Conventions, markets and industry evolution: the example of the wind turbine industry in Germany 1977–2021 Open Access

Abstract

Markets are drivers of evolutionary change. They link social norms to the evolution of an industry. The study uses a process tracing approach to identify two mechanisms of this link. The first mechanism describes how different conventions change the construction of markets. The second mechanism describes how the changing principles of evaluation influence the resource construction of firms and thus the life cycle of the industry. The empirical example is the wind turbine industry in Germany from 1977 to 2021. The development of the industry can be divided into different phases shaped by different conventions, each affecting market construction and industry evolution in different ways.

Introduction

A perspective on markets is crucial to investigate economic transformations (Geels, 2004; Mazzucato, 2016). Markets are a driver of change as they involve norms and conventions about what is considered valuable (Aspers, 2009). Boschma and Martin (2010, 6–7) described the aim of Evolutionary Economic Geography as ‘the processes by which the economic landscape - the spatial organisation of economic production, circulation, exchange, distribution and consumption - is transformed from within over time’. In many cases, innovation, technological change, path dependency, relatedness or routines were considered important for these transformations (Boschma and Frenken, 2006; Essletzbichler and Rigby, 2007; Trippl et al., 2020). The importance of markets for these transformation processes has received less attention so far.

Studies have already shown how social norms enter into the construction of markets (Boltanski and Chiapello, 2005; Berndt and Böckler, 2009; Çalışkan and Callon, 2010). Others have also shown how quality standards in markets that ­follow different norms lead to the differentiation of global production networks (Ponte and Gibbon, 2005). Studies on sustainability transitions have applied a temporal perspective. These studies have shown how social change is an impetus for policies that contribute to the formation of markets (Wüstenhagen and Bilharz, 2006; Geels and Schot, 2007). Social values can change and, with them, the corresponding policies (Lauber and Jacobsson, 2016), which also has an influence on the evolution of the corresponding industry.

This study aims to identify the mechanisms by which changes in societal values and goals affect the evolution of industries. The wind turbine industry in Germany serves as example. In the 1990s, political measures in Germany created the world’s largest market for wind energy. With this market, a globally competitive industry developed. Yet, policies were contested (Lauber and Jacobsson, 2016) and also the industry often vacillates between boom and crisis (Bruns and Ohlhorst, 2011).

The mechanisms that link the change of social values with the evolution of industries are elaborated with a process-tracing approach. A process-tracing approach serves to prove causalities in qualitative case studies: a variable X leads via a mechanism M to a variable Y (Beach and Pedersen, 2016). Variable X is described here by renewable energy policies, variable Y by industry development. Two mechanisms subsequently leading from X to Y are examined.

The first mechanism is through the construction of the market. Policies integrate into a market social norms about how to value something. These social norms are conceptualized via convention theory (Ponte and Gibbon, 2005; Boltanski and Thévenot, 2006; Diaz-Bone, 2016). Conventions refer to shared expectations (Diaz-Bone, 2016) that can be distinguished according to different ‘modes of evaluation’, like efficiency, reputation or creativity (Boltanski and Thévenot, 2006). This concept has already been used to study the differentiation of global production networks (Ouma, 2010). Here, this concept is used to capture temporal change in markets (Boltanski and Chiapello, 2005).

The second mechanism is through firms. The industry life cycle (ILC) model is used as a perspective on industry (Klepper, 1997). The ILC describes the evolution of an industry based on firm routines. It hypothesizes that ­markets develop due to the firms’ enhancement in their productivity. However, this article assumes the causality in the other direction, i.e. markets influence the ILC. The thesis of this study is therefore that the prevailing convention in a market influences the way in which companies develop their routines. Accordingly, different conventions in markets should also have different influences on firm routines and result in different patterns of the ILC.

The data to investigate how valuations in markets alter the evolutionary dynamics of the wind turbine industry in Germany base on the one hand important elements of market construction like laws, programmes as well as secondary literature that describes the political objectives behind these programmes (Jacobsson and Lauber, 2006; Wüstenhagen and Bilharz, 2006; Lauber and Jacobsson, 2016). The analysis of industry evolution bases on the database ‘The Wind Power’ (thewindpower.net). The database contains information on both wind farms and the manufacturers that supplied turbines for these farms.

We start the study with the first wind energy related projects and policies in 1977 and end in 2021. According to Henning (2019), this would make this study one of the few studies in Evolutionary Economic Geography that goes further back than the 1990s and 1980s. Henning (2019) further disclosed that especially historical research is required to examine the evolution of novelty, retention and selection, whereas the present work focuses on the latter.

The article is organized as follows. The next section describes the conceptualization of markets in evolutionary economics, as well as neighbouring areas. The third section introduces convention theory as a means of conceptualizing selection dynamics. In the fourth section, these selection dynamics are combined with the ILC. The fifth section presents the case, methodology and data. The sixth section investigates the market and its construction, whereas the seventh section analyses industry evolution.

Markets in- and outside evolutionary economics

Evolutionary Economic Geography builds on the basic dynamics of Evolutionary Economics. Foster and Metcalfe (2001) described the basic evolutionary economic model, which comprises two essential elements of variation and selection. Variation and selection do not take place randomly but have a particular directionality. This directionality increases the rate of economic progress compared with random novelty generation.

However, studies with an evolutionary perspective usually focus on one of these elements, namely, variation, with emphasis on concepts like innovation, learning or organisational routines (Geels, 2004). Accordingly, evolutionary economic geography studies are predominantly concerned with questions of how the creation of novelty drives the emergence of new industries (Fornahl et al., 2010), transforms existing regional industries (Hassink et al., 2019) or serves as the basis for regional diversification (Boschma, 2017) and regional growth (Frenken et al., 2007).

Markets describe the selection side. In a neoclassical perspective, markets are the most efficient allocation mechanism for coordinating supply and demand. If markets can develop undisturbed, a market equilibrium is achieved. Evolutionary Economics contrasts this equilibrium perspective on markets with a dynamic perspective (Potts, 2001). Markets change with the dynamics of firms and routines (Klepper, 1997). Yet, markets are not investigated as a distinct entity, as competition occurs either through price and innovation, where firms with better routines displace firms with worse routines (Klepper, 1997), or are seen as institutionalised in the decision of agents (Foster and Metcalfe, 2001).

