Expectation dynamics and niche acceleration in China’s wind and solar power development

This paper addresses the question of how alignment dynamics between niche and regime actors shape niche acceleration. We develop a conceptual framework that focuses on the role of expectations as a necessary precondition and even as a key proxy for strategic collaboration be- tween niche and regime actors. Based on actors’ expectations, we conceptualise three alignment patterns of strong, medium-strong and weak alignment. We propose a 16 % threshold of niche technology adoption for substantial niche acceleration. We explore our conceptual framework in two contrasting case studies of wind and solar power development in China between 2000 and 2017. Both cases experienced niche acceleration but followed different paths. Our research findings indicate that the three proposed alignment patterns between niche and regime actors’ expectations can be seen as a good proxy for explaining these different paths. Strong alignment between niche and regime actors’ expectations does go hand in hand with niche acceleration.

national new installed market (see Fig. 5). Both cases are suitable for developing the framework not only because wind and solar power have taken off rapidly but also because they followed different trajectories we can usefully compare. We can thus observe niche acceleration in different time periods within the same context (China) and explore whether we can relate these periods within and across both cases to our projected expectation alignment patterns between niche and regime actors. We acknowledge that China may be a specific case, because both types of actors may have a particular relationship due to the specific role of the state in the country. This issue will be discussed in the final part of the paper.
The paper is structured as follows. Section 2 introduces the concepts of expectations and alignment building in sustainability transition and sociological studies of expectations literature, followed by the introduction of the three conceptualised alignment patterns, and a definition of three niche acceleration phases. Section 3 introduces the operationalisation and methodology. Some of the key aspects of the framework need contextualisation. For example, to apply our conceptual framework, we have to specify who niche and regime actors are and define the phases of niche acceleration in the two cases. We then present how we organised the datagathering process. Section 4 presents a historical and comparative account of alignment patterns between niche and regime actors' expectations in relation to the niche acceleration phases. Section 5 provides a discussion of the results. Section 6 offers concluding remarks.

Alignment between niche and regime actors through expectations
The sociology of expectations and sustainability transitions literatures recognise expectations as playing an essential role in guiding the emergence of new technologies and niches. When niche innovations emerge, actors generally hold various and often contradictory visions of the future (Garud and Ahlstrom, 1997;Rip and Talma, 1998;Van Lente and Bakker, 2010). This is especially true for niche (new entrants) and regime (incumbent) actors. When niche and regime actors' expectations become aligned, they begin to drive socio-technical system change in new directions. Van Lente and  argue that these expectations serve as prospective socio-technical structures for actors. Konrad (2006) argues similarly that widely shared expectations become a social repertoire for a specific community and the public in general. Such a repertoire has force and helps to build a shared agenda for further actions. Furthermore, the collective expectations tend to attract other actors, who do not necessarily share the expectations, to expand the social network. In this sense, expectations could be seen as strategies deployed by the actors to enrol other actors. In his seminal work on the role of expectations, Van Lente (1993) introduces a promise-requirement cycle to explain the performative power of expectation sharing. In such a cycle, promises (expectations) are translated into requirements for socio-technical change.
How do we know whether expectations are shared or aligned? Based on strategic niche management (SNM) literature we propose two dimensions to measure alignment between niche and regime actors (Schot and Geels, 2008). First, the breadth of alignment, i.e. how many niche and regime actors are aligned. When expectations are more widely shared, it is more likely they will be translated into actors' shared goals and collective activities. Second, the depth of alignment, which relates to what is called in the SNM literature "the quality and specificity" of the shared expectations (Schot and Geels, 2008). We operationalise this dimension by mobilising a multi-level perspective (MLP) understanding of expectations, building upon the work of Truffer et al. (2008) and other scholars in sociological studies of expectations and sustainability transitions literature.
Van Lente (1993) distinguishes three different levels of expectations: micro, meso and macro. For him, micro-level expectations refer to the specification of the artefacts, systems or process to be developed. They function as heuristics and guide the search processes. Meso-level expectations are less specific. They tend to express functions that the technology presumably will fulfil. Macrolevel expectations are broad and general. They take the form of scenarios about the technology as a whole to fit societal trends and provide legitimacy for technology development. This distinction is similar to the three levels identified by Geels and Raven (2006): project-specific expectations, technology field perspective and societal developments. Ruef and Markard (2010) and Van Lente et al. (2013) indicate that expectations at the three levels follow different hype-disappointment patterns. This implies that the nature of actors' expectations (positive or negative) may be different at the three levels. Budde et al. (2012) illustrate this in their case study of Germany's mobility systems. Although actors had positive expectations about hydrogen and fuel cell niche technologies, the German government anticipated less positive landscape-level development. This led to a reduction in investment in these technologies. This case illustrates that it is crucial to measure the nature of actors' expectations across different levels to understand their strategies. Kriechbaum et al. (2018) elaborate on how these multi-level natures of expectations contribute to the divergent niche development of solar PV in Germany and Spain. Their analysis confirms that it is useful to unpack the interaction dynamics across three levels to understand niche development.
The above studies mainly articulate the expectations of emerging technologies, niches and socio-technical structures, while neglecting expectations about regime resilience. Drawing on MLP, Truffer et al. (2008) suggest actors' expectation structures for system transformation could be mapped in accordance with MLP levels: niche, regime and landscape. They argue actors' strategies and activities are influenced by their expectations not only about the emergence of niches and landscape-level development but also about regime resilience. Moreover, in their analysis, they distinguished individual actors' expectations and collective expectations at each level. The prospective socio-technical structure is shaped by actors' collective expectations at the three levels. Budde and Konrad (2019) suggest that these three levels of expectations may support and reinforce, or contradict and weaken each other, with direct impact on the transition dynamics. In their analysis, Budde and Konrad (2019) expand the focus of actors beyond the conventional research, industry and social actors to include policy actors' expectation dynamics. They prove that policy also responds to the changing expectation dynamics at three levels. Building on these findings from the literature we can construct a theoretical framework for alignment dynamics between niche and regime actors' expectations that includes a notion of breadth as well as depth of alignment.

