER, this might induce a shift in social norms and thus a significant social pressure to decide on ER. The overwhelming complexity of the ER process, the opaque and frequently changing funding ‘jungle’, and the uncertainty about the effectiveness of ER seem to be major current challenges. Accordingly, the actual (polarising) societal debates in Germany about ER are counterproductive. Instead, policies are needed to create trust in ER and to help a variety of homeowners identify with the goals of the national heat transition (see the variable ‘belief in national heat transition’ in the CLD).
Shifting homeowners’ perceptions of benefits and efforts
ER increases the financial value of the real estate and thermal comfort and reduces heating costs (Lovell 2005). The authors see no need for additional policies if the carbon pricing and energy prices rise as expected in Germany. The CLD variable ‘own contribution for climate change’ as a benefit only plays a significant role if ER is an important issue in the associated social group. From the CLD decision-making model, the most important leverage is seen on the effort side. At present, the perceived complexity of the ER process is a major barrier (Eker & Zimmermann 2016). The idea of implementing one-stop-shops for ER, as promoted by the European Union (Boza-Kiss & Bertoldi 2018), or approaches to reduce the ER duration to a few weeks are innovative and promising approaches to reduce the perceived difficulty. A central campaign giving unbiased and easy-to-understand information about ER (the process, actors and measures) can significantly reduce the effort for knowledge acquisition and simultaneously enhance the trust in ER. Another key effort is the accessibility of the different available subsidies. Homeowners often complain about the complicated application process and the frequently changing conditions of financial support (Mayer et al. 2022). When homeowners are confronted with these obstacles before reaching the Rubicon, it can hinder ER (Heckhausen & Gollwitzer 1987). Therefore, economic policies should be less bureaucratic, constant in their availability and user-friendly.
Improving capacity
If a significant increase in homeowners opt for ER, there will be a lack of capacity in skilled workers, planners, energy consultants and ER materials/components. It takes time to educate and train people, as well as scale-up factory production (e.g. ER materials such as insulation). It is important for politicians to avoid a bottleneck that would lessen the motivation of homeowners. Additionally, such delays can lead to a missing the suitable window of opportunity (left side of the CLD, grey text), which means that a change of ownership or a general need for refurbishment can be a short-term situation to decide for ER (Abreu et al. 2019).
The policies suggested above originate from a very aggregated and systemic view based on the developed CLD which is not exhaustive and not sufficient for suggesting a final policy mix to enhance the ER dynamics. The focus is on analysing the individual decision-making process and its societal influences while technological aspects are mainly considered as external variables. However, combining complex system approaches with psychological theories and thus investigating individual decision-making, including societal influences, can deliver systemic insights to design more effective policies (Homer-Dixon et al. 2014).
4.2 Combining psychological and systems modelling
Social systems are far more complex and harder to understand than technological systems. Why then do we not use the same approach of making models of social systems and conducting laboratory experiments before adopting new laws and government programs?
The developed CLD model shows how combining psychology and complexity science can help one understand how societal transformation and transitions can be achieved. The presented approach shows how a general but individually based psychological model (the Rubicon model) can be extended to a systemic decision-making model by including the societal impacts on the decision-making process in an aggregated fashion. Such approaches should be adopted to inform policy, especially for policies that are intended to trigger transformation and transition processes.
The strength of a CLD is its visual representation and capability to explain, in contrast to an empirical black box model. A drawback of the CLD might be seen that it is a qualitative, and not a quantitative, simulation. However, such social models include a large number of social variables and interconnection which are hard to quantify and parametrise and thus lead to results that are hard to validate or interpret (Schünemann et al. 2024). Thus, the authors advocate the use of qualitative system models to consider the social impact of policies, which is still a niche in policy advice. In this context, it is important to consider that human decision-making is often not fully rational and under limited knowledge.