The analysis of markets has gained considerable prominence in the study of sustainability transitions (e.g. Environmental Innovation and Societal Transitions, Volume 42). Central to many studies in this field is the notion of the socio-technical system. In contrast to other approaches, such as the sectoral innovation system, the socio-technical system incorporates market-driven and user-led selection, and therefore, the selection environment as well (Geels, 2004). A central argument for the ­importance of markets is that sustainability transitions require the creation of new markets as niches (Geels and Schot, 2007), in which new technologies can be further developed. Accordingly, research on sustainability transitions investigates the processes involved in market formation, i.e. the transition of a market from a niche to a mass market (Geels and Schot, 2007; Dewald and Truffer, 2012). In this vein, studies have demonstrated how the state-led creation of markets has resulted in the emergence of industries, particularly in renewable energies (Dewald and Truffer, 2012; O’Sullivan, 2020).

Studies show that the further development of these markets is also subject to different norms, e.g. if market construction focus on private or social costs (Jacobsson and Lauber, 2006). Studies have additionally emphasized the significance of mission-oriented policies in determining the directionality of industrial and technological development (Mazzucato, 2016), the causal relations and feedbacks that determine market formation processes (Nijhof et al., 2022), and the misalignments, bottlenecks and undesirable developments that coincide with market formation (Boon et al., 2022).

The geography of these market formations in protected niches has been identified early on. Dewald and Truffer (2012) employed the example of photovoltaics to describe the evolution from a nurturing market that is protected by a spatial niche, in the form of regional feed-in tariffs, to a multiscale mass market. Moreover, the geography of market formation has an impact the geography of industries: Bednarz and Broekel (2020) revealed that the wind turbine industry in Germany emerged primarily in spatial proximity to the early markets in northern Germany.

While transition studies investigate markets as part of a socio-technical system, scholars of economic sociology argue that markets are entities in their own right (Callon, 1998; Muniesa et al., 2007) and ‘marketization’ as a distinct process (Berndt and Böckler, 2009). Çalışkan and Callon (2010, 3) ‘define the study of marketization as the entirety of efforts aimed at describing, analysing and making intelligible the shape, constitution and dynamics of a market socio-technical arrangement’. This perspective focuses on how markets are socially and physically constructed (Callon, 1998; Muniesa et al., 2007; Çalışkan and Callon, 2010).

A fundamental assumption in this literature is that markets need knowledge to function (Aspers, 2009). Scholars, such as Çalışkan and Callon (2010) or Muniesa et al. (2007), suggested that the information upon which buyers and sellers make their decisions is not only embedded in social relations but also in the relations something has to other entities (Callon, 1998) like standards (Ponte and Gibbon, 2005), the status of other agents (Aspers, 2009), or market devices (Muniesa et al., 2007). For example, quality standards communicate information about product qualities and in doing so disembed coordination and evaluation from social relations (Ponte and Gibbon, 2005).

Berndt and Böckler (2009) established a particular geographical perspective on markets by arguing that their ordering of markets is done via bordering, i.e. drawing geographical borders around what belongs to the market and what does not. Other studies have examined how the construction of markets affects the relations between the industrialised North and less developed South (Berndt and Wirth, 2019), or the interdependencies of local and global markets (Haisch and Menzel, 2023).

In conclusion, evolutionary economics argues that markets are a crucial element of selection, and that selection influences the directionality of evolution. However, the market itself and thus selection through markets is not fully conceptualized. Research conducted on sustainability transitions and studies on marketization treat markets as a distinct entity, whose shape and how they select depends on norms, the objectives of actors and also power relations. The task then is to conceptualise markets as changing entities and how these changes also alter what is valued in a market.

Conventions and selection

One approach to examining the construction of a market and the integration of knowledge and norms within it is through the lens of conventions (Boltanski and Thévenot, 2006). Diaz-Bone (2016, 49) defined conventions as follows: ‘Conventions are socio-cognitive resources actors rely on to achieve shared interpretations, evaluations and valuations of situations and the value of objects, persons and actions.’

Conventions are a suitable way to conceptualize selection as they rely on equivalence between actors, i.e. to ‘converge towards a common definition of the relevant object in this situation [….] they must share a common capacity to see what fits the situation and under what relation’ (Boltanski and Thévenot, 1999, 361). Thus, conventions describe shared ‘modes of evaluation’ (Diaz-Bone, 2018b). These shared ‘modes of evaluation’ have two primary functions. First, they allows for the investigation of the ‘collective assignment of worth’ (Diaz-Bone, 2016, 49) to something. Consequently, they enable an agreement on whether something is deemed small or great based on a particular convention. Second, they serve to coordinate actions and thereby as the foundation for achieving collective goals (Diaz-Bone, 2016). Storper (1996), for instance, highlighted how specific products necessitate particular innovation systems that are shaped by specific conventions to mitigate uncertainty.

Convention theory distinguishes various ‘modes of evaluation’ (Boltanski and Thévenot, 1999), which give rise to different ‘orders of worth’ (Boltanski and Thévenot, 2006), from which conventions can be derived: the domestic convention bases on tradition and craftmanship and evaluates reputation; the industrial convention bases on planning and standardization and evaluates efficiency; the market convention bases on supply and demand and evaluates price; the inspired convention bases on creativity and evaluates originality; the opinion convention bases on renown and evaluates recognition and the civic convention bases on collective interest and evaluates relevance for society (Diaz-Bone, 2018a). Thus, convention theory emphasises that transactions in markets can take place based on different principles and that a textbook ­description of markets (Callon, 1998) is only one possibility of how transactions in a market are organized.

Convention theory is used to analyse situations where actors utilize different conventions in to justify their practices (Boltanski and Thévenot, 1999). Yet, convention theory describes also ‘investments in forms’ (Diaz-Bone, 2015, 28). With respect to markets, a particular form were quality standards, labels and certifications, as described in the literature on global production networks (Ponte and Gibbon, 2005; Ouma, 2010). But these ‘investments in forms’ can also relate to laws (Diaz-Bone, 2015), programs and support systems that serve as crucial pillars in market construction (Mazzucato, 2016).