Typology of alignment dynamics between niche and regime actors' expectations
Alignment is not a 0/1 dichotomy; in reality there is a wide spectrum between no alignment and complete alignment, and actors may change position over time. We argue that the alignment between niche and regime actors is a dynamic process with shifts along this spectrum. However, how can we measure this dynamic? In this section we develop a theoretical framework that contains a typology of three basic alignment patterns. To get there, we have to take a number of steps that we will explain in detail in order to make the framework credible. The steps themselves are complex, but they lead to simple end-results: three basic alignment patterns. All building blocks are based on existing literature. Our contribution is the specific way we put them together.
We first recognise that both niche and regime actors are heterogeneous. As indicated by Geels and Schot (2007), regimes are often semi-coherent, not all regime dimensions are fully aligned and they carry internal tensions and contradictions. These tensions could be utilised by niche actors to build connections and to provide windows of opportunity for niche empowerment (Smith and Raven, 2012;Bui et al., 2016). When confronted with pressure or crisis, regime actors perceive different opportunities and hold a variety of expectations about niche, regime and landscape developments (Smith, 2007). Similarly, niche actors may have different expectations about options for niche development and the obduracy of the prevailing regime and landscape developments. The heterogeneity of both niche and regime actors and their expectations generate multiple options for alignment between niche and regime actors.
In order to map all possible options, we have built a typology of alignment patterns between niche and regime actors' expectations in three steps: Step 1: identification of expectations at three different levels. We argue that if the expectations of both niche and regime actors converge for all three levels, there is an in-depth alignment.
Step 2: measuring breadth of alignment between niche and regime actors at each separate level. In Step 3 we systematically combine steps 1 (depth of alignment) and 2 (breadth of alignment) in 27 theoretically possible different types of alignment between niche and regime actors' expectations. We then show that these 27 types can be reduced to 12 basic types.

Step 1. Distinguish actors' expectations for future developments at three levels
Following Truffer et al. (2008) and Budde and Konrad (2019) we distinguish between three levels of expectations by both niche and regime actors: 1 Landscape-level expectations: these refer to actors' perceived future of the external environment, such as perceptions of climate change or environmental issues, which influences the long-term development of the sector or system. These provide external momentum to guide the direction of transition. Landscape-level expectations tend to be more general compared to the other two levels.
Regime-level expectations: these are expectations of the regime's incapability of adapting to internal tensions and crises or to external pressures. If regime-level expectations become positive, i.e. actors start to question regime resilience, it will lead to regime destabilisation and thus may contribute to niche development. Expectations of regime incapability cover all dimensions of the dominant socio-technical system. For example, for our case study it includes technology performance of thermal power, its policy support and market environment.
Niche-level expectations: these are expectations of future performance of the specific socio-technical configurations of emerging technologies, such as the role of wind power in meeting energy demand, technology performance or expected market competitive advantages. When positive, they will contribute to niche acceleration. Expectation at this level is often more specific, and visible, compared to the other two levels' expectations, as niche actors generally mobilise their expectations and express them as strategies to attract other actors.

Step 2. Define breadth of alignment between actors' expectations at each level
For a transition to happen, niche and regime actors need to align their expectations at each separate level. In other words, transitions require coordination of niche and regime actors' expectations at the landscape, regime and niche levels. We thus measure alignment for each of the three levels separately.
In our proposed theoretical framework, breadth of alignment is defined by three types of alignment between niche and regime actors: (1) sparse alignment, towards the end of the spectrum where no regime actors align with niche actors; (2) broad alignment, towards the spectrum end where all of the regime actors align with niche actors; (3) selective alignment, where some regime actors align with niche actors, an intermediate state between sparse and broad alignment. We thus propose to measure the three degrees of breadth by counting how many regime actors align with niche actors in terms of their expectations at a specific level.
To measure and define selective alignment we need to know what number constitutes 'some' actors. But as this depends on the context and structure of the socio-technical system under study, the framework does not provide an absolute rule on how many regime and niche actors need to align; this needs to be defined for each case study, as we do below.
We are now able to define breadth of alignment between niche and regime actors' expectations at each of the three levels. 1) Breadth of alignment between actors' expectations at landscape level When regime actors begin to share the perception that changes at the landscape level challenge future regime resilience, more pressure is generated to open the way for a regime shift (Smith, 2007;Turnheim and Geels, 2013). This highlights the importance of scrutinising regime actors' expectations towards the landscape level for understanding the transition process. Meanwhile, niche actors could leverage narratives of needed change (expectations about future developments of the landscape level) to create cultural legitimacy for niche technologies and ensure they are accepted by the broader public (Geels and Verhees, 2011). When such narratives created by niche actors are being articulated and acknowledged by the regime actors, it could potentially bring niche technologies into regime actors' searching sphere (Turnheim and Geels, 2013). For example, renewable energy could be labelled as a promising solution to social or environmental issues (e.g. climate change or air pollution). This may create cultural and political legitimacy for the sector. When this happens regime actors may consider investing seriously in renewable energy as a necessary step for a future clean, low-carbon power supply. They feel under pressure to respond to what they now perceive as a serious threat to their business created by climate change at the landscape level. However, they will not invest in regime change if they still believe in the resilience of their seasoned strategies to respond to future threats and opportunities.
2) Breadth of alignment between actors' expectations at regime level When there is sparse alignment between niche and regime actors' expectations about future developments at regime level, it refers to a situation in which neither niche nor regime actors question the regime's resilience to respond to internal crises and/or external pressures. In such a case, niche actors may aim for limited niche development because acceleration is not seen as a viable strategy. Niche development is mainly regarded as an add-on to the mainstream markets: a small market niche at best. When the opposite situation begins to emerge and niche and regime actors broadly share expectations that the dominant regime not only at risk but may fall apart because it can no longer respond to future threats and opportunities, it indicates that regime actors have started to question the regime's resilience. This also means that they are searching for alternatives, which could open spaces for niche acceleration. This search process will have to become focused on specific paths of niche acceleration.
3) Breadth of alignment between actors' expectations at niche level The measurement of expectations at niche level plays a crucial role and has been discussed mostly in comparison with the other two levels in transition studies. Only when niche and regime actors share expectations about the viability of specific niche technology will regime actors mobilise resources to support the development of the niche . SNM studies have identified the robust alignment of expectations as an essential way to enrol other actors for niche acceleration (Geels and Raven, 2006;Schot and Geels, 2008).