4.3 Critical reflection and outlook
Why does the innovative approach of systemic decision-making modelling not lead to more innovative results in the context of social complexity? This might be a reasonable question here. As discussed, the developed systemic decision-making model should be understood only as a showcase to demonstrate how psychological theories and system-modelling approaches can be integrated to better consider social complexity in transformative policy design. As the aim is to map the whole decision-making process of homeowners in an understandable presentation format, the CLD model is highly aggregated, only containing the most important identified variables and interconnections. The presented CLD is a summary of previous CLDs containing many more variables and connectors. Accordingly, the systemic decision-making model is only able to explain the complex reality of how homeowners decide to a very limited extent. Deeper insights can be expected if the existing CLD captures the whole decision-making process, but with significantly more detailed variables and interconnections, resulting in a very large system model. Alternatively, subsystems, i.e. parts of the decision-making process, can be modelled in more detail. As an example, the social norm and personal attitude of homeowners towards ER could be analysed in much more detail. Analysing subsystems has the advantage that the model remains in a size that is explainable and thus better usable for policy advice (Ghaffarzadegan et al. 2011). A drawback might be that some feedback loops are not considered if only subsystems are considered. This trade-off is well-known in system modelling and is a critical issue in defining the most suitable boundary conditions, which are needed to fully capture the problem to be analysed, but not to become overwhelmed by the inherent complexity of such systems. Although the parametrisation of the complete CLD to a simulation model (e.g. a system dynamics simulation) holds the promise of more specific explorations, the authors are sceptical because of the large number of soft variables (Schünemann et al. 2024).
For the specific case, the aims of the developed systemic decision-making model were:
to create an overview of the decision-making process and the main variables to understand where the critical processes and leverages are to motivate homeowners to perform ER and to derive some first general (high-level) policy recommendations
to compare the presented CLD for single-family homeowners with those for owners of apartments in condominium owners’ associations (occupied or rented) and real estate companies, aiming for a better understanding of the different motivational factors (Abreu et al. 2019) and
to use the identified variables and connectors of the developed CLD to create an online survey for apartment owners, considering the most relevant factors and a deeper understanding of the most relevant connections and variables in the CLD.
The overall survey results are now available from Schünemann (2025). These will be used, together with the systemic decision-making model, to better understand homeowners’ perceived ER barriers and motivations. More specific policy recommendations can then be developed.
5. Conclusions
Although the barriers and drivers of energy retrofit (ER)—especially for single-family homeowners—are well-known, the existing policies did not increase the ER dynamics required to meet the climate mitigation goals in most countries. This is partly because most available counselling tools to support the creation of public policies are focused on technological issues. There has been a lack of systemic analyses of the psychological decision-making process of homeowners, including the bounded rationality for ER. The present study is a first attempt to combine complexity science and psychological knowledge as a basis for a systemic decision-making model. The evolved Causal Loop Diagram (CLD) model focuses on the interdependencies of the psychological factors that influence the decision-making process of single-family homeowners for ER.
Based on the developed system structure and identified feedback loops, the authors distilled three general leverages that policy mixes should address, including tangible policy suggestions: (1) raising the enthusiasm for and trust in ER in society; (2) shifting the perception of ‘own benefits’ relative to ‘own efforts’ for homeowners to be more advantageous; and (3) reducing the bottleneck of insufficient capacity of craftsman, planners, energy consultants and ER material.
The presented combination of psychology and complexity science can be a vital extension of the existing policy assessment toolbox. Through the visualisation of the interconnectedness of the multiple variables, the model explains how policies might affect behavioural change, including unintended side-effects. This kind of complexity-informed policy advice is especially important for policies triggering transformational processes with high uncertainty. The general structure of the CLD model can also be transferred to other problems where social norms and uncertain transformative processes play an important role in decision-making. However, the presented aggregated system model should be seen as a first attempt to understand decision-making processes in the context of transformations.
Data accessibilty
The discussed qualitative system model (Causal Loop Diagram—CLD) is deposited in the following repository: https://zenodo.org/records/14645883.
Notes
[4] For the revised CLD model, see https://zenodo.org/records/14645883/.
[5] For a more detailed ranking, including a deeper analysis of barriers, see Rentrop (2018).
Acknowledgements
The authors acknowledge the valuable feedback from diverse experts and stakeholders on the Causal Loop Diagram (CLD) draft from very different fields (see Section 2.4). They also thank their former colleagues Paula Leutert, Ronja Paleit and Paolo Mamusa for contributions to the literature research for the project; and Sandra Volken, University of Bern, for support in creating the system model.
Competing interests
The authors have no competing interests to declare.
Author contributions
C.S.: conceptualised the article, developed the system model, led the project and literature review, and wrote most of the article. M.D. and S.S.: contributed to system model development, writing, reviewing and editing, especially from the psychological perspective.
Supplemental data
Supplemental data for this article can be accessed at: https://doi.org/10.5334/bc.534.s1