By coordinating activities, especially when transformed into laws, conventions share similarities with institutions (Diaz-Bone, 2018b). Yet, there are several differences. First, conventions are ‘deeper culturally established knowledge frames’ (Diaz-Bone, 2018b, 69) which form around basic ‘orders of worth’ (Boltanski and Thévenot, 2006). Institutions, on the other hand, can take a variety of forms and be based on different conventions. Second, institutions describe a manifest and even formal mode of external coordination, whereas conventions are applied in situations and serve to interpret institutions. Third, conventions and institutions can coincide, but when conventions change, institutions must also change, since a mismatch between conventions and institutions would lead to a crisis (Diaz-Bone, 2018b). Fourth, as conventions describe forms of evaluation, they allow a ranking of objects, actors or practices.

Conventions exert a selection effect as they allow one to evaluate what is great or in accordance with a specific ­convention. As this selection effect differs between conventions, different conventions result in different beneficiaries and disadvantaged parties (Ponte, 2016). Accordingly, a diversity of conventions would lead to a market fragmentation. Indeed, the literature on quality conventions describe how markets differentiate themselves via different conventions (Ponte, 2016). Furthermore, conventions are subject to change over time, often due to societal changes (Boltanski and Chiapello, 2005). For instance, Ponte (2016) pointed out that numerous studies describe how changing conventions in global value chains has transformed their governance from organized industrial mass production to a trust and place-based relationship. Consequently, alterations in conventions enable the description of shifts in selection patterns over time, which, in turn, can facilitate the investigation of why actors in a given field exhibit varying performance.

Table 1 presents a summary of the key conventions described in the literature. It amalgamated the original categories by Boltanski and Thévenot (1999) with the advancements in the Global Value Chains literature that highlight various levels of evaluation (Ponte and Sturgeon, 2014), as well as the different product qualities that arise from various conventions (Diaz-Bone, 2018a). The ‘Mode of Evaluation’ (Boltanski and Thévenot, 1999) refers to the higher principle of equivalence, which serves as the basis for ranking. ‘Focus’ (Ponte and Sturgeon, 2014) considers that evaluations may not necessarily be based solely on an object itself, but rather on its integration into a system of controls, coverage in social media, or forms of redistribution. ‘Measurement’ describes how something is measured based on both the ‘Mode of Evaluation’ (Ponte and Sturgeon, 2014). Lastly, ‘Product’ describes the different product qualities that may arise from each convention (Diaz-Bone, 2018a). When analysing the impact of ­conventions on ­selection, this table demonstrates that conventions vary not only in terms of their mode of evaluation but also in terms of what is valued and the resultant product.

Convention theory serves to differentiate modes of evaluation in markets. It makes no assumptions about how different conventions affect the growth or decline of markets. However, it can be assumed that markets grow, particularly when they are characterised by an industrial or civic convention. In the first case, through the increase in productivity, which goes hand in hand with falling prices (Boltanski and Chiapello, 2005). In the second case, additional resources are channelled into these markets in order to have a social impact.

Markets and the ILC

In evolutionary terms, the evaluation along the modes of evaluation described by convention theory defines the selection mechanism. This section combines convention theory with evolutionary concepts. The basic assumption is that conventions affect markets, which again affect industry evolution. The ILC (Klepper, 1997, 2007) is used as a baseline model.

The ILC predicts an initial increase in the number of firms in an emerging industry, which is followed by a sharp decrease in the number of firms after a shakeout occurs. The shakeout phenomenon refers to the competitive environment in which some firms become so successful that others are forced out of the market (Klepper, 1997).

The ILC is based on three different arguments. The first argument is that firms differ in their routines, which affects their performance. The second argument is that better-performing firms attract better employees and are able to invest more in research and development (R&D), which further improves their routines and practices. The third argument connects the individual firm with industry dynamics. Specifically, firms investing more in R&D increase their productivity more strongly and create more competitive pressure. Increasing differences between the most and least competitive firms ultimately leads to a shakeout of less competitive firms. As the shakeout was brought about by the strong productivity of some firms, the shakeout is usually accompanied by strong growth in the market.

The model has also been used to explain spatial variation in industries. New entrants benefit from experience in related industries. Therefore, firms in new industries tend to locate in regions with related industries (Boschma and Wenting, 2007), which is one driver of geographic fragmentation. In industries characterised by spin-off processes, geographical concentration tends to occur, as spin-offs tend to originate from firms with better routines and the iterative inheritance of routines through spin-offs leads to the concentration of firms with superior routines (Klepper, 2007).

Conventions are expected to impact the ILC by defining the creation of value in markets. This is done in several ways. The first is the size of the market. The ILC hypothesizes an increase in the market size due to the firms’ enhancement in productivity (Klepper, 1997). However, the causality might also go into the other direction. Conventions could influence the market size by directing more resources into it, which affects the ILC. An increase in market size enables firms to invest more in R&D, which allows better firms to invest even more and thereby widen the productivity gap between the most and the least competitive firms. As a result, larger markets are expected to accelerate the ILC, but might also give more firms the ability to stay in the market.

Second, certain conventions can lead to the fragmentation of markets into sub-markets (Storper, 1996). Studies on the ILC show that fragmented markets can limit the investment in R&D, thus slowing down the ILC (Bhaskarabhatla and Klepper, 2014). ILC studies argue for a technological reason for market fragmentation. But market fragmentation can also result from conventions. A domestic convention is an example of this, resulting in the production of customized products. Fragmented markets can also result, when different conventions shape a market (Ponte, 2016), thereby affecting the ILC.

Third, conventions have the potential to affect the resources allocated to R&D. In general, the ILC presupposes that investing in R&D leads to an enhancement in productivity. However, it is worth noting that R&D can be directed towards either product or process innovation (Abernathy and Utterback, 1978). Conventions such as the inspired convention, which emphasizes novelty, are likely to foster investment in product innovation, whereas conventions such as the industrial convention, which emphasizes productivity, are inclined to promote process innovation. Therefore, it is anticipated that markets that encourage investment in R&D towards process innovation, rather than product innovation, will grow stronger than markets that support investments in product innovation.