Step 3. Building alignment patterns
This third step introduces the systematic combination of steps 1 and 2. Now we have finished the assessment of actors' expectations at three separate levels (step 1) and the assessment of three degrees of breadth at each level (from broad, to selective, to sparse). Theoretically, we are able to distinguish 27 (3*3*3) different types of alignment combinations between niche and regime actors' expectations during the transition process. These are presented in Fig. 2.
In reality, however, not all of these options will be relevant for our research question that aims to understand the connection of alignment between niche and regime actors to niche acceleration. Drawing on sustainability transitions literature, we can reduce the 27 to 12 possible types by considering the following. First, we can exclude the sparse alignment between niche and regime actors' expectations at all three levels (type 1, in Fig. 2) as this type does not contribute to niche acceleration. Second, landscape-level expectations are more general than expectations at the other two levels, and therefore actors are more likely to share such general expectations (Konrad, 2006). In other words, we may consider such sharing as a precondition for alignment of expectations at the two other levels. Based on this observation we can exclude alignment patterns for types 2, 3, 4, 5, 6, 7, 8, 9, 16, 17 and 18 where actors share broader alignment at niche and regime level than at landscape level. These are dismissed as unrealistic scenarios. Third, sustainability transition literature indicates that regime actors are generally locked into their existing routines. Regime actors may invest in some niche development, for window dressing or exploration of future opportunities, but certainly not in niche acceleration. For this to happen, regime actors first have to begin questioning the regime's resilience. Therefore, we exclude the types of alignment 12 and 21, where regime actors agree on the strategic importance of specific niches, not just for the sake of new opportunities, but also as a serious future to invest in, but they do not agree on the ability of the regime to respond to sustainability challenges. They are dismissed as unlikely scenarios. For a similar reason alignment type 15 is excluded: it is unlikely to have broad alignment at niche level (i.e. all niche and regime actors share expectations of niche development) while holding selective alignment at the landscape and regime level.
For the remaining 12 types of alignment (types I-XII in Fig. 3), based on the proposed two dimensions (breadth and depth of alignment), we can distinguish three different basic alignment patterns: Weak alignment refers to a situation in which niche and regime actors have selective alignment at the landscape level and various alignments but never a broad alignment or a simultaneous selective alignment at the two other levels (types I-III); or a situation when there is broad alignment at the landscape level but this has not resulted (yet) in selective alignment at either regime or niche level (type V). For all these types, niches are invisible or less attractive to regime actors. For the alignment types I and V, niche and regime actors share limited expectations of both niche and regime's development. Regime actors are deeply embedded in their K. Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 routines and believe optimisation is a viable way to proceed. They generally do not share expectations with niche actors, or they do not recognise niches as a threat to the regime's future. Moreover, at the early stage of niche development, niche actors may focus on the niche and have no clear visions of how the process of regime destabilisation may happen. Niche actors have limited social networks, which are less stable, and the niche technology improves within a protected space where it is isolated from the dominant selection environment. Niches may expand if there is leeway outside the mainstream market, but growth is limited. Alignment type II may evolve from a situation in which some regime actors built a network with niche actors; however, they see the niche as a small market instead of a threat to the regime. This pattern leads to very limited niche development, especially when there is insufficient pressure from the landscape level. Alignment type III emerges when regime actors start to question the regime's resilience and expect that it will be unable to adapt to external pressures. However, this expectation does not necessarily lead regime actors to move towards investing in a potential new regime if they are not convinced of the performance of niche technologies or opportunities for niches to expand. In this situation they feel they need to stick to a regime optimisation pathway or shift to other more convincing niches. As we will discuss below, in our case study, when the coal power regime actors faced questions about their capability to fulfil the fast-growing electricity demand, they anticipated that the potential of wind and solar power development was limited compared to competing alternatives such as hydropower and nuclear power. Therefore, the limited alignment of expectations between niche and regime actors at the niche level indicates that their limited resources could not be mobilised towards the expansion of niches, thus hampering niche acceleration.
Medium-strong alignment refers to a wide range of situations, including ones where niche and regime actors have selective alignment at all three levels, but not broad alignment (type IV); or broad alignment at landscape level and selective alignment at niche level, but regime actors still maintain the resilience of the dominant regime, resulting in sparse alignment at this level (type VI); or broad alignment of expectations about future developments of the landscape level, which has resulted in selective or even broad alignment about regime incapability but not yet any alignment about specific niche acceleration (types VII, X). In all of these situations there are aligned expectations between niche and regime actors, but it is limited to specific levels or actors. Expectations are not aligned across all three levels.
In alignment types IV and VI, some regime actors begin to express expectations about a bright future for a niche technology. Niche actors also begin to envision the future regime they aim to build, providing an alternative to the dominant socio-technical system. This imagining, for example, the renewable energy (RE) penetration of China's future energy system in our case, could act as a platform for aligning niche and regime actors' expectations at all levels, building the conditions for niche acceleration. However, limited questioning of the regime's resilience and the consequences of landscape pressures may restrict large-scale investment in niche development Geels, 2012, 2013). But even when regime actors begin to question regime resilience and are starting a "more distant search and exploration of technical alternatives" Geels, 2013, p.1754), they may invest in multiple niches leaving limited resources for specific ones (as for alignment types VII and X).
Strong alignment refers to alignment types VIII, IX, XI and XII, which have broad alignment at landscape level and at least selective alignment of expectations at both niche and regime level. In this situation niche acceleration is highly probable. As argued K. Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 by Smith (2007), an "influential niche enlists a broad network of actors in support of its socio-technical practice and the future regime it prefigures. Supportive actors must include producers, users, third parties (e.g. regulators, standards institutes, investors) and policy-makers" (p.430). When regime and niche actors align their understandings of landscape developments, it provides an opportunity for niche actors to mobilise landscape pressure as a resource for articulating concrete regime pressures (for example, the perception of climate change exerts strong pressure on the fossil-fuel-dominant regime towards RE). Moreover, strong alignment between niche and regime actors' expectations at regime level indicates regime destabilisation, which contributes to the further breakthrough of niches (a hypothesis developed by Schot and Geels (2007) and supported by Turnheim and Geels (2012)).