In summary, conventions direct more or less resources to a market, create a fragmented or homogeneous market, or support investment to increase either productivity or product variety. Table 2 presents an overview of the expected impact of conventions on markets on and how these markets affect the ILC. The ILC is usually applied to manufacturing industries. As a result, Table 2 is confined also to those conventions that usually are also applied to these industries, i.e. the domestic, industrial, market and civic conventions.

The ILC itself is not associated with a particular convention. Due to the widespread connection of a market with a market convention (which is strongly countered by the convention theory) and the ILC assumption that the market is a result of the ILC, the association of the ILC with the market convention serves as a reference model.

Table 2 therefore describes the deviations from the market convention. The industrial convention supports investments in increasing productivity, which leads to stronger market growth. Two opposing processes are hypothesized to impact the ILC. Firstly, when the industrial convention aids in enhancing productivity and efficiency for a wide range of firms, it may result in an increase in the number of firms, as even less competitive firms improve. Secondly, firms with better routines may particularly benefit from the larger market size and increased production scaling, which in turn intensifies competition for other firms. Combined, it is assumed that a shakeout in the industrial convention will not diverge in time from a market convention but occur at a larger size of the market. In general, the industrial convention is connected with increasing output per firm. Due to reduced uncertainty in production and the facilitation of industry wide knowledge spillovers, this convention may promote spatial dispersion (Storper, 1996).

An evaluation within a domestic convention takes place based on reputation and market exchanges take place between actors with specific assets. Craft-based production is prevalent, with products often being unique and made for specific purposes (Diaz-Bone, 2018a). As there are few increasing returns to scale, the market remains fragmented. It is expected that a growing number of firms would coincide with a growing market and vice versa. In this market, increases in productivity are less rewarded than in a market or industrial convention. As a result, a domestic convention-defined market would not result in a shakeout at the industry level. Firm output is expected to be smaller than compared to the market convention. Consequently, spatial concentration would not occur either, as this requires a shakeout according to the ILC model.

In a civic convention, evaluation is based on the relevance for society (Boltanski and Thévenot, 2006; Ponte and Sturgeon, 2014). The justification of actions is often based on rules or a redistribution system that serve the collective or public interest. Therefore, the implementation of a civic convention can mobilize various resources towards achieving these goals. This convention would coincide with the growth of the market, allowing for competitive firms to invest more in R&D and increase competitive pressure. Yet, the implementation of a civic convention may attract new entrants, resulting in a larger number of firms in comparison to both the market and the industrial convention. However, the average output per firm does not increase more than under the market convention because it allows more firms to stay in the market. Since the civic convention allows more firms to survive and the ILC predicts spatial concentration resulting from shakeout, it is expected that firms will be spatially more dispersed than in a market convention.

Although industries are often organized around dominant conventions (Storper, 1996), there are also often different and competing conventions in a market (Ponte, 2016). It is assumed that the existence of multiple conventions in a market can lead to fragmentation and a blocked ILC. Additionally, conventions change. It is expected that the emergence of new conventions could facilitate the entry of new firms and provide new opportunities for those firms that previously struggled to catch up under the old convention to thrive under new ones.

Case, Method and Data

This section describes the development of wind energy in Germany, the method of a process-tracing approach and the data.

The case: wind energy in Germany

After early markets in Denmark and the US, Germany became the world’s largest wind energy market in the 1990s. A competitive industry also formed with several manufacturers among the global top ten manufacturers (Menzel, 2021). Studies point in particular to the policies that have led to these developments (Wüstenhagen and Bilharz, 2006). Important milestones are certainly the early technology support starting from 1977, the 100 MW wind programme of 1989 (introduction of test programmes and feed-in tariffs for turbines), the StrEG of 1990 (Electricity Feed Act, Stromeinspeisegesetz, introduction of feed-in tariffs for all renewable energies), the EEG of 2000 (Renewable Energy Sources Act, Erneuerbare Energien Gesetz; feed-in tariffs and setting of an expansion corridor) (Jacobsson and Lauber, 2006; Lauber and Jacobsson, 2016).

Studies usually describe the phases of the wind energy industry’s development using these programmes and laws. Bruns and Ohlhorst (2011) describe the early years as a pioneering phase characterised by technology programmes; the phase of the 100 and later 250 MW wind programme and the StrEG as a phase of adapting to the framework conditions; the subsequent phase, which lasted (with interruptions) from 1991 to 2002, as a breakthrough and boom phase; and a consolidation phase characterised by slower market growth from 2002.

Lauber and Jacobsson (2016) describe the different phases of renewable energy policy as alternating between ‘fit and conform’, with a focus on short term private costs, and ‘stretch and transform’, with a focus on long-term social costs. They described the StrEG as preparation for a policy of ‘stretch and transform’, which became effective through the EEG 2000. They argue that the amendment of the EEG 2014 could already be a turning point towards a ‘fit and conform’ policy.

The market for wind turbines can be measured via yearly installed wind energy capacity. Wind farms are typically owned by private companies such as utility companies, banks, project developers or civic associations. The wind turbine industry includes in addition to manufacturers also suppliers of components nacelles, gearboxes, towers, foundations and rotors. There are highly vertically integrated manufacturers, as well as manufacturers that source the majority of components through suppliers (Menzel and Adrian, 2018). The deployment service chain, which deals with deployment, operation and maintenance should be distinguished from this.

While wind turbine manufacturers are mainly located in northern Germany, many suppliers are located in the established industrial centres of North Rhine-Westphalia or southern Germany (MacKinnon et al., 2019). However, no decided centre has emerged in the production of onshore wind turbines (Kammer, 2011). This spatial pattern fits to the ILC assumption that industries shaped by diversification do not necessarily spatially concentrate (Klepper, 2007). Recent data from Menzel (2021) indicates that the industry comprises of 17 diversifiers and 5 spin-offs. Yet, there are smaller concentrations of manufacturers and suppliers in offshore wind turbine production in Bremerhaven (Fornahl et al., 2012) and Cuxhaven; and Hamburg emerged as a centre for organisation of global value chains (Menzel and Adrian, 2018). However, Bremerhaven has experienced a decline in recent years due to the bankruptcies and relocations of several companies.

Method: process tracing approach

Methodologically, a process-tracing approach is employed (Collier, 2011). Process tracing methods are utilised to establish causality in single case studies, i.e. how a variable (X) influences another variable (Y) via a mechanism (M). The elaboration of the mechanism can be done in different ways. Accordingly, Beach and Pedersen (2016) distinguish between three types of process tracing: theory testing (testing a causal mechanism derived from theory), theory building (establishing causal mechanisms for mid-range theories) and explaining outcome (explaining the causes that result in a particular outcome). In this study, theory-testing process tracing is used, and the expected mechanisms are derived from theory.

In this case study, the X (the independent variable) are policies that influence the development of the industry (the dependent variable Y). Assumptions were made about two mechanisms. The first mechanism (M1) describes the ‘principles of evaluation’ that are integrated into a market and according to which the market selects. The second mechanism (M2) describes how firms build up resources according to the construction of the market.

Since a mechanism often cannot be observed directly, the analysis is carried out using various variables that speak for or against the presence of this mechanism. Table 1, for example, shows the different variables that should be present for a particular convention, whereas Table 2 describes the assumptions on how particular conventions affect the ILC. A conclusion about whether a mechanism is present depends on an assessment of the strength of the effect of each variable and whether other conclusions can be ruled out (Collier, 2011).

In order to work out the influence of the change in the mechanism, the case is divided into different temporal phases in which different forms of evaluation exist. This division into different phases, or ‘temporal bracketing’ is also suggested by Langley et al. (2013) as analytical heuristics for process studies. Temporal brackets are ‘separated by identifiable discontinuities in the temporal flow’ (Langley et al., 2013, 7), which are changes in conventions, in our case.

Data

The conventions that shape markets are analysed via laws or large public support programs for wind energy that ­impact a significant number of firms. The primary governmental programs, laws and regulations are the GROWIAN project, which aimed to develop large-scale wind turbines, the 100/250 MW Wind Program of 1989, the Electricity Feed Act of 1990 (StrEG), the Renewable Energy Sources Act (EEG) of 2000 and its subsequent amendments.

These laws and programs are analysed based on the categorization presented in Table 1. The connection between acts, programmes and conventions can often be directly linked to a ‘mode of evaluation’. Conventions rely on ‘justification’ (Boltanski and Thévenot, 2006) and laws usually include such a justification that is either explicitly stated in the act or in a memorandum attached to the act. Additionally, the processes that lead to the creation of acts and programs, as well as how they are contested, for example, through lawsuits have been described in detail in several studies (e.g. in Jacobsson and Lauber, 2006; Lauber and Jacobsson, 2016).

The development of the market and industry are described quantitatively. The basis for this is the ‘The Wind Power’ data base (thewindpower.net) in the May 2022 version. This database includes data from wind farms worldwide, the installed capacity, the types of wind turbines installed and their manufacturers. This database is also used in other studies on the wind turbine industry (O’Sullivan, 2020).

This data base includes data on 33,903 wind farms. First, data of the wind farms in Germany are extracted from this database. This extraction serves to analyse market development in Germany. In total, information is available on 11,401 wind farms in Germany between 1983 and 2021. Of these, 45 have no information on installed capacity (sometimes because they are still under construction), 1902 have no data on manufacturers and 444 have no clear date of installation. Moreover, some of the missing values overlap, resulting in a total of 2,199 wind farms with missing values and 9,202 wind farm data records that are complete. The Wind Farm Database contains 62,275 MW as sum of yearly installed capacity. This number does not account for decommissioned wind turbines.

The analysis of the industry is limited to the manufacturers, in line with the studies on the ILC (Klepper, 2007; Bhaskarabhatla and Klepper, 2014). To analyse their performance of German wind turbine manufacturers, data on wind farms equipped with turbines from German manufacturers are extracted from this data base. On a global scale, there were 10,502 wind farms with turbines from a German manufacturer. Yet, given that the database does not cover the early phase of the wind turbine industry in particular, data on firm entries and exits were taken from Kammer (2011) and Menzel (2021). Firms have been defined as German manufacturers if the headquarters of production is located in Germany. For example, Siemens is not defined as a company headquartered in Germany, because it entered the market via the acquisition of Danish firm Bonus in 2004 and, accordingly, less than 10% of wind turbine capacity installed by Siemens and Siemens-Gamesa (which resulted from merger with Spanish producer Gamesa in 2017) was installed in Germany. Data on wind turbine manufacturers are analysed based on the expectations of Table 2.

As wind turbine manufacturers do not form distinct concentration in space, for the purpose of this analysis, the division between coastal and inland regions is used as an indicator of spatial distribution. Coastal regions are the federal states of Bremen, Hamburg, Lower Saxony, Schleswig-Holstein and Mecklenburg-Western Pomerania (see also Bednarz and Broekel, 2020).

Conventions in the German wind energy market

This section outlines the temporal phases and the conventions that shape these phases, which are summarized in Table 3. The dates in the table refer to the year of introduction of these most important laws and programs. Additionally, the table includes the objective of programs and laws as indicator for the mode of evaluation as well as the dimensions according to which conventions were operationalized in Table 1.

1977: Domestic convention

The first developments began in the 1970s. In 1974, the then Federal Ministry of Research and Technology (BMFT) decided to support research into the use of wind energy in response to the 1973 oil crisis (Heymann, 1998). The most important, but later unsuccessful, project was GROWIAN (GROßWIndANlage, large scale wind turbine), which aimed to build a 3 MW wind turbine. This project started in 1977. GROWIAN was erected in 1983 and uninstalled in 1987.

Involved in the project were leading national companies from the mechanical engineering and aviation industries such as MAN, MBB and Dornier. These companies already had relationships with public agencies and were directly supported through technology funding (Heymann, 1998). Despite commonly regarded as failure, a follow-up project with the same consortium, GROWIAN II, was launched in the mid-1980s (Heymann, 1998). The GROWIAN project ­focussed on specific companies with established relations, as well as producing an individual showcase turbine correspond to a domestic convention.