Relating alignment patterns to niche acceleration
Our theoretical framework aims to connect alignment patterns to niche acceleration in the following way: we would expect niche acceleration to happen following strong alignment, but this process may gain some momentum during the medium-strong alignment phase. This still raises the question of how we establish whether niche acceleration has happened. Niche acceleration is not just about adopting new products. They are part of a transition process that leads to the emergence of a new socio-technical system. A core aspect of such a new system is the development of new rules; in other words, it is a regime formation or institutionalisation process (Fünfschilling and Truffer, 2014). Such a process implies that a new system gains momentum or moves from a situation of fluidity to a more stable one. Schot and Geels (2007) have argued that such a stabilisation of rules is a necessary precondition for niche acceleration, and this hypothesis has been confirmed in historical analysis of the development of the automobile regime (Kanger and Schot, 2016).
But how do we know whether institutionalisation is happening? Measuring this can be complex (see discussion of different stages of institutionalisation by Tolbert and Zucker (1999) and Fünfschilling and Truffer (2014)). For our case study, we use a simpler measurement building on innovation diffusion studies, in particular the work of Rogers (2010). These studies are focused on diffusion of products, which is different from system diffusion (Rotmans et al., 2001). Yet by focusing on the diffusion of a focal technology of a new system, innovation diffusion studies may still contain relevant insights (Geels and Johnson, 2018;Van der Kam et al., 2018) and diffusion curves are often used in sustainability transition studies (Rotmans et al., 2001;Elzen et al., 2012;De Haan et al., 2016;Kanger and Schot, 2016). Rogers (2010) distinguished five groups of buyers with different personal profiles adopting new technology at different sequences of time. Moore (1991) argued that there is a chasm in the diffusion process around a 16 % threshold, since it is very difficult to move from the early adopters into the early majority group (see Fig. 4). Early adopters are visionaries; they want what others do not have and are happy to promote a discontinuity between old and new ways and are prepared to champion these against entrenched resistance. People and organisations in the early majority group want to rely on a well-established reference and support infrastructure and follow a social norm. When the early majority start to adopt a new product, it indicates this new product or technology is becoming part of the mainstream. This is a very good description of what happens in a niche, and in the process of moving from a niche to a regime (Schot and Geels, 2008). We argue that the 16 % threshold is based on the idea that adopters at that point move from being driven by specific conditions (as in a niche) to accepting the use of technologies as a consequence of a new social norm and a system being put in place to support this norm. So, adopters become more rule-driven because the niche innovation begins to stabilise.
Based on the above considerations we are able to specify the notion of niche development. When the market share of wind or solar energy is below 2.5 % (group of innovators) we assume a slow niche development. When the market share is between 2.5 % and 16 %, we assume a moderate niche development (group of early adopters), and when the market share is above 16 %, we assume a substantial niche acceleration (moving into the group of early majority).  (2010) K. Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 3. Methodology

Specifying the framework for our case studies: niche acceleration
Diffusion studies express the market share of new technologies in terms of number of adopters; however, we think relative market share is a better indicator because it automatically takes into account market shares of competitors (other niches) and the decline of the dominant regime. We have used the market share of annual newly installed capacity for wind and solar and included the figures of other niches and installed capacity of coal power plants (see Fig. 5). We could have also taken the increasing rate of electricity generation or cumulative installed capacity, but data are lacking.
When we apply these thresholds to our two cases, the following picture emerges. For the wind power case we can distinguish three stages of niche acceleration:

Specifying niche and regime actors
Our framework focuses on alignment between a heterogeneous set of niche and regime actors in a socio-technical system but does not specify how many actors need to be aligned. This needs to be done for each case study separately. Therefore, we first have to identify the main stakeholders for each case by looking at the entire value chain, including generation, transmission, distribution and retail (Stenzel and Frenzel, 2008). For our case studies, we have identified the actors after the reform of China's electricity sector in 2002. In this reform, China's planning-based, centralised electricity sector was transformed into a substantially more market-based system with more diversified actors (Ma and He, 2008;Williams and Kahrl, 2008). The State Power Corporation, which was in charge of generation, transmission and distribution, was split into 11 new corporations: two grid operators (State Grid Corporation of China and China Southern Power Grid) in charge of transmission and distribution across China (apart from the western part of Inner Mongolia); 'Big Five' power generators; and four other auxiliary corporations (Ma and He, 2008). China's current electricity sector still has the same structure (Zhao et al., 2016).
The key stakeholders in our two cases include: central government; research institutes; manufacturers; the grid company; thermal power companies; the financing agency; wind and solar power generators; industry associations; users; NGOs and green organisations (Zhao et al., 2016;Mori, 2018). We acknowledge that the key actors may change over time along with the development of wind and solar power. For example, the wind and solar power industry association and large industrial users started to play a role at a later stage of development.
To define the medium-strong alignment pattern, the threshold that we used in our two cases largely depends on the shifting of key actors' expectations. In our cases, the key regime actors are the central government, coal power generators and the grid company; the key niche actors are wind and solar power generators and the manufacturing industry (as depicted in Fig. 6). For example, when any two of the three key regime actors share expectations with the niche actors, we categorise this as selective alignment at the niche level. Sparse alignment at the niche level refers to fewer than two of the key regime actors aligning with the key niche actors. Selective alignment at regime level refers to one of the two key niche actors sharing expectations with regime actors. Sparse alignment at the regime level refers to none of the key niche actors sharing expectations with regime actors. Broad alignment at niche/regime level refers to all of the key regime actors aligning with the key niche actors' expectations. K. Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 3.3. Data collection and analysis 3.3.1. Data collection Data collection included: (i) 31 semi-structured and 6 informal interviews with relevant actors; (ii) a workshop 2 with 22 participants, both niche and regime actors; and (iii) desk-based research, in particular retrieval of news from relevant websites, professional journals and organisation reports.
The interviews were conducted by the first author between October 2017 and March 2018. Using interviews to collect data on actors' expectations has several challenges. First, the interviewees may have implicit expectations that they do not easily express. Second, they may hold retrospective bias when asked about their perceptions of historical events.
To overcome these challenges, multiple experts from similar groups were interviewed to reveal expectations and limit individual bias (Eisenhardt and Graebner, 2007). For example, the study included four interviewees from central government so they could validate each other (see Appendix Table A1). Moreover, the interviews were designed to include cross-checking questions. For example, wind and solar power investors were asked about the grid company's expectations of wind and solar power at certain development stages, and vice versa. This cross-checking was also important to identify alignment patterns among actors. If actors expressed different expectations or expectation alignment was unclear, additional data was sought through archival data sources.
To develop the interview questions, we used the items presented in Table 1. We asked interviewees questions related not just to their own expectations, but also to expectations of other actors for triangulation purposes. In order to allow the interviewees to speak relatively openly, they were guaranteed confidentiality. All of the semi-structured interviews were audiotaped, and each interview lasted around one hour. Most interviews were conducted in Mandarin, then transcribed and translated from Mandarin to English. The six informal interviews were conducted at a later stage of the fieldwork (January-February 2018). They were conducted in an unstructured way and used to discuss sensitive issues, such as expectations from coal and grid companies, and to query inconsistencies. These were not recorded.
The workshop took place in March 2018 with all authors present. The aim was to discuss the historical development (through selected key events) of wind and solar power development between 2000 and 2017 and agree on niche development phases and the relationship between main actors during the development process, in a setting designed to build consensus. The first author presented reports on the wind and solar case studies. During the workshop, we collected data through presentations, plenary discussions and facilitated group discussions. Specifically, we conducted two focus group discussions on wind and solar power.
The archival data included: articles from China's largest professional electric power news website, 3 BJX: http://www.bjx.com.cn/; institutional reports, such as the annual report of China's electric power development produced by the China Electricity Council from 2001 to 2017, and reports produced by the State Grid from 2015 to 2017; professional journal articles, including China's professional journals on RE, < Solar Energy > , < Wind Power > , < State Grid > ; and key government policy documents, such as the < These keywords were also applied to identify relevant journal articles. 4 Note: generally, CWEA data are higher than NEA data because NEA data cover grid-connected installed capacity, while CWEA data refer to wind turbines that have been installed but may not be connected to the grid.

Data analysis
Our data analysis aimed to produce an assessment of alignment patterns between niche and regime actors' expectations at different niche development stages. Alignment patterns had to be identified at three different levels (niche, regime and landscape). For the niche and regime levels we looked at five dimensions: Science and Technology (S&T), Political, Industry, Market and Culture. Our analysis started with coding using keywords from Table 1. These keywords cover the five dimensions for both niche and regime levels based on a selective literature review, which includes (Konrad, 2006;Truffer et al., 2008;Turnheim and Geels, 2013;Kriechbaum et al., 2018;Budde and Konrad, 2019).
Coding led to two sets of results: self-evaluated results and other stakeholders' evaluated results. This made it possible to crosscheck these two sets of results and see whether they were aligned. If they aligned with each other we put the results (positive/ negative/null) into the data result table (see Tables 2 and 3). If they diverged, we looked at secondary data to reach a conclusion. When the end-result was not consistent we concluded that no alignment had taken place.
Secondary data were also used to fill in the gaps. For example, during the interviews and the workshop, the thermal power companies did not specify their expectations of the landscape before 2006, and secondary data was used to trace their views. For secondary data we constructed a database of relevant articles, which we then coded.
The results are the outcome of a disciplined coding process coupled with triangulation of various sources and interpretation. An important interpretation problem arose because we looked at expectation dynamics across different dimensions of the energy system, from future S&T development to political developments and so on (see Table 1). This meant that a range of actors had to agree on expectations concerning each specific dimension. In cases of difference we gave more weight to actors who were evaluating expectations in their own area of work.
The results are presented in the next section, with a table each for wind and solar (Tables 2 and 3) and a narrative. In the tables we present three types of results for the nature of expectations of each niche and regime actor: "√"as "positive"; "×" as "negative"; "−" as "no information"; we also show how actors' expectation dynamics at three levels match the niche acceleration stages we identified beforehand and converge with a specific alignment pattern.

Alignment dynamics between niche and regime actors' expectations of wind power
As indicated in Table 2, the alignment patterns between niche and regime actors' expectations of wind power have not been static between 2000 and 2017. Both the content and nature of actors' expectations of the three different levels are changing over time. There are several actors for whom there is insufficient information on their expectations. However, they do not influence the threshold of alignment patterns between niche and regime actors' expectations, as their expectations will not strongly influence other actors' expectations, or they did not explicitly concern themselves with the future of that dimension of the socio-technical system. For example, wind turbine component suppliers are generally less concerned about the future of the coal regime; instead their  Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 expectations are more closely connected to the potential future of the market of niche development.

Stage 1: 2000-2007 weak alignment
Between 2000 and 2007, there was weak alignment between niche and regime actors' expectations (pattern I in Fig. 3 with selective alignment at the landscape level and sparse alignment at regime and niche levels). Although at this stage several incumbent actors, such as the central government and the electric power association, started to realise the unsustainability of coal power and energy security issues, there was limited articulation of the landscape pressures by niche actors.
Generally, at this stage, renewable energy took place in the niche market, rural areas with less access to electricity or remote areas with weak grid infrastructure. Niche technology experts articulated the market potential of wind power technology, which could be domestically commercialised and industrialised with the government's policy support (Shi, 2001). However, even the central government and the electric power associations started to pay attention to wind power development, although less priority was given compared to hydropower and nuclear power. Wind power did not attract significant attention from other industry regime actors, in particular from power generators and the grid company.
There was widespread sharing of expectations among different actors of the short-supply issues of China's electricity system, which would be accelerated by rapidly increasing electricity demand to fuel economic growth. Pessimistic views about the levels of domestic coal reserves in China, which could fulfil demand for 20 years at most, were widespread in the public news (BJXnews, 2005). This led to large investment in hydropower construction rather than wind power (BloombergNEF, 2018). Narratives criticising the unsustainability and environmental impact of coal power emerged (China Electricity Council, 2002). However, values around environmental protection and sustainability were not explicitly or strategically shared among niche and regime actors (Urban et al., 2012).