During the 1980s, several other programs were introduced with a lesser focus on the established firms. The main objective of these programs was to facilitate domestic companies in the production of dependable wind turbines. The duration of these programs ranged from 1 to 3 years (Hoppe-Kilpper, 2003). An example of such a programme was the special demonstration programme for wind turbines up to 250 kWh from 1986 to 1988. The erection of 48 wind turbines from 13 German manufacturers was supported. These wind turbines had to be part of a measurement programme to gain information about the reliability of wind turbines (Hoppe-Kilpper, 2003). In contrast to the GROWIAN project, which involved few selected companies and had the aim to generate a technologically sophisticated large-scale win turbine, these programs were open to other companies and had a focus on measurement systems to make wind turbines reliable. Thus, these programs, although usually small scale, fit to the industrial convention.

In conclusion, the initial phase is characterized by various conventions. GROWIAN I and II projects played a defining role in this phase and emerged from a domestic convention. However, the small-scale demonstration programs hinted at a shift at the end of this phase.

1989: Industrial convention

In 1989, the 100 MW Wind Programme was launched. It was designed to promote the installation of 100 MW of wind energy. Due to its success, it was expanded into the 250 MW Wind Programme. The programme had two components. The first was to provide financial assistance to buyers to install wind turbines. The second component was a feed-in tariff of 4 and later 3 cents per kW/h. Producers of electricity from renewable energy sources received surcharges for the electricity fed into the grid (Wüstenhagen and Bilharz, 2006).

Funding through the 100/250 MW programme required participation in the Scientific Measurement and Evaluation Program (Wissenschaftliches Mess- und Evaluierungsprogramm, WMEP) (Hoppe-Kilpper, 2003). The aim of WMEP was to gain extensive knowledge about the reliability of different wind turbines under different conditions. The findings of this programme were made publicly available to enable companies to improve their turbines (Langniss, 2006).

The two programmes ran until 1996. Support was provided especially in the early phase of the programme. In the years 1990–1992, about 80% of the installed plants received support through the 250 MW programme (Hoppe-Kilpper, 2003). Additionally, the programme had a focus on German manufacturers. The market then was dominated by Danish manufacturers (Oelker, 2005). Yet, Langniss (2006) stated that 67% of supported wind turbines came from German based manufacturers, whereas 31% from Danish and 2% from Dutch manufacturers.

In addition to these programmes, the StrEG was introduced in 1990. The explanatory memorandum to the StrEG explicitly stated that this law was designed as a departure from market mechanisms. This law obliged the large energy supply companies to feed the electricity produced by small generators into their grid and to pay for it. The price set in the StrEG for wind energy was 90% of the price for end consumers. In the years 1990–2000, this was 8.25–8.84 cents per kW/h (Hoppe-Kilpper, 2003).

The unique condition of this phase was the large-scale integration of wind turbines into a system of test and control to objectively measure their reliability, combined with financial support through a distribution system of feed-in tariffs. Such a test and control system are a crucial element of the industrial convention. Additionally, large shares of installed wind turbines were part of this program. The WMEP program was in particular directed at German companies, thus policy aimed to develop a domestic industry (Langniss, 2006).

Other elements were the StrEG feed-in tariffs and the 100/250 MW programme. Feed-in tariffs serve to redistribute energy costs, which would be a necessary condition for a civil convention. However, there are several reasons to believe that in this case the feed-in tariffs were only necessary to achieve a certain number of installed wind turbines for the WMEP programme to work. First, feed-in tariffs were intended to ‘level the playing field’ (Jacobsson and Lauber, 2006, 264) for renewable energy by allowing it to compete with, rather than substitute for, fossil forms of energy production. Second, a hardship clause was introduced, which excludes electricity companies from purchasing renewable energy if they would have to raise their electricity prices significantly (StrEG 1990 Section 2 Abs. 2). Thirdly, it was assumed that the feed-in tariffs would not have a significant impact on energy prices (Lauber and Jacobsson, 2016). Thus, these programmes would limit the expansion of installed wind turbines but would support enough wind turbines for the scientific measurement and evaluation programme to be effective in creating a competitive industry.

When the 100/250 MW programme came to an end in 1996 and the StrEG was still in place, there was a discrepancy between the industrial convention that characterised this phase and the remaining StrEG, which is indeed a redistribution system in its own right. In addition, the redistributive effects, which were initially thought to be negligible, began to affect the large utilities. In response, incumbent utilities began to lobby at the level of government and EU competition policy and to challenge the law in court (Lauber and Jacobsson, 2016). This phase, described by Jacobsson and Lauber (2006) as one of crisis and uncertainty, only ended with the 1998 amendment of the StrEG, which confirmed the existing feed-in tariffs. Thus, while the beginning of this phase was characterised by an industrial convention, the end of this phase is characterised by contested conventions.

2000: Civic convention

The following phase commenced in 2000 with the implementation of the EEG. The EEG is recognized as a significant policy shift in supporting renewable energies (Lauber and Jacobsson, 2016). The EEG expanded on the feed-in tariffs that were already in place under the StrEG. In comparison to the StrEG, a fixed remuneration of 9.1 cents per kW/h for wind energy was established, and feed-in tariffs were provided for a period of 20 years, making investment estimable. Additionally, the EEG removed the restrictions on renewable energy production that were imposed by the StrEG. The hardship clause, which limited the expansion of renewable energy, was eliminated, and large utility companies were permitted to benefit from the feed-in tariffs. Furthermore, renewable energies were given priority when they were supplied to the grid (Wüstenhagen and Bilharz, 2006).

The first EEG from 2000 already promoted electricity from offshore wind turbines. Until 2008, however, due to the high costs and risks of offshore wind power generation, almost no capacities were installed. With the revision of the EEG in 2009, the feed-in tariffs for electricity from offshore wind power increased to consider the high costs of installing offshore wind turbines. The tariffs were now 13–15 cents per kWh for the first 12 years of operation.