Stage 2: 2008-2010 medium-strong alignment
There is medium-strong alignment between niche and regime actors' expectations at this stage (pattern VI in Fig. 3, with broad alignment at the landscape level, sparse alignment at regime level, and selective alignment at niche level). Alignment at the niche level was broader than during the former stage.
This stage witnessed the nascent shift of China to low carbon development. Green and low carbon emerged as values for economic growth. There was an increasingly high expectation of renewable energy's bright future among different actors in China after the Renewable Energy Law was introduced in 2005. The central government showed increased enthusiasm and commitment to wind power, which was endorsed as the most advantageous renewable energy (Li et al., 2008;Han et al., 2009). In 2009, the central government positioned the renewable energy industry as the strategic emerging industry, one of the engines for China's future green economy growth. It soon became a 'hot spot' for social investment, with a rapidly growing number of wind power manufacturers. The central government introduced a renewable energy mandatory policy in 2007 and large power generators started to invest in wind farms as a long-term development strategy (Wang, 2010). Power generators' commitment to wind power deepened at a later stage following long-term tensions in the coal industry about high coal prices (Wang, 2007;Liu, 2013). This tension weakened their faith in the competitive advantages of the coal power regime. Furthermore, from 2008 onwards, with decreasing wind power plant costs and a belief in long-term positive government support for wind power, power generators started strategically setting up subsidiaries for  Legend: "√"as "positive" expectation; "×" as "negative" expectation; "−" as "no information; "→" indicates changes occur. Shadowed areas depict actors who have not played a significant role at that specific stage. a As an example, a positive checkmark for coal power companies' expectations at the landscape level indicates that they perceive pressure from future development at landscape level which may undermine regime stabilisation.
K. Yang, et al. Environmental Innovation and Societal Transitions 36 (2020) 177-196 wind power businesses (Chen, 2012). However, wind power was treated as an add-on to the market with both niche and regime actors less explicitly showing belief that thermal power would be substituted by RE (Iizuka, 2015). Moreover, wind power was regarded as "rubbish electricity" by the grid company, which stated that because of the intermittency of wind power, its large-scale integration in the grid would undermine the safety of the electricity system (Yuan et al., 2012). This lack of support from the grid company led to China's wind power suffering from high curtailment rates at a later stage (Zhao et al., 2012).

Stage 3: 2011-2017 strong alignment
At this stage there was broader alignment between niche and regime actors' expectations at three levels (pattern XII in Fig. 3). Actors' perceptions of pressure from the landscape level became clearer than previously. Expectations were that the future of the energy system should be "clean, low carbon, safe and efficient" (NDRC and NEA, 2016a). There was a deep congruent understanding of the urgency to restructure and transform China's current coal-dominated energy supply system to mitigate climate change and domestic air pollution issues (NDRC and NEA, 2016b). This policy document sets 15 % and 20 % as minimum targets for non-fossil fuel in the energy mix by 2020 and 2030 (see Appendix Table A2). Increasing the proportion of renewable energy in the energy mix was reframed as necessary to achieve the central government's carbon emissions reduction targets. Wind power technology was regarded as one of the main technologies that could help China achieve a low carbon strategy (Shi, 2014).
Government and industry actors regarded the wind power industry as mature enough for the technology to be scaled up and put into commercial application across China without subsidies by 2020 (He, 2016;NDRC, 2016). The big coal power companies started to invest strategically in renewable energy, especially wind power. Since 2016, there has been a fast shift of regime actors' expectations about the coal power regime's resilience to external pressures: "the more foresighted companies… such as SDIC Power (the State Development and Investment Corporation), are already disposing of coal-fired power assets. China's five major power companies are much less inclined to invest in new capacity and are speeding up divestment from some old or poor quality assets" (Zhang, 2017). Furthermore, with the large increase in installed wind capacity, the grid company improved its infrastructure capabilities and dispatch practices to integrate more renewable energy. The State Grid Corporation started issuing a white paper < Promote the Renewable Energy Development > every year since 2015. Clean and low carbon became core values of its business strategies.

Alignment dynamics between niche and regime actors' expectations of solar power
The alignment between niche and regime actors' expectations of solar power at the three levels has been changing between 2000 and 2017 (see Table 3). The strong alignment between niche and regime actors' expectations about future developments of the niche level formalised almost at the same time at the regime level, distinguishing it from wind power. For the latter, the build-up of broad alignment at the niche level took place before it achieved the same breadth of alignment at regime level. We present the storyline of actors' expectations for solar power development, drawing attention mostly to the evolution of expectations at the niche level as the regime actors' expectations about the other two levels (regime and landscape) have been largely presented in the wind power case. However, we will present how the niche actors perceive the future of landscape-level and regime-level development.