The EEG 2000 transformed the industrial convention into a civic convention. The primary objective of this policy was to facilitate the shift from fossil fuels to renewable energy sources, thus achieving a social objective (Lauber and Jacobsson, 2016). The key mechanism used to realize this goal was the adoption of feed-in-tariffs, which are a form of redistribution system that is typical of a civic convention. In comparison to the StrEG, the EEG intended to have a more significant societal impact, as it involved an increase in the level of feed-in-tariffs, an expansion of the group of actors who could benefit from the tariffs, and the elimination of any restrictions on installed wind turbine capacities. However, the 2014 amendment to the EEG already imposed a (yearly increasing) cap on installed capacity and prepared the introduction of a tender system into the next amendment (Lauber and Jacobsson, 2016).

2017: Market convention

In 2017, the EEG changed from a law based upon civic convention towards a market convention. Several reasons apply for this. First, the redistribution system was fundamentally altered. A change from fixed feed-in tariffs to a tendering procedure occurred. Prospective wind park operators compete with others for subsidies via an auction procedure. Funding was provided for projects that required the smallest share of subsidies. Thus, the price is the decisive element of evaluation, which is the main principle of a market convention. In this phase, feed-in tariffs reduced to 5.25–5.78 ct/kWh (Handelsblatt, 2017), which were far lower than before. Second, the goal of keeping the costs for the production of renewable energy low was explicitly formulated in Section 4 para 2 EEG 2017, which changed the focus from the collective impact (which is important in the civic convention) to the price (which is the crucial element in the market convention).

The reason for this change was due to an increase in electricity prices, caused by the compensation system distributing the higher prices of renewable energy to consumers. While the price increase may have been overstated (Lauber and Jacobsson, 2016) it nevertheless resulted in a fundamental change, which was anticipated already by Lauber and Jacobsson (2016) as one from ‘stretch and transform’ to ‘fit and conform’ policies.

Development of wind turbine markets over time in Germany

Four different phases with different market conventions were identified. The market is expected to grow in particular in the industrial convention and the civic convention. Corresponding market developments would be indicators of these conventions. Figure 1 describes the development of the annual installed capacity of wind turbines and of this the offshore capacity based on the ‘The Wind Power’. The installed capacity based on DEWI (1989–1995), BTM Consult (1996–2004) and Global Wind Energy Council (2005–2021) serve to assess the quality of the data. The figure starts in 1989 due to the availability of data and the low-installed capacity. In the previous phase of the domestic convention, only a few MW of wind power were installed. Studies speak of 20 (Jacobsson and Lauber, 2006) to 30 MW (Oelker, 2005)’. The Wind Power’ data base shows 1.3 MW installed by then. The market for wind energy was therefore small at that time. The figure shows that there are gaps in the data. Till 1995, the database represents less than 50% of installed capacity. Only after 1998, this share was higher than 90%. Fluctuations above and below the comparative data apparently reflect points in time when a wind farm was connected to the grid.

During the industrial convention from 1989 onwards, the market grew strongly, especially since 1993. In 1999 alone, a total of 1,568 MW was installed. With the implementation of the EEG in 2000, a civic convention was enforced. However, the flattening of the annually installed capacity does not correspond to the expected further growth. The reason is that areas for wind turbines were designated more restrictively due to citizen protests against wind farms (Bruns and Ohlhorst, 2011). With the increased feed-in-tariffs for offshore wind farms in the EEG 2009, the corresponding offshore expansion increased considerably. By 2008, installed offshore capacity consisted mainly of test turbines. At the peak in 2015, the yearly installed offshore capacity marked ca. 2,400 MW and therefore ca. 40% of total installation.

From 2017 onwards, the market convention has been reflected in a decline of the annual installed capacity, with the installation of offshore wind turbines being particularly affected. In 2021, there was no newly installed capacity, which amounted to exactly 0 MW. The market convention and increased price competition was thus accompanied by a reduced market size. In summary, the market developments fit the assumptions. The only exception is the temporary decline during the civic convention, which can, however, be explained by more restrictive land use designations (Bruns and Ohlhorst, 2011).

Development of the wind turbine industry in Germany

This section investigates the development the industry. Figure 2 illustrates the number of German wind turbine manufacturers over time as well as the logarithm of the average installed capacity per firm worldwide. The number of firms increases during the domestic and industrial convention. The peak happened during the industrial convention with 18 companies by 1993. Menzel and Kammer (2019) show such a peak for the industries in the USA and Denmark in 1981 and 1987, respectively, and thus much earlier.

During the civic convention, it took 14 years for the number of companies to get halved to 9. This time span is slow compared with the American and Danish wind turbine industry. In the Danish industry, a comparable shakeout from 19 to 9 companies took place in 4 years. In the US industry, this shakeout occurred from 18 to 4 companies in 5 years (Menzel and Kammer, 2019). The figure further shows an increase in the number of companies from 2008 onwards, caused by the increased promotion of offshore wind energy since 2009 that led to new entries. From 2016 onwards, there is a sharp decline in the number of companies, mainly during the market convention. These slower shakeout and the again rising number of entries indicate that especially a civic convention affects the ILC, as it keeps more firms in the market.

Figure 2 also describes entry and exits. The domestic phase exhibited an average of 1.3 entries and 0.25 exits per year. The industrial phase exhibited an average of 1.5 entries and 1.5 exits per year. The civic phase exhibited an average of 0.9 entries and 1.4 exits per year. Actually, the ILC would expect an increase of exits during the shakeout that took place in this phase. Yet, it seems that the number of exits decreased by the civic convention. The market phase exhibited an overage of 0.6 entries and 2 exits per year.

The average installed capacity per firm increased strongly during the industrial convention phase. This growth slowed down during the civic convention but was still strong. For example, the average installed capacity of the 14 active companies in 2000 was 68 MW and the average installed capacity of the 8 active companies in 2008 was 800 MW. This number also grew during the market convention, to an average of 1,070 MW per firm. However, the number of active companies decreased to three during this period. To summarize, the domestic convention was shaped by a strong increase in number of firms, the industrial convention was shaped by stable number of firms and strong growth in productivity, the civic convention was shaped by a declining number of firms and still strong productivity growth, whereas in the market phase productivity growth seem to be mostly driven by many exits of less productive firms.