Stage 1: 2000-2012 weak alignment
There was a weak alignment between niche and regime actors' expectations of solar PV (pattern I in Fig. 3), with selective alignment at the landscape level, sparse alignment at the regime level and niche level.
In the early 2000s, private entrepreneurs (such as the CEO of Trina Solar) articulated that renewable energy would be a potential substitute for fossil fuel in the long term (Huang et al., 2016). However, there were relatively low expectations of the market potentials of solar PV, as it was widely regarded as too expensive to be deployed in the country at scale and not competitive in the market compared with conventional power in the short term. "When the founder of Suntech told us that he would like to build up 10 MW solar PV production line in 2001, we feel like it is impossible, there won't have market for that massive production" (former national policymaker, Beijing, 12 Dec. 2017). The domestic deployment of solar PV was predominantly targeted at remote areas without electricity access, for example, in western China. As a stand-alone energy system, solar PV was believed by the central government to be suitable for areas with limited access to electricity and weak grid infrastructure capability (NDRC, 2007), but too expensive to be widely used in the Chinese electricity market. Meanwhile, solar power was believed to be less competitive than other clean technologies such as hydropower, nuclear power, wind and biomass (Li et al., 2007). According to the < Medium-long term development plan for the RE (2007) > , the total capacity of solar power (PV and thermal together) was set at 300 MW by 2010, reaching 1800 MW by 2020, while the targets set for wind power were 5000 MW and 30,000 MW respectively. There was less explicit articulation of the strategic role that solar PV could play in China achieving a low carbon future.
With the fast take-up of China's solar power industry because of the global market, especially the expanding European market (Marigo et al., 2008), industry actors started to believe that the potential domestic market would increase in the near future with the continuous reduction of solar panel costs. Especially after the then biggest manufacturing company, Suntech Power Holdings, went public on the New York Stock Exchange in 2005, showing that solar PV could bring great wealth, the private enterprises started to flood into the solar PV manufacturing industry because of its bright future. This high expectation further burgeoned when it was labelled as a strategic emerging industry by the central government in 2009. However, these expectations did not translate into domestic deployment (Fischer, 2012).

Stage 2: 2013-2015 medium-strong alignment
The alignment between niche and regime actors' expectations at the three levels was medium-strong (see Fig. 3, pattern VII), with broad alignment at landscape level, selective alignment at regime level and sparse alignment at niche level. With increasing concern about climate change and domestic air pollution, the thermal power regime started to be questioned by both niche and regime actors. However, collective expectations between niche and regime actors were less strong compared to the later stage. As solar panel costs fell, solar PV was perceived as a potential option for future clean energy in China. Especially after the Fukushima nuclear accident in 2011, solar PV was seen as a safer alternative. Moreover, it was perceived that there would be limited potential to increase the market for hydropower in China. Industry actors believed that solar PV was a sunrise industry with great potential to fuel the future green economy. This expectation was mobilised to lobby the central government to support the domestic market (Huang et al., 2016). Furthermore, solar PV was perceived to have a potential large market with diversified applications, not just for centralised power plants but also for distributed solar PV panels. The flexibility of solar PV systems and the multiple emerging business models further reinforced expectations for fast increasing domestic market.

Stage 3: 2016-2017 strong alignment
The alignment between niche and regime actors' expectations at the three levels was strong (see Fig. 3, pattern XII), with broad alignment at all three levels.
From 2016, the coal power regime began to destabilise. Coal power was criticised as unsustainable, with negative impact on air quality and water consumption (Greenpeace, 2017). Along with emerging oversupply issues in the electricity market, there were increasingly high expectations that coal power in China would peak in 2020 (Zhang, 2016). The central government showed determination to cap coal power plants. During a roundtable discussion of the transition of China's electricity system for the 13th Fiveyear Plan in January 2016, the experts agreed that the golden age of coal power had passed (NRDC, 2017). The successful decoupling of China's economic growth from coal power was considered to have ushered the country into a post-coal era (Duan, 2016;Qi et al., 2016). In December 2017, NEA convened the 2018 national energy conference, during which, for the first time, it officially declared the overcapacity problems of coal power plants in China and that the development of thermal power was entering a "defusing the risk of overcapacity" stage (Cableac.com, 2018). NEA made a clear statement that with the transformation of the energy system, the future for coal power was to provide a dispatch auxiliary service for renewable energy and to make space for renewable energy generation. Previously, the function of thermal power was believed to be the dominant power "to guarantee the supply of electricity". Solar PV has been regarded as an important strategy for big utilities and conventional coal power investors to transform their business towards a clean and low carbon future. With the further decrease of solar panel costs, it has been perceived that by 2020 solar PV panels will be competitive in the conventional power market. The Solar PV Manufacturing Industry Association argued that with the achieving of grid parity, solar PV would become the dominant RE power in China's energy market. Solar power has a low requirement for physical infrastructure and can be built as a stand-alone energy system, which does not need a large piece of land. It can fit onto the rooftops of existing buildings, a huge advantage over traditional large-scale power plants. These characteristics make it suitable for relieving the energy supply pressure in large electricity loading areas, such as in southern China (Senior experts in solar PV industry association, interview, 2017). The development of solar power is believed to aid the development of a low carbon and clean energy system in China. Moreover, the government has mobilised the development of a distributed solar PV energy system as a strategy to alleviate poverty in China with an objective of adding 10 G W capacity to benefit households and villages across the country by 2020 (Geall and Shen, 2018). With the emerging of new business models, financing mechanisms, and further ICT and energy storage technology development, industry actors believe that solar PV will become the dominant new electric technology in China.

Discussion
Our two cases have demonstrated that expectations play a crucial role in coordinating the alignment process between niche and regime actors. Oriented by their shared expectations, they work collectively to shape the prospective socio-technical structures. Different alignment patterns shape different phases of niche development.

Weak alignment and slow niche development
The two cases have illustrated that weak alignment between niche and regime actors' expectations matches low speed and scale of niche development. Before 2007, when there was weak alignment between niche and regime actors, there was relatively low take-up of wind and solar power (see Fig. 5). The narrowly shared expectations about niche development could also explain why at this stage, the policy goals set up by the central government for the wind power could not be achieved.
The correspondence of weak alignment to slow niche development is also validated by the comparative insights drawn across two cases. We see there was comparatively weak alignment between niche and regime actors' expectations towards solar power compared with wind power between 2007 and 2012, resulting in limited take-up of solar power compared with wind power. With higher expectations of wind power, regime actors such as the big utility giants showed more interest in investing in wind power plants when they were confronted with expectations of further stringent policy regulations requiring a shift towards clean and low carbon energy.

Medium-strong alignment and moderate niche development
The two cases evidenced that medium-strong alignment contributes to moderate niche development. However, niche and regime actors held different types of alignment in the two cases. In the wind power case, there was broader alignment between niche and regime actors at the niche level compared with the regime level, while in the solar power case the two sets of actors had similar breadth of alignment at both levels. Between 2008 and 2010 the development of wind power could be seen as an add-on to the market. Because of the narrower alignment at the regime level, wind power experienced high levels of curtailment. However, as we see in the later stage, when the vision of "clean and low carbon" was widely shared, wind power was further legitimised in the electricity system, and niche actors mobilised this legitimacy to argue for more institutional support to guarantee its generation. In the case of solar PV, the build-up of shared expectations at the niche level encountered a different process. Along with weakened expectations of competing technologies, such as hydropower and nuclear power, rapidly decreasing solar panel costs escalated the expectations of the competitive advantages of solar PV in the market.

Strong alignment and substantial niche acceleration
As demonstrated in the two cases, when there is strong alignment between niche and regime actors' expectations, their expectations could be translated into concrete goals and requirements of other actors. For example, in 2011, the Energy Research Institute under NDRC issued the < Roadmap to 2050 for China's wind power development > , establishing long-term development targets for installed capacity of wind power to achieve 400 G W by 2030 and 1000 G W by 2050. Moreover, we see explicit articulations of connecting wind and solar power development with sustainable and clean development at the landscape level. When the clean and low carbon ethos was widely shared in society, with the prospective visions that China needs for a low carbon transition, the scenarios with high proportions of wind and solar power in the electricity system were generated to urge further actions from corresponding actors (Energy Research Institute under NDRC, 2015).
Furthermore, when the central government share these expectations, it is more likely to implement supporting institutions for niche development. For example, to further stimulate support from the grid company for wind and solar power, in 2016 the central government set minimum annual generation hours for wind and solar power to encourage the utilisation of RE in the electricity mix. As evidenced in the two cases, when there is strong alignment between niche and regime actors, we see their shared expectations being more stabilised and able to be mobilised for more institutional support for wind and solar power's further development. For example, the central government implemented a stringent policy to cap coal power to create space for RE deployment to meet the targets set for non-fossil fuel in the energy mix in 2020. This institutional change contributed further to the fast wind and solar power deployment and the reduction of thermal power plants.
This strong alignment between niche and regime actors created self-reinforcing mechanisms which further contributed to fast system transformation. We see that when the regime is under pressure niche actors start to argue for the necessity of increasing RE to promote clean and sustainable energy revolution. Furthermore, niche actors specifically attempt to reduce coal power regime since 2016 to further increase space for wind and solar power generation in the country's electricity mix. As we see from Fig. 5, the percentage of yearly new installed capacity of thermal power has dropped by 11.5 % (from 82.4 % in 2011 to 70.9 % in 2017) in seven years. This rapid decreasing of the market share of coal power further weakened coal power investors' expectations of the strategic role of thermal power in future market. Furthermore, with stringent policy regulation from the central government, the coal power regime actors started to question coal power regime resilience. This provided further opportunities for niche actors to articulate potential solutions through the two RE technologies. This strong alignment explains why China's wind power and solar PV installed capacity surpassed its 2020 goals three years ahead of schedule (Finamore, 2019).

Conclusion
This paper endeavours to make a first contribution to unpacking the alignment dynamics between niche and regime actors' expectations, and how their alignment dynamics contribute to niche acceleration. Our contribution is fourfold. First, we conceptualise three alignment patterns between niche and regime actors' expectations: strong alignment, medium-strong alignment and weak alignment. Second, we define three phases of niche accelerations based on technology adoption lifecycle studies and relate them to the three alignment patterns. Third, we operationalise our conceptual framework by specifying different phases of niche accelerations and corresponding niche and regime actors in our cases and set thresholds to define different alignment patterns. Moreover, we offer a quasi-quantitative method to map out the alignment patterns between niche and regime actors. Fourth, we illustrate the alignment dynamics between niche and regime actors' expectations in two cases of wind and solar power development in China between 2000 and 2017.
Overall, this paper provides a theoretical framework that clarifies how the expectation alignments between niche and regime actors contribute to niche development, including its acceleration. Based on our results we would even argue that alignment dynamics between niche and regime actors' expectations can be seen as a good proxy for expected niche development. These research findings have significant policy implications. They suggest that to accelerate niche development, policy can play a crucial role in shaping the process of building shared expectations between niche and regime actors.
Moreover, our research results challenge the dominant state-led understanding of China's fast RE development and support recent research that has shown tensions and competitions between actors during China's wind and solar power development (Luo et al., 2012;Dai, 2015;Dent, 2015;Luo et al., 2016;Cai and Aoyama, 2018;Shen and Xie, 2018). For example, although the central government held ambitious goals for wind power, these goals were not achieved before 2007. This can be explained through our framework as a result of relatively weak alignment between niche and regime actors' expectations. Post-2007, alignment between the two increased, leading to surpass the central government's targets. Thus, our research framework can offer useful insights to illustrate the evolving coordination and alignment processes between different stakeholders. Our research findings also suggest this dynamic process is crucial for understanding the niche acceleration process in a country such as China. We argue that the proposed conceptual framework can also be used for other cases outside of China, an area for new research.

Funding
The first author would like to acknowledge the funding from the China Scholarship Council (CSC)/ University of Sussex Joint Scholarships, the Chinese Academy of Science and Technology for Development (CASTED) for the Transformative Innovation Policy Consortium (TIPC) project and the Science Policy Research Unit (SPRU) Doctoral Research.

Declaration of Competing Interest
The authors report no declarations of interest.  Non-fossil fuel More than 15 % of primary energy consumption by 2020.
More than 15 % of primary energy consumption by 2020 Non-fossil fuel in the energy mix should be higher than 20 % by 2030; Non-fossil power generation account for more than 50 % of total generation;

Coal powers
Reduce the share of coal power in the electricity mix to lower than 62% by 2020.
Reduce the share of coal power to less than 58% by 2020. Primary energy consumption of coal power should be capped below 6 billion tce; a It includes solar PV and solar thermal together.