Figure 3 describes the share of German manufacturers in the domestic market and the share these manufacturers installed abroad, as indicators of the manufacturers’ position in a global industry. Since hardly any data is available for the domestic convention phase, the figure starts with the industrial convention phase from 1989 onwards. The industrial convention phase begins with a low international orientation and a low domestic market share of German companies. Although the reliability of the data in the first half of the 1990s is limited, this result is consistent with descriptions of low competitiveness and weak market position of German manufacturers in this period (Heymann, 1998; Oelker, 2005). Both indicators increase until the end of the 1990s, indicating increased competitiveness of German manufacturers.

During the period of the civic convention, there was a notable increase in the proportion of German manufacturers operating in the home market, as well as their activity in foreign markets. This upward trend peaked between 2008 and 2013, with market in the domestic market and share the producers installed abroad over 70%, meaning that not only did German manufacturers produce 70% of the installed capacity in Germany, but this capacity represented only 30% of the total production capacity of German manufacturers. Towards the end of this phase, there was a decrease in the proportion of German manufacturers in both the domestic market and their activity in international markets. This decline can be attributed to the rapid expansion that was occurring during this period. Between 2013 and 2017, the annual installed capacity more than doubled, as depicted in Figure 1, and potentially grew at a faster pace than the corresponding manufacturing capacity in Germany.

When conventions changed to market and yearly installed capacity strongly decreased, the share of German manufacturers in Germany declined and their share of activities in international markets increased. The increase in international markets is a statistical artifact, resulting from the strong decline in production for German markets. Production for foreign markets declined as well in absolute numbers.

The increasing market and international shares of German manufacturers during the industrial conventions met the assumption that an industrial convention increases productivity. The further increasing domestic and international share during the time of the civic convention, however, is less expected. Especially the strong increase in international share indicates growing competitiveness of German manufacturers. There could be two explanations for this. First, the productivity increases during the industrial convention had a longer lasting effect on the productivity of manufacturers. Second, the resource transfer into the German market during the civic convention allowed German manufacturers especially to increase their productivity. In contrast, the market convention phase shows the assumed developments. The smaller share of German producers in the home market indicated a rise in competitive pressure.

Figure 4 illustrates the distribution of the share of coastal state firms and their yearly installed wind energy capacity, in comparison to the firms located in inland states. The ILC anticipates a rise in concentration during the shakeout for industries influenced by spin-offs (Klepper, 2007). However, the wind turbine industry is shaped by diversifiers. Accordingly, concentration of firms in Northern Germany fluctuated predominantly between 60% and 70%, with a few exceptions. The first exception occurred during the early stages of the industry when a domestic convention influenced the low number of firms from coastal states. This was due to the research funding allocated to southern German companies within the GROWIAN projects. The second exception was observed during the industrial convention phase when the share of southern German firms increased once again. The occurrence of spatial dispersion would be expected during an industrial convention, particularly in cases where knowledge is disseminated throughout the industry, such as the Scientific Measurement and Evaluation Program. The third exception occurred during the beginning of the market phase, where the proportion of Northern German companies rose to over 80%, which aligns with the anticipated outcomes of a market convention.

Conclusion

Many studies in evolutionary economic geography examine innovation. This article aimed to further develop an evolutionary perspective on markets as drivers of the transformation of the economic landscape (Loasby, 2000; Potts, 2001). In doing so, the article applied convention theory to the evolution of the wind turbine industry in Germany. Conventions define principles of evaluation (Boltanski and Thévenot, 2006) and thus ultimately the principles according to which a market is constructed and selection takes place (Diaz-Bone, 2016). Two mechanisms were elaborated for this purpose. One describes how the integration of social values into markets, differentiated by convention, defines the principles of evaluation in these markets. The second describes how these evaluation principles influence how companies build up resources and thus how the industry develops. Evidence was found for the assumed mechanisms, their change over time and how they affect industry evolution.

This article had the advantage that it could define conventions that affect the whole industry, as these conventions found their way into laws. Yet, the study had several limitations. One limitation refers to data. The data were incomplete, especially during the early development of the ­industry. Another limitation refers to geographical scale. The industry was also subject to conventions in other markets, which could only be taken into account only indirectly via market shares in the home market and internationalization of firms. Additionally, different conventions only slightly altered the geography of the industry. Yet, analyses in this respect were limited due to the focus on location of headquarters and not production plants, the propensity of industries shaped by diversifiers to be distributed in space (Klepper 2007) and the limitation to manufacturers and not suppliers.

What also influenced the study were different temporalities. Firstly, the transitions between the phases were smoother than the classification suggests. The transitions between phases were often preceded by crises. Diaz-Bone (2018b) defines crises here as a mismatch between institutions, i.e. the conventions formalised in laws and programmes, and the conventions considered important in ­social and political debates. Second, this article also showed the different temporality of the conventions, as already described in other studies (Diaz-Bone, 2018b). While the industrial convention had a long-term impact, the market convention had a short-term impact. Furthermore, while the study used conventions to describe markets, it did not go into more detail into the construction of these markets, for example via investigating feedback loops between legislators and market participants (Nijhof et al., 2022), or the social and physical infrastructures that are necessary for these markets to function (Çalışkan and Callon, 2010). Strictly speaking, also in this study, markets were not part of ‘the processes by which the economic landscape [….] is transformed from within over time’ (emphasis added, Boschma and Martin, 2010, 6–7). This neglect was justified as the intention was to show the effects of markets.

Therefore, further research might be fruitful to conceptualize the spatial evolution of markets themselves. A stronger integration of markets in Evolutionary Economic Geography is obvious for three reasons. First, markets include knowledge, technologies, infrastructures, standards and institutions (Çalışkan and Callon, 2010), which are fundamental dimensions of evolutionary thinking (Loasby, 2000). Secondly, markets have a geography. Markets can be local, such as trade fairs (Haisch and Menzel, 2023) or global (Berndt and Böckler, 2009). Thirdly, formation and evolution of markets affect the location of new industries (Bednarz and Broekel, 2020) and the changing locational pattern of established industries (Menzel and Adrian, 2018).